Using Force to Protect Civilians: Successes and Failures of United Nations Peace Operations in Africa 019285710X, 9780192857101

Table of contents :
cover
titlepage
copyright
Acknowledgments
Contents
Preface
List of figures
List of tables
1 Protection by military force in UN peace operations
The book in a nutshell
Defining success in UN military protection operations
Unpacking UN military protection operations
Understanding the perpetrators to tailor effective operations
So what?
Outline of the book
2 Understanding the utility of force to protect civilians from violence
Four causal condition candidates
Causal mechanisms and hypotheses
Further deliberations on the matching theory
Summary of causal condition candidates
3 Exploring characteristics and outcomes of UN military protection operations
Characteristics of UN military protection operations
Summary of descriptive statistics
Estimating outcomes of UN military protection operations (1999–2017)
4 Analyzing empirical patterns of UN military protection operations
Operationalization of causal condition candidates
Bivariate cross-tabulations
Summary of cross-tabulations
Multivariate linear probability analysis
5 Discovering causal pathways to successful protection outcomes
Qualitative Comparative Analysis
The outcome
Causal condition candidates
Necessary conditions for successful outcomes
Causal recipes toward successful outcomes
6 Protecting Civilians from the M23
Matching the M23
Pre-empting the M23
Troop-to-perpetrator ratios
Willingness to accept risk
Additional explanations for the successful outcome
Conclusion
7 Overwhelmed by the White Army
Failing to match the White Army
The absence of relevant responses to Murle revenge attacks
Pre-emption
Troop-to-perpetrator ratios
Willingness to accept risk
Additional explanations for the unsuccessful outcome
Conclusion
8 Increasing the utility of force to protect
Summary of main findings
Implications for theory, policy, and practice of UN military protection operations
Appendix
References
Index

Citation preview

Using Force to Protect Civilians

Using Force to Protect Civilians Successes and Failures of United Nations Peace Operations in Africa ST I A N K J EK SRU D

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Stian Kjeksrud 2023 The moral rights of the author have been asserted All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2023930765 ISBN 978–0–19–285710–1 DOI: 10.1093/oso/9780192857101.001.0001 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

Acknowledgments In early 2020, I reached out to Professor Lisa Hultman at Uppsala University to ask for her opinion of my first ever book proposal. Lisa had recently published Peacekeeping in the Midst of War—with Jacob D. Kathman and Megan Shannon—a brilliant analysis of the conflict reducing effects of United Nations peace operations. That book had greatly influenced my latest ideas about Blue Helmets’ ability to protect civilians from violence. I could not think of a better peer to evaluate the survivability of my approach. Although she barely knew me, Lisa responded kindly with encouragement and support. Unselfishly, she even offered to contact Dominic Byatt—political science and international relations editor at Oxford University Press (OUP)—who accepted to read my draft. He has guided me skillfully and empathically ever since. I am deeply grateful to both Lisa and Dominic for believing in a first-time author. Thank you so much. No other person has influenced this project more than Jacob Aasland Ravndal. Thank you for your mentoring and our friendship. I would also like to thank friends and colleagues who have provided generous feedback and guidance over the years, including Alexander W. Beadle, Paul D. Williams, Annika S. Hansen, Scott Gates, Gunhild Hoogensen Gjørv, Stuart Griffin, Francesca Refsum Jensenius, Andreas Forø Tollefsen, Elin Marthinussen Gustavsen, Per Martin Norheim-Martinsen, Tore Nyhamar, Petter H. F. Lindqvist, Victoria Holt, Alison Giffen, Aditi Gorur, Haidi Scarlett Willmot, Ralph Mamiya, Walt Kilroy, Sukanya Podder, Cedric de Coning, Allard Duursma, Tom Røseth, Sine Holen, Therese Klingstedt, Unni Rørslett, Øystein Tunsjø, Jo Sannem, Elizabeth Hunter, and Nicola Trier. I am also grateful for the constructive feedback and much needed critique from the four anonymous reviewers from OUP, which greatly improved the manuscript. Vicki Sunter—academic editor at OUP—skillfully held my hand throughout the entire review and production process. Thank you. The project has been funded by the Norwegian Ministry of Defence and the Norwegian Defence University College, where I now work. I would like to recognize John Otto Pedersen, Rune Solberg, and Bjørn Inge Ruset—consecutive commanders of the Norwegian Defence International Center (NODEFIC)

vi

ACKNOWLEDGMENTS

over the past few years—for providing time and encouragement to finish the book. I also thank my parents Eli and Jan for being there when I needed assistance. My in-laws Ingeborg and Roar also helped whenever I asked. Thank you. I dedicate the book to Silje, Ada, and Karla, my awesome family. I admire your patience and love you dearly. Oslo, November 2022

Contents Preface List of figures List of tables

1. Protection by military force in UN peace operations The book in a nutshell Defining success in UN military protection operations Unpacking UN military protection operations Understanding the perpetrators to tailor effective operations So what? Outline of the book

2. Understanding the utility of force to protect civilians from violence Four causal condition candidates Causal mechanisms and hypotheses Further deliberations on the matching theory Summary of causal condition candidates

3. Exploring characteristics and outcomes of UN military protection operations Characteristics of UN military protection operations Summary of descriptive statistics Estimating outcomes of UN military protection operations (1999–2017)

ix xii xiii

1 6 8 13 20 25 26

28 32 35 48 59

61 63 82 84

4. Analyzing empirical patterns of UN military protection operations 89 Operationalization of causal condition candidates Bivariate cross-tabulations Summary of cross-tabulations Multivariate linear probability analysis

5. Discovering causal pathways to successful protection outcomes Qualitative Comparative Analysis The outcome Causal condition candidates Necessary conditions for successful outcomes Causal recipes toward successful outcomes

89 94 100 101

104 105 107 110 125 127

viii

CONTENTS

6. Protecting Civilians from the M23 Matching the M23 Pre-empting the M23 Troop-to-perpetrator ratios Willingness to accept risk Additional explanations for the successful outcome Conclusion

7. Overwhelmed by the White Army Failing to match the White Army The absence of relevant responses to Murle revenge attacks Pre-emption Troop-to-perpetrator ratios Willingness to accept risk Additional explanations for the unsuccessful outcome Conclusion

8. Increasing the utility of force to protect Summary of main findings Implications for theory, policy, and practice of UN military protection operations

Appendix References Index

130 133 144 145 147 148 152

154 156 164 166 166 168 170 172

174 175 178

187 208 222

Preface As a soldier and officer and deployed to wars and armed conflicts in Afghanistan, the Balkans, and the Middle East from the mid-1990s to the mid2000s, I have seen firsthand that military forces often fail to reduce civilian suffering. In fact, much civilian harm is caused by the same forces that are meant to protect them. Yet, I still believe that protecting civilians from violence sometimes demands the use of force. But how can force be used most effectively to protect, without causing more harm in the process? In this book, I combine the military practitioner’s field experience with a researcher’s tools for scientific investigation to seek answers to that question. Toward the end of the 2000s, protecting civilians from violence in armed conflicts—if necessary, with the use of lethal force—became the priority task for almost every United Nations (UN) military peacekeeper. Acutely aware of the UN’s spectacular protection failures of the not-so-distant past in Rwanda, Bosnia, and Somalia, I found it astonishing that the United Nations Security Council (UNSC) yet again relied on military forces to protect those threatened with violence. For decades, one of the bedrock principles of UN peacekeeping had been to avoid using force altogether, except in self-defense. I was one of those Blue Helmets that stood idly by when civilians were threatened with violence. Now, the UNSC expected the Blue Helmets to directly counter armed groups that deliberately killed, raped, and otherwise harmed civilians as part of their warfare. Conventional military doctrine did not capture this new understanding of protection, as it still narrowly related to the rules laid out by International Humanitarian Law (IHL), i.e., to minimize harm to civilians during military operations to avoid so-called collateral damage. Irrefutably, the way the UN now understood protection of civilians in peacekeeping operations constituted a new task for all military forces. Intrigued by these observations, I traveled to the eastern parts of the Democratic Republic of the Congo (DRC) in 2008, where myriad armed groups continued to target civilians even in the presence of the largest and militarily most robust UN peacekeeping operation at that time. Interviewing UN practitioners, I found that there were few—if any—guidelines tailored to military troops on how to protect civilians from violence. Existing scholarship on

x

PREFACE

this topic was marginal at best. On my return, I established a new research activity at the institute I then worked—the Norwegian Defence Research Establishment (FFI)—to investigate how UN troops could protect civilians from violence in contemporary armed conflicts. I have devoted my time since then to answer that question. This book captures most of what I have learned about the utility and limitations of force to protect, hopefully generating new and useful knowledge in an understudied field of inquiry. Although it treats a grim and distressing aspect of contemporary armed conflict, I hope you find the book thought-provoking, informative, and useful. I also recognize that my findings cannot form a general theory on the utility of force to protect, beyond pointing toward some insights about the ability of UN peace operations to protect civilians. As this project came to an end, Russia invaded Ukraine with such devastating effects on civilian security only armed state actors can inflict. Within the first few weeks of fighting, millions of Ukrainians were displaced, thousands were killed or otherwise physically harmed. Scores of others will carry mental scars for life. Russia’s warfare in Ukraine has blatantly failed to distinguish between military targets and the civilian population. More worryingly, the targeting of civilians significantly overstepped the principle of proportionality. Worst of all, some of Russia’s operations—including in Mariopol and Bucha—indicate that civilian targeting may be intentional, seeking to influence the outcome of the war through direct targeting of civilians. It shows that the regulations in IHL are only effective if its signatories obey its letter and intent. This observation does not indicate that IHL is irrelevant. Indeed, it is especially in times of war IHL is crucial to spare civilian life from unlawful and unnecessary harm. When ignored, however, IHL does not provide much protection to those civilians that are under direct threat of violence. Russia’s targeting of civilians—however awful—should come as a surprise to no-one. Deliberate targeting of civilians—or ignorance of the rules laid out in IHL—is a recurring characteristic of modern war. Russian warfare elsewhere—including in Chechnya, Georgia, and Syria—shows us that civilian targeting form part of Russian forces’ modus operandi. Violence against civilians serves many purposes. The use of imprecise heavy weapons in urban areas not only makes such neighborhoods inhabitable, but it accelerates the speed of flight of civilian populations. Millions on the run makes a country more difficult to govern and can sow discontent and political bickering in neighboring countries. Extrajudicial killings of civilians can be used to influence and control parts of the population that Russia seeks to rule in the future. However

PREFACE

xi

bizarre, violence against civilians can also serve Putin’s idea of “de-nazifying” Ukraine, expelling or killing parts of the population. Protecting civilians from such deliberate violence—as envisioned in UN, and since 2016, NATO policies—remains utopian when great power war looms. To avoid a destructive nuclear war, NATO cannot intervene directly to protect the Ukrainian population. Despairingly, despite three decades of increased concern for human security in armed conflict, war crimes and crimes against humanity can still become part of the wartime repertoire for belligerents that can threaten nuclear war. However painful it is to spectate the suffering of Ukrainians without being able to act—beyond providing weapons to aid the country’s defense—we should follow closely how Russia targets civilians in war. First, by understanding why and how Russia use violence against civilians to achieve wartime aims, we can improve the future defense of European countries and their populations. Second, future war tribunals will depend on systematic reporting and analysis of potential war crimes and crimes against humanity. Mostly, military analyses of possible wartime counterparts relate to the number and qualities of weapon platforms and their doctrinal ideas of how they aim to fight other armies. Russia’s war in Ukraine (and the analysis found within these pages) show that it also is important to understand how, and for what purposes, weapons are used to target civilian populations. Only when we have a deeper understanding of perpetrators of violence can we tailor military operations that stop or reduce physical threats to civilians, without causing more harm in the process.

List of figures 1.1. Total peacekeepers deployed by type, 1990–2017

15

3.1. Distribution of reported UN protection operations per UN mission, 1999−2017

64

3.2. Protection operations per year per UN mission (averaged), 1999 ̶ 2017

65

3.3. Reported UN military protection operations in Africa per year, 1999–2017

68

3.4. Total number of uniformed UN peacekeepers deployed by type

68

3.5. Total number of UN military protection operations per year in Africa compared with protection operations only performed by MONUC/MONUSCO, 1999 ̶ 2017

69

3.6. Distribution of scenarios across 200 military protection operations in Africa, 1999 ̶ 2017

71

3.7. Distribution and prevalence of functions of force across UN protection operations in Africa, 1999–2017

74

3.8. Total civilian fatalities per year deduced from 97 cases where UN troops used force to protect in Africa, 2004–2017

76

3.9. Average number of civilians killed per case per year deduced from 97 cases where UN troops used force to protect in Africa, 2004–2017

77

3.10. Total number of UN fatalities from malicious acts (1999–2017) in UN missions represented in UNPOCO (solid line), UN fatalities from malicious acts in Mali (2013–2017) (broken line), and UN fatalities captured in UNPOCO (dotted line). Data taken from United Nations and UNPOCO (United Nations 2018)

82

3.11. Distribution of outcome estimations of UN military protection operations over time (1999–2017), with trend lines

87

5.1. Venn diagrams portraying the logic of sufficiency and necessity

106

5.2. Venn diagrams portraying the logic of a trivial (1) and non-trivial (2) necessary condition

126

List of tables 2.1. Perpetrator’s use of violence vs. protector’s use of military force

47

2.2. Eight generic threat-scenarios

50

3.1. Threat scenarios structured according to the perpetrators’ main rationale for attacking civilians, with examples from armed conflicts

71

3.2. Outcome estimations of 200 UN military protection operations captured in UNPOCO

87

4.1. Cross-tabulation between absolute troop numbers and outcomes

95

4.2. Chi-square significance test of absolute troop numbers

95

4.3. Cross-tabulation between troop-to-population ratios and outcomes

96

4.4. Chi-square significance test of troop-to-population ratios

97

4.5. Cross-tabulation of willingness to accept risk and outcomes

97

4.6. Chi-square significance test of willingness to accept risk

98

4.7. Cross-tabulation of pre-emption and outcomes

98

4.8. Chi-square significance test of pre-emption

99

4.9. Cross-tabulation of matching and outcomes

99

4.10. Chi-square significance test of matching 4.11. Linear probability model

99 102

5.1. Estimated outcomes of 126 UN military protection operations in Africa, 1999 ̶ 2017

110

5.2. Outcome calibrations with fuzzy scores, qualitative descriptions, number of cases, and case IDs from UNPOCO

111

5.3. Troop-to-population ratio calibrations with fuzzy scores, descriptions, number of cases, ratio thresholds, and case IDs from UNPOCO

115

5.4. Calibration of TCCs’ willingness to use force to protect

118

5.5. Calibration of TCCs’ willingness to use force to protect, including constellations of willing/hesitant TCCs, fuzzy scores, description, number of cases and case IDs from UNPOCO

119

5.6. Calibration of the pre-emptive/reactive character of UN military protection operations, including fuzzy scores, description, number of cases, and case IDs from UNPOCO

122

5.7. Functions of force (protector) versus types of violence (perpetrator)

123

xiv

LIST OF TABLES

5.8. Calibration of UN troops’ ability to match the perpetrators by force, including fuzzy scores, description, number of cases, and case IDs from UNPOCO

124

5.9. Analysis of necessary conditions for the presence of positive outcomes

125

5.10. Relevant rows from the truth table deduced from the analysis of the QCA matrix

127

5.11. Intermediate solution analysis of the truth table deduced from the QCA matrix

128

6.1. Perpetrator’s use of violence vs. protector’s use of military force to protect

134

6.2. Five phases of FARDC-FIB operations against the M23

135

6.3. Phase 1 (March–April 2013): Resolution 2098 and M23 threats

136

6.4. Phase 2 (May): First Mutaho Hills battle

138

6.5. Phase 3 (June–July 29): Second Mutaho Hills battle

139

6.6. Phase 4 (July 30–August): Goma red line and the Kibati battle

141

6.7. Phase 5 (September–November): Three-pronged attack defeating the M23

144

7.1. Four phases of SPLA-UNMISS operations against the White Army in 2011–2012

157

7.2. Perpetrator’s use of violence vs. protector’s use of military force to protect

158

7.3. Phase 1 (December 5–12): White Army mobilization, early warning, and UN deployment

159

7.4. Phase 2 (December 13–22, 2011): Deterrence fails as the White Army moves south

161

7.5. Phase 3 (December 23–30, 2011): Multiple White Army attacks

162

7.6. Phase 4 (December 31, 2011–January 4, 2012): White Army Pibor attack, SPLA/UNMISS defense, and circumnavigation

165

1 Protection by military force in UN peace operations It is now more dangerous to be a woman than to be a soldier in modern wars. —Major General (R.) Patrick Cammaert, Former Commander of the Eastern Division, United Nations Mission in the Democratic Republic of the Congo, 2008 (Cammaert 2008)

This book sheds light on two interrelated questions: To what degree have UN military troops protected civilians under imminent physical threats in African armed conflicts? What determines UN military troops’ ability to protect civilians from physical violence? The answers I provide indicate that Blue Helmets are more effective protectors than commonly thought and that physical threats to civilians vary greatly as does the utility of force to protect them. Together, the answers hold potential building blocks for a theory on the utility of force to protect civilians from violence in UN peace operations. Violent targeting of civilians is a disturbingly common feature of contemporary armed conflicts. When Dutch General Patrick Cammaert was deployed to the eastern parts of the Democratic Republic of the Congo (DRC) toward the end of the 2000s, he observed that in Ituri, militias continued to attack civilians, killing men, women, and children, and burning down their houses. In North Kivu, regular clashes between political and military factions were reported. In South Kivu, FDLR-elements [Forces démocratiques de libération du Rwanda] continued to terrorise civilians, committing such atrocities as locking them up and burning them alive. (Cammaert 2008, 68)

The General’s grim accounts of armed conflicts fought among, for, and against civilians mirror contemporary reports from the DRC and reflect worrying

Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0001

2

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

trends in global civil war violence (Pettersson and Öberg 2020; R. Smith 2008; United Nations 2020a, para. 6, 2020c, paras. 11–20). Protecting civilians from physical violence has been a task for United Nations peace operations¹ (UNPOs) since 1999 and prioritized above other tasks since 2009 (United Nations 1999a, 2009, para. 19). Today, multidimensional United Nations (UN) missions employ what is known as a comprehensive and integrated approach to the protection of civilians (POC), which basically implies a combined effort of all mission components—civilian, police, and military—to keep civilians safe from harm (United Nations 2019b, para. 8). However, if other measures fail, protecting civilians may demand the use of deadly military force, which can only be wielded by military troops (Cammaert 2008, 70; dos Santos Cruz, Phillips, and Cusimano 2017; United Nations 2019b, para. 18). In UNPOs, uniformed military personnel—better known as Blue Helmets—manage the application of force on behalf of the United Nations Security Council (UNSC). Formally, the Blue Helmets’ use of force is guided by mandates—in the form of Security Council resolutions—specific UN protection policies, tailored implementation guidelines for military components, bedrock principles of peace operations, concepts of operations (CONOPS), and rules of engagement (RoEs) (United Nations 2008, chap. 3, 2017a, 2019b). In practice, however, there is also a significant portion of meddling by the capitals of troop-contributing countries, which often counters official UN guidance and challenges command and control² ¹ In 2019, the UN restructured its peace and security pillar, establishing the Department of Peace Operations (DPO) to succeed the better-known Department of Peacekeeping Operations (DPKO). Consequently, the term “peace operations” now reflects a continuum of UN responses to armed conflict, not limited to peacekeeping operations. Although this book reflects this change, I use the terms “UN peace operations,” “UN peacekeeping operations,” and “UN missions” interchangeably, since most of my data and sources originate from before 2019. I broadly relate to the UN’s definition of this activity as presented in the Capstone Doctrine from 2008, where peacekeeping is described as “a technique designed to preserve the peace, however fragile, where fighting has been halted, and to assist in implementing agreements achieved by the peacemakers. Over the years, peacekeeping has evolved from a primarily military model of observing cease-fires and the separation of forces after inter-state wars, to incorporate a complex model of many elements—military, police and civilian—working together to help lay the foundations for sustainable peace” (United Nations 2008, 18). ² The term “Command and control” is defined in UN peace operations as “The authority delegated to a Military or Police Commander for the direction, coordination and control of uniform personnel under his or her command. Operational command and control includes the authority to assign tasks, designate objectives and give direction to individual uniformed personnel, units and sub-units necessary to accomplish the mission” (United Nations 2019a, para. 101). Moreover, “administrative control” has a distinct meaning in UN peace operations, according to the same policy, and is “the authority over subordinate or other organizations within national contingents for administrative matters such as personnel management, supply, services and other non-operational missions of the subordinate or other organizations. Administrative Control is a national responsibility given to the NCC [National Contingent Commander] in a peacekeeping mission.” As alluded to in the text, many times capitals of troop contributing countries interfere with UN Force or Police commanders command responsibilities, potentially undermining the effectiveness of operations.

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

3

arrangements, and leads to less-than-optimal outcomes of operations (United Nations Office of Internal Oversight Services 2014). Implicitly, policies and guidelines tailored to UN troops seek to provide answers to a fundamental yet paradoxical question: How can the threat or use of force reduce violence in armed conflict? Worryingly, none of the existing UN documents provides answers to how military force can be used most effectively to protect civilians. This may immediately appear an overly harsh judgment. To clarify, I am not downplaying the significant practical, conceptual, and policy developments that have occurred in the field and within the UN Secretariat to make UN troops’ protection efforts more effective. These processes are welcome, valuable, and necessary. They are, nevertheless, not sufficient. The main reason for their insufficiency is simply that few convincing answers exist. Protecting civilians from perpetrators of violence is a new task for military forces. When compared to other tasks, for example, counterinsurgency—resting on decades of practical experience, lessons learned, doctrinal development, scientific studies, and even “classics”—there is only marginal practice to lean on and almost no systematic thinking on how force can be used most effectively by UN troops to protect civilians from deliberate targeting. These shortcomings are not unique to the UN. Most militaries still confuse protection of civilians—as understood in both UN and North Atlantic Treaty Organization (NATO) policies—with merely adhering to the Law of Armed Conflict (LOAC), also known as International Humanitarian Law (IHL). It is worth elaborating on the key differences between these concepts. Essentially, LOAC is concerned with regulating belligerents’ behavior to avoid unlawful and unnecessary harm to civilians during armed conflict (International Committee of the Red Cross 2004). In military lingo, such harm to civilians is often described with a rather clinical term: “collateral damage.” Protection of civilians—as understood in this book—is about the role, utility, and limitations of force in protecting civilians from the violence against civilians committed by others, the perpetrators of violence. While there are overlapping concerns in these concepts, they address two very different phenomena. LOAC limits and regulates the use of military force to make war more humane. POC encourages—and sometimes demands—the use of force to stop or reduce targeted violence against civilians. I will use a case from NATOs operations in Afghanistan to illustrate this difference further. Toward the end of the 2000s, the International Security Assistance Force (ISAF) Commander, U.S. General Stanley McChrystal, issued new counterinsurgency guidance to his troops in Afghanistan, emphasizing that “protecting the people is the mission” (McChrystal 2009). This shift

4

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

toward a human-centric approach came in response to a substantial number of collateral damage events caused by ISAF troops, or more often, by their partners in the Afghan National Army (ANA). General McChrystal’s guidance aptly reflects the maturity of counterinsurgency logic and contains thoughtful arguments about the dilemmas of using military force to counter insurgent groups in a war that largely plays out among the civilian population. His guidance is nonetheless largely anchored in LOAC considerations. Consequently, when NATO sought to gather lessons learned from Afghanistan on efforts to protect civilians, the alliance limited investigations to collateral damage events (Keenan and Beadle 2015; NATO Joint analysis and lessons learned centre 2015; Shortland and Bohannon 2014; Shortland, Sari, and Nader 2019; United Nations Assistance Mission in Afghanistan 2020). In other words, the analyses rested on cases where NATO troops—or their local Afghan partners—were responsible for inflicting civilian harm. As such, the lessons learned activity did successfully capture the fact that ISAF/ANA had reduced their percentage of civilian casualties quite significantly from 2009 to 2014. However, NATO failed to recognize that the number of civilian casualties in Afghanistan increased significantly—in fact it nearly doubled—during that period, and that most civilian casualties were caused by Taliban and other non-state armed groups (Shortland, Sari, and Nader 2019; United Nations Assistance Mission in Afghanistan 2020). Although the mission was to protect the people, people became less secure. Being unable to provide physical protection to populations under threat in areas where you deploy military forces will influence your ability to succeed with any mission, even when protection of civilians is not a formal task from the North Atlantic Council (NAC), the UNSC, or your own government. While LOAC seeks to spare civilians from the negative consequences of armed conflict, the task of protecting civilians from violent perpetrators implies something entirely different. NATO has now realized that the role of the military in providing human security is much broader, as reflected in its 2022 strategic concept and its recent policy on the protection of civilians, which also mirrors the UN’s definition of POC in peace operations (NATO 2016, para. 18, NATO 2022). The UN defines POC as: without prejudice to the primary responsibility of the host state, integrated and coordinated activities by all civilian and uniformed mission components to prevent, deter or respond to threats of physical violence against civilians within the mission’s capabilities and areas of deployment through the use of all necessary means, up to and including deadly force. (United Nations 2019b, para. 18)

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

5

Beyond taking the utmost care to avoid harming civilians during operations, we now expect UN—and NATO—troops to protect civilians from perpetrators of violence that target them as part of their warfare. For military planners and practitioners, this expansion of tasks implies that they not only need to analyze potential negative consequences of their own actions but also to understand and counter the violence inflicted by others. This is not merely a subtle change in day-to-day operations but rather a new paradigm in thinking about the role and utility of military force in contemporary armed conflict, with consequences for the planning, conduct, and outcome of military operations across the world. Although effective physical protection sometimes demands the use of force, we know that any forceful intervention on the part of the Blue Helmets places severe strain on the bedrock principles of UNPOs: impartiality, consent, and the non-use of force, except in self-defense and defense of the mandate.³ If UN troops ignore these well-established principles, they could jeopardize their protected status under international conventions, struggle to retain the hoststate consent required for mandate implementation, and find themselves at the receiving end of threats from both state and non-state armed actors (Benson 2016; Cammaert and Blyth 2013; Macura 2020; Sheeran and Case 2014; Tull 2016). Understandably, many troop contributors to UN missions hesitate to accept such risks and seldom intervene militarily to protect civilians under threat (Bode and Karlsrud 2018; United Nations Office of Internal Oversight Services 2014, 2017, 2018a, 2018b). More than two decades after the UNSC issued its first protection mandate, there are still significant implementation gaps in providing security to civilians in armed conflict, despite the strategic importance attached to this task (Brahimi 2000; Everett 2017; Giffen 2010; Holt and Berkman 2006; Thakur 2013; United Nations 1999b, para. 14). Notwithstanding severe political, legal, conceptual, and practical limitations in turning ambitious human security policies into practice, the UN protection record is far from dismal. An emerging scholarly consensus strongly indicates that the presence of uniformed UN peacekeepers significantly reduces civilian suffering in armed conflicts (Cil et al. 2019; Di Salvatore and Ruggeri 2017; Fjelde, Hultman, and Nilsson 2019; Hegre, Hultman, and Nygård 2019; Hultman, Kathman, and Shannon 2019; Phayal and Prins 2020). These findings mostly rest on quantitative studies that compare detailed monthly deployment data on uniformed personnel with authoritative datasets on organized ³ For a brief analysis of the evolution of the three bedrock principles, see the introductory chapter in United Nations Peace Operations and International Theory (Oksamytna and Karlsrud 2020).

6

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

violence in armed conflict (United Nations 2021b; Uppsala University 2020). Although providing immensely valuable insights into the conflict-reducing effects of peace operations across time and UN missions—countrywide as well as local—these studies do not capture empirically what UN troops do to protect civilians in situations where they are under threat. In short, they cannot help us to understand what works, why, and when, from a military perspective. Interestingly, albeit somewhat worryingly, the quantitative studies referred to above point to a strong deterrent effect of UN uniformed presence in large numbers (meaning thousands). We already know that Blue Helmet presence is in many cases not enough to protect civilians from violence, as UN troops often fail to deter potential perpetrators. For example, in the summer of 2010, largescale attacks against civilians occurred near UN troops in eastern DRC. Over the course of four days—along the Kibua–Mpofi axis in Walikale territory— “at least 387 civilians were raped, including 300 women, 23 men, 55 girls, and 9 boys” by hundreds of militia members (MONUSCO 2011, para. 25). The perpetrators also abducted more than 100 civilians, while looting and destroying some thousand houses and shops (MONUSCO 2011, paras. 30–2). The Blue Helmets located near Kibua—on that very road axis—even patrolled the area while attacks were still underway, failing to recognize the severity of the situation and unable to deter the perpetrators of violence (MONUSCO 2011, paras. 46–7). This event is unfortunately not unique (United Nations Joint Human Rights Office 2011, paras. 37–40). In the not-so-distant past, UN troops were involved in even more spectacular protection failures involving botched attempts to deter determined perpetrators of violence, including in Rwanda, Bosnia, and Somalia in the mid-1990s (Annan 1999a; Carlsson, Han, and Kupolati 1999; Ngulube, Hagglund, and Erskine 1994). I am not in any way blaming the troops deployed in these extreme cases, rather pointing to the fact that it matters what troops do (or not) to protect. It is not enough just to be there. These observations form the starting point for this book.

The book in a nutshell This book is guided by two different but interrelated questions. Firstly, I want to know more about the outcomes of UN military protection operations across time and missions: To what degree have UN military troops provided protection to civilians under imminent physical threat in Africa between 1999 and 2017? To the best of my knowledge, this is the first attempt to analyze

THE BOOK IN A NUTSHELL

7

the outcomes of UN military protection operations systematically across time and place, providing a much-needed starting point for systematic studies of the anatomy of successes and failures. Secondly, I seek to understand what constitutes successful conditions for successful protection outcomes: What determines UN military troops’ ability to protect civilians from physical violence? The answers hold potential building blocks for a theory on the utility of force to protect civilians from violence in UN peace operations. In pursuit of answers to these questions, I build on new and unique data, which I have captured and coded in the openly accessible United Nations Protection of Civilians Operations dataset (UNPOCO) (Kjeksrud 2019). UNPOCO contains both the quantitative and qualitative characteristics of 200 military protection operations across ten UN missions in Africa from 1999 to 2017. This is the only dataset to date that contains systematic event data on UN military protection operations across time and UN missions. The data collection begins in 1999, which is when the UNSC first mandated a UN mission to protect civilians from violence. It ends in 2017—somewhat arbitrarily—allowing time to analyze the substantial amount of data and to write and publish this book. Based on the UNPOCO data, I firstly estimate the outcomes of UN military protection operations, enabling an analysis of the Blue Helmets’ overall success rate across time and missions when they apply force to protect. I will return shortly to how I go about this analysis. Secondly—based on a sub-set of 126 cases derived from the UNPOCO data—I systematically search for causal conditions⁴ that may help explain the utility of military force to protect civilians across time and UN missions. I cast the net wide to capture promising explanations from existing literature, including troop numbers (deterrent presence), willingness to accept risk, pre-emption, and tailored operations (matching perpetrators of violence). Finally, I dig deeper into military protection operations in the DRC and South Sudan to test some of my cross-case findings on two particularly interesting cases, and to search for additional explanations

⁴ Much of the qualitative literature on UN peacekeeping use case study methods that employ socalled mechanism-based explanations, recognized by the term “causal mechanisms” when exploring causal relations. According to Hedstro¨m and Ylikoski, mechanism-based explanations “implies that proper explanations should detail the cogs and wheels of the causal process through which the outcome to be explained was brought about” (Hedstro¨m and Ylikoski 2010, 50–1). Perhaps confusingly, I use the term “causal conditions” systematically throughout this book, which should not be mistaken to mean a causal mechanism. “Causal conditions” is a term linked to the key method employed in this book, Qualitative Comparative Analysis (QCA). QCA is used as a method to identify logic (combinations of ) necessary and sufficient conditions to explain specific outcomes, underpinned by the logic of Boolean algebra (Schneider and Wagemann 2012).

8

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

for effective UN military protection operations. There will be more on the mixed-methods approach in the next section. My findings indicate that UN troops succeed about as often as they fail, countering dominant views of the chronic ineffectiveness of UN troops in providing security to populations under threat. Given the challenging task of stopping violence before it occurs—or reducing its severity while underway—often in inhospitable terrain with only marginal military capabilities, I would even dare to say that UN troops perform quite well when they first decide to intervene. More worryingly—conforming to the findings of other studies—I find that UN troops almost never intervene with force when perpetrators threaten civilians with violence. As such, there is much room for improvement. Moreover, my findings indicate that to be more successful, UN troops must tailor their operations to match and pre-empt perpetrators of violence, moving beyond the UN’s blueprint approach of spreading troops thinly throughout the area of operations with the hope of deterring would-be perpetrators of violence. The case studies of operations in South Sudan and the DRC further point to additional explanations that could facilitate matching and pre-emption, including host-state support, “good” troop-to-perpetrator ratios, force mobility and projection, operational art, and improved threat-analyses to understand the perpetrators of violence. These findings question current policies, guidelines, principles, and practices of UN peace operations. Most importantly, they speak to the need to think systematically about what UN troops can and cannot do to protect civilians from violence. With this book, I aim to unpack the problem of protection by force and provide insights that can help to inform future studies and practices of protection in UN peace operations and beyond. We should nevertheless carefully avoid becoming too ambitious in our hopes to provide definitive answers to what works for every situation, as wars and armed conflicts tend to skew our theoretical assumptions about their dynamics, demanding continuous theoretical and practical tweaking tailored to particular situations for the best possible outcome.

Defining success in UN military protection operations This book is part of a larger debate on peacekeeping effectiveness. How can we know if peacekeeping works? Given such a complex social phenomenon which varies through an immense number of variables produced by several

DEFINING SUCCESS IN UN MILITARY PROTECTION OPERATIONS

9

actors—global and regional organizations, governments, non-governmental organizations, communities, and individuals—influenced by policies, cultures, history, perceptions, power, force, violence, and risk, there is obviously not one definite way to answer that question. Other scholars have done a terrific job of explaining where we stand in this field; what we measure, why, with what methods and data, including the strengths and weaknesses of our analyses (Autesserre 2014; Di Salvatore and Ruggeri 2017; Diehl and Druckman 2015; Doyle and Sambanis 2006; Fjelde, Hultman, and Nilsson 2019; Fortna 2007; Fortna and Howard 2008; Hegre, Hultman, and Nygård 2019; Hultman, Kathman, and Shannon 2019, chaps. 2–3). I will not attempt to outdo these efforts, but rather provide you with an idea of how I go about analyzing the outcome of UN military protection operations. Importantly—although I do stand on the shoulders of the eminent scholars referred to above—I do not seek to analyze the overall effectiveness of UN peace operations. As such, I am not adding to debates about positive or negative peace or other absolutist conceptualizations of peacekeeping success and failure (Hultman, Kathman, and Shannon 2019, 28–32). Rather, I am inspired by Hultman and her colleagues’ evaluation of peacekeeping effectiveness in relative terms, seeking to explore variation in nuances and outcomes of the activities of Blue Helmets on the ground (Hultman, Kathman, and Shannon 2019, 32). While this is perhaps a less ambitious research agenda, I avoid many of the difficulties of studying global peacekeeping’s effectiveness on war and peace. By “going micro”—choosing to narrow down to a niche phenomenon in peacekeeping and conflict—I hope to contribute, bottom-up, to theoretical building blocks that may ultimately make peacekeeping an even more effective tool to protect civilians from violence (Autesserre 2014). Building on—but also moving beyond Hultman and her colleagues—my starting position is that it must matter what UN troops do to protect civilians against different types of threats, not least because we know that Blue Helmet presence does not deter all types of perpetrators. Deterrence is not a universal phenomenon, rather its effectiveness varies along a wide range of conditions (Sundberg 2020). Unfortunately, most qualitative analyses of UN protection efforts rely on single-case study designs seeking to explain particular protection failures. This approach omits knowledge from successful cases, which could provide stepping stones toward a theory on the utility of force to protect. Comparative studies of military protection efforts across time and missions are rare, creating wanting knowledge on how UN troops have fared overall when they intervene with force to protect civilians. We simply do not know the distribution of successes and failures of UN military protection operations.

10

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

This knowledge gap undermines our attempts to understand this phenomenon and limits our ability to make UN troops more effective. The main reason for these knowledge gaps is a lack of systematic and updated event data that can form the basis for such comparative analyses (Clayton 2016). Consequently, there are few theoretical building blocks to assist the search for the utility of force to protect civilians in UN peace operations. With the help of the UNPOCO dataset, I attempt to address this shortcoming, capturing 200 military protection operations across 10 UN missions from 1999 to 2017 according to the following four criteria: i) perpetrators physically threatened or harmed civilians; ii) UN military troops, with a mandate to protect civilians, deployed to the location where civilians were threatened or harmed; iii) UN troops used military force to protect civilians, by applying one or more of the four functions of force: amelioration, containment, deterrence/coercion, and destruction; iv) the UN Secretary-General’s reporting to the UNSC captured the event. How can we best evaluate the outcomes of these military UN protection operations across time and different UN missions and search for generalizable causal explanations for military protection successes? While it would be possible to explore and compare a limited number of these cases in depth—which I do in Chapters 6 and 7—a traditional qualitative case-study strategy alone is not able to analyze potential causal explanations across a large number of cases. It is also possible to search for relevant statistical relationships between potential causal conditions and outcomes—which I do in Chapter 4—but the number of relevant cases is still quite low for a large-N study, they are not randomly selected, and the findings will therefore remain uncertain. In addition, the cases occurred in different countries, at different times, involving different troops faced with different perpetrators targeting different civilian populations in different ways. These observations point to a phenomenon characterized by causal complexity, which is challenging to capture merely with statistical analyses. Causal complexity, as understood in this book, refers to social phenomena portraying characteristics such as equifinality (different causal paths leading to the same outcome), conjunctural causation (causal conditions that only in combination lead to the outcome of interest), and asymmetry (that the occurrence and non-occurrence of any given social phenomenon might require different explanations involving different causal conditions) (Schneider and

DEFINING SUCCESS IN UN MILITARY PROTECTION OPERATIONS

11

Wagemann 2012, 5–6). Therefore, there is a need for a middle way, using different methods for different purposes at different stages. Consequently, I use a mixed-method research design. The aim is to exhaust the methods’ complementarity, while compensating for weaknesses in each (George and Bennett 2005, 4–8). I employ three methods: (i) quantitative analysis, including descriptive statistics, statistical cross-tabulations, Chi-square tests, and multivariate linear probability analysis; (ii) fuzzy set Qualitative Comparative Analysis (fsQCA); and (iii) qualitative comparative case studies, using process tracing of causal mechanisms. I am also guided by counterfactual reasoning when I analyze outcomes of operations. While this method often unconsciously colors social science research, it is seldomly applied openly and systematically. Although I am inspired by the core tenets of “possible-worlds”-thinking, I cannot claim to perform a fully systematic counterfactual analysis within these pages. I have developed the UNPOCO dataset from the openly available reporting from the UN Secretary-General to the UNSC. Chapter 3 first provides descriptive statistics of the data captured and coded in the UNPOCO dataset. The rationale for including this basic quantitative analytical step is to start covering the gap in systematic event data on military protection operations across time and UN missions (Clayton 2016). Knowing more about the location, year, casualty figures, the type of threat civilians faced, and how force was used to protect, will help us recognize this phenomenon’s most basic characteristics. This step is of value of itself but adds little to our understanding of how UN forces have fared in protecting civilians over the past two decades in particular situations. Therefore, responding to the first research question—to what degree have UN military troops provided protection to civilians under imminent physical threat in Africa between 1999 and 2017—I estimate variations in outcomes of these operations based on counterfactual reasoning toward the end of Chapter 3. Counterfactual reasoning commonly appears in political science studies, although it often remains an implicit part of the research strategy (Fearon 1991). Attempts to define the role of counterfactuals in the study of causal relations have been around since Hume, but it was not until the late 1960s that rigorous counterfactual theories and methods emerged (Menzies 2014). In 1973, David Lewis first published his theory on counterfactuals, which has been revised on several occasions over the past decades (Lewis 2001).⁵ Lewis’s ⁵ For a fascinating insight into David Lewis’s life, seek out Season 5, episodes 1–4, of the podcast Hi-Phi Nation—A Show about Philosophy that Turns Stories into Ideas. https://hiphination.org/

12

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

elaborations still influence contemporary debates about the value of counterfactuals. Jon Elster has also provided insights into how counterfactuals can be of value in interpreting historical events (Elster 1978, 175–218). Despite disagreeing on what makes counterfactual reasoning valid, a core element of both Elster’s and Lewis’s theorizing is linked to the idea of possible worlds. Put simply, counterfactual reasoning must rest on alternative outcomes—possible worlds—that resemble the actual event as closely as possible. When introducing a counterfactual condition, it should make minimal change to the real situation and must therefore be close in time to where the real world and the counterfactual possible world branched off. Although the literature on counterfactuals is rather intricate—often resting on lengthy formal logical arguments—the basic idea underpinning counterfactual reasoning “is that the meaning of causal claims can be explained in terms of counterfactual conditionals of the form ‘If A had not occurred, C would not have occurred’” (Menzies 2014). The basic tenets of counterfactual reasoning are useful for answering the first research questions posed in this book. First, I ask in each case what is likely to have occurred without a UN military intervention. I try to establish a counterfactual possible world where the UN did not intervene, to be able to analyze what effect the Blue Helmet intervention had in real life. I try to minimize the changes to the possible world by only removing one condition, i.e., the UN military effort to protect. To be able to analyze the outcomes, I use case-specific knowledge about the modus operandi of each particular perpetrator of violence, and lean on the threat scenarios developed by Beadle that capture generic traits from a wide range of similar perpetrators (Beadle 2014, 2015). Second, I compare the possible world with the actual outcome following a UN intervention, leading to an analysis of whether a few or many civilians were protected in each case. I do not attempt to explore the longer-term effects of the protection operation, which would potentially undermine the value of the counterfactual reasoning by introducing second-order effects. Each analytical chapter (Chapters 3–7) provides a more detailed explanation of how the outcomes are operationalized to fit each method. Although this approach provides rigor to the analysis of the effect of UN military efforts to protect, it remains challenging to determine counterfactually how and to what degree the UN military response influenced the perpetrators of violence. One challenge is that a military protection operation consists of many variables. Although the ideal of counterfactual methods is to remove only one variable to say something meaningful about possible worlds, UN military protection operations are often quite complex. Even though I treat

UNPACKING UN MILITARY PROTECTION OPERATIONS

13

operations analytically as one event—limited in time and scope—many of the cases captured in UNPOCO contain several variables that are not accounted for. As such, it is only possible to broadly estimate the outcomes of UN protection operations. The findings in the book should be read in light of the above methodological issues. The mixed methods approach certainly provides nuances that would otherwise remain unaccounted for, which strengthens the validity of the findings. However, both the statistical analysis (Chapter 5) and the Qualitative Comparative Analysis (Chapter 6) rest on the same data from UNPOCO, which we know have weak spots. Given that I have systematically captured events over a long time period—using the same case selection criteria—we can be reasonably certain that I have found a representative selection of cases to analyze these phenomena. The empirical data presented in Chapter 3 is therefore also likely the best available description to date of these phenomena across time and place. The counterfactual approach to assessing outcomes of operations comes with its own set of limitations, as discussed above. To facilitate critique of my analyses, the qualitative reasoning linked to each case is included in the UNPOCO dataset (Kjeksrud 2019). I would encourage other scholars to test my findings to see if other methods better can analyze the effectiveness of Blue Helmets in operations, as my approach can only bring us thus far. I still feel confident that my main findings provide relevant knowledge to make protection operations more effective: they must be tailored to particular threats and be employed in time to reduce or stop violence against civilians. Exactly how such tailoring is best designed, and what “in time” means in particular situations, may vary across cases.

Unpacking UN military protection operations Military force is a controversial topic for the UN. As we have seen, most troopcontributing countries to UN peace operations remain hesitant to use force for any purpose. Forceful interventions by uniformed peacekeepers can also go against the grain of other functions performed by the global organization, such as promotion of human rights, social progress, and respect for international law (Karlsrud 2015, 49). Such reluctance to embrace force as part of the UN’s conflict management arsenal is quite understandable, not least since as a rule, the UN Charter prohibits the use of force in international relations (United Nations 1945, chap. I, article 2). Nevertheless, military force—or at least the idea of its potential utility—has occupied a prominent role in the UN toolbox ever since the organization’s

14

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

inception. Importantly, the UN Charter provides two exceptions to the nonuse of force clause. Firstly, a country has the right to defend itself by force if attacked (United Nations 1945, chap. VII, article 51). The Russian invasion of Ukraine in February 2022 provides clear examples of how force is prohibited (the Russian invasion) and how the use of force under certain conditions is allowed (Ukraine’s self-defense efforts against the Russian invasion, including calls for help to the outside world). Secondly, the UNSC can determine that a threat to international peace and security demands a forceful response (United Nations 1945, article 42.).⁶ This latter exception is also known as the UNSC’s collective security responsibility (Willmot 2016). At the time of writing—toward the end of 2022—the world is understandably mostly concerned with Russia’s war in Ukraine. Since the end of the Second World War, however, the frequency of wars between states—or interstate wars—has dwindled. Since the end of the Cold War, the UNSC has been mostly concerned with addressing threats to international peace and security emerging from within countries in conflict—often referred to as intra-state wars—a phenomenon that has seen a significant upswing in the same period (Pettersson and Öberg 2020, 600). It is within some of these civil wars that governments in weak or failing states invite a UN peacekeeping presence—in theory voluntarily, but in practice often under significant international and regional pressure—to avoid further bloodshed (Antonini 2009). Today, ambitiously mandated UN missions aim to stabilize post-conflict situations, implement peace agreements, and even rebuild societies in the aftermath of armed conflict to accomplish a more democratic, prosperous, and peaceful longer-term development. It is also within these armed conflicts that the need for military force is most pronounced, since “[i]ncreasingly, missions are being deployed where there is no peace to keep, and no peace agreement to defend, where grave abuses are being committed against civilians” (Ki-moon 2014). Although UN troops are expected to use deadly military force to protect civilians—if all else fails—UN mandates have not always authorized the use of this controversial tool (Koops et al. 2015). In the not-so-distant past, the Security Council deployed lightly armed UN missions to oversee cease-fire and peace agreements, merely expecting UN troops to stand in-between former belligerents, report any transgressions of agreements to the outside world, and only defend themselves from harm. This less ambitious approach was well ⁶ I will not go into Security Council working methods, procedures, and processes, as great resources on these topics can be found elsewhere (Malone 2004; Security Council Report 2020; Weiss and Daws 2018).

UNPACKING UN MILITARY PROTECTION OPERATIONS

15

suited to an era where we still saw several armed conflicts between states, and where Cold War logic restrained the effectiveness and authority of the Security Council and the UN Secretariat. Indeed, we might experience a revival of these ideas in the aftermath of the war in Ukraine. The end of the Cold War, however, brought with it two significant developments of relevance to the use of force in UN peace operations and protection of civilians. Firstly, in lieu of great power competition, the UNSC became more active and involved in issues related to international peace and security (Malone 2004). As seen in Figure 1.1, there was a significant increase in the number of peace operations and personnel deployed at the beginning of the 1990s. 100,000 80,000

Troops Police Experts

60,000 40,000 20,000 0

1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017

Figure 1.1 Total peacekeepers deployed by type, 1990–2017. Source: Providing for Peacekeeping (2018).

Secondly, the characteristics of modern intra-state conflict quickly laid bare the UN’s inability to implement ambitious mandates, including massive failures to protect civilians from mass atrocities in Rwanda and Bosnia. In short, the conduct of wars among, for, and against the people strained the existing concepts and practices of peacekeeping. As a result, the UN entered a period of introspection in the mid-to-late 1990s, where few UN missions were deployed. Two reports by the then Secretary-General Boutros Boutros-Ghali aptly describe this roller-coaster ride in the 1990s. His Agenda for Peace from 1992 mirrors optimism on behalf of the UN as a promoter of international peace and security (Boutros-Ghali 1992). Remember, this was before mass killings took place in Srebrenica and Kigali. In his Supplement to the Agenda for Peace from 1995—shortly after genocide and ethnic cleansing had caught the UN completely off guard—the Secretary-General reevaluated the UN’s role and ability to protect civilians from violence (Buotros-Ghali 1995). These horrific events and accompanying protection failures induced significant shame in the UN Secretariat and among several Security Council members (Keating 2014).

16

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

A never-again mantra took hold in the UN, not least influenced by the soul-searching of the late Secretary-General Kofi Annan, who oversaw the Department of Peacekeeping Operations (DPKO) at the time of both Rwanda and Srebrenica (Annan 1999b, 1999c, 2000). The UN leveraged these human-made catastrophes to spur significant and impressive reform efforts that are still ongoing. And already by the end of the 1990s, the UN returned to addressing threats to international peace and security through highly ambitious projects, best exemplified by running the day-to-day governance of both Kosovo and East Timor. Since then, we have witnessed an ever-expanding peacekeeping agenda. The UN has taken on highly complex state- and peacebuilding tasks, guided by exceedingly ambitious mandates, seeking to rebuild—or even create—legitimate, functional, and democratic governments, bureaucracies, and security sectors. Beginning in 1999, the UNSC also tasked UN peace operations to protect civilians from physical violence, if necessary, with the use of deadly military force (United Nations 1999a)). An important reason for us even discussing POC in UN peace operations today is the consistent UNSC decisions from 1999 until the present day that provide the protection mandate for UNPOs (Security Council Report (SCR) 2015). Quite an impressive feat given the great-power rivalry on just about every other issue on the Council’s table. Other milestones of relevance to UN military protection operations are the Capstone doctrine from 2008, the POC policy—first issued in 2015 and then revisited in 2019—and implementation guidelines for UN military components on POC and the use of force (United Nations 2008, 2017a, 2019b). I will return to these in more detail in Chapter 2. The most important take-away at this point is that UN troops are only allowed to use deadly force at the tactical level, which implies that force should not be used to influence the overall dynamics of armed conflicts. Further, force is just one of many tools available to multidimensional and integrated UNPOs, the others including political and diplomatic efforts, as well as policing and civilian protection expertise. Adding the bedrock principles of peace operations—consent, impartiality, and the non-use of force—it becomes clear that many restrictions apply to the use of force in UNPOs, despite it playing a prominent role in fulfilling their most important task. Finally, UN troops are only meant to protect civilians from physical threats. According to the updated POC policy from 2019, such threats: encompass all hostile acts or situations which are likely to lead to death or serious bodily injury of civilians, including sexual violence, regardless of

UNPACKING UN MILITARY PROTECTION OPERATIONS

17

the source of the threat. This includes, inter alia, threats posed by non-state armed groups, self-defence groups, domestic and foreign state defence and security forces and other state agents and state-sponsored armed actors, as well as extremist groups and communities. It includes both direct and indiscriminate attacks, and attempts to kill, torture, maim, rape or sexually exploit, forcibly displace, starve, pillage, abduct or arbitrarily detain, kidnap, disappear or traffic persons or recruit and use children. It also includes harm associated with the presence of explosive ordnance including mines, explosive remnants of war and improvised explosive devices. “Threat” includes both violence against civilians which has materialised and is ongoing and violence which has the realistic potential to occur. The threat need not be imminent, unless the specific Security Council mandate requires this. (United Nations, 2019b, para. 23)

Evidently, this definition captures a wide variety of threat situations. Nevertheless, we know that UN troops cannot always protect every civilian under threat. In fact, we know that UN troops very seldom intervene at all when danger looms (United Nations Office of Internal Oversight Services 2014). Despite much talk about a “robust turn” in UN peace operations, there is very little application of force in practice (Hunt 2016). One reason for the perception that the UN has become (too) militarized is the lack of trustworthy and openly accessible data on the conduct and outcome of UN military efforts to protect. My systematic collection of cases provides a rich empirical universe in which to begin searching for the utility of force to protect. I will present the full depth of the empirical data in Chapter 3 but will briefly mention some of the main observations here. Unsurprisingly, I find that UN troops have been involved in very different situations on the ground across time and operations. Recall that the data only captures the period from 1999 to 2017. In the not-so-distant past, UN troops and civilians encountered different and more serious situations than we have seen during the last two decades, such as in Rwanda (genocide) and in the Balkans (ethnic cleansing and an act of genocide (Srebrenica)), with casualty figures in the tens and even hundreds of thousands. In the future, new types of threats may emerge. Therefore, if we expect UN peacekeepers to effectively use force to protect civilians, they must be prepared to address both limited attacks—with relatively few killed and injured—and mass atrocities, where the civilian casualty figures are significantly higher. Furthermore, I have already emphasized that using military force to protect remains a marginal phenomenon in UNPOs. UNPOCO data reconfirm

18

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

that Blue Helmets seldom respond with force when perpetrators of violence physically threaten civilians. However, while the early 2000s saw almost no protection operations across UN missions, the trend since 2008 seems to point toward a steady increase in the annual number of protection operations. The increase may reflect a slow turn toward more consistently robust military responses to protect those under threat of violence. It could also reflect an increasing number of UN troops on the ground, greater demand for physical protection where UN troops deploy, or even changing reporting practices in the organization. Based on data captured in UNPOCO, I find that only a handful of UN missions have undertaken most military protection operations, with the UN mission in the DRC (MONUC/MONUSCO) standing out as the top-ranking mission. Indeed, many of the best practices and lessons learned underpinning current UN policies and guidelines stem from peacekeeping experiences from the DRC. However, the data captured in UNPOCO indicate that there is value in casting the net wider to capture lessons from other missions as well. The most common threats UN forces and civilians faced between 1999 and 2017 in African conflicts were predatory armed groups, communities in conflict, and insurgent rebel groups, although other both more (ethnic cleansing) and less serious threat scenarios (mob violence) have occurred during this period. While insurgent violence does not usually lead to a large number of civilian casualties, both predatory violence and communal violence often do. The data show that UN forces must be prepared to protect civilians against a wide variety of threats, but that some types of threats should probably receive more attention in training and preparation for deployment. Faced with these threats, UN military troops have employed the whole spectrum of force to protect civilians—including destructive use of force— but they have relied most heavily on deterrence and coercion. This underlines that UN forces can operate forcefully, and have done so on quite a few occasions, although UN missions are most seen to employ defensive and reactive responses to threats against civilians. Civilian casualty figures from these operations remain challenging to collect and analyze. A sub-set of cases from UNPOCO indicates great variation, with a few cases that are very violent (casualties in the hundreds), while most cases involve few civilian casualties. In comparison, datasets that are more comprehensive show that the civilian death toll between 1999 and 2017 across these conflicts vastly outnumbers the civilian fatalities captured by UNPOCO (ACLED 2020; Uppsala University 2020). Clearly, UN forces are not able to address most of the violence against civilians occurring in these conflicts.

UNPACKING UN MILITARY PROTECTION OPERATIONS

19

UN fatalities from what are known as “malicious acts” relative to troop numbers remain low. Previous research has shown that peacekeepers more commonly die from illness and accidents (Henke 2016; Rogers and Kennedy 2014; van der Lijn and Smit 2015). The United Nations Multidimensional Integrated Stabilization Mission in Mali (MINUSMA) is an outlier, where UN troops are countering insurgent rebel groups together with, and sometimes on behalf of, the Malian government (Kjeksrud and Vermeij 2017). Although insurgent groups regularly attack MINUSMA and government forces, the UN mission, thus far, performs few military protection operations. As the situation in central Mali is changing, demanding regular protection efforts from MINUSMA, the UN mission’s approach is also bound to change (Gallagher, Lawrinson, and Hunt 2022). Hence, few UN troops die because of participating in this task. The data captured in UNPOCO indicate that as few as 14 military peacekeepers have been killed by malicious acts while actively protecting civilians during the 200 operations from 1999 to 2017. These fatality figures indicate that there may be a potential to be somewhat more risk accepting. As argued by others, it is plausible that a more robust stance could even make UN troops more secure (dos Santos Cruz, Phillips, and Cusimano 2017). Finally, poor data quality and limited availability have been recurring challenges throughout the data collection process. This is because of the limited scholarly attention paid to the phenomenon, the lack of systematic reporting and analysis of UN military protection operations, and the challenges of obtaining reliable information from the conflict areas to which UN troops are deployed to protect. Online sources—the UN Secretary-General’s openly available reporting to UNSC in particular—have been the most important for developing the UNPOCO dataset. In general, proper triangulation of the data sources has not been possible. To counter some of these weaknesses, I also rely on written sources, personal observations, and data collected from semi-structured interviews that I conducted during fieldwork. Due to these limitations in data quality, it is challenging to convey exactly how much you should trust the findings within these pages. My approach is therefore to lay bare the weaknesses and invite you to consider how they may impact the validity of my analyses. In my humble opinion, there are good reasons to pay attention to the main insights of designing timely and tailored operations to particular threat situations. How that should be reflected in the planning and conduct of operations must be considered against a wider variety of considerations in each case. From a scientific standpoint, I am certain that other scholars will find better ways at analyzing these phenomena, which I will sincerely welcome.

20

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

Understanding the perpetrators to tailor effective operations Equipped with a better idea of the characteristics of UN military protection operations—before we move on to the data, analyses, and findings—I will discuss one key theoretical premise for finding the utility of force to protect; the need to understand the rationale and modus operandi of perpetrators of violence. The only existing theory that seeks to explain how force can be used with more utility to protect argues that the most important premise for increasing military forces’ effectiveness is to better understand why and how perpetrators attack civilians as part of their strategy (Beadle 2014, 2015). I put Alexander Beadle’s theory to the test later in the book, but I accept his premise as a starting point for my analyses.⁷ Without in-depth knowledge of the rationales driving perpetrators to target non-combatants, how they translate these rationales into violence, and what capabilities they need to do so, it seems unlikely that any military protection effort will succeed. We can even recognize elements of Sun Tzu’s eternal insights on the conduct and outcomes of war echoed in Beadle’s commonsensical principle, “[i]f you know the enemy and know yourself, you need not fear the result of a hundred battles. If you know yourself but not the enemy, for every victory gained you will also suffer a defeat. If you know neither the enemy nor yourself, you will succumb in every battle” (Griffith 1963). Military planners and practitioners are of course aware of the need to understand their environment in any type of operation to increase their chances of success. Most military planning processes include systematic threat assessments and war-gaming, including what is known as red teaming, where part of the planning group simulates a free-thinking opponents’ potential responses to the developing plan (Appleget, Burks, and Cameron 2020). Critically, protecting civilians from violence inflicted by others implies an important change in military mindsets. Commonly, military forces plan, prepare, and conduct operations against armed actors that oppose them. When the mission is to protect civilians from physical violence, military forces must tailor operations to influence why and how perpetrators target civilians. This conceptual difference in thinking about human security in operations leads to

⁷ Alexander Beadle and I were colleagues at the Norwegian Defence Research Establishment (FFI) when he developed his theory, and we have worked closely on these issues together over several years. I am aware that this professional relationship and friendship could bias my analysis of his theory, although I have made every effort not to let this color the scientific approach found within these pages.

UNDERSTANDING THE PERPETRATORS TO TAILOR EFFECTIVE

21

implications that are less understood in the literature and practice of protection operations thus far. One of the ambitions of this book is to unpack some of these implications and point to possible ways forward to increase the utility of force to protect. Importantly, I am not interested in whether UN troops should use more or less force, although I understand the controversies involved in providing advice on when to do what. Rather, this book is about understanding when and how force may be used with utility, and when we can expect force to have little or no utility, based on empirical data and scientific methods. Puzzlingly, the nature of threats to civilians does not necessarily decide where or how UN troops are deployed and operate within missions, although protecting civilians from such threats is the prioritized task for almost every military UN peacekeeper. Usually, each troop-contributing country is responsible for a particular geographical area in which they set up camps—more or less evenly distributed—depending on national infrastructure and UN logistics capacity, both of which can be quite limited. Daytime patrolling from these hubs usually ensues. This modus operandi is commonly mirrored in the UN Secretary-General’s reporting to the UNSC, which tends to provide a tally of the number of patrols performed in a given period (United Nations 2020b, para. 43). Worryingly, there is usually neither measurement nor analysis of the effect of these patrols on civilian security. Again, the underlying premise seems to be that the presence of UN troops will deter would-be perpetrators of violence, although we know that this approach has failed many times in the past. The reliance on “protection by numbers” is also the dominant thinking in quantitative cross-case studies. This book rather seeks to understand what those troops do and to what effect on civilian security, forming part of emerging efforts to analyze the effects of UN military activities with the help of military theory (Williams 2023). Recently, we have indeed seen attempts to change UN deployment patterns to provide dynamic protection responses, including Temporary Operating Bases/Company Operating Bases (TOB/COB) in the DRC, South Sudan, and Mali. Although the UN has identified and sometimes acted on the need for more flexible troop postures, the dominant approach is still to spread troops thinly across the area of operation. While recent research indicates that UN troops actually do deploy to “the frontline: they go where conflict occurs … there is a notable delay in their deployment. Furthermore, peacekeepers tend to be deployed near major urban areas” (Ruggeri, Dorussen, and Gizelis 2018, 1005). These findings appear to support the observation that UN troops often find themselves several steps behind the perpetrators of violence, although they do respond eventually.

22

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

Deploying troops to where civilians have already been attacked can in theory deter future attacks against members of that community, although there is no guarantee that such a deterrent effect will materialize. Unfortunately, UN troops are seldom able to hold and control areas over time given deficiencies in troop numbers, logistical support, and willingness to accept risk, which often leaves communities vulnerable to repeated attacks. Moreover, many perpetrators attack civilians merely for profit, largely circumnavigating UN deployment patterns, easily moving to the next undefended target. Critical to understand this phenomenon is the fact that responding after civilians have been attacked matters little to those already killed, expelled, or otherwise harmed by perpetrators of violence. These observations indicate that military force may be more effective in seeking to influence the perpetrators’ will and ability to attack civilians in the first place, rather than responding in the aftermath, seeking to alleviate the consequences of attacks already instigated. However, we still know little about what types of military responses will work in different situations. Countering would-be perpetrators before attacks have materialized would also be controversial, although the UN POC policy points in that direction with the objective of becoming more effective (United Nations 2019b, para. 23). Observations from past conflicts and operations strongly support the insight that understanding threats matters to protect civilians more effectively. I have already mentioned NATO’s failure to recognize the most serious threat to civilians in Afghanistan, even skewing the lessons learned approach in the aftermath of that conflict. This is not the first time NATO has struggled to account for the nature of threats to civilians. During the Kosovo war, in the spring of 1999, NATO’s strategy largely relied on aerial attacks against Serbian infrastructure, regular military units, and command and control hubs inside Serbia proper to stop the unlawful targeting of civilians in Kosovo. Nevertheless, while NATO bombed Belgrade, paramilitaries nearly emptied Kosovo of Albanians (Independent International Commission on Kosovo 2000, 88–9). Moving village to village and door to door in Kosovo, these irregular militias sought to ensure that the local population never returned to their homes, creating fear through occasional massacres, and destroying homes and buildings of cultural significance. As such, the ethnic cleansing campaign mostly succeeded, although NATO’s bombing campaign did eventually bring Milosevic to the negotiation table. This points to another important aspect of human security and the role of military force; the concern for civilian life during armed conflict sometimes counters other strategic aims of military operations. The UN has also struggled to understand the rationale driving the perpetrators of violence. For example, toward the end of 2011, in one particularly

UNDERSTANDING THE PERPETRATORS TO TAILOR EFFECTIVE

23

violent event in South Sudan, the Lou Nuer White Army amassed some 8,000 militia fighters in the northern parts of Jonglei⁸ to exact brutal revenge on the Murle, their long time enemy (UNMISS Human Rights Division 2012). A few months earlier, Murle fighters had killed hundreds of Nuer civilians and displaced many more in Pieri, also in Jonglei (BBC News 2011). Within a few days, the White Army killed more than 600 Murle civilians during their violent southward march (UNMISS Human Rights Division 2012). In response, numerous small bands of Murle fighters launched immediate revenge attacks against the now ill-defended Nuer communities, killing another 275 civilians over the next month UNMISS Human Rights Division 2012). These figures are probably conservative estimates, and others have claimed much higher fatality numbers (Gettleman 2012). Not only did these intense armed clashes produce many civilian fatalities, but they also led to tens of thousands of people forcibly displaced, abduction of numerous women and children, extensive material damage to homes and livelihoods, and countless heads of stolen cattle, to the economic detriment of both communities (UNMISS Human Rights Division 2012). This time around, the White Army instigated the main attack, but the roles could well have been reversed. Amid the extreme violence, the role of the United Nations Mission in South Sudan (UNMISS) was mostly that of a frustrated spectator to the civilian suffering, although it made substantial efforts to provide early warning and to deploy extra forces to defend Pibor town, the Murle’s main population center. However, UNMISS failed to recognize that Murle civilians—wherever they lived—were the center of gravity for the perpetrators of violence, and not their infrastructure, towns or armed forces. UNMISS also failed to foresee the immediate revenge attacks perpetrated by the Murle, leading to many fatalities. When armed communities attack each other to avenge previous attacks and to deter future attacks against them, the role of victim and perpetrator is constantly shifting (Beadle and Kjeksrud 2018). The main reason for the seemingly endless cycle of violence is that none of the communities possesses sufficient military capabilities to end the conflict decisively. Instead of creating more security, this deadly balancing act of deterrence and revenge between weak enemies usually leaves each community vulnerable to new attacks in the future. The UN has also achieved successful protection outcomes. In 2012, the armed rebel group Mouvement du 23 March (M23) in the DRC caused ⁸ South Sudan has changed its administrative structure several times since 2011. Jonglei is now significantly smaller than in 2011. Moreover, Jonglei partly seceded from South Sudan in 2011, and today enjoys some level of autonomy.

24

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

regional and international uproar when it captured and held Goma, the main city in the eastern DRC and the main hub for UN peace operation activities in the country. After having overrun Goma, the M23 even threatened to march toward the capital Kinshasa (BBC News 2012; Jones and Smith 2012). While this latter threat was no more than hot air, the M23 had already caused significant civilian suffering during its reign. The M23 nonetheless used violence more sparingly and for different purposes than the White Army in South Sudan. According to the United Nations’ Joint Human Rights Office in the DRC, the M23 is thought to be responsible for killing 116 persons, physically harming 351 persons (including 161 rapes), as well as many instances of forced labor, arbitrary arrests, and illegal detentions (United Nations Joint Human Rights Office 2014, para. 14). Many of the killings were perpetrated when local authorities failed to cooperate with the M23 (United Nations Joint Human Rights Office 2014, para. 16). There were indeed instances of predatory violence committed by the M23—for personal benefit in some form—but overall, the rebel group mostly spared civilian populations in areas under its control from extreme violence. The aim of the M23 was thus not to maximize violence against civilians to avenge previous attacks and deter future attacks against them, a motivation clearly displayed by the White Army in South Sudan. Rather, the M23 used limited violence against civilians as a tool to challenge the government’s and local power brokers’ grip on the eastern DRC, seeking to increase its political influence and dominate local governance in areas under its control. The M23 even started training governors and its own police forces, in preparation for a future position of political power. This is typical for political insurgencies, seeking to rule populations in the future (Beadle and Kjeksrud 2018). In response to M23 exactions, UN troops in the DRC employed attack helicopters, sniper units, and artillery to dislodge and defeat the M23 in and around Goma, in support of the Forces armées de la république démocratique du Congo (FARDC), the national army (Kjeksrud and Vermeij 2017). Although this outcome is novel, the Force Intervention Brigade (FIB) and FARDC quite aptly analyzed the modus operandi of the M23 and its capabilities, leading to effective counteroperations. We will return to both cases in Chapters 6 and 7, as they provide ample evidence of variations in perpetrator motivation and tactics, as well as outcomes in terms of civilian suffering. They also reflect very different UN military protection responses, providing insights that are helpful to uncover variations in the utility of force to protect. Curiously, the UN responded with unprecedented force against a relatively “benign” perpetrator in the DRC,

SO WHAT?

25

while struggling to muster a relevant military response to a massive threat to the civilian population in South Sudan. Besides providing relevant empirical data to answer the puzzles that drive the analyses in this book, these two cases highlight that being present is in many cases not enough to deter perpetrators from instigating attacks against civilians (Sundberg 2020). It matters what UN troops do to protect civilians from different types of threats. If we accept that understanding the perpetrators of violence is a critical aspect of increasing the utility of force to protect, the question is then how can we best understand them? Insights from the existing literature indicate that perpetrators attack civilians in many ways and for many different purposes (Chenoweth and Lawrence 2010; Kaldor 2007; Kalyvas 2006; Slim 2007). To capture this variation in threats, I employ a threat typology—also developed by Beadle, with a minor contribution from me—spanning from genocide to mob violence (Beadle 2014; Kjeksrud, Beadle, and Lindqvist 2016). With the help of this analytical tool, I can describe systematically the most typical situations facing UN troops when they are tasked to protect and analyze how UN troops have fared in different types of situations. Although the typology necessarily simplifies the full variation in threats facing civilians over the past two decades in African conflicts, it provides a useful tool to categorize and make sense of similar threats. I believe this to be a significant step forward in the study of protection of civilians in UN peace operations.

So what? Why should we care about how UN troops fare in protecting civilians from violence in far-away places and struggle to unearth hard-to-come-by-data to seek explanations for successful military protection outcomes that few seem to care about? I believe there are at least three convincing reasons why we should pursue a better understanding of this phenomenon. Firstly, UN peace operations are arguably the most effective tool for international conflict management, despite all their many flaws. Emerging evidence, as I have referred to, points to significant conflict-reducing effects of UN peace operations. Not only does the presence of large peace operations decrease the intensity of armed conflict in terms of civilian casualty figures, the presence of large multidimensional operations also decreases the duration of war, increases the longevity of peace, and reduces the chances of armed conflicts spreading. If maintaining international peace and security is important, improving UNPOs is important. Secondly, despite the positive macro-effects on armed conflict identified in the existing quantitative literature, there is a vast potential for improving

26

PROTECTION BY MILITARY FORCE IN UN PE ACE OPERATIONS

our understanding of what works in different situations to protect civilians from violence. Notwithstanding a range of innovative conceptual and political developments over the past decade, the practice of UN peace operations still largely rests on lessons learned and silent knowledge about what does and does not work. While this approach may work in some situations, it fails to employ systematic knowledge from studies of successes and failures across time and UN missions. Such systematic studies—of which this book is one representative—can lead toward a tailored theory of the utility of force to protect civilians from violence. While a future theory will not solve every protection challenge, it beats testing myriad hypotheses through practice, where no result means more harm to civilians. Thirdly, based on a more altruistic argument—and perhaps the only argument we need—it is worth continuing our investigation of how to improve the life, dignity, and well-being of civilians caught in armed conflict. Obviously, we cannot expect Blue Helmets to always protect civilians under threat, but we can expect them to do more. Today, the dominating modus operandi of UN peace operations is more akin to firefighting than fire-prevention. As every firefighter knows, it is difficult—if not futile—to anticipate where the next fire will ignite. Rather, it makes much more sense to prevent fires from breaking out in the first place through best practices and systematic knowledge about what causes fires. Similarly, since perpetrators of violence can choose the time and place of attack, it is incredibly difficult to estimate where the next attack on civilians will materialize. However, it is possible to use military means to influence those that are known to target civilians before they do so. With this book, I hope to contribute to a better understanding of how UN troops more effectively can influence the perpetrators of violence, before they launch their attacks, or reduce their ability to inflict harm. I believe it is possible to minimize the insecurity of more people if we maximize the utility of force to protect.

Outline of the book Following this introduction, I explore existing literature in Chapter 2, searching for theoretical insights to guide my pursuit of the utility of force to protect civilians from violence in UN peace operations. I capture four promising causal condition candidates: deterrent presence, willingness to accept risk, pre-emption, and matching the perpetrators of violence. Chapter 2 also elaborates on the threat-based approach to protection of civilians that provides a holistic and systematic approach to analyze physical threats to civilians in

OUTLINE OF THE BOOK

27

armed conflict—capturing the whole spectrum of threats to civilians—and derives potential military response options and courses of action that are likely to lead to better outcomes. In Chapter 3, I provide descriptive statistics based on the UNPOCO dataset, explaining when, where, and how UN troops have used force to protect across 10 UN missions in Africa from 1999 to 2017. I conclude Chapter 3 by estimating outcomes of operations across time and locations with the help of counterfactual reasoning, answering the first of two research questions guiding this book. Given the invariably poor quality of data on the conduct and outcomes of operations in UN reporting, the outcome estimations must be treated with caution. However, my results point to an almost equal number of successes and failures across time and missions, which indicates that UN troops may be more effective than previously thought. Nevertheless, there is room for improvement, as UN troops very seldom intervene with force to protect those under threat. In Chapter 4, I employ statistical analyses to investigate potential relationships between the four causal condition candidates and the outcomes of operations across time and UN missions. Surprisingly, I find that troop numbers and willingness to accept risk—preconditions for the use of force—do not systematically influence the outcomes of operations. Matching the perpetrators and pre-emptive efforts—the modus operandi of UN troops—however, seem to have a more sustained effect on positive outcomes. In Chapter 5, I investigate whether the most promising causal conditions unite in causal pathways to explain outcomes across cases with the help of fuzzy set Qualitative Comparative Analysis (fsQCA), a method based on set theory. While finding that matching and pre-emption combined emerge as the most promising causal recipe to explain outcomes across cases, many nonetheless remain unexplained. In Chapters 6 and 7, I therefore continue to explore the most promising causal conditions in local contexts with the help of process tracing of causal mechanisms, while searching for alternative conditions in deeper case studies of events from the DRC (2013) and South Sudan (2011–2012). The case studies point to additional explanations that could facilitate matching and pre-emption, including host-state support, troop-to-perpetrator ratios, force mobility and projection, operational art, and understanding the perpetrators of violence. Finally, I conclude with a summary of the findings and their relevance to academic literature, UN protection policies, and the practice of using military forces to protect civilians from violence.

2 Understanding the utility of force to protect civilians from violence “[…] the failure to act in the face of mass killings of civilians is not simply a function of political will or legal authority; the failure also reflects a lack of thinking about how military forces might respond.” —Sarah Sewall, former U.S. Under Secretary for Civilian Security, Democracy, and Human Rights (Sewall, Raymond, and Chin 2010, 5)

How can military forces protect civilians from violence most effectively? Unfortunately, the question is still painfully relevant in armed conflicts across the world, a decade and more after Sarah Sewall and her team explored how militaries might respond to mass atrocities. While research performed over the past decade has uncovered an overall conflict reducing effect of UN peace operations, we still know little about how effective military protection operations come about. Moreover, key peacekeeping scholars still question the utility of using force in UN peace operations. Lise Morjé Howard’s seminal book on power in peacekeeping finds that Blue Helmets indeed do reduce death and destruction in civil war, convincingly explaining that they do so mostly through verbal persuasion, economic and institutional incentives, and “military power short of offensive military force” (Howard 2019, x). Howard acknowledges that peacekeepers sometimes compel actors in conflict through coercion but finds that “UN peacekeepers do not draw on compellent force as their main means of power” (Howard 2019, 1). While I share Howard’s understanding of Blue Helmets’ overall modus operandi and her skepticism about a forceful UN, I stray from her logic at one crucial point. Compellent force— meaning coercion and destruction—is sometimes necessary to stop violence against civilians in armed conflict. Without it, the UN will again risk losing all credibility as a peacekeeper, as it did in the mid-1990s. In this book, it also becomes evident that Blue Helmets sometimes do wield kinetic force—and many other types of force—effectively, with positive

Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0002

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

29

outcomes for human security. I want to uncover how Blue Helmets may use force successfully to stop or reduce physical violence against civilians. My theoretical starting point builds on Rupert Smith’s insights that there is a great difference between deploying military forces and employing force (Smith 2008). We need to understand how force—in all its facets—may be utilized with greatest utility to alleviate the negative consequences of armed conflict on civilian life and when the use of force is outright counterproductive. Contrary to Howard, though, I cannot say anything about the overall success of UN peacekeeping, rather indicate how Blue Helmets may become better protectors at the operational and tactical levels when force is indeed called for. While we know that UN peacekeeping operations have conflict reducing effects—despite an array of political, conceptual, and practical challenges— I approach my task humbly aware of the inherent limitations of military interventions in civil wars. Although being the only true global conflict management mechanism with a promising track record, it is not given that UN peacekeeping’s relative success thus far will extend into future armed conflicts. Based on the premise that we are moving toward a multipolar world system— where we may find a multitude of military and political power hubs competing for influence and security—Barry Posen warns outside interveners to expect success in future efforts to manage civil wars (Posen 2017). Using the cases of Bosnia and Syria, Posen strongly indicates that outside intervention in a multipolar world will negatively influence the interveners’ chances of success: […] This new constellation of power seems likely to magnify disagreements about how states suffering civil wars should be stabilized, limit preventive diplomacy, produce external intervention that will make for longer and more destructive wars, and render settlements more difficult to police. (Posen 2017, 167)

While we are transitioning from a unipolar moment toward a bi- or multipolar system, some of these obstacles are already visible in the practice of UN peacekeeping, although still largely driven by local conflict dynamics. Allard Duursma shows how the Sudanese government balanced between international sovereignty—accepting the presence of UNAMID in Darfur to protect civilians—and internal sovereignty—pinioning the same mission to facilitate more effective counterinsurgency efforts, which sometimes included targeting civilians (Duursma 2019, 2021). Other scholars show how armed groups sometimes target the Blue Helmets themselves in order to undermine incumbent regimes that the peacekeeping efforts are there to support

30

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

(Fjelde, Hultman, and Bromley 2016). The utility of force to protect civilians is very much dependent on the context in which it is meant to have effect. To some degree, I avoid these conundrums by studying cases where force actually has been used to protect, paying little heed to the political context. We already know that peacekeepers deploy to where conflict is most intense—despite clustering “around transportation networks, densely populated areas, surfacebased resources, and international borders”—although they seldom intervene militarily for any purpose (Powers, Reeder, and Townsen 2015; Townsen and Reeder 2014, 70; United Nations Office of Internal Oversight Services 2014). I am interested in understanding what peacekeepers do when they do act and to what effect on civilian security. The United Nations Security Council has asked Blue Helmets to protect civilians from physical threats since the end of the 1990s but has largely failed to guide UN troops on how this might be done in practice. This has led to a substantial “implementation gap,” where civilians on the ground are not much better protected despite the strategic importance attached to protection in language and rhetoric of Security Council resolutions and statements. Spurred on by the Council’s directions to protect civilians “with all necessary means”— which is how it expresses its intention to use deadly military force to protect if needed—Blue Helmets indeed used force to protect civilians during the early to mid-2000s, most noticeably in the eastern parts of the Democratic Republic of the Congo (DRC) (Cammaert 2008; see e.g. Isberg and Tillberg 2011). Nevertheless, UN military protection operations remained few and far between for the better part of that decade. UN missions were largely left to invent their own approach to this new and highly demanding task for military forces (see Chapter 3). A whole decade passed until the United Nations Secretariat responded with preliminary guidelines for Blue Helmets on how to protect, largely based on problem solving practices from precisely the UN mission in the DRC (United Nations 2010b). The DRC was and remains the UN’s protection laboratory, where myriad armed groups target civilians as part of their warfare and where Blue Helmets are asked to prioritize physical protection of civilians before other tasks. Experimental treatments in this highly deadly lab include everything from joint civil–military protection teams with a light footprint to heavy-handed intervention brigades, resulting in occasional successes and many more setbacks. In the second decade of the twenty-first century, however, the UN’s conceptual development took flight. A formal policy led the way in 2015—updated more recently in 2019—followed by a range of guidelines, training packages,

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

31

a handbook, and much more (UN Integrated Training Service 2018; United Nations 2015a, 2019b, 2020d). Surprisingly, despite the unprecedented strategic importance attached to providing better protection to civilians and the plethora of guidance for Blue Helmets, civilians still do not seem much better protected on the ground. One reason is that the operational questions of “protecting from what” and “how to protect by force” have not yet been answered, despite the remarkable efforts of the Secretariat over the past decade to make UN missions more effective protectors. Furthermore, existing knowledge provided by the literature on the utility of force to protect is largely overlooked in guiding documents from the UN Secretariat, failing to introduce potential causal links (or lack thereof ) between military protection efforts and reduced threats to civilians based on systematic data and analysis. I will return with more substantial arguments to back this claim later in the chapter. One reason for this “unscientific” approach may be that the literature we do find is rather diverse and largely bereft of dominant theories or debates. Historically, peacekeeping studies “closely followed the practice of peacekeeping,” which only described military peacekeepers’ efforts to protect occasionally (Fortna and Howard 2008, 283). After the protection failures of the mid-1990s, most of the literature naturally turned to address the limits of peacekeepers’ ability to use force to protect civilians under threat (Findlay 2002; Schmidl 1997). Since using force for any purpose remains highly controversial for the UN, much of the more recent literature is still—for many convincing reasons—mostly concerned with the UN’s inherent limitations to using force (Berdal and Ucko 2015; de Coning, Karlsrud, and Aoi 2017; Howard 2008; Howard and Dayal 2018; Karlsrud 2015; Nadin 2018, 201; Tardy 2011). Only by the mid-2000s did the literature begin to concern itself “with any variation between success and failure” (Fortna and Howard 2008, 284). Today, the most dominant strand in the literature—broadly referred to here as quantitative civil war studies—has come far in establishing that peacekeeping “works,” finding that it has an overall positive and conflict reducing effect in the aftermath of civil war (Di Salvatore and Ruggeri 2017; Fortna 2007; Fortna and Howard 2008; Goldstein 2012; Hegre, Hultman, and Nygård 2019; Howard 2019). Nevertheless, the field still suffers from a lack of comparative analyses of actual military efforts to protect based on reliable and systematic event data, undermining the potential for generalization (Autesserre 2014; Clayton 2016). The UN does not provide systematic reporting on its military protection efforts. Another fundamental challenge is that Blue Helmets almost never use military force to protect, limiting the number of cases to learn from and

32

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

challenging efforts to pool cases for systematic reviews (United Nations Office of Internal Oversight Services 2014). Another strand in the literature has been concerned with bridging critical empirical gaps, developing typologies of different types of missions across time, and discussing their core characteristics and their most pressing tasks (Bellamy, Williams, and Griffin 2004, 2010; Durch 2006; Koops et al. 2015; Williams and Bellamy 2021). Although these do address how particular missions have attempted to protect civilians, they do not provide theory-driven cross-case comparisons of military protection operations. Furthermore, thematic literature on UN peace operations—mostly from US-based think tanks—seeks to influence ongoing policy debates, producing timely responses to particular policy developments (or lack thereof ), reviews of UN peacekeeping, or descriptions of particularly damaging protection failures (Cammaert and Blyth 2013; Center for Civilians in Conflict 2015; Friedrichs 2011; Gorur 2013; Willmot 2017). While providing a wealth of empirical information about contemporary protection challenges and what they may mean for ongoing UN reform, they provide few stepping-stones for theory development or generalizable explanations for why military peacekeepers succeed or fail to protect. Finally, the UN itself has produced a range of strategic documents from key actors with significant impact on how we think about the potential role of military peacekeepers in protecting civilians (Annan 1999b, 1999c, 2000; Boutros-Ghali 1992; Brahimi 2000; Buotros-Ghali 1995; High-Level Independent Panel on Peace Operations 2015; Holt, Taylor, and Kelly 2009). However, they do not present theoretical ideas of what might work when, nor do they base their analyses on systematic event data. I will return to some of the specific guidance developed by the UN Secretariat below, which is indeed important conceptually, but again lacks theoretical foundations pointing toward what may work in specific situations.

Four causal condition candidates Given that existing theory and literature is diverse and unorganized I have cast the net wide to find promising causal condition candidates that I find likely to influence UN troops’ ability to protect civilians from violence: (i) deterrent presence, (ii) willingness to accept risk, (iii) pre-emption, and (iv) matching the perpetrators of violence. In more colloquial terms: in order to protect effectively, deploy enough forces who are willing to take risks to save others from harm, and do so in time and in ways that are tailored to the particular

FOUR CAUSAL CONDITION CANDIDATES

33

threat civilians are facing. The first two conditions—deterrent presence and willingness to accept risk—capture preconditions for military protection operations, i.e., troop numbers and caveats of troop-contributing countries, while the latter two conditions—pre-emption and matching—capture variations in the modus operandi of UN troops in actual operations. There may be a range of other conditions that could influence the outcomes of operations. As long as we lack an encompassing theory on the utility of force to protect, however, these four conditions make out potentially relevant stepping-stones toward a future theory. One of these conditions is frequently mentioned as the key explanatory factor of UN missions’ effects on civilian targeting: deterrent presence. In fact, deterrent presence is perhaps the most dominant existing explanation in quantitative cross-case studies (Sundberg 2020, 214–15). Reform debates in the UN and review reports—such as HIPPO (The High-Level Independent Panel on Peace Operations)—keep returning to the quality of troops and their general unwillingness to take risk to protect when they seek to improve the effectiveness of UN missions’ protection efforts. Pre-emption is largely derived from the current UN guidance on how to protect. But it is also chosen due to its logical connection to the phenomenon force is meant to reduce. Deploy Blue Helmets too late and civilians will already be killed, harmed, or displaced. The matching theory is drawn directly from Alexander Beadle’s work, which has been published in various book chapters and reports, but which to date has received little attention. However, it is a rare effort to specifically explain the role and utility of force to protect against different physical threats. The first condition—deterrent presence—is quite simply linked to troop numbers’ potential deterrent effect on civilian security. One of the strongest findings from quantitative civil war studies implies that the presence of many uniformed UN peacekeepers decreases the intensity of armed conflict, including civilian targeting (see e.g. Hultman, Kathman, and Shannon 2019). The underlying hypotheses is that the presence of many UN troops will increase the cost of continued fighting and that it will facilitate credible commitment to peace negotiations and agreements (Hultman, Kathman, and Shannon 2019, 50). While these premises are logical, and the findings are important, there are some challenges in relying on correlations between the number of troops and the number of civilian casualties to explain how effective UN troops are in protecting civilians from perpetrators of violence. The deterrent effect of Blue Helmets is uneven at best. We know that many armed groups continue to attack civilians also in the presence of large peacekeeping deployments. Moreover, UN troops almost never strike back militarily, even when civilians

34

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

are under imminent physical threat, questioning their ability to deter committed perpetrators (United Nations Office of Internal Oversight Services 2014). Rupert Smith argues that deterrence will only work if there is a credible threat of coercion when it fails (Smith 2008). Surely, peacekeepers report exactions against civilians to the outside world every day, but many of the perpetrators I study portray few worries about their global reputation. Relying too much on the number of troops to improve the effectiveness of peacekeeping also risks missing crucial insights from military theory. It must matter what troops do, rather than being present in large numbers. Russia’s war in Ukraine also reminds us that numbers indeed do not explain all outcomes of operations. For these reasons I will investigate whether the deterrent effect of many uniformed personnel systematically explains variations in outcomes at the tactical and operational levels. I suspect that uniformed presence is many times not enough to protect effectively. The second causal condition candidate captures variations in UN troop contributing countries’ willingness to accept risk to protect civilians from violence. While this aspect is largely overlooked in existing academic literature, UN policies, external reviews, and reporting from the field indicate that the ability to protect civilians sometimes demands troops to accept significant risk to succeed (dos Santos Cruz, Phillips, and Cusimano 2017). We already know that most UN troops almost never use force for any purpose, indicating widespread risk aversion. Meanwhile, we also know that there is great variation amongst troop contributing countries. While many of the “traditional” contributors— including Bangladesh, India, and Pakistan—largely are seen as averse to risk in peacekeeping, some of the relative “newcomers”—including Mongolia, South Africa, and Rwanda—seem to represent a more forward-leaning approach, portraying a higher acceptance to risk when they deploy. I am interested in exploring variations in outcomes of operations where troop contributors’ willingness to accept risk differs. The underlying hypothesis is that troops coming from countries that are more willing to accept risk systematically perform better than those deployed from more risk-averse countries. The third causal condition candidate—pre-emption—is all about when UN troops intervene to protect. This condition is critical, since responding too late may jeopardize any chance to influence the outcome, leading to death, injury, trauma, and displacement among civilian populations. Commonly, UN peace operations are quite static, mostly responding to violence against civilians after the fact. Interestingly, the UN policy on protection of civilians in UN peace operations highlights the need for pre-emptive operations to protect effectively (United Nations 2019b). Furthermore, some strands in the literature

CAUSAL MECHANISMS AND HYPOTHESES

35

are interested in understanding and improving the UN’s early warning systems and intelligence structures, which can facilitate pre-emptive operations (Willmot 2017). Few, however, have systematically studied whether UN troops are more effective in pre-emptive or reactive modes of operation. My theorizing is that we should expect to see more successful outcomes when UN troops seek to pre-empt perpetrators of violence, based on the hypothesis that reactive responses will often make it challenging to influence the outcome, and therefore lead to more harm to civilians. The fourth condition—matching—rests on the only existing theory on how military force can be used with utility to protect civilians from violence (Beadle 2015). The basic assumption is that to protect effectively military forces must tailor their use of force to match the ways particular perpetrators use violence against civilians. I perform the first systematic test of Beadle’s matching theory, based on comparisons of cases across time and locations. I expect to see more successful outcomes when UN troops have matched the perpetrators of violence, tailoring their responses to fit various threats. Each of the four causal condition candidates will be presented in depth in the next section. Although they cover much of our (rather limited) knowledge about this phenomenon, I recognize that other conditions may form part of the explanations of outcomes. I will search for new or omitted explanations in the deeper qualitative case studies of protection operations in the DRC and South Sudan (see Chapters 6 and 7).

Causal mechanisms and hypotheses In this section, we will delve deeper into the four causal condition candidates for successful military protection operations: (i) deterrent presence, (ii) willingness to accept risk, (iii) pre-emption, and (iv) matching. To facilitate the mixed-methods approach that constitutes the research design, each condition will be operationalized and tailored to fit the analysis in each related chapter, while this section explains why and how the conditions may be relevant to answer the research questions in the first place.

Deterrent presence According to one important strand in the literature, deploying sufficiently high UN troop numbers have positive effects on civilian targeting. Large, uniformed deployments correlate with a reduction in violence. The underlying

36

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

hypothesis is that the presence of (enough) uniformed personnel will increase the cost of civilian targeting, thereby deterring perpetrators from attacking civilians. This idea of a protective—or deterrent—presence is also reflected in the formal UN policy on Protection of Civilians (POC) (see e.g. United Nations 2019b, para. 54). Based on rigorous quantitative methods, these studies have greatly contributed to our understanding of the macro effects of UN peacekeeping on civilian security. Moreover, this literature finds a range of other positive effects of large peacekeeping deployments. More troublingly, they also leave core questions unanswered. Specifically, Lisa Hultman, Håvard Hegre, Håvard Nygård, Jacob Kathman, Megan Shannon, and Reed Wood have compared conflict-related civilian deaths in Africa with deployment data of uniformed UN peacekeepers, in search of possible conflict-reducing effects (Hegre, Hultman, and Nygård 2010, 2015, 2015; Hultman 2016; Hultman, Kathman, and Shannon 2013b, 2013a, 2019; Kathman and Wood 2014). They find that UN peace operations score very well, reducing the intensity of conflict and thereby civilian targeting. Importantly, they also find that uniformed UN deployments reduce the duration of armed conflict, expand the peaceful period following conflict, and reduce the risk of armed conflicts spreading. Some of the findings have been confirmed by several other studies, even capturing a reduction in civilian targeting by rebel groups at the local level, adding nuance to the insight of the overall conflict-reducing effect of UN peacekeepers’ presence (Fjelde, Hultman, and Nilsson 2019; Phayal and Prins 2020). These findings are significant, telling us that “peacekeeping works” (Hegre, Hultman, and Nygård 2015). These effects, however, only seem to appear when the UN deploys large operations—meaning those missions that deploy thousands of uniformed personnel—rather than more limited observer missions and traditional peacekeeping operations. However, the authors do not pinpoint an exact threshold number of troops that triggers this deterrent effect. More problematic is that the literature says little about the mechanisms producing these positive protection effects. To be clear, Hultman, Kathman, and Shannon provide valuable insights into what UN peace operations do in general terms: separating, demobilizing, and reintegrating combatants, maintaining buffer zones, verifying compliance, and “policing behind the front lines” (Hultman, Kathman, and Shannon 2019, 58). However, the authors do not capture variations in these activities empirically and they therefore remain theoretical assumptions about what exactly makes UN troops effective. My critique is that they rely too heavily on only one variable—troop numbers— rather than empirically studying how UN troops respond to different threats

CAUSAL MECHANISMS AND HYPOTHESES

37

and to what effect on civilian security in particular situations. Recently, Ralph Sundberg also added valuable insights on peacekeeping’s effect on other forms of victimization during conflict from South Sudan, where he found that UNMISS had little—if any—deterrent effect on displacement numbers (Sundberg 2020). Although based on a single case study, it seems that the presence of peacekeepers not always equates to deterrence. In addition, there could be other confounding variables which may explain why violence decreases when the UN deploys (large) peace operations. One of the bedrock principles of UN peacekeeping demands the consent of the host nation and the main parties to the conflict before any mission is conceived. One would therefore expect a certain willingness to solve the conflict with other means, and that the level of violence will decrease as a result, regardless of how many UN troops are deployed. Only when we combine the knowledge from these meta-studies of troop numbers and one-sided violence, with studies of what UN troops do with the violence that remains after troops deploy, we may move closer to a more holistic understanding of what works. Along the lines of Hultman et al.—but more numerically specific—military studies have also been concerned with the effect of troop numbers on civilian security and stability. Several scholars have attempted to identify a particular threshold troop ratio—either relative to the opposing belligerent or to the population number—that correlates to positive outcomes of stabilization and counterinsurgency operations. Debates about ratios have been around at least since the early 1960s, when David Galula elaborated on the need to outspend the insurgent at a rate of 10 or 20:1 (Galula 1964). The most influential in recent times, however—and perhaps the most controversial—is James Quinlivan’s 1:50 troop-to-population ratio (Quinlivan 1995). It is influential because it seems to form the basis of some of the considerations in US counterinsurgency doctrine (Headquarters Department of the US Army 2006). It is also somewhat controversial because Quinlivan’s findings are based on only a handful of cases that obtained the 1:50 ratio. Steven Goode has challenged Quinlivan findings by proposing a ratio of 1:357, based on a much larger set of cases (Goode 2009, 46). Goode is, however, more skeptical about the explanatory power of such ratios, as “having enough forces does not equate to victory” (Goode 2009, 56). He urges caution in using ratios, as success also depends on several other factors. Riley Moore rejects this troop-to-population way of thinking altogether, as a “relic of conventional warfare that has been brought into COIN [counterinsurgency]” (Moore 2013, 857). He argues that the manner in which forces are used and their adaptability are the imperatives for success, not particular ratios

38

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

(Moore 2013, 870). Jeffrey Friedman also opposes the idea that a particular ratio can determine the outcome of operations (Friedman 2011). He agrees with Moore that the “manner and context in which [troops] are employed” is more important than the number of troops (Friedman 2011, 588). Recent comparative studies of counterinsurgency operations also echo this notion, highlighting different approaches, practices, and principles that make for successful operations (Paul, Clarke, and Grill 2010). This latter understanding of the utility of force(s) is also echoed in this book, although tailored to UN peace operations and the physical protection of civilians. Troop numbers are still present in UN reform debates, seeking to attract more (and better) troop contributors to take part in UN peace operations (Patrick 2015; White House 2015). Although most UN peace operations deploy relatively few troops to large areas with huge populations (e.g. Mali and the DRC)—a far cry from fulfilling the troop-to-population ratios presented as relevant in the literature—some UN missions have in fact met the predicted favorable ratios, including the UN missions in Abyei, Sierra Leone, and Liberia. At first glance, these latter three missions have also fared quite well in their protection efforts. In addition, peacekeeping should arguably be less demanding of troops than counterinsurgency operations, because UN troops operate as impartial interveners with the consent of the host state and main parties to the conflict, seldom expected to employ force beyond self-defense (United Nations 2008, para. 3.1). That said, available data does not allow for a fine-grained analysis of troop-to-population numbers. For example, although MONUSCO is mainly deployed in the eastern parts of the DRC, we do not hold reliable population numbers for the relevant provinces over time. Likewise, while Abyei is part of Sudan I have treated it as one entity since this is where United Nations Interim Security Force for Abyei (UNISFA) is deployed. Despite apparent difficulties of identifying specific troop-to-population ratios that would lead to more positive protection outcomes, it seems worthwhile investigating whether they can in fact be a precondition for successful operations and part of the explanation of what determines UN military troops’ ability to protect civilians from physical violence. However, troopto-population ratios at the country level—which I have used as a proxymeasurement for investigating possible deterrent effects of Blue Helmets—says nothing about the actual troop-to-population ratios at the local level, where violence takes place. I have not found a viable way of systematically narrowing down UN deployment numbers to specific regions or provinces. Likewise, there is just not enough granularity in population numbers across all countries at the local level. Therefore, my measurements remain a rough approximation

CAUSAL MECHANISMS AND HYPOTHESES

39

of possible deterrent effects of the presence of Blue Helmets. It is not, however, a strong enough test to refute the hypothesis on the importance (or irrelevance) of troop numbers. Therefore, I add another test in the qualitative case studies, where I investigate troop-to-perpetrator ratios and the possible effects on outcomes of operations. Keeping these limitations in mind, I investigate whether “deterrence by numbers” appears in the UNPOCO data. Are there more positive outcomes in cases with “good” troop-to-population ratios? In both literatures referred to above, there is scarce examination of potential causal relations at work leading to improved protection or stability with the help of sufficient troop numbers. Likewise, I have not succeeded in developing potential causal mechanisms linked to troop numbers. In Chapters 4 and 5, however, I investigate if mission size can indicate the rate of protection successes or failures. Still, we know that perpetrators of violence also target civilians even in the presence of large UN peace operations. As such, deterrence by numbers does not always work. In the case studies found in Chapters 6 and 7, I analyze troop numbers relative to the number of perpetrators, which seems to form part of the explanation for outcomes of those particular operations. More than relying on these proxy measurements of a potential deterrent effect, I believe it is necessary to explore other causal explanations to protection successes and failures, including analyzing how and to what effect UN troops use force toward different types of perpetrators.

Willingness to accept risk Soldiering is not risk-free. Intervening to protect civilians from perpetrators that deliberately target them does necessarily demand considerable risk taking. Many UN troop contributors are seldom willing to take such risks (Berdal and Ucko 2015, 11). An internal UN review from 2014 found that Blue Helmets mostly shy away from using force altogether, failing to respond when civilians are under imminent threat, or responding too late when civilians have already been attacked (United Nations Office of Internal Oversight Services 2014). HIPPO also addressed challenges related to the ineffectiveness of UN troop-contributing countries (High-Level Independent Panel on Peace Operations 2015, paras. 30, 108, 120). Nevertheless, HIPPO did not name or shame troop contributing countries—which are also UN member states— as this would be highly controversial. UN mission reviews also point to the challenge of the static and passive postures of some troop contingents, again avoiding spelling out exactly which ones (United Nations 2014e). The UN has

40

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

created the Office of Peacekeeping Strategic Partnerships to review military capabilities and particular troop contingents (United Nations 2015d). However, records and assessments are not made public, as public shaming at this level is understandably not considered constructive. We need to look beyond official UN processes to learn more about what aversion to risk may mean in practical terms. Brigadier General Jan-Gunnar Isberg—former Deputy Force Commander of MONUC and subsequently commander of the Ituri (and later Kivu) brigade in the eastern parts of the Democratic Republic of the Congo from 2003 to 2005—is one of few who have actually led Blue Helmets in combat operations at the brigade level. Together with Lotta Victor Tillberg he has written a book that explains in detail how different battalions approached the robust operations, which indeed included significant risks for everyone involved (Isberg and Tillberg 2011). Importantly, Isberg shows how willingness to accept risk varies—sometimes even within a few weeks—and within one contingent of troops. In one example, he writes about Uruguayan troops that had been under massive pressure to do something about the rebel groups in and around Bunia, receiving much criticism for being too passive when civilians were massacred in large numbers (Isberg and Tillberg 2011, 35–7). With the help of scenario-specific training and some level of “shaming,” the contingent switched stance and became more willing to take on risky assignments and to deliver on the mandate to protect civilians ((Isberg and Tillberg 2011, 35–7). At one point, the Uruguayans even moved beyond their rules of engagement—again according to Isberg—when they employed deadly sniper fire to stop an oncoming group of rebels (Isberg and Tillberg 2011, 80). Shortly after—in Bukavu, further south—Uruguayan troops largely refused to contribute to the defense of the town (Isberg and Tillberg 2011, 106). This shows how fluid willingness to accept risk can be, demanding continuous management to make sure deployed troops are indeed up to implementing mandated tasks. We know that some troop contributors are willing to take more risks than others, such as the Chadian contingent deployed to Mali and the Mongolian contingent deployed to South Sudan (Karlsrud 2015, 47; Mold 2017). There are great variations in how different troop contributors relate to risk and the use of force to protect in UN peace operations (Providing for Peacekeeping 2020). Since the most important task for UN troops is to protect civilians from violence, and successes seem few and far between, it is necessary to move beyond national sensitivities and investigate aspects that may be controversial. Due to potential controversy, it is better if these types of analyses fall to research performed outside the UN system. However, there is

CAUSAL MECHANISMS AND HYPOTHESES

41

still not much literature and theory to draw on to investigate whether the willingness of UN troops to accept risk systematically leads to better protection outcomes. It is also challenging to credibly designate different levels of willingness to different troop contributors. Yet, I try to systematically analyze whether national caveats—restrictions closely linked to the will and ability to use force to protect—can be part of what explains variations in protection outcomes on the ground. Even the effect of national caveats on operational effectiveness largely evade scholarly studies—with some exceptions—since many national restrictions remain classified and are not easily accessible (Frost-Nielsen 2017; Kingsley 2018). I therefore use official caveats and statements from debates in the General Assembly as a proxy measurement for willingness to accept risk (Bellamy and Williams 2013; Providing for Peacekeeping 2020). Lacking more detailed and systematic knowledge about contingents’ hidden restrictions, I balance this information with insights on how different troop contributors are perceived by others. Practitioners hold a great deal of insight about the willingness and effectiveness of different UN troop contributors, while this knowledge often remains tacit. Therefore, I have tested my scoring of troop contributors’ willingness to accept risks on an expert group, consisting of highly experienced UN practitioners. I am confident that my scoring reflects overall trends, but it remains a crude measurement until we have access to more systematic and detailed reporting on how UN troops perform. If willingness to accept risk influences protection outcomes, we should expect to see that protection operations including troops from risk-accepting countries are systematically more successful than those operations including troops from countries that are less willing to put their troops in harm’s way. I should also mention that the aim is to measure willingness to accept risk rather than risk aversion, meaning that those with low scores are not necessarily unwilling, only not among the most willing to accept risk. In fact, those troop contributors that are categorized as risk averse have still performed many of the operations captured in my dataset, as they provide most troops. We already know that they both fail and succeed. My task is rather to explore whether this condition appears systematically across cases.

Pre-emption Timing is everything. When civilians are under imminent threat of violence, pre-emptive protection operations—launched before perpetrators attack— may become necessary. Ideally, pre-emptive operations will deny perpetrators

42

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

the opportunity to attack civilians altogether or at least significantly reduce their ability or willingness to inflict harm on civilians. This condition is dependent on a well-functioning intelligence capability for timely detection of threats, an effective command-and-control system to direct capable forces, as well as the ability to amass sufficient forces, and move those forces in a timely manner to where they are most needed (Duursma and Karlsrud 2019). None of these traits is common to most UN peace operations, which undermines their effectiveness. Over the past few years, there has been a series of demonstrations criticizing UN missions for failing to respond in time to exactions against civilians (Al Jazeera 2019; France 24 2016; The Guardian 2021). Instead, the UN’s default approach to deployment in peace operations is to assign different geographical areas to different troop-contributing countries and then to spread camps and troops evenly—and often thinly—across the theatre of operations, regardless of the particularities of the situation on the ground. This deployment pattern strongly limits opportunities to perform dynamic, pre-emptive operations, even if the will to do so had been there. Again, the idea is that a uniformed and armed UN presence in many locations can deter wrongdoing against civilians. This is fundamentally a defensive, static, and reactive approach to operations. It has more to do with adherence to the bedrock principles of peacekeeping—which undeniably were tailored to another era—than to military reasoning based on systematic analysis of the threats civilians are facing in contemporary armed conflict (de Coning, Karlsrud, and Aoi 2017). Deployment patterns are slowly beginning to change, seeing the introduction of temporary operating bases—better placed to react quickly—in some of the most challenging conflicts, such as in some parts of the eastern parts of the DRC and the Central African Republic. In addition, we also know that the UN occasionally amasses forces to go on the offensive to pre-empt perpetrators of violence, such as during the fall of 2013, when the UN Force Intervention Brigade attacked the M23 alongside national Congolese armed forces. The empirical mapping performed as part of this book has also revealed several other cases that vary in each type of operation—pre-emptive and reactive—as well as the outcomes. While pre-emption is overlooked in the academic literature, UN policy and guidelines for military components indicate how UN peacekeepers should use force to protect most effectively (United Nations 2015c, 2017a, 2019b). First, the policy contains a three-tiered approach, categorizing different measures to protect: protection through dialogue and engagement (Tier I); provision of physical protection (Tier II); and establishment of a protective environment

CAUSAL MECHANISMS AND HYPOTHESES

43

(Tier III) (United Nations 2019b, sec. D.3). The role of military peacekeepers is most prominent under Tier II. Keep in mind that the overarching definition of POC in UN peace operations concerns physical protection, and while all three tiers are meant to contribute toward improved physical protection for civilians under threat, although only Tier II states so specifically. The 2019 policy provides further nuance, dividing each tier into four response phases: (i) prevention, (ii) pre-emption, (iii) response, and (iv) consolidation. It does not follow this structure systematically, though, and the conceptualization creates quite a bit of confusion. While the 2015 version of the POC policy only divides Tier II into four phases (prevention, pre-emption, response, consolidation), the 2019 version of the policy divides all three tiers into the same four phases. However, it does not follow this structure when it describes what is supposed to happen within each tier/phase, which is quite confusing. Moreover, the updated conceptualization creates some bizarre categories. What is preventive or pre-emptive dialogue (Tier I)? Or pre-emptive establishment of a protective environment (Tier III)? Regardless, the first three phases still dovetail with three of the four causal condition candidates selected for further scrutiny in this book: deterrent presence (to prevent attacks on civilians from happening); pre-emption (to actively hinder a specific threat before it occurs); and matching (responding in ways that match the perpetrators of violence). However, I add theoretical depth and nuance from the existing literature to explain how each mechanism is meant to work in practice. Phase four—consolidation—is only relevant in the aftermath of violence and is therefore not of direct relevance to the research questions posed here. I have already introduced ideas underpinning the potential deterrent—or preventive—effect of having a sufficient uniformed presence, like the ideas underpinning Phase 1. The next section (matching) will cover UN military responses to different types of threats, similar to the ideas presented in Phase 3 of the UN’s POC policy. The 2019 version of the UN’s POC policy describes the logic of pre-emption in the following way: When a concrete threat of an attack against civilians is identified, proactive and timely measures must be taken to eliminate or mitigate the threat before violence occurs, including through credible deterrent actions such as reinforced presence and patrolling, show of force, securing key sites, interpositioning, psychological operations and proactive military and police operations which may extend to pre-empting and neutralising the source of the threat in accordance with the mandate, ROE and DUF. Contingency plans shall be developed in advance to enable rapid response. (United Nations 2019b, para. 56)

44

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

While this description includes many different activities, it highlights “proactive and timely measures,” which is also how I understand the basic logic of pre-emption in military terms: to do something in time to stop imminent threats from materializing. These aspects are now regularly reflected in the United Nations Secretary-General’s (UNSG) reporting to the UNSC. For example, in a 2017 report from the UN mission in the DRC, the United Nations Secretary-General (UNSG) states that: “[…] All MONUSCO troops must be ready and willing to use force against armed groups that pose a threat to the civilian population, and to do so pre-emptively” (United Nations 2017c, para. 59). By referring to the necessity of pro-active measures to protect, the UN further expands the role of military force in peace operations, which originally followed a basic principle of non-use of force, except in self-defense. Accordingly, I seek to investigate variations in outcomes of both reactive and pre-emptive UN protection operations, expecting more successes resulting from pre-emptive operations. In addition, it remains critical to explore under what combinations of conditions successes are more frequent. Consequently, I code all cases as either “pre-emptive” or “reactive” and examine them as potential pieces of the puzzle, in combination with the other promising causal condition candidates.

Matching Modus operandi matters. Military theory offers the most promising insights into how UN troops can better protect civilians under threat, besides being present. Most notably, General (retired) Rupert Smith has written an account of how military force can be employed more wisely in intra-state conflicts “amongst the people” to increase its utility (Smith 2008). It is one of the few works that combine deliberations on the utility of force with concern for civilian life during contemporary armed conflict. Smith criticizes today’s military interventions for their “deep and abiding confusion between deploying a force and employing force” (Smith 2008, 6). To remedy this confusion, Smith demonstrates how to increase the utility of military force by better understanding its four main functions and the contexts in which each function is most relevant: (i) Amelioration: Troops assist in delivering humanitarian aid, put up refugee camps, observe ceasefires etc., and military force is only employed in self-defense (Smith 2008, 323). Tasks such as operating observation posts and checkpoints, patrolling, outreach, and

CAUSAL MECHANISMS AND HYPOTHESES

45

engagement would be included in this function of force. Traditional UN peacekeeping falls into this category. Examples are the UN Disengagement Observer Force, the UN Interim Force in Lebanon (UNIFIL), the UN Military Observer Group in India and Pakistan, and the UN Peacekeeping Force in Cyprus. (ii) Containment: Military forces prevent something, such as arms, planes, or troops from spreading or passing through a barrier (Smith 2008, 324). This can be done through maintaining arms embargos and no-fly zones. This category would also include positioning UN forces between armed opponents, or between perpetrators and a civilian population, as well as the establishment of demilitarized buffer zones and safe areas. Rules of engagement usually limit and control the use of force. One example is the United Nations Protection Force in Croatia, Bosnia, and Herzegovina. (iii) Deterrence/coercion: This function involves a “wider use of force,” according to Smith (2008). Military forces are used to pose or carry out a threat to change or form the opposition’s intentions. When force is employed, it is used to coerce. UN military forces sometimes undertake cordon and search operations and perform joint military operations with host-state security forces, targeting opposing forces, both of which actions would fall into this category. Mostly, UN peace operations rely heavily on being present in many locations, implicitly expecting a deterrent effect from the presence itself. According to Smith, however, deterrence does not work unless there is a credible threat of coercion when it fails. Examples from UN missions are MONUSCO, the United Nations Multidimensional Integrated Stabilization Mission in Mali (MINUSMA), and the United Nations Multidimensional Integrated Stabilization Mission in the Central African Republic (MINUSCA). (iv) Destruction: This implies using military force “to attack the opposing force in order to destroy its ability to prevent the achievement of the political purpose,” according to Smith (2008, 324–5). This traditional understanding of military force is a rare function of force in UN peace operations. However, the FIB, which is an integrated part of the UN mission in the DRC, has a mandate to neutralize armed groups through targeted offensive operations. As I will demonstrate in the case study from the DRC, destruction was part of the UN operations against the insurgent group M23 in 2013. In addition, the UN Operation in the Congo (ONUC, 1960–64) ended up destroying the

46

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

ability of the Katangan secessionists to break away from the Congo, including through the use of fighter jets (O’Brien 1962). These four functions of force would seemingly be instructive principles for devising a military strategy to protect civilians from violence. However, Smith seems to underplay the agency of the opponent—the perpetrators of violence—by mostly focusing on what his own forces can and cannot do. Alexander William Beadle has therefore used Smith’s four functions as a point of departure for developing his own theory on the utility of force to protect civilians, which brings in the agency of the perpetrators (Beadle 2011, 2014, 2015). Beadle argues that knowing why and how armed groups use violence against civilians is a critical first step to determine the appropriate and proportionate military countermeasures—or functions of force—to protect them. Mirroring Smith’s four functions of force, he finds that perpetrators can employ four types of violence against civilians (Beadle 2014, 10): (i) Impairment: Fostering insecurity by threatening civilian life without physically targeting civilians. Perpetrators may impair civilian security by virtue of their threatening presence or by using civilians as human shields. One example is the M23 (DRC). (ii) Incitement: Using violence against civilians to spread fear and insecurity, including through improvised explosive devices (IEDs) and suicide bombers. Perpetrators are not seeking to kill as many civilians as possible, but rather to undermine the government’s ability and/or credibility to protect its own citizens. Examples are the Taliban (before their takeover of Afghanistan in 2021), Boko Haram (Nigeria), al Shabaab (Somalia), and Tuareg rebels (Mali). (iii) Deterrence/coercion: Using violence to change civilian behavior, often to deter collaboration with the opposition or to coerce populations into compliance. Examples: M23 (DRC), Forces démocratiques de libération du Rwanda (FDLR, DRC), Sudan People’s Liberation Army (SPLA, South Sudan), Sudan People’s Liberation Army-In Opposition (SPLA-IO, South Sudan). (iv) Destruction: Using violence to destroy civilians (or civilian installations) directly, such as during genocide and mass killings. Examples are the Interahamwe (Rwanda) and Bosnian Serb forces in Srebrenica (Bosnia and Herzegovina). Beadle’s core argument is that, to find utility of force to protect, the function of force employed by the protector must match the type of violence applied

CAUSAL MECHANISMS AND HYPOTHESES

47

by the perpetrator (Beadle 2011, 35–6). The core phrase—“matching”—needs to be unpacked. He explains that, if a perpetrator aims to “destroy” an ethnic group, the protector will not find utility of force by “ameliorating” the situation by merely supporting the delivery of humanitarian aid. In this situation, greater utility of force is found in matching the perpetrator, by destroying his ability to conduct mass killings. Conversely, if a perpetrator uses “incitement” or “impairment” against civilians to undermine the legitimacy of a government, using coercive or destructive force against them is likely to lead to stronger incentives to scale up attacks against civilians. In addition, if the most violent functions of force are applied, they risk causing more harm during operations than otherwise would occur in these less violent situations, challenging well-established rules of proportionality in International Humanitarian Law. Instead, “containment” and “amelioration” are better suited to protect civilians in such scenarios. Consequently, to maximize the utility of force, protectors must match the perpetrator’s violence against civilians. Only then will protectors be able to influence the will and ability of the perpetrator to attack civilians. Table 2.1 illustrates how military forces ideally can match these four ways perpetrators use violence against civilians, to protect civilians more effectively.

Table 2.1 Perpetrator’s use of violence vs. protector’s use of military forcea Perpetrator violence against civilians

Protector use of military force to protect

Impairment (e.g. presence of armed actors and constant threat of armed clashes)

Amelioration (e.g. presence of military observers reporting human rights violations)

Incitement (e.g. indiscriminate attacks by insurgents in government-held areas)

Containment (e.g. creation of weapon-free zones, counter-IED operations)

Deterrence or coercion of civilians (e.g. threats or retaliatory attacks against civilians associated with the enemy, or demonstrative violence to make people flee)

Deterrence or coercion of the perpetrators (e.g. threats or actual use of force to alter the willingness to target civilians through robust show of force or punishing attacks)

Destruction of civilian life or property (e.g. massacres or scorched earth policies)

Destruction of perpetrator capabilities (e.g. neutralization of rebel armed forces)

a.

This table was published in a book chapter authored by Alexander William Beadle and me, “The Utility of Force to Protect in UN Peace Operations,” in The Use of Force in UN Peacekeeping, edited by Peter Nadin (Nadin 2018).

48

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

In a first empirical test of Beadle’s theory, I map different types of violence against civilians (impair, incite, deter/coerce, and destroy) as well as the function of force used to protect (ameliorate, contain, deter/coerce, and destroy). This enables an analysis of the degree to which UN forces have been able to match perpetrators of violence in each case, and whether that appears to influence outcomes across operations. If this theory holds explanatory power, we should expect to see more successful outcomes whenever UN troops have matched the perpetrators of violence. Some words of caution are still needed. First, I am only evaluating specific situations on the ground, not the overall logic of how each perpetrator applies violence as part of their strategy. As such, if some armed groups that regularly “destroy” civilian lives and infrastructure are involved in a less violent situation, where they “only” impair the security situation for civilians, their actions will be coded as “impairment,” although they might act much more violent in other situations.

Further deliberations on the matching theory How does violence against civilians differ between cases? In this section, we move from theories on potential causal conditions to theories related to the outcomes of operations. To facilitate the scoring of the outcomes of operations in later chapters, I rely on counterfactual reasoning (see Chapter 3). A part of this analysis seeks to estimate the expected outcome if the UN had not intervened militarily in particular cases. To facilitate this analytical step, I am relying on a perpetrator typology—also developed by Alexander Beadle— as well as case-specific knowledge. The typology is also helpful in developing descriptive statistics of the most common types of threats civilians and UN troops face and to theorize about how military force can be used with utility to protect civilians from different types of perpetrators (Beadle 2011, 35–6). Beadle has disaggregated the scope of possible threats to civilians into a typology consisting of seven scenarios. He first developed this typology to help military planners differentiate between different types of threats to civilians, facilitating systematic input to military planning procedures. It is structured according to core characteristics representing “the most commonly discussed aspects of perpetrators of violence against civilians”: (i) the rationale for attacking civilians; (ii) the type of actor attacking civilians; (iii) strategies and tactics used to attack civilians; (iv) the capabilities perpetrators require to attack; and (v) the expected outcome if they succeed (Beadle 2014, 12). The criteria selection is not exhaustive, as many other characteristics may also describe

FURTHER DELIBERATIONS ON THE MATCHING THEORY

49

differences in types of perpetrators. They are listed in Table 2.2—from the least violent to the most violent—according to the main rationale for attacking civilians. Based on observations from UN protection operations in Africa in the period 1999–2017, I have developed an eighth scenario: mob violence.¹ The typology satisfies many of the criteria for good typologies, as it: (i) clearly defines the overarching concept of the typology, (ii) specifies whether the typology is descriptive or explanatory, (iii) describes in detail how the types are (inductively and/or deductively) constructed, (iv) proposes an intuitive model or matrix of the typology; and (v) considers a simpler solution with mutually exclusive types (Ravndal 2015, 65). Beadle clearly states the overarching concept of the typology, i.e.: “[…] to capture the range of situations in which military forces may be deployed and expected to protect civilians. The scenarios are meant to help military staffs identify the particular nature of threat facing civilians and the military responses that are most likely to work” (Beadle 2014, 3). He also thoroughly describes how the types are constructed by immersing himself in the literature on violence against civilians and inductively building parameters for the scenarios (Beadle 2014, 12–21). Further, the typology is visualized through a simplified matrix (see Table 2.2). However, Beadle does not specify whether the typology is descriptive or explanatory, and, in some sense, it is both. He aims both to describe empirical differences in types of threats and to explain perpetrators’ motivations for using violence. This probably has to do with the purpose of the typology in the first place, which was to assist military planners with both a planning tool and theories explaining why and how such phenomena occur. Finally, although he aims to produce threat categories that are mutually exclusive, he acknowledges that many times the threats civilians are facing defy the sharp delineation between categories (Beadle 2014, 22–3). The typology remains useful for the analysis in this book for three reasons. First, it seeks to encompass the full spectrum of threats civilians have been faced with in the past. This provides a coherent analytical framework to support a systematic recording of the types and frequency of threats civilians and UN troops have faced over time (see Chapter 3). As far as I am aware, no other typology seeks to capture the entire spectrum of threats to civilians. Second, it seeks to provide mutually exclusive threat categories, which are useful

¹ The scenario descriptions were developed by and first published in Protection of Civilians: Military Planning Scenarios and Implications by Alexander Beadle (Beadle 2014). The additional “mob violence” scenario was developed by me and first described in a guide tailored to UN peace operations (Kjeksrud, Beadle, and Lindqvist 2016).

Table 2.2 Eight generic threat-scenarios Scenario

Actor type

Rationale

Strategies and tactics

Necessary capabilities

Expected outcome

Mob violence Liberia (’04, ’05, ’09, ’11, ’15) Ivory Coast (’04) Sierra Leone (’00, ’02) DRC (’13, ’15) Post-conflict revenge Kosovo (post-’99) Iraq (post-’03)

Individuals or mobs demonstrating discord

To exploit mob dynamics for personal gain, revenge, or political influence

Non- or semi-organized criminal acts, such as murder, looting, arson, rape

Freedom of movement

Few killed, but possibly extensive material damage to property and general perception of insecurity

Individuals or mobs

To avenge past crimes on a personal basis

Tit-for-tat score-settling through criminal acts of violence, such as murder, arson, kidnapping, looting

Freedom of movement for individuals and small groups to access victims

Insurgency Mali (’13–’15) DRC (’12–’13) S. Sudan (’12–’13)

Rebel groups (classic insurgents with political or ideological objectives) Rebel groups (predatory behavior)

Selective and indiscriminate violence, through threats, targeted killings, bombings, retribution, depending on their level of control Coerce civilians into compliance through plunder, taxation, forced recruitment, opportunistic rape, brutality, especially against “easy targets”

Freedom of movement to pick time and place of attack, access to indiscriminate and explosive weapons

Predatory violence DRC (’99–’15)

To control populations upon which they depend and undermine trust in their rivals To survive or make a profit by exploiting civilians

Few killed (dozens, hundreds), but groups associated with previous perpetrators may flee, following relatively little violence Fewer killed and injured than in many other scenarios, most due to indiscriminate weapons; gradual displacement from areas of heavy fighting Temporary, but large-scale, displacement in affected areas, disproportionate to the number of people actually attacked; many abductions, especially of young adolescents

Freedom of movement to pick time and place of attack, operational secrecy, often central command

Communal conflict Mali (the Tuareg vs. Fulani) South Sudan (the Lou Nuer vs. Murle) Abyei (Misseriya vs. Ngok Dinka) DRC (Hema vs. Lendu)

Whole tribal, ethnic or sectarian communities (possibly with outside support)

To avenge a previous attack and to deter further retribution in order to protect their own community

Attempts to coerce other community into submission through massacres, abductions, raids, destruction of homes and means of survival, often seeking to maximize violence

Freedom of movement to reach other community, access to deadlier weapons is associated with higher number of deaths

Government repression Ivory Coast (’10–’11) Syria (’12–present)

Authoritarian regimes, or de facto authorities in an area

To control restless populations, on basis of real or perceived affiliation with opposition

Repress population, through selective and indiscriminate violence, threats, detention, rape as terror, destruction, occasional massacres

Command and control for governments, freedom of movement for regular forces, heavy weapons, support of special/ irregular units

Ethnic cleansing Bosnia (’92–’95) Central African Republic (’14)

States, or the militarily superior actor

To expel a certain group from a specific territory

Force targeted group to leave through threats, highly visible killings, brutality, mass rape, destruction of property

Command and control, freedom of movement for irregular units, regular units for military control

Genocide Rwanda (’94) Srebrenica in Bosnia (’95)

States, or the militarily superior actor

To exterminate a certain group

Destroy existence of a group through several, simultaneous mass killings, deportation, camps, systematic rape to prevent reproduction

Command and control, freedom of movement for special/ irregular units, sufficient small arms

Relatively high number of people killed and abducted on both sides, especially women and children; livelihoods stolen or destroyed; temporary displacement in homogeneous areas, gradual withdrawal to “their own” in mixed areas Mostly combatant deaths, gradual increase in civilian deaths due to heavy weapons and in accordance with intensity of fighting, large-scale displacement, widespread destruction of population centers Only a few percent killed, but the vast majority of the targeted population expelled (~90%); destruction of victim homes and cultural buildings Majority of members of the targeted group killed (50+ percent), in relatively short time

52

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

when describing systematic variations in this phenomenon. Although situations on the ground are sometimes difficult to define, the typology brings rigor to the process of determining what type of threat civilians were facing in each case. Third, the characteristics enable deeper comparisons between threats that seem similar in motivation, useful for the qualitative case studies. All cases identified in the UNPOCO dataset are coded according to these eight scenarios. However, only five scenarios were identified in the period between 1999 and 2017 in African conflicts where UN troops deployed to protect. These five are summarized below, appearing in the order of how common they are based on the data taken from UNPOCO, starting with the most common: predatory violence.² The summaries provide insights into how these categories of violence against civilians are developed. However, for a full description of each scenario and a better understanding of the research underpinning them, readers should refer to Beadle’s original scenario report (Beadle 2014). Predatory violence occurs when armed groups attack the local population to ensure their own survival or their own profit. Perpetrators are typically rogue security forces or insurgents who have failed to achieve their original political objectives. Rogue armed forces will typically lack logistical support and finances to feed and pay their own troops. Failed insurgents can continue their existence by preying on civilian populations, refusing to demobilize or disarm. Often physically removed from the geographic areas where they may gain support, they have little strategic incentive to limit predatory behavior (pillage, forced recruitment, illegal taxation). Both will typically lack popular support and alienate the population in the areas in which they operate. All civilians are potential victims of predatory violence. The use of excessive violence beyond what is required to kill, harm, or instill fear in the population is common. Severe torture and mutilation may be used to deter resistance in the future. Attacks are usually launched based on opportunity, preferring “easy,” undefended targets—with high deterrence value—especially women and children. Relatively few people may be killed, but the number of abductees and displaced will often be high, due to the brutality and unpredictability of attacks. There are often no geographical patterns of where attacks take place, other than in areas where rewards are high, and resistance is low. If predatory actors are also involved in communal conflict, this scenario can become very violent.

² The summaries are drawn from Protecting Civilians from Violence: A threat-based Approach to Protection of Civilians in UN Peace Operations (Kjeksrud, Beadle, and Lindqvist 2016).

FURTHER DELIBERATIONS ON THE MATCHING THEORY

53

Communal conflict is potentially a very violent situation and occurs when whole communities engage in continuous cycles of violence, driven by a combination of revenge and self-protection. Precisely because both sides are organized as communities, rather than as organized armed actors, they are unlikely to possess the means to settle conflicts permanently. There will usually be a relatively even balance of power between the communities in terms of armed strength. However, they cannot afford not to retaliate, as this will invite further attacks upon themselves. Each community’s own acts of violence will be perceived and portrayed as necessary for self-defense. Conflict may persist for years or even decades, with periodic escalations in violence, beyond “normal” patterns of cattle raids, skirmishes, etc. This can be linked to rising tensions over access to land, water, cattle, or other means of survival. Civilians are primary targets for both sides, as the roles of perpetrator and victim shift with each cycle. Civilians are targeted based on their communal identity or their perceived communal identity. Women and children are often singled out, since they often are easy targets, and because the perpetrators are not primarily seeking to destroy the adversary’s military capability (since this is seldom possible), but rather to deter him from future attacks. Open declarations of intent to exterminate or expel the other community as the only viable solution to defend themselves may occur. The outcome may be a high number of casualties relative to the community’s total population, and rapid displacement of entire communities that flee impending attacks. The tactics used are often destructive, despite their limited means (e.g. killing rather than capturing people, destroying homes and means of survival, indiscriminately targeting the most populated locations). If a communal actor gains the upper military hand, this scenario can escalate into ethnic cleansing or even (acts of ) genocide, as the objective may be to expel or exterminate the opposing community. But if a balance of power is maintained, they lack the means to do so. This has particular importance during disarmament campaigns. In insurgencies, civilians are targeted as a tactic to control the population itself or to undermine the control exercised by the government or other armed groups. The perpetrators are typically armed insurgent groups fighting for political power. Government forces or rival groups are usually the primary targets, but insurgents still employ a combination of selective violence (e.g. assassinations) to prevent the population from collaborating with the enemy and indiscriminate attacks (e.g. IEDs) against civilians and rivals alike. Attacks may also be intended to prompt an overreaction from government forces. Often, perpetrators express their justification for targeting individuals or groups of people who are legally defined as civilians (e.g. government

54

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

officials, governors, police forces who do not partake in military operations). Physical violence is only likely to be the main concern for civilians in contested areas, whilst most have other grievances (e.g. unemployment, corruption). The number of people killed will be relatively low compared to “reversed” scenarios, i.e., where governments, not rebel groups, are the main perpetrators. When people flee, they will usually flee the presence of fighting rather than the presence of insurgents per se, depending on their affiliation with either of the parties. Military operations against insurgent groups may increase the threat they pose to civilians, due to risks of retaliation for perceived collaboration. The scenario may see peaks in acts of brutality in certain areas or at certain times. In addition, case-specific variations, especially where the conflict runs along ethnic divisions, may lead to more bloody insurgencies, such as in Syria (Bostad 2018). However, if the perpetrator retains a political objective, it is likely that their use of violence against civilians will be relatively limited. Mob violence occurs when rioting individuals and/or loosely organized groups of civilians (in mobs) destroy property and harm civilians. Personal gain, revenge, and political discord may all form part of the rationale in this primarily opportunistic form of violence against civilians. Small-scale skirmishes prior to elections and/or major political events can trigger mob violence. Planned demonstrations may also get out of hand. Civilians can be targeted randomly or intentionally, but relatively few people are likely to be killed. The presence of violent mobs will often lead to a general perception of instability, although national and international security forces are often able to defuse these situations quite quickly. Individuals or loosely organized groups will have limited means to threaten military UN peacekeepers. Freedom of movement is essential to perpetrators, as is the ability to come together. This scenario rarely leads to more strategic violence against civilians but can be manipulated by political actors and escalate into more deadly situations. If violence occurs along ethnical divisions, this scenario can evolve into larger scale communal conflict. Government repression occurs when rising political pressure on the government in power or a de facto authority leads to violent repression of threats against its own survival. Civilians are primarily targeted according to presumed or real affiliation with political opposition, not based on ethnic or sectarian identity, although communal identity may increasingly be used as a proxy for targeting political opposition. Violence will be most severe where opposition is perceived to be strongest, e.g. where known opposition members hide and/or operate. The principal threat to civilians comes from the

FURTHER DELIBERATIONS ON THE MATCHING THEORY

55

indiscriminate tactics and means used to suppress both armed and unarmed resistance (e.g. conventional weapons against civilian areas). The number of people killed or displaced will vary, according to the local level of fighting. Combatants are equally, or, at times, even more at risk of being killed as are civilians. This scenario may lead to ethnic cleansing or even acts of genocide if the government’s survival continues to be threatened. Although I am not seeking to analyze violence against civilians committed by governments, I have included the description of the scenario here, since it appeared as part of the mapping of cases.

The utility of force to protect from different types of threats The utility of force to protect civilians from these different types of threats will vary substantially. In communal conflicts, where communities attack each other motivated by both revenge and self-defense, military force can deter attacks from either side to prolong the “pause” in between revenge/self-defense attacks. Positioning UN forces between warring communities is one way to achieve this. If UN forces can decrease the frequency of attacks, more civilians can be protected in the interim until a more peaceful situation is established by other means. Furthermore, military force may be necessary to deter or coerce occasional large-scale attacks, which often turn deadly for civilians in this scenario (Small Arms Survey 2012; UNMISS Human Rights Division 2012). During such attacks, both sides are likely to attack “soft targets” in the opposing community, such as children or the elderly. Again, the purpose is to deter future attacks and to avenge previous attacks. Perpetrators often seek to inflict maximum damage, degrading the opposing community’s ability to fight back, aiming to settle scores for the last time. Yet they often lack the military capabilities to do so, a typical characteristic of this scenario, which facilitates another round of revenge. In addition, actively engaging one community militarily to defend the other may lead to a loss of perceived impartiality for UN forces. Large-scale attacks may therefore be stopped by force, but potentially at a significant cost for UN forces. Predatory actors, in comparison, target civilians to survive (e.g. to attain food, water, or livestock) or to obtain other benefits, such as access to weapons, financial means, and different sorts of loot. These types of perpetrators are also commonly responsible for forced recruitment of children and youth to function as fighters, temporary porters, and/or sex slaves (Gates and Reich 2010).

56

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

Professional military forces can often deter this largely opportunistic type of perpetrator from attacking civilians, if they are present at the right time, as there is no need for the predatory actor to win military battles, and because their interest in a particular target is incidental. Instead of seeking military confrontation, predatory perpetrators can shift to the next undefended population center and still achieve their goal. However, as they are likely to continue attacking civilians to survive, the only way to provide longer-term protection will be to decrease their ability to target civilians. This implies using force offensively, either to defeat the perpetrator or to significantly degrade his military capabilities. Other means, including inducing defections from predatory armed groups, have also had effect in the past (Gates 2002). Insurgent armed groups target civilians to undermine ruling authorities in attempts to achieve their political purpose. Thus, insurgents often limit their violence against civilians, although peaks in violence may occur. Protecting civilians from insurgent violence is challenging for UN forces, primarily because this type of conflict is highly political. As UN missions most often support the incumbent authorities, any military effort by UN forces against insurgents will be highly likely to be perceived as partial, and therefore put Blue Helmets at risk of becoming targets themselves. The conflict in Mali underlines this point, where 95 UN peacekeepers were killed between April 2013 and March 2018, almost half the total number of such fatalities across all peacekeeping missions in the same period (204) (United Nations 2018). Therefore, the role of military force in protecting civilians from physical violence in insurgencies will be limited. In fact, physical safety is unlikely to be the primary concern for most civilians since the threat of violence is generally quite low and selective. If military means are used to protect, UN military forces must be able to hold and control areas that are wrested from insurgent control until local security forces can establish a permanent presence. If control is not feasible, clearing insurgent strongholds will only increase the threat to civilians, as retaliation against perceived collaborators is likely to occur if insurgents recapture the area. UN military efforts to protect in this scenario should be distinguished from counterinsurgency efforts, usually undertaken by other actors (High-Level Independent Panel on Peace Operations 2015; Razza 2018). Dealing with mob violence is primarily a policing task. Indeed, if an initial demonstration is lawful, local and UN police forces will facilitate such protests. However, if public order management fails, military UN forces may be the only units available to regain control and order. Military forces are not meant for law enforcement tasks, but this scenario can challenge the existing

FURTHER DELIBERATIONS ON THE MATCHING THEORY

57

capabilities of the host-nation’s security forces and UN police. When all other responses have failed, courses of action could include deterring, controlling, and dispersing crowds, preferably through non-lethal means and with gradual use of force. While this scenario may become the sole responsibility of military UN forces, close cooperation with police forces remains essential. Key infrastructure, locations, and buildings must be defended, as there is potential for significant material damage to civilian property. Employing a gradual response when countering mob violence is of utmost importance, as a military response with the application of basic military principles may escalate a situation and spark increased violence. Finally, robust deployment of military forces, primarily through a show of force, may be required to contain and control violent crowds (which themselves often consist of civilians) when civilians are threatened. Finally, met with government repression, any UN military operation to protect is likely to be controversial, as it involves violations committed by the host-nation government, or the de facto authorities. This scenario might require responses that fall outside the UN’s military capabilities. However, UN troops may still be deployed when such a situation develops. In that case, a UN military presence can limit the government’s ability or willingness to use heavy weapons against its own population, which is often a primary cause of civilian casualties in this scenario. Areas of strong civilian opposition support and elites should be defended and the government’s ability to crush its own population must be reduced. Any effort that may hinder or reduce the perpetrator’s access to heavy weapons will have an effect. Doing so, however, may jeopardize the consent of the host state, and lead to an early exit of the UN operation. A highly visible UN military presence in protection hot spots may deter attacks, although this approach may significantly increase the risk to the UN personnel involved and attacks may just shift elsewhere. Offensive capabilities (Special Forces, attack helicopters, etc.) may be used to target what is being used to attack populated areas, but, again, this is a highly risky approach for most UN missions. There are only two cases in UNPOCO scored as part of this category, when UN and French troops used force to hinder a recently deposited government in the Ivory Coast to crack down on its own population. They are included to portray the full variation in types of threats between 1999 and 2017. Furthermore, the main perpetrators had indeed lost their formal position, and were as such a non-state armed group, fulfilling the criteria for how a perpetrator is defined. I still recognize that the perpetrators in these events still saw themselves as the governing party when the events unfolded.

58

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

The strategic versus the tactical level Some may question the wisdom of applying Beadle’s theory to the universe of UN peace operations, as force is seldom used to protect, and the UN is not really a military actor in a traditional sense. The following responds to possible critique, and reasons why the approach remains useful, despite these inherent challenges. First, military practitioners proficient in planning and execution of military operations may claim that the theory does not provide enough detail of exactly what military forces are supposed to do in different situations to protect civilians most effectively. However, the functions of force are instead meant as broader analytical categories, each encompassing a wide range of potential military activities. Most importantly, Smith’s categorization seeks to describe the main purpose of each function of force, to remedy the confusion between deploying forces and effectively employing force. Combined with Beadle’s description of how perpetrators use violence against civilians, and for which purpose, it provides logic to how force—in principle—can protect most effectively. Planning and execution aspects rather belong in doctrinal publications and practical guidelines.8 Still, there is a need to explore more ways to respond militarily to different types of threats to protect civilians better. This book adds to this last aspect through deeper qualitative analyses of how matching and mismatching have unfolded in recent operations (Chapter 6 and 7). Second, both Beadle and Smith discuss how force may be used at the strategic level to influence parties to a conflict. This seemingly represents a conundrum for this study, as UN forces are not expected or allowed to influence the overall dynamic of a conflict or shift the balance of power by using forceful means. The Capstone Doctrine guided UN peace operations only to use force to protect at the tactical level, either for self-protection purposes or in the defense of mandate implementation, including protecting civilians (United Nations 2008). While this has been expanded in recent guidelines to include the operational level, this differentiation between levels is seldom clear-cut (United Nations 2017a). Moreover, the UNSC has singled out certain armed spoilers for targeted offensive operations by the FIB in the DRC (United Nations 2013f, 7). In Mali, UN troops are meant to deter armed groups from re-entering population centers (United Nations 2013g, 7). It follows that UNSC protection mandates sometimes order UN military peacekeepers to use force beyond what is commonly understood as the tactical level. Thus, Beadle’s theory does offer relevant insights for UN military operations on how to protect, although force is seldom applied strategically in a conventional sense.

SUMMARY OF CAUSAL CONDITION CANDIDATES

59

This latter aspect links in with a third, which indicates that, although the UN is primarily a political actor, there are many examples of UN operations displaying strategic rationales that fulfill the purposes of the four functions of force. Most traditional peacekeeping operations were meant to deter former warring parties from reverting to attacking each other, such as in Cyprus. Others, such as ONUC in the Congo, set out to contain the conflict from spreading from Katanga to the rest of the country. Later, ONUC coerced Katanga into revising its secessionist plans and even destroyed its capabilities to do so (O’Brien 1962, 2011). Yet other operations, such as in South Lebanon and in Bosnia and Herzegovina, mostly ended up ameliorating the consequences of armed conflict for civilian life, although they initially were meant to have a deterrent effect. The use of military forces in different UN missions serves different strategic rationales at different times. Therefore, it seems useful to employ a theory that also considers these strategic concerns. To sum up, I hold that the logic driving UN military operations to protect and perpetrators’ violence against civilians can responsibly be studied by adhering to Smith’s and Beadle’s functions of force, types of violence, and matching theory. In addition, the case studies in Chapters 6 and 7 move beyond the stringent categorization of functions of force and types of violence. They unpack what amelioration, containment, deterrence/coercion, and destruction mean in practical terms, and dig deeper into cases where force has been used to protect to search for more finely grained causal explanations of successes and failures. Finally, the book seeks to add nuance to Beadle’s theory, to better fit the characteristics and limitations of UN peace operations and develop insights as to what works when in a UN setting.

Summary of causal condition candidates I have found that existing knowledge of the causes of UN military protection successes across missions is scarce. I therefore introduce four causal condition candidates frequently mentioned as relevant for successful outcomes: (i) deterrent presence; (ii) willingness to accept risk; (iii) pre-emption; and (iv) matching. Consequently, we would expect that UN missions that deploy many troops relative to population size would succeed more often in their military protection operations than those missions that do not. Similar expectations are linked to those protection operations that involve risk-accepting troops from countries that impose few caveats. Further, if those troops have been able to match the perpetrators of violence, they should also succeed more often than

60

UNDERSTANDING THE UTILIT Y OF FORCE TO PROTECT CIVILIANS

those operations that do not. Finally, pre-emptive operations are sometimes needed to protect, and we should expect more successful outcomes when UN troops intervene before perpetrators launch attacks on civilians, while we must also expect that UN troops may succeed and fail in both offensive and defensive modes of operations. Chapters 6 and 7 further investigate these claims, anticipating other explanatory factors that are not captured in the literature.

3 Exploring characteristics and outcomes of UN military protection operations If we are interested in understanding how peacekeeping works—or does not work—to protect civilians, we need to improve our data on what peacekeepers do once deployed. —Professor Lisa Hultman Department of Peace and Conflict Studies Uppsala University (Clayton 2016, 29)

What does a typical UN military protection operation look like? And how have Blue Helmets fared when they used force to protect civilians, across time and UN missions? Incredibly—after more than two decades of UN military protection efforts—we still do not have a common idea of the core characteristics of this phenomenon and no systematic insights on the operations’ effectiveness across time and place. This chapter draws contours of possible answers to both questions. The first part of this chapter provides descriptive data from the UNPOCO dataset. While I primarily developed this dataset to explore to what degree UN troops have protected civilians from physical threats and to explain what determines positive outcomes, it also provides an opportunity to describe systematically the key characteristics of the operations where UN used force to protect (before diving into analyses of variation in outcomes and potential causal relations). This purely descriptive exercise is valuable to better understand the phenomenon at the center of this book, the application of force by Blue Helmets. I therefore describe where operations took place, the number of operations across time and place, the types of threat facing civilians (based on Beadle’s typology. See Chapter 2), how force has been used to protect (according to the four functions of force also described in Chapter 2), and civilian and UN casualty figures over time, before summarizing the descriptive statistics to say something about the “typical” protection operation (which is quite hard to describe).

Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0003

62

E XPLORING UN MILITARY PROTECTION OPERATIONS

UN protection efforts can be understood in many different ways. As a reminder, recall that I have used the following four criteria when I have selected cases for further study: i) perpetrators physically threatened or harmed civilians; ii) UN military troops, with a mandate to protect civilians, deployed to the location where civilians were threatened or harmed; iii) UN troops used military force to protect civilians, by applying one or more of the four functions of force: amelioration, containment, deterrence/coercion, and destruction (see Chapter 2); iv) the UN Secretary-General’s reporting to the UNSC between 1999 and 2017 captured the event. When I present the descriptive data, I draw from a pool of 200 military protection operations that I captured from the UN Secretary-General’s reports to the Security Council (Kjeksrud 2019). Even when guided by these criteria, the depth and detail of the information available vary significantly. The UN Secretary-General’s reporting to the UN Security Council is delivered in a wide variety of ways. I have captured cases that are limited to a certain geographical area and that are limited in time. However, some cases cover larger geographical areas, or evolve over time from one area to another. I have not set a particular limit to what constitutes one case. As long as it is treated as one event in the Secretary-General’s reporting, I have chosen to add it to my collection of data. To illustrate the variation in data, we mostly know where and when UN military protection operations took place, but I could only find specific civilian fatality figures for fewer than half of the 200 cases (97). While UNPOCO provides a first-of-its-kind overview of events where UN forces have used force to protect in Africa between 1999 and 2017, it has not been possible provide a comprehensive description of every case considering all its characteristics. Despite that, there is a wealth of available information that aids our search for a deeper understanding of this phenomenon. In the second part of the chapter, I estimate outcomes of UN military protection operations, answering the first research question posed in this book: To what degree have UN military troops protected civilians under imminent physical threats in African armed conflicts? While we mostly hear about those cases where UN troops failed to provide physical protection to populations under threat, I find that Blue Helmets are indeed able to protect civilians. I find that UN troops succeed about as often as they fail when they use force to protect, indicating that there is reason for optimism in their ability to

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

63

provide physical protection to threatened populations under the right conditions. However, these findings should be treated with caution, as the data underpinning the analysis come with weaknesses, as does the counterfactual method supporting the assessment of outcomes (Mahoney 2021).

Characteristics of UN military protection operations The chapter’s first part delves into the core characteristics of UN military protection operations across time and place. It rests on data derived from UNPOCO—capturing 200 cases across 10 UN missions from 1999 to 2017— and describes where operations took place, how many, the type of threat civilians were facing, the function of force used to protect, as well as civilian casualty figures.

Location Knowing where most UN military protection operations take place is primarily of value in pointing out which missions hold relevant experiences from which we may learn. Location data could of course indicate where civilians are most under threat, but we already know that UN troops seldom respond with force when civilians are threatened. Therefore, UNPOCO-data cannot— and should not—be used to indicate where civilians need protection most. I include a comparison of UNPOCO civilian casualty data with the Uppsala Conflict Data Program Georeferenced Events Dataset (UCDP GED) and the Armed Conflict Location Events Dataset (ACLED). The aim is simply to display the discrepancy between the number of UN responses captured in UNPOCO and the number of times civilians have needed physical protection, according to these more authoritative datasets on civilian casualties in armed conflict. Quantitative studies have demonstrated that there is potential in combining geographical deployment data of UN troops with conflict data from other datasets (Di Salvatore and Ruggeri 2017; Ruggeri, Dorussen, and Gizelis 2017, 2018). Still, even with the best available data, there is potential for misrepresenting, or failing to capture, the actual effects of uniformed peacekeeping deployments (Clayton 2016; Diehl and Druckman 2015). I alleviate some of these shortcomings in this book by studying actual protection operations and their immediate effects on civilian security, which of course also comes with its own trade-offs.

64

E XPLORING UN MILITARY PROTECTION OPERATIONS 120 100 80 60 40 20 No. of operations

U

N

IS FA

U N

M

O

N

U

C

/M

O

N

U SC

O

(2

00

020 17 (2 ) 01 AM 1 ID 20 17 (2 U ) 00 N M 720 IL 17 U (2 N ) 00 M 3I 20 SS M 18 IN (2 ) 01 U SM 12 M 01 IN A (2 7) 01 U SC 320 A 17 (2 U 01 ) N 4O 20 CI 17 (2 U ) 00 N 4M 2 IS U 01 N ( 7) AM 200 5SI 20 L 11 (1 ) 99 920 05 )

0

Figure 3.1 Distribution of reported UN protection operations per UN mission, 1999−2017.

One immediate observation leaps out from Figure 3.1. One mission is responsible for more than half of the total number of reported cases. In fact, 55 percent of the reported UN protection operations captured in UNPOCO—110 out of 200 cases—occurred in the DRC, where the UN mission (MONUC/MONUSCO) has been mandated to protect civilians since 2000. In fact, many conceptual innovations on how to protect have been taken from the experiences and best practices of MONUC/MONUSCO (Kjeksrud and Ravndal 2011). In comparison, the next highest UN mission on the list—UNISFA in Abyei, which is a contested area situated between Sudan and South Sudan—only carried out 20 operations, or 7.5 percent. While UNISFA has been deployed to Abyei since 2011—potentially holding many valuable experiences—few have studied UNISFA’s military protection operations (Osterrieder, Lehne, and Kmec 2015). From a military theoretic perspective—and definitely from a peacekeeping policy angle—there is probably much to be learned from UNISFA, not least because the mission largely consists of military troops (Kjeksrud 2014). The list continues with: UNAMID (Darfur) 16 operations; UNMIL (Liberia) 15 operations; UNMISS (South Sudan) 11 operations; MINUSMA (Mali) 9 operations; MINUSCA (Central African Republic) 8 operations; UNOCI (Ivory Coast) 6 operations; UNMIS (Sudan) 4 operations; and UNAMSIL (Sierra Leone) 1 operation.

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

65

7.00 6.00 5.00 4.00 3.00 2.00 1.00

Operations per year (avg,)

U N

M

O

N

U

C

/M

O

N

U SC

O (2 00 I 0 M SF I N A ( -20 17 20 U ) 11 M SC A IN (2 201 U 7) SM 01 A 4-20 U ( N M 201 17) U ISS 3-20 N 17 AM (20 ) 1 ID 1-2 U 0 ( 1 2 N M 007 7) IL U (2 201 N 0 7 0 M 3- ) I U S (2 201 N 8) 00 O 5 U N CI ( -20 AM 20 1 04 1) SI -2 L 0 (1 99 17) 920 05 )

-

Figure 3.2 Protection operations per year per UN mission (averaged), 1999 ̶ 2017.

Since the missions have different life spans, it makes sense to include the average distribution of operations per year per UN mission. Figure 3.2 shows that MONUC/MONUSCO predictably remains on top, performing on average 6.1 protection operations per year. That is still twice as many as UNISFA—remaining second—performing on average 2.9 protection operations per year. MINUSCA jumps up from seventh to third place, performing an average of two protection operations per year, followed by MINUSMA (1.8), UNMISS (1.6), and UNAMID (1.5). In real life, protection operations do not occur in yearly averages. Later, I will show how protection operations are distributed across time. Keep in mind that the events captured in UNPOCO only represent a sub-set of all protection operations performed by UN forces in this period. The actual distribution may, therefore, differ from what I have found. For example, the UN missions in the Ivory Coast, Sudan, and Sierra Leone—the bottom three on the list—have most probably conducted more protection operations than this collection of cases captures. Most of their efforts to protect occurred long before policy developments declared protection to be the main priority for UN peace operations. Varying reporting practices over time are therefore probably part of the explanation for the suspiciously low count.

66

E XPLORING UN MILITARY PROTECTION OPERATIONS

Context-specific explanations are also likely to influence the distribution. The UN mission(s) in the DRC are probably at the top of the list since Eastern DRC “is littered with dozens of foreign and Congolese armed groups of all shapes and sizes” (Stearns, Verweijen, and Baaz 2013, 13). The actual number of non-state armed groups active in the east remains elusive. During spring 2015, it fluctuated somewhere around 70, including predatory armed groups such as the FDLR, Allied Democratic Forces (ADF), and various Mai-Mai militias (Allen and Vlassenroot 2010; Fahey 2015; Stearns 2012a, 2012b, 2013a, 2013b; Verweijen and Iguma 2015). In 2018, the number had increased to more than 100 (Human Rights Watch 2018). These predatory perpetrators mostly target civilians as part of their strategy to survive, or for personal profit and benefit (Beadle 2014, 51). The pool of armed actors in the east also included the insurgent rebel groups Congrès national pour la défense du people (CNDP) (2006–2009) and the M23 (2009–2013). Both, arguably, used violence against civilians more sparingly to influence political opponents, although predatory behavior toward civilians did indeed occur (BBC News 2012; Beadle 2014, 57; Stearns 2012a; United Nations Joint Human Rights Office 2014). Consequently, demand for UN military efforts to protect civilians from various types of threats in the eastern DRC has consistently remained high since MONUC’s inception at the end of the 1990s. In recent years, MONUSCO has also been provided with the Force Intervention Brigade (FIB)—a militarily robust, brigade-sized unit—which has facilitated a substantial increase in military protection operations. Since the FIB became operational toward the end of August 2013, MONUSCO has performed 56 protection operations, constituting almost half of the registered operations performed by MONUC/MONUSCO between 2000 and 2017. I will return with a deeper analysis of FIB/FARDC-operations against the M23 in Chapter 6. Finally, MONUSCO is also one of the larger missions, which could influence the number of protection operations. Military protection efforts from UNISFA are still in demand, as threats to civilians peak at least twice each year when the migratory patterns of the Misseriya collide with the water and grazing areas used by the predominantly pastoralist Ngok Dinka (Kjeksrud 2014, 119–20). Until civil war re-ignited in South Sudan in mid-December 2013—severely curtailing the UN mission’s ability to operate beyond its own camps—UNMISS ran military operations to protect civilians from several different types of threats. UNMISS addressed large-scale communal conflicts, as well as local insurgencies in Jonglei, violent cattle raids in Bahr el Ghazal and the so-called Tri-State area, and armed clashes between the SPLA and anti-government groups in the Upper Nile.

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

67

We will take a closer look at one of these operations in Chapter 7. In Liberia, UNMIL was quite active in the first few years of operations, when national and UN police often needed UNMIL military intervention to curtail mob violence in urban areas. In later years, the demand for protection operations by UNMIL declined, due to an improving security situation. In 2018, UNMIL ceased its operations altogether. This statistic will keep changing. Some UN missions captured in UNPOCO are already part of history, while new peacekeeping missions may be established in Syria, Libya, Yemen, Ukraine, or elsewhere, in the future.

Number of operations Mapping the number of reported protection operations per year aids our understanding of how this phenomenon has evolved over time. It could indicate the level of intensity of civilian targeting in the conflicts to which UN forces deployed to protect. However, reported UN protection operations do not mirror well the actual number of threats to civilians. Instead, they merely point to the prevalence or absence of these types of protection efforts over time. These could be influenced by policy developments, reporting practices, or actual numbers of operations performed. As we can see in Figure 3.3, reported instances of the use of force to protect had a slow start since the UN Security Council first mandated a UN mission to protect civilians from violence in 1999 (United Nations 1999a). A slight increase was registered in the mid-2000s— immediately followed by a dip around 2007 and 2008—with an increasingly growing prevalence from 2010 to a new peak in 2016. The dip in 2007 and 2008 may be linked to political sensitivities about the protection of civilians as a mandated task. Another possible explanation for this trend line is the increasing number of military UN peacekeepers deployed since 1999. Figure 3.4, produced by the Providing for Peacekeeping project at the International Peace Institute (IPI), presents deployment trends of uniformed personnel from 1991 to 2018, of which the vast majority is UN troops (Providing for Peacekeeping 2018). Toward the end of the 1990s, the UN deployed fewer than 15,000 uniformed peacekeepers. This correlates with findings in UNPOCO, which recorded no reported protection operations in 1999—understandably so—since UNAMSIL was the only UN mission with a POC mandate at the time. In addition, UNAMSIL was not fully operational until early 2000. Furthermore, only one military protection operation was recorded in the Secretary-General’s reporting in 2000, none in 2001 or 2002, and only two in 2003. In this period,

68

E XPLORING UN MILITARY PROTECTION OPERATIONS

35 30 25 20 Number of operations

15 10

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

0

1999

5

Figure 3.3 Reported UN military protection operations in Africa per year, 1999–2017.

there was in fact a significant bump in uniformed deployments, and one would have expected that the increased number of troops would have influenced the number of protection operations. In 2004, 2005, and 2006, however, the UN reports more protection operations—6, 15, and 5 respectively—with 2005 representing the first “peak”

100,000 80,000

Troops Police Experts

60,000 40,000 20,000 0

1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017

Figure 3.4 Total number of uniformed UN peacekeepers deployed by type. Source: Providing for Peacekeeping (2018).

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

69

in Figure 3.4. In 2005, MONUC was responsible for 13 of the 15 protection operations. MONUC operations included several cordon and search operations in Ituri against the Front nationaliste intégrationiste, large-scale joint operations with the Armed Forces of the DRC (FARDC) against FDLR in South Kivu, offensive operations against predatory Mai-Mai groups in Virunga national park, as well as joint FARDC–MONUC operations against the ADF/National Army for the Liberation of Uganda (NALU) in North Kivu. The temporary increase in protection operations from 2004 to 2006 correlated with increasing troop numbers, with more than 60,000 uniformed UN personnel deployed in 2006. However, a sharp decline in reported protection operations immediately followed. The number of reported operations reached a low in 2007 and 2008, with only three and two, at the same time as troop numbers were steadily increasing. A rise in the number of protection operations since then was observed, until a new peak in 2016 with 33 reported protection operations, at a time when there were eight UN missions with an active protection mandate in Africa, and about 100,000 uniformed UN personnel deployed. MONUC/MONUSCO in the DRC was responsible for more than half of these (19), MINUSCA in CAR ran four operations, MINUSMA in Mali three, UNAMID in Darfur three, while UNISFA, UNMIL, UNMISS, and UNOCI each ran one. Again, the UN missions in the DRC are clearly the 35 30 25 20 Number of operations

15

MONUC/ MONUSCO

10 5

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

0

Figure 3.5 Total number of UN military protection operations per year in Africa compared with protection operations only performed by MONUC/MONUSCO, 1999–2017.

70

E XPLORING UN MILITARY PROTECTION OPERATIONS

driving force behind these numbers. As Figure 3.5 portrays, the dashed line of MONUC/MONUSCO closely follows the solid line portraying total numbers from 2000 to 2010, when MONUC changed into MONUSCO. In the following years, new UN missions were established in Abyei (2011), South Sudan (2011), Mali (2013), and CAR (2014), which spread military protection efforts across more UN missions, but MONUSCO still remained responsible for a significant part of the UN’s protection operations. At first glance, the UN seems to launch more protection operations today than in the recent past. However, keep in mind that reporting practices have also changed considerably over time. Policy reforms on POC do coincide with a growing prevalence of reported operations. These data may therefore reflect changing reporting practices, rather than an actual increase in protection operations. Furthermore, the number of troops deployed does not consistently correlate with the number of reported protection operations. Finally, protection operations were mostly performed by MONUC until 2010, while the addition of new missions from 2011 onward—all with robust protection mandates—have led to a more diverse phenomenon.

Type of threat The main theory tested in this book rests on the idea that using force to protect effectively demands a better understanding of the perpetrators of violence, i.e., why, how, and with what capabilities they target civilians as part of their warfare (Beadle 2014). The details and depth of this argument were presented in Chapter 2, and it will be further analyzed in later chapters, alongside other potential causal conditions. Here, however, I will present an overview of the different categories of perpetrators that UN troops have been up against in their operations when seeking to protect civilians. Accordingly, all cases in UNPOCO have been coded according to Beadle’s seven threat scenarios—adding an eighth scenario of mob violence—which I have developed (Kjeksrud, Beadle, and Lindqvist 2016). Beadle structures the scenarios according to the perpetrators’ rationale, i.e., their dominant logic for attacking civilians. They are listed in Table 3.1 from the most dangerous to the least dangerous in terms of physical threat to civilians, adding examples of each. Historically, UN peace operations have encountered all eight scenarios. However, in Africa between 1999 and 2017, data in UNPOCO indicate that almost all reported cases fall within only three of the eight scenarios: predatory violence, communal conflict, and insurgency. Together, as portrayed in Figure 3.6, these three scenarios account for 94.5 percent of all reported

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

71

Table 3.1 Threat scenarios structured according to the perpetrators’ main rationale for attacking civilians, with examples from armed conflictsa Scenarios

Rationale

Examples

Genocide Ethnic cleansing Government repression Communal conflict Predatory violence Insurgency Post-conflict revenge Mob violence

Exterminate a certain group Expel a certain group from a specific territory Control populations affiliated with opposition Avenge previous attacks and deter retribution Survive or profit Control populations and undermine rivals Avenge past crimes Profit, revenge, or voice political discord

Rwanda Bosnia Libya South Sudan DR Congo Somalia Kosovo Liberia

a

This table was published in a book chapter authored by Alexander William Beadle and me, “The Utility of Force to Protect in UN Peace Operations.” in The Use of Force in UN Peacekeeping, edited by Peter Nadin (Nadin 2018).

situations where the UN has used military force to protect civilians at the tactical and operational levels. Compared to the more extreme scenarios in the mid-1990s—genocide in Rwanda, an act of genocide in Srebrenica, and ethnic cleansing elsewhere in the Balkans—the period between 1999 and 2017 seemingly presented UN troops with less serious threats to civilians. I do not in any way downplay the gravity of the situation for those civilians caught up in 90 80 70 60 50 Number of cases

40 30 20 10 0 Predatory violence (82)

Communal corflict (81)

Insurgency (26)

Mob violence (9)

Government repression (2)

Figure 3.6 Distribution of scenarios across 200 military protection operations in Africa, 1999 ̶ 2017.

72

E XPLORING UN MILITARY PROTECTION OPERATIONS

armed conflicts in this period—or the risks involved for those Blue Helmets set to protect them—but rather to provide context to explain what types of situations UN forces need to be prepared to tackle based on openly available data. Different missions face different threats to civilians. In the DRC (MONUC/MONUSCO—still the most prominent feature in the dataset— most reported protection operations between 2000 and 2017 were related to predatory violence (77 of 110 cases), although communal conflict (18 cases) and insurgency (15 cases) also occurred. Communal conflict is by far the most dominant scenario for the UN mission in Abyei (UNISFA), constituting 19 of 20 reported cases. The Liberian (UNMIL) cases vary; they include communal conflict and predatory violence, but unusually there are also several instances of mob violence (7 out of 15 cases). However, the data in UNPOCO do not provide the full picture. For example, the conflict in Darfur—where the primary threat to civilians came from the Sudanese government’s regime crackdown—has arguably escalated to ethnic cleansing during its most violent phase. Some scholars have found that 57 percent of the targeted population in one area of Darfur had been ethnically cleansed, and 48 percent of their villages destroyed, between 2003 and 2008 (Olsson and Valsecchi 2010). The Secretary-General’s reporting to the Security Council reveals that the African Union/UN Hybrid operation in Darfur (UNAMID) largely used military force to protect civilians in response to communal conflicts, suggesting that the mission has not used force to address the most serious threat to civilians. This is also quite understandable, as any attempt to confront government forces would possibly lead to loss of hoststate consent. Similar trends can be found in South Sudan, where the regime violence against civilians arguably includes ethnic cleansing, but where UN troops are unable to address the larger conflict dynamics (UN Commission on Human Rights in South Sudan 2016). In addition, the ethnic cleansing of Muslims by Anti-balaka in western parts of CAR in early 2014—a few months before MINUSCA was deployed—underlines that more violent scenarios are indeed possible (Øen 2014, 27–30). Sometimes, it has been challenging to assign the proper scenario to each perpetrator captured in UNPOCO. Many perpetrators portray more than one rationale for targeting civilians. For example, I have coded ADF/NALU in the DRC as a predominantly predatory actor, preying on the local population to survive and for profit. The armed group, however, also portrays other motivations for targeting civilians (Congo Research Group 2017). At its inception decades ago, ADF/NALU was a political insurgent group, seeking to

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

73

overthrow the government in Uganda. Lately, the armed group has turned toward confronting Congolese government forces, as well as MONUSCO troops, through direct attacks. This is a rather uncommon feature for most predatory perpetrators. I have not been able to discern a clear political objective driving ADF/NALU’s attacks on government forces and MONUSCO troops, which could have indicated a renewed insurgent rationale. Such multiple motivations are common for some perpetrators. I am still confident that UNPOCO captures the main motivations for most perpetrators. In addition, the UNPOCO dataset is publicly available, facilitating critique of the coding choices I have made throughout this mapping exercise (Kjeksrud 2019). To sum up, in the past 30 years UN military forces in Africa have faced the entire spectrum of violence against civilians as described by the eight threat scenarios. From 1999 to 2017, however, only three types of perpetrators constitute the vast majority of threats to civilians in Africa when UN forces have intervened to protect: predatory armed groups, communal militias, and insurgent rebel groups. These three constitute almost 95 percent of the 200 cases in UNPOCO. According to Beadle’s theory, while force may be used with utility in the first two scenarios, political insurgency is both more limited in threats to civilians as well as more challenging to cope with for military forces (see Chapter 2). Chapter 5 will analyze how well UN forces have fared in protecting civilians from these different types of threats across cases. Chapters 6 and 7 provide insights into the causal mechanisms leading to successes as well as protection failures against different types of perpetrators, with the help of two qualitative case studies from the DRC and South Sudan.

Functions of force According to existing theory, four functions of force can be used with utility in contemporary wars amongst the people, including to enhance security for civilians under physical threat: amelioration, containment, deterrence/coercion, and destruction (Beadle 2014; Smith 2008) (see Chapter 2). In this section, I describe the prevalence of each function across operations, regardless of outcome (see Figure 3.7). The aim is simply to provide an empirical foundation for how UN troops apply force to protect across time and operations. I have chosen to code Smith’s function “deterrence/coercion” as two separate functions of force. According to Smith, this function involves a “wider use of force,” where military forces are used to pose or carry out a threat, in order to change or form the opposition’s intentions

74

E XPLORING UN MILITARY PROTECTION OPERATIONS

140 120 100 80 Number of cases

60 40 20 0

Amelioration (38)

Containment (33)

Deterrence (119)

Coercion(87) Destruction(41)

Figure 3.7 Distribution and prevalence of functions of force across UN protection operations in Africa, 1999–2017.

(Smith 2008, 323). When force is actually employed, it is used to coerce. Since the use of coercive force is still a controversial issue for the UN, it makes sense to separate these into two functions of force in this overview. I find that, while deterrence remains the most prevalent function of the force used by UN missions in Africa between 1999 and 2017, UN forces have in fact employed the entire spectrum of the functions of force to protect, including the most forceful—coercion and destruction—on quite a few occasions. Of the 200 cases in the dataset, it was possible to assign functions of force to 191 cases. The two dominant functions of force employed by UN peacekeepers were deterrence (occurred in 62.3 percent of the cases) and coercion (occurred in 45.6 percent of the cases). Destruction (21.5 percent), amelioration (19.9 percent), and containment (17.3 percent) were less prevalent. While deterrence clearly plays a major role in UN peacekeeping operations, UN troops employed the entire spectrum of force to protect. While Figure 3.7 displays how many times each function of force occurs across the 191 cases, it does not account for combinations. In many cases, UN troops applied more than one function of force, in different phases of an operation. The analyses in Chapters 6 and 7 will elaborate on this aspect, by comparing the most dominant function of force in each case with the type of violence employed by the perpetrators, and dig deeper into particular cases, analyzing how UN forces used force in different phases of the operations.

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

75

Casualty figures Measuring civilian casualties in civil war and armed conflict is both highly contested and practically and methodically challenging (Kalyvas 2006; Seybolt, Aronson, and Fischhoff 2013; Våge and Beadle 2014). Reliable and consistent data are hard to come by, as “relevant information is frequently concealed by parties to the conflict, destroyed in the course of the war, or never recorded at all” (Lacina and Gleditsch 2005, 146). Still, comprehensive global datasets on civilian casualties in civil war and armed conflict do exist, continuously improving the quality of casualty figures, facilitating better analyses and improving our understanding of this phenomenon (ACLED 2020; Allansson, Melander, and Themnér 2017; Melander 2015; Pettersson and Öberg 2020; PRIO 2016; Uppsala University 2020). Nevertheless, it remains methodically challenging to develop concepts and models that appropriately capture the complexities of civilian casualties, even when data are available (Gates and Strand 2004). The data in the UNPOCO dataset are no exception. In fact, I could only find reported civilian fatality figures for 97 of the 200 cases. Furthermore, only 57 cases contain information about non-lethal violence against civilians. In addition, from 1999 to 2003, I found no cases with civilian casualties via the Secretary-General’s reporting. This period also saw few UN protection operations, and this observation is therefore no big surprise. As a result, I only portray the period from 2004 to 2017 in this section. Challenges riddle the available information; how many were killed; who, how, by whom; how many were injured, raped, displaced, or subjected to forced labor? Sometimes, the only assessment of casualty figures states “many killed,” “several harmed,” “unconfirmed number abducted,” etc., with no numerical estimate provided. Mostly, even this type of information is absent. I am not primarily concerned with trends in global civilian casualty patterns in armed conflict. Rather, I seek to trace the effects of UN military protection efforts and understand variations in outcomes of tactical level protection operations to counter immediate threats to civilians. What types of situations do UN troops face when they employ force to protect? Reporting to the UN Security Council does in fact include casualty figures for quite a few of the cases. Ad hoc assessments by troops on the ground, as well as investigations in the aftermath of attacks by NGOs or civilian sections of the UN missions, do provide some information. I will therefore briefly summarize and analyze reported data of civilian casualties in situations where UN forces were involved

76

E XPLORING UN MILITARY PROTECTION OPERATIONS

in protection and compare this information with available casualty figures from other datasets. Civilian death toll The 97 cases in UNPOCO that contain specific information report from zero to 612 civilian fatalities. These are all included Figure 3.8. Two cases involved the use of terms such as “many” and “dozens” to describe the numbers of civilians killed. It follows that civilian death toll data remains unknown in 101 of the cases. Data on non-lethal violence will be treated in the next section. Two observations leap out from Figure 3.8. First, civilian fatalities in operations where UN troops use force to protect seem to be increasing. Second, 2011 stands out as a significant outlier. This was when the Lou Nuer White Army marched on the Murle population in Jonglei state, South Sudan, and Murle fighters subsequently launched revenge attacks on Nuer communities in the immediate aftermath. An estimated 888 civilian lives were lost (UNMISS Human Rights Division 2012), or about 32 percent of the total number of 2,772 civilians reportedly killed from 2004 to 2017 in situations where UN troops used force to protect. This case will be treated as a separate case study in Chapter 7. The average number of civilians killed per year in situations where UN troops used force to protect is 198. However, as we can read from Figure 3.9, there is great variation between years. As mentioned, the mean is driven 1000 900 800 700 600 500

Civilians killed

400 300 200 100 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Figure 3.8 Total civilian fatalities per year deduced from 97 cases where UN troops used force to protect in Africa, 2004–2017.

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

77

60 50 40 30

Civilians killed per case per year

20 10 0

04 005 006 007 008 009 010 011 012 013 014 015 016 017 2 2 2 2 2 2 2 2 2 2 2 2 2

20

Figure 3.9 Average number of civilians killed per case per year deduced from 97 cases where UN troops used force to protect in Africa, 2004–2017.

upward due to the highly deadly events in Jonglei in 2011–2012. Distributed evenly across cases, the fatality numbers indicate that, on average, 28 civilians were killed per case. The median, however, remains at three civilians killed. In fact, in 32 cases no civilians were killed, according to UN reports. Although there is great variation between cases, a handful of very violent events significantly influence these averages. For example, a suspected Lord’s Resistance Army (LRA) attack in Mabanga ya Talo in Orientale province in the DRC in 2009 reportedly took more than 100 civilian lives (United Nations 2010c)). In February 2013, Aballa and Beni Hussain communal groups clashed in El Sireaf, Darfur, leaving approximately 80 dead and “upward of 100 wounded” (United Nations 2013c, 255). These numbers and examples indicate that UN troops in Africa have faced several quite severe situations from 2004 to 2017. In addition, the overall number of civilians killed per year and per case where UN forces have been involved to protect seems to be increasing. Again, we cannot reliably infer from the UNPOCO data that the situation for civilians and those set to protect them is indeed worsening, since reporting practices vary, and we lack information on many cases. For those preparing to protect, the safest bet would be to acknowledge that there is great variation between years and cases, and they must, therefore, prepare for both extreme cases where civilians are killed in the dozens and hundreds, as well as more common situations where none or “only” a handful of victims are killed.

78

E XPLORING UN MILITARY PROTECTION OPERATIONS

Comparison with civilian fatality numbers from other datasets We already know from internal UN evaluations that the number of attacks against civilians greatly outnumbers the times UN forces have applied force to protect them (United Nations Office of Internal Oversight Services 2014). It is not surprising that civilian fatality data compiled by the Armed Conflict Location & Event Data Project (ACLED) and Uppsala Conflict Data Program (UCDP) vastly outnumber fatality figures in UNPOCO (ACLED 2022; Davies, Pettersson, and Öberg 2022; Sundberg and Melander 2013; Uppsala University 2022).¹ It is worth reiterating just how little of the violence against civilians is addressed by UN troops. On the one hand, that confirms that there is a significant “protection deficit” where UN troops deploy to protect, and, on the other hand, that the so-called robust turn in UN peace operations is seemingly more present in UNSC mandates, UN policies, and debates than in actual operational practices. The comparisons that follow are rough. First, the typical caveats found in UNSC mandates—guiding UN missions only to protect civilians “within areas of responsibility” and “within capabilities”—are not reflected in the comparison. Neither is the fact that protecting civilians is primarily the responsibility of the state. I have simply compiled, described, and compared country–year data on civilian fatalities from 2004 to 2017, as they are presented in ACLED and UCDP databases. Some of the events captured by ACLED and UCDP are not formally the responsibility of the UN. Second, there are challenges with both ACLED and UCDP’s methods and outputs (Eck 2012). ACLED has been criticized for weak quality-control mechanisms, as well as a broad approach to coding events as violence against civilians (Eck 2012, 126–7). While Uppsala’s dataset only includes events that lead to fatalities—relating to a well-established understanding of what constitutes armed conflict—ACLED also includes non-fatal and non-violent events (Diehl and Druckman 2015). I expect that ACLED reports many more instances of violence against civilians than UCDP does. Between 2004 and 2017—collecting evidence from the eight African conflicts that are represented in UNPOCO in this same period—ACLED finds 11,404 events of violence against civilians. In total, these events reportedly resulted in 35,103 civilian fatalities. However, 1,342 events were coded as blanks. Of the remaining 10,062 events actually containing information about ¹ Uppsala University’s Conflict Data program (UCDP) regularly updates its datasets, also updating past events when new sources come available. As such, my numbers may differ slightly from the numbers fund in the latest available version of the UCDP Georeferenced Event Dataset, as I ran the analysis in 2018. It does not alter the observation that Blue Helmets only are involved in a small percentage of the total amount of one-sided violence.

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

79

civilian fatalities, 5,262 events were coded with zero civilian fatalities. In more than half of the events recorded as violence against civilians, no civilians were killed (meaning that the median rests at 0). In the remaining 4,800 cases, where one or more civilians were killed, ACLED’s fatality figures vary between 1 and 750. The mean (not including the cases with zero fatalities) is found at 7.3. Uppsala University’s geo-referenced dataset finds 3,399 events coded as “one-sided violence”—meaning violence against civilians—in the same period (2004–2017) in the same eight African conflicts. In total, 25,959 civilian fatalities were recorded. Recall that UCDP only records events with one or more fatalities. The mean is found at 7.6, while the median is at two civilians killed. As expected, ACLED codes many more cases as violence against civilians, than does UCDP. However, many of those events contain either no information about civilian fatalities (blanks) or are events with no civilian fatalities. UCDP provides a more conservative estimate. I have not investigated potential differences in the quality of the information, which has been done by others in the past (Eck 2012). Interestingly, while ACLED reports more than 1,400 more cases and 10,000 more fatalities in the same period than UCDP does, both datasets end up with similar figures on the average civilian death toll per event (about seven civilians killed per event on average). While this rudimentary comparison reconfirms that all civilian casualty data should be treated with caution, these two datasets are the best—and sometimes the only—sources available for comparisons across time and place. For the purposes of this book, both datasets are still able to display the glaring discrepancy between the “total” amount of violence against civilians, and the UN’s efforts to protect by military force. UNPOCO only captures 200 cases where UN forces were deployed to protect civilians under threat between 1999 and 2017. Of those 200 cases, almost all (197) occurred between 2004 and 2017. Civilian fatality figures (including 0) only exist for 97 cases. UNPOCO only provides data for 65 cases where one or more fatalities occurred, with a total of 2,772 fatalities reported. Compared to the thousands of violent events recorded by ACLED and UCDP—recording somewhere between 26,000 and 35,000 fatalities—UN troops in Africa have been involved in only a small percentage of the total amount of violence against civilians. Even when all precautions are considered, this observation points to a significant deficit in the ability or willingness of the UN to protect civilians from violence. The numbers taken from UNPOCO also point to another interesting observation, in that the averages and median numbers could indicate that UN forces often deal with more serious violence against civilians. The mean for the 65 cases in UNPOCO with one or more civilian fatalities is 42.6 and the median

80

E XPLORING UN MILITARY PROTECTION OPERATIONS

rests at 17. In short, UNPOCO captures events that differ from the most typical events found in ACLED and UCDP, but that constitute a small subset of all violence against civilians in Africa. Non-lethal violence Available data for civilians subjected to non-lethal violence across the 200 protection operations are scarce. Specific numbers only exist for 57 cases in the reporting to the Security Council, varying from none (29 cases) to 387 civilians injured. In an additional four cases, the reports include qualitative descriptions of the number of wounded, such as “many,” and “some.” These four cases are not included in this analysis. In 139 cases, the number of injuries remains unknown. Abductions leading to forced labor and others forms of involuntary relations to the perpetrator are not included, as the quality of data for these phenomena is clearly inadequate. According to the reports, 1,235 civilians were wounded, or otherwise physically harmed (including rape), across the 28 cases with one or more injuries reported. Distributed evenly, this indicates more than 44 civilians injured per case on average. The median remains at 10.5 civilians injured. Significant outliers in the available data drive the mean upward. For example, in July 2010: “at least 387 civilians, including 300 women, 23 men, 55 girls and 9 boys, were raped by a coalition of combatants from the Democratic Forces for the Liberation of Rwanda and the Mai-Mai Sheka, as well as by residual elements led by Lieutenant Colonel Emmanuel Nsengiyumva. In addition, at least 923 houses and 42 shops were looted, and 116 civilians were abducted and subjected to forced labor by the assailants” (MONUSCO 2011, 4). The UN mission was not aware of the attacks and did not respond until several days had gone by, when most of the attacks were over. MONUSCO was harshly criticized for failing to respond in time to protect civilians (United Nations 2010d, para. 8). Furthermore, in October 2004, 208 civilians were wounded—and 19 killed—in and around Monrovia, the capital of Liberia. Some of the casualties from this event may include combatants and security forces. Large-scale rioting initially triggered by land disputes “assumed ethnic and religious dimensions involving members of the predominantly Mandingo ethnic group” (United Nations 2004, para. 3). The situation further deteriorated when “disgruntled combatants awaiting reintegration, loyalists of former President Charles Taylor and some elements of the opposing factions within LURD ethnic dimensions […] repeatedly exploited [the riots] for their own ends” (United Nations 2004, para. 3). UNMIL troops and UN police were critical in bringing the situation under control, according to the UN reports.

CHARACTERISTICS OF UN MILITARY PROTECTION OPERATIONS

81

Again, UNPOCO cannot provide a comprehensive review of non-lethal violence against civilians. The sub-set of cases referred to above merely underlines that protecting civilians in UN peace operations is not always about countering unlawful murders of civilians, but also about reducing different types of non-lethal harm. UN casualties There is an ongoing debate amongst scholars and UN practitioners about whether or not UN peace operations have become more dangerous (dos Santos Cruz, Phillips, and Cusimano 2017; Guterres 2018; Henke 2016; van der Lijn and Smit 2015). From a practitioner’s perspective, the turn toward socalled robust peacekeeping in UNSC mandates and UN policies—in which peacekeepers are expected to be more proactive in their efforts to protect civilians—can expose peacekeepers to greater risks. In addition, some contemporary armed groups clearly ignore the Law of Armed Conflict, directing their attacks toward UN peacekeepers (UN News 2017). However, when controlled for monthly deployment numbers, existing research does not find that UN peacekeeping has become more deadly over time. In fact, a recent study found “a notable decrease in [UN] deaths between 1990 and 2015, both in absolute and relative terms” (van der Lijn and Smit 2015). Supporting these findings—using newly released monthly data on UN deployments to analyze fatality ratios at the national contingent, UN mission, and global level—Henke, Rogers, and Kennedy also found that, rather than being killed by armed attacks, the number of UN personnel dying from illness is increasing (Henke 2016, 10; Rogers and Kennedy 2014). Overall, UN fatalities from so-called “malicious acts” remain quite limited and thus largely unaffected by the robust turn in peacekeeping. These findings are also reflected in this book. According to UN statistics, so-called malicious acts led to 315 UN fatalities across the ten UN missions from 1999 to 2017 that also feature in UNPOCO (United Nations 2018)). Only 14 of these fatalities occurred in connection with the 200 military protection operations in UNPOCO (see Figure 3.10). During the same protection operations, 42 uniformed UN personnel were wounded, again according to the numbers in UNPOCO. The UN does not provide openly available information on numbers of wounded and otherwise harmed. If UNPOCO is a reasonable reflection of those situations where UN troops use force to protect, it does not seem that this task puts UN troops in particular danger. Although the fatality ratios from malicious acts are stable or slightly decreasing, Figure 3.10 shows that in absolute terms—without controlling for monthly

82

E XPLORING UN MILITARY PROTECTION OPERATIONS

70 60 50 UN fatalities/ malicious acts

40

UN fatalities Mali/ malicious acts UNPOCO fatalities

30 20 10

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

0

Figure 3.10 Total number of UN fatalities from malicious acts (1999–2017) in UN missions represented in UNPOCO (solid line), UN fatalities from malicious acts in Mali (2013–2017) (broken line), and UN fatalities captured in UNPOCO (dotted line). Data taken from United Nations and UNPOCO (United Nations 2018).

troop deployments—more peacekeepers are targets of violence today than in the recent past. The UN mission in Mali (MINUSMA) is responsible for many of these events (Karlsrud 2015; Kjeksrud and Vermeij 2017). Few UN troops are killed while taking part in military operations to protect civilians from violence, as depicted by the scattered green line at the bottom of Figure 3.10. One reason is that there are few of these operations in the first place; but it must also mean that they seldom result in much physical damage to UN troops. Earlier in this chapter, I did find a slight increase in the number of reported UN military operations to protect over time, which might indicate that these types of protection responses are becoming more common. However, despite much talk about UN peace operations becoming more dangerous, there is little evidence of this in the available data, with MINUSMA as an outlier.

Summary of descriptive statistics It is challenging to respond to the first question that opened this chapter: what does a typical UN military protection operation look like? The descriptive statistics taken from UNPOCO—despite containing many uncertainties— indicate that UN troops have been involved in very different situations on the

SUMMARY OF DESCRIPTIVE STATISTICS

83

ground across time and operations. Recall also that the data only capture the period from 1999 to 2017. In the not-too-distant past, UN troops and civilians have faced more serious situations—think Rwanda (genocide) and the Balkans (ethnic cleansing and an act of genocide (Srebrenica))—with casualty figures in the tens and even hundreds of thousands. In the future, new types of threats may emerge. Therefore, if we expect Blue Helmets to use force to protect civilians effectively, they must be prepared to address both limited attacks—with relatively few killed and injured—as well as mass atrocities, where the civilian casualty figures are significantly higher. Furthermore, we know that protecting civilians by using military force is still a marginal phenomenon in UN peace operations. The data I have captured in UNPOCO reconfirm that UN troops seldom respond with force when perpetrators of violence physically threaten civilians. However, while the early 2000s saw almost no protection operations across UN missions, the trend since 2008 seems to point toward a steady increase in the yearly number of protection operations. The increase possibly reflects a slow turn toward more consistent robust military responses to protect those under threat of violence and it could also reflect an increasing demand for physical protection in contemporary armed conflicts. However, it could also point to more trivial explanations, such as changing reporting practices in the UN. We also know that only a handful of UN missions have undertaken most military protection operations, with MONUC/MONUSCO standing out as the top-ranking mission by far. Indeed, many of the best practices and lessons learned underpinning current UN policies and guidelines stem from experiences from the DRC. However, the data captured in UNPOCO indicate that there is value in casting the net wider to capture lessons from other missions as well. The most common threats UN forces and civilians faced between 1999 and 2017 in African conflicts were predatory armed groups, communities in conflict, and insurgent rebel groups, although both more and less serious threat scenarios have occurred in this period. While insurgent violence usually does not lead to many civilian casualties, both predatory violence and communal violence often do. The data show that UN forces must be prepared to protect civilians against a wide variety of threats, but that some types of threats should probably receive more attention in training and preparation for deployment. Met with these threats, UN military troops have employed the whole spectrum of force to protect civilians—including destructive use of force—but have relied most heavily on deterrence and coercion. This underlines that UN forces can operate forcefully—and have done so on many occasions—although UN

84

E XPLORING UN MILITARY PROTECTION OPERATIONS

missions most commonly are employing defensive and reactive responses to threats against civilians. Keep in mind that these statistics do not address the effectiveness of the functions of force applied to protect. The next chapters delve into how UN forces have responded—and with what outcome—across cases and UN missions. Casualty figures from these operations remain challenging to collect and analyze. A sub-set of cases from UNPOCO indicates great variation, with a few cases that are very violent—with civilian casualties in the hundreds— while most cases involve few civilian casualties. In comparison, datasets that are more comprehensive show that the civilian death toll between 1999 and 2017 across these conflicts vastly outnumbers the civilian fatalities captured by UNPOCO. Clearly, UN forces are not able to address most of the violence against civilians occurring in armed conflicts. However, this should not undermine our efforts to understand the utility of force to protect in UN peace operations; rather, it should be a spur to the identification of those operations that may provide valuable lessons for future protection efforts. UN fatalities from so-called malicious acts relative to the number of troops deployed remain low. Existing research shows that peacekeepers more commonly die from illness and accidents. An outlier is the UN mission in Mali, where UN troops are countering insurgent rebel groups together with—and sometimes on behalf of—the Malian government. Although insurgent groups regularly attack MINUSMA and government forces, the UN mission in Mali performed few protection operations in this period. Hence, few UN troops die because of taking part in this task. The data captured in UNPOCO indicate that as few as 14 military peacekeepers have been killed by malicious acts while actively protecting civilians in 200 operations from 1999 to 2017. These fatality figures indicate that there is probably a potential to be a bit more risk accepting. As argued by others, it could be that a more robust stance could even make UN troops more secure (dos Santos Cruz, Phillips, and Cusimano 2017). However, I am not a proponent of using more force in UN peace operations. Rather, I aim to explore how force can be used most effectively against different types of threats, and when force should be avoided altogether.

Estimating outcomes of UN military protection operations (1999–2017) Descriptive statistics taken from UNPOCO help us unpack the characteristics of UN military protection operations across UN missions in Africa between 1999 and 2017. But how have UN troops fared in these operations?

ESTIMATING OUTCOMES OF UN MILITARY PROTECTION OPERATIONS

85

Consistently estimating the outcomes of the 200 protection operations in UNPOCO is challenging. In this section, I relay the results of coding each case as either “many” civilians protected (a successful outcome) or “few” civilians protected (an unsuccessful outcome). The coding rests on a qualitative estimate of the outcome of each case, guided by counterfactual reasoning, knowledge about the perpetrator of violence, and the UN military efforts to protect. I also rely on the generic scenarios developed by Beadle, which all draw on existing knowledge about the modus operandi of similar types of perpetrators (see Chapter 2). Accordingly, I consider the modus operandi of the perpetrators of violence to indicate what is likely to have occurred without the UN response, i.e., the expected outcome without intervention. Then, I hold this expectation up against the actual outcome of each case. These evaluations remain approximate, as there is not enough information available across cases for more systematic approaches. Ideally, I would like to know more about the population at risk—who and how many were potential targets for each threat—but this information is not readily available. In Chapter 5, I develop a more finely tuned estimation of outcomes—including partial successes and partial failures—of a sub-set of 126 cases that contain more information. In addition, the qualitative assessments of the outcomes can all be found in UNPOCO, opening up for critique of my results (Kjeksrud 2019). The most successful outcome of a military protection operation is that no civilians are killed, harmed, or displaced, i.e., that an imminent attack is deterred or pre-empted altogether. Some cases in UNPOCO do in fact have this fully successful outcome, where all potential victims are protected. For example, on March 14, 2013, UNISFA swiftly intervened, blocked, and deterred between 4,000 and 5,000 armed Misseriya fighters from marching on Makir Awed in Abyei (United Nations 2014c). In this case, it is quite straightforward to assume that an armed group of this size marching toward enemy territory represents an imminent and significant threat to civilians, and that UNISFA’s military intervention was the main reason why they chose not to attack. We know that large-scale revenge attacks between these two warring communities are quite common, have been so for decades, and are typical of their modus operandi (Johnson 2016, 8). However, we cannot be certain that harming civilians was their main objective during this event since violence did not in fact occur. Perhaps the Misseriya understood that UNISFA would intervene, and the main purpose was merely to instill a sense of fear amongst the opposing Ngok Dinka community in Abyei. This shows that even the most clear-cut cases with seemingly successful outcomes are up for debate, as it remains challenging to analyze and measure outcomes of events that did

86

E XPLORING UN MILITARY PROTECTION OPERATIONS

not occur. For an overall estimation of UN military efforts to protect, however, it seems safe to code this type of event in the “many protected” category. In other events, UN troops responded late or had little impact on the outcome. In an extreme case from the DRC over four days in July and August 2010, at least 303 people were systematically raped in 13 villages on the Mpofi-Kibua axis in Walikale territory by FDLR and Mai-Mai Sheka elements. At least 923 houses and 42 shops were also looted, and 116 civilians were abducted and subjected to forced labor by the assailants. Reports of the attacks did not reach MONUSCO until several days after they had begun, and MONUSCO patrols and protection mechanisms in this case were unable to detect the gravity of the situation (United Nations 2010d). Like the first example, it does not seem problematic to code the outcome as “few” (if any) civilians protected in this situation, as we know much more about the actual outcome. Many cases in UNPOCO are less clear-cut than these two initial examples. Sometimes, UN forces respond when threats or attacks against civilians are already underway but are still able to influence the outcome positively. In January 2012: “a group of nomads carrying guns arrived in Leu, which is predominantly inhabited by Ngok Dinka returnees, with about 6,000 cattle. This unexpected development generated fear and panic in the local community and even led some newly arrived returnees to go back to neighboring villages. With reinforcements from mission headquarters, the UNISFA company deployed at Leu managed to control the situation and convince the Misseriya migrants, after several hours of discussions, to vacate the village” (United Nations 2012b, para. 24). In this event, the UN troops failed to pre-empt the arrival of armed fighters and some civilians fled, but Blue Helmets still probably prevented further bloodshed through their deterrent and ameliorating efforts. Conversely, in other events UN troops intervened after the fact, but with only marginal influence on the violence against civilians. For example, in the DR Congo: “[…] on 27 November in Luhanga, Lubero territory, […] some 50 Mai-Mai Mazembe elements attacked a camp of internally displaced persons, killing 30 and injuring 21. MONUSCO troops exchanged fire with the Mai-Mai elements, killing one and injuring two. MONUSCO attended to the wounded and evacuated at least 15 injured displaced persons to local hospitals. After the attack, FARDC and MONUSCO reinforced their positions around Luhanga” (United Nations 2016b, para. 20). In this event, many civilians were killed and harmed, and, although the UN troops did engage the perpetrators using force, that did not significantly influence their willingness and ability to attack the camp. Of course, without any UN intervention at all, more civilians may have been victimized. Still, I have chosen to code this case

ESTIMATING OUTCOMES OF UN MILITARY PROTECTION OPERATIONS

87

as “few protected,” as the opposite would seem less convincing. These two cases highlight the dilemma of determining success from failure. However, I believe it is valuable to estimate outcomes of these operations, despite the challenges involved. The outcome estimations are presented in Table 3.2. Out of the 200 cases in UNPOCO, I have coded 83 cases as “few” protected, and 76 cases as “many” protected. In the remaining 41 cases, I have not been able to designate a particular outcome. Table 3.2 Outcome estimations of 200 UN military protection operations captured in UNPOCO Outcomes

Few protected

Many protected

Unknown

No. of cases Percentage

83 41.5

76 38.0

41 20.5

How are these outcomes distributed over time? Figure 3.11 tells us that successful outcomes were more frequent, or as prominent, as failures, from 2002 to 2014, while failures were more prominent from 2014 to 2017. The trend lines indicate that all categories are slightly increasing over time, as is the number of operations overall. Although we should be careful in reading too much into 35 30 25 20

Total no. of operations Few protected

15

Many protected Unknown outcame

10 5

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

0

Figure 3.11 Distribution of outcome estimations of UN military protection operations over time (1999–2017), with trend lines.

88

E XPLORING UN MILITARY PROTECTION OPERATIONS

these numbers, there are perhaps reasons to investigate the period from 2014 to 2017 in order to understand better the “spike” in operations with unsuccessful outcomes in this period. It is time to provide an answer to the first research question: to what degree have UN military troops provided protection to civilians under imminent physical threat in Africa between 1999 and 2017? Somewhat unsatisfactorily, albeit still interestingly, the data I have collected in UNPOCO portray a mixed picture. First, UN troops are certainly able to protect civilians from violence. On—at least—76 out of 200 occasions between 1999 and 2017, UN forces have protected civilians from imminent threats. This is perhaps not an impressive number of successes, given how many times civilians have been under threat in these conflicts, but it counters prevailing perceptions that UN troops are not up for their most prioritized task. Second, UN troops seem to fail about as often as they succeed. I have captured 83 failures amongst 200 cases, where UN forces had little or no effect on civilian security, despite having intervened militarily. We cannot expect that UN troops will always succeed, but there is probably a potential to improve this record. We cannot be certain that the distribution between successes and failures is representative of this phenomenon, as more than 20 percent of the cases still have unknown outcomes. In addition, there are probably many more cases that were never reported through the Secretary-General’s reporting to the Security Council. Again, these observations point to the need to improve reporting practices. A more systematic approach to reporting—always including basic information such as: location, time, type of perpetrator, the number and type of casualties (when possible), and a description of the UN and host-state efforts—would make for better comparative studies, improving our understanding of what works and what does not in different situations. When possible, this information should be shared with the wider research community with an interest in UN peace operations, as the UN is not able to study all cases of interest.

4 Analyzing empirical patterns of UN military protection operations “Given that protection crises are complex and conditions can change quickly on the ground, guidance and training on the use of intelligence, posture, and use of force is critical in the protection of civilians up and down the chain of command.” —Alison Giffen Director UN portfolio Civilians in Conflict (Giffen 2010, 9)

How do successes in UN military protection operations come about? In 2010, Alison Giffen—a peacekeeping and protection expert—accurately observed that better guidance for the entire chain of command is needed to improve military protection efforts (Giffen 2010). Worryingly, we still do not know what works—from a military perspective—when civilians are threatened with violence. In this chapter, I analyze potential causal stepping-stones that can help us explain successful protection outcomes across time and UN missions, which may point to what is needed in terms of better guidance and training to Blue Helmets. I do this with the help of basic statistical analysis of categorical data derived from UNPOCO, before employing cross-tabulations, Chi-square tests, and multivariate linear probability analysis. The aim is simply to search for statistically significant relationships between causal condition candidates and variations in outcome, that may point in the direction of more consistent explanations of what works. While statistical analysis of datasets may seem far removed from the complexities of protection crises in real life, systematic investigations of the best available data is the only viable way forward to start building generalizable explanations to inform what to do in different situations.

Operationalization of causal condition candidates This chapter revisits briefly the four causal condition candidates for successful protection outcomes drawn from existing literature—deterrent presence, Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0004

90

EMPIRICAL PATTERNS OF UN MILITARY PROTECTION OPERATIONS

willingness to accept risk, pre-emption, and matching—explaining how they are operationalized for the statistical analyses.

Deterrent presence I have identified the number of UN troops present in the country in each case to investigate whether the presence of many uniformed personnel is related to more successful protection outcomes, perhaps pointing toward an overall deterrent effect. We know that large uniformed UN deployments have significant conflict-reducing effects—both at the country and local level—but it remains unknown whether this effect is reflected in the outcomes of actual protection operations (Di Salvatore and Ruggeri 2017; Hultman, Kathman, and Shannon 2019; Phayal and Prins 2020). Earlier, I discussed how countrylevel data on troop deployment can only indicate possible correlations, as it does not capture local variations in each case. As such, I cannot fully test the “deterrence by numbers” hypothesis. Ideally, the troop-to-population ratio should build on data on how many civilians were under threat in each case, and how many UN troops were used in that particular operation. Unfortunately, openly available data does not provide such granularity. Another angle would be to explore troop numbers related to the size of territory UN troops are responsible for in-country. However, I found it impossible to provide consistent and accurate estimations of the size of operational areas, as the maps provided by the UN only pinpoint camp locations, rather than the actual areas of responsibility. To make consistent comparisons of troop-to-population ratios across countries, missions, and time, I rather relate to the total population number of each country and the number of troops deployed in the country as each case occurred. I rely on population numbers from the World Bank and monthly deployment numbers of uniformed personnel published by the UN (United Nations 2021b; World Bank 2018). In Chapter 5, I add to this analytical approach using Qualitative Comparative Analysis (QCA)—based on set-theory—to move beyond statistical correlations and point toward potential causal relationships of sufficiency and necessity. The ratios range from 22 inhabitants per uniformed peacekeeper in Abyei to 5,649 inhabitants per uniformed peacekeeper in the early days of MONUC in the DRC. In some cases, this ratio is somewhat misleading since UN troop deployments are concentrated in specific areas of a country. For example, the majority of MONUSCO’s troops are concentrated in the eastern DRC. In

OPERATIONALIZ ATION OF CAUSAL CONDITION CANDIDATES

91

South Sudan, however, Blue Helmets are more evenly distributed throughout the country. Although not fully satisfactory, this analytical approach facilitates a systematic investigation of whether “good” troop-to-population ratios at the macro-level are correlated with better outcomes at the tactical and operational level across time and missions. I would use caution, though, when interpreting these findings, as the analytical approach holds significant shortcomings.

Willingness to accept risk I have coded all troop-contributing countries involved in each case according to their willingness to accept risk in using force to protect civilians from violence. There is great variation in troop-contributing countries’ policies toward—and acceptance of the use of—force in peacekeeping operations (Providing for Peacekeeping 2020). However, some troop contributors are less risk-averse than others. I have not been able to develop a finely grained ranking of all contributors involved in the events captured in UNPOCO, but rather determined which countries display few caveats and a positive attitude toward the UN’s expanding peacekeeping agenda, including the use of force to protect civilians. These troop contributors are termed as “willing” (scored one (1) in the cross-tabulation analysis), while all others are termed “hesitant” (scored zero (0)). In those cases where several troop contributors are involved—including one or more “willing” contributors—I have scored the constellation as 0.5, i.e., not fully “willing,” but more willing than those constellations deemed hesitant. This third category is meant to capture an expected positive contribution by the involvement of troops that are less risk averse. The scores are based on existing literature, official national policies, statements in the UN General Assembly, and expert opinions on how they tend to operate on the ground (Bellamy and Williams 2013; Chesterman 2004; Providing for Peacekeeping 2020; United Nations 2010a, 2015b). While this approach does not capture all potentially relevant variables about each troop contributor’s willingness to accept risk, it distinguishes between countries on one of the most important aspects and facilitates comparisons across cases.

Pre-emption I distinguish between cases where UN troops were more forward-leaning— seeking to pre-empt attacks on civilians—and more passive approaches, where UN troops reacted to situations where attacks are already underway. Each case

92

EMPIRICAL PATTERNS OF UN MILITARY PROTECTION OPERATIONS

is coded as either “pre-emptive” (scored one (1) in the cross-tabulation analysis) or “reactive” (scored zero (0)). This condition is quite straightforward to operationalize, but real-life situations are certainly not always clear-cut. In some cases, UN troops responded pre-emptively against perpetrators that had repeatedly attacked civilians in the not-so-distant past (and where there was no UN intervention). It can be challenging to determine whether the case is in fact a response to former exactions—in other words a reactive operation—or a pre-emptive effort to protect.

Matching Finally, I investigate whether the UN troops’ use of force matched the ways and means with which the perpetrators used violence against civilians. This condition has been described in detail in Chapter 2. It has been operationalized by first ascribing one or more functions of force to the protector in each case and then assessing the type of violence committed against civilians by the perpetrator, before comparing the two to evaluate whether the use of force matches the violence by the perpetrator. Those cases where matching occurred are coded as one (1), and, conversely, those cases where the use of force was a mismatch received the score zero (0). For example, on January 6, 2013, “a MONUSCO armored personnel carrier based at the mobile operating base in Mambasa–Wikipedia supported FARDC with heavy machine gun fire and jointly pushed back the several hundred Mayi-Mayi Simba combatants who had entered Mambasa town the previous day and caused FARDC to temporarily withdraw from the town […]” (United Nations 2013a, para. 41). In this case, an imminent threat to civilians was coerced by robust operations, matching the coercive violence employed by the perpetrators of violence. The perpetrator was quickly removed, although the Mayi-Mayi Simba had remained in the village for a day before being dislodged by FARDC and MONUSCO. If they had stayed on for longer, it is likely that many more civilians would have been threatened, not least because the FARDC had previously held the town, which could have led to reprisals against perceived collaborators. In another case from the DRC, in January 2012, “an estimated 45 people were killed by suspected FDLR combatants and approximately 2,700 persons were internally displaced during attacks on several remote villages in Shabunda territory, South Kivu. On January 5 MONUSCO conducted a reconnaissance mission to the area, and on January 6 the Mission established a

OPERATIONALIZ ATION OF CAUSAL CONDITION CANDIDATES

93

mobile base at Lubimbe and dispatched patrols to other affected areas” (United Nations 2012a, para. 24). In this case, the UN response merely sought to ameliorate the situation, mismatching the coercive and destructive use of force by the suspected FDLR combatants. It has been challenging to determine which functions of force were employed by UN troops and what type of violence the perpetrators used. Many potential cases were left out of the analysis due to lack of information on how force/violence was applied. In addition, in some cases, more than one function of force or violence was in play. In that type of case, I have only related to the most violent function of force or violence to evaluate whether matching occurred or not.

The outcome Finally, each case is coded on the outcome, as either “many” (1) or “few” (0) civilians protected. The coding rests on a qualitative estimate of the outcome of each case, guided by counterfactual reasoning, knowledge about the perpetrators of violence, and the UN military efforts to protect. I also rely on the generic scenarios developed by Beadle, which all draw on existing knowledge about the modus operandi of similar types of perpetrators (see Chapter 2). Accordingly, I consider the modus operandi of the perpetrators of violence to indicate what is likely to have occurred without the UN response, in other words, the expected outcome without intervention. Then, I hold this expectation up against the actual outcome of each case. These evaluations remain approximate and should be read in that light. Due to the poor consistency in data quality across cases, I have only been able to use 126 cases of the 200 captured in UNPOCO to perform the analysis in this chapter. This implies that the analyses do not rest on a random sample and can only be used to indicate interesting data patterns. And although pointing in similar directions, the analyses provide quite different interpretations of the data. Furthermore, judging by what I know about this empirical universe, the results do not fully capture the complex causal relationships that characterize this phenomenon. For example, although matching appears to be the most interesting condition, I know that matching also has occurred on many occasions when UN troops failed to protect. Similarly, although pre-emption seems like an important part of the puzzle, I know that UN troops also succeeded in protecting when they responded after civilians were attacked. Similar inconsistencies also appear in UNPOCO regarding troop-to-population ratios, where UN troops have succeeded even though they were few in numbers. Given these

94

EMPIRICAL PATTERNS OF UN MILITARY PROTECTION OPERATIONS

complex causal patterns and empirical inconsistencies there is a need to pursue answers with the help of other methods, which I do in Chapters 5, 6, and 7. The cross-tabulations do not inform us about possible interaction effects between the causal conditions. I therefore add a multivariate linear probability analysis, which investigates if any of the four causal conditions stand out as particularly relevant for positive outcomes when accounting for the simultaneous presence of the other conditions (which also may influence the outcome). From the multivariate analysis, it appears that matching and pre-emption again stand out as the most relevant conditions. In fact, troop-topopulation ratios no longer appear relevant, while willingness to accept risk portray similar—non-significant—results as in the cross-tabulations.

Bivariate cross-tabulations In this section, each condition is cross-tabulated with variations in the outcome before the results are tested for statistical significance with the help of Chi-square tests. This step is helpful to study the relationship between categorical conditions and various outcomes. Three of the conditions (matching, willingness to accept risk, and pre-emption) are already coded with categorical values (1 or 0)—as is the outcome—which facilitates such cross-tabulations. Deterrent presence (troop ratios and absolute troop numbers) is not, and therefore needs to be, recoded in categories. To search for general trends in the data, I decided to group troop ratios into three categories and absolute troop numbers into five categories. A more finely grained categorization would possibly provide more detail but would also make the cross-tabulations rather unwieldy, countering the aim of this analytical step. Keep in mind that the data is a sub-set of cases from UNPOCO—selected based on sufficient quality of data—and does not fulfill the requirements of a random sample from a larger population of cases. The results can therefore only provide indications of possibly interesting relationships. Statistical Package for the Social Sciences software—or only SPSS—was used to calculate the cross-tabulations and perform robustness tests of the results.

Absolute troop numbers To facilitate the cross-tabulations of absolute troop numbers and the outcomes (see Table 4.1), I have grouped different troop numbers into five categories:

BIVARIATE CROSS-TABUL ATIONS

• Category deployed • Category deployed • Category deployed • Category deployed • Category deployed

95

1: cases with fewer than 4,999 uniformed peacekeepers 2: cases with 5,000 to 9,999 uniformed peacekeepers 3: cases with 10,000 to 14,999 uniformed peacekeepers 4: cases with 15,000 to 19,999 uniformed peacekeepers 5: cases with 20,000 to 24,999 uniformed peacekeepers

According to the analysis, the relationship between absolute troop numbers and outcomes is not significant (see Table 4.2). In categories 3 to 5—cases where 10,000 to 24,999 uniformed peacekeepers were deployed—the combined 88 cases portray an almost even distribution between successes (many protected) and failures (few protected). Somewhat surprisingly, categories 1 and 2—where fewer than 10,000 uniformed personnel were deployed— portray a better distribution between successes and failures, but this is not enough to make the relationship between troop numbers and outcomes significant. One reason for the high frequency of successes for categories 1 and Table 4.1 Cross-tabulation between absolute troop numbers and outcomes

Table 4.2 Chi-square significance test of absolute troop numbers

a

0 cells (0.0%) have expected count less than 5. The minimum expected count is 7.56.

96

EMPIRICAL PATTERNS OF UN MILITARY PROTECTION OPERATIONS

2 could be that more than half of those 38 cases involve operations in Abyei, where UNISFA has been quite effective in protecting civilians. Nevertheless, these patterns also indicate that large, uniformed components country-wide do not necessarily translate to a higher frequency of successful protection outcomes at the tactical and operational levels.

Troop-to-population ratios A qualitatively different aspect of the deterrent presence of UN uniformed personnel—rather than absolute troop numbers—is the ratio of troops to inhabitants. To facilitate the cross-tabulations, I have grouped troop-topopulation ratios into three categories: • Category 1: cases with fewer than 1,999 inhabitants per uniformed peacekeeper • Category 2: cases with 2,000 to 3,999 inhabitants per uniformed peacekeeper • Category 3: cases with 4,000 to 5,999 inhabitants per uniformed peacekeeper According to the cross-tabulation and the significance test in Tables 4.3 and 4.4, there is a significant relationship between troop-to-population ratios and the outcome across cases. There is a 0.6 percent chance that this distribution is random (Table 4.4). In the 65 situations with the “best” troop-to-population Table 4.3 Cross-tabulation between troop-to-population ratios and outcomes

BIVARIATE CROSS-TABUL ATIONS

97

Table 4.4 Chi-square significance test of troop-to-population ratios

a

0 cells (0.0%) have expected count less than 5. The minimum expected count is 7.11.

ratio—with fewer than 1,999 inhabitants per uniformed peacekeeper (category 1)—we see successful outcomes in 61.5 percent of the cases. I find an almost identical distribution of successes and failures for the 45 cases coded as Category 2 (2,000 to 3,999 inhabitants per uniformed peacekeeper), with 60 percent successful outcomes. Finally, the 16 cases with the “worst” troopto-population ratios—ranging from 4,000 to 5,999 inhabitants per uniformed peacekeeper—saw as few as 18.8 percent successful outcomes. Thus, troop numbers may still matter after all, and there are good reasons to investigate the condition further, alongside the other potential causal conditions.

Willingness to accept risk There is a significant relationship between willingness to accept risk and outcomes of operations. There is only 1 percent chance of the distribution portrayed in Table 4.5 being a random occurrence in the data (Table 4.6). However, the distribution of success frequencies is somewhat unexpected. Of the 54 cases only including troops from countries that are coded as willing to accept risk (category 1), 61.1 percent saw successful outcomes. Conversely, Table 4.5 Cross-tabulation of willingness to accept risk and outcomes

98

EMPIRICAL PATTERNS OF UN MILITARY PROTECTION OPERATIONS Table 4.6 Chi-square significance test of willingness to accept risk

a

0 cells (0.0%) have expected count less than 5. The minimum expected count is 13.78.

the 41 cases including only Troop Contributing Countries (TCCs) that are scored as hesitant (category 0) saw 65.9 percent successful outcomes. Furthermore, 31 cases including at least one troop-contributor willing to accept risk but also including troops from more risk-averse TCCs have only 32.3 percent successful outcomes. This analysis seems to indicate that both willing to accept risk and riskaverse troops have almost similar frequency of successes. There could be underlying conditions that explain this distribution. Perhaps risk-averse troops score rather well because they take on “easier” operations, with better chances for success. However, I find little support for that in the UNPOCO dataset. Nevertheless, statistical significance does not equal causal importance. To further explore these indications, I will also include willingness to accept risk alongside other causal condition candidates in further analyses.

Pre-emption There is also a significant relationship between pre-emptive operations and outcomes. The significance test indicates that there is a very low probability of the distribution of cases being a random result (0.000 percent) (see Table 4.8). Of 43 pre-emptive operations—coded as category 1—86 percent led to successful outcomes (see Table 4.7). When responses were reactive—category Table 4.7 Cross-tabulation of pre-emption and outcomes

BIVARIATE CROSS-TABUL ATIONS

99

Table 4.8 Chi-square significance test of pre-emption

a

0 cells (0.0%) have expected count less than 5. The minimum expected count is 19.11.

0—successes only occurred in 39.8 percent of the cases. There are therefore good reasons to include pre-emption in further investigations of causal conditions leading to successful outcomes, as it may well be an important part of the puzzle.

Matching Finally, there is a significant statistical relationship between matching the perpetrators of violence and outcomes (see Table 4.10). Like pre-emption, there is a very low probability of this distribution of successes and failures being a random pattern (0.000). Of 99 cases where matching occurred, 68 led to successful outcomes (see Table 4.9). This also means that in the remaining 31 cases where matching occurred, few civilians were protected. Conversely, when UN troops mismatched the perpetrators of violence, failures occurred in 92.6 percent of Table 4.9 Cross-tabulation of matching and outcomes

Table 4.10 Chi-square significance test of matching

a

0 cells (0.0%) have expected count less than 5. The minimum expected count is 12.00.

100

EMPIRICAL PATTERNS OF UN MILITARY PROTECTION OPERATIONS

the cases. Although matching seems to be present in almost every successful outcome, it also appears often when UN troops fail to protect. Emerging as a potentially relevant building block for successful protection operations, this condition will also be pursued with the help of other methods in later chapters.

Summary of cross-tabulations To sum up, I find no significant statistical relationship between absolute troop numbers and outcomes of operations. This is interesting. It indicates that the overall conflict-reducing effect of the deterrent presence of large, uniformed contributions—one of the strongest findings from quantitative studies of UN peacekeeping deployments—does not necessarily impact on how uniformed components fare at the tactical and operational levels against particular perpetrators. These two phenomena can of course co-exist. We already know that civilians continue to be targeted even in the presence of large UN missions. We also know that UN troops seldom apply force to protect, and that they often struggle to deal with violence that occurs in many conflict areas where these missions deploy. Importantly, this initial analysis indicates that—alongside being present in large enough numbers—other conditions must determine when and how UN troops protect civilians effectively from different types of threats. However, troop numbers may still matter. In fact, I find that troop-topopulation ratios do have a significant relationship to the outcomes. Where troop-to-population ratios are “good,” we also see better outcomes. While this condition is merely a different way to investigate the potential deterrent effect of troop numbers, it does introduce a qualitative difference. Furthermore, although willingness to accept risk does have a significant relationship to outcomes, the distribution of success frequencies possibly indicates that this is a trivial relationship. Willingness to accept risk and riskaversion lead to a similar distribution of successes and failures. Again, it could indicate that, when force is applied, it matters more what troops do to protect, rather than where they are from. This observation finds support in the analysis of pre-emption and matching, as both portray significant relationships with the outcome, and both seem to lead toward better outcomes. However, matching also occurs quite often when UN troops fail to protect. There is a need to investigate these conditions further. These results must be treated with caution. Although they do reflect general patterns in the UNPOCO data, it is not possible to use them to infer causal

MULTIVARIATE LINE AR PROBABILIT Y ANALYSIS

101

relations. Ideally, I should have been able to control for potential confounding variables that may influence these relationships. For example, I have not controlled for type and intensity of conflict, variation in military training and equipment of UN troops, and so forth. These ideas are indeed candidates for any future revisions of the UNPOCO dataset and analyses. However, it may also be that the four conditions chosen for the analyses in the present study mutually influence each other and their effect on the outcome. In the next analytical step, I will therefore control for these potential effects with the help of a multivariate linear probability model.

Multivariate linear probability analysis Moving beyond the bivariate analysis presented in the last section, I will now explore the impact of the conditions on the outcome in a multivariate statistical analysis. More specifically, I investigate if any of the four causal conditions stand out as particularly relevant for positive outcomes when accounting for the simultaneous presence of the other conditions (which also may influence the outcome). When the dependent variable is dichotomous—which is the case for the outcome condition in the analysis performed here—two models can be used: a logistic regression analysis or a linear probability model (LPM). I choose to present the results of the latter, while I have performed a robustness test with the help of a logistic regression analysis. Results from both analyses point in the same direction, where matching and pre-emption stand out as the two most promising conditions for further analysis, supporting their potential relevance emerging from the cross-tabulations. Table 4.11 portrays the results of the linear probability analysis. In short, the analysis calculates the impact of an increase in the independent variables on the probability of seeing a favorable outcome—many civilians protected (scored 1)—when the value of a condition increases by one, while at the same time keeping all other conditions’ values constant. This way, we can measure the additive effect of an increase in each condition when controlled for the presence of other effects. The results can be interpreted as percentages, describing how the probability of the outcome increases in a particular condition. Based on the adjusted R2 value—which is a statistical measure of how close the data are to the fitted regression line—the model explains about 42 percent of the variation in outcomes. While this result is quite decent for most models, I am more interested in how the independent conditions fare given the presence of others.

102

EMPIRICAL PATTERNS OF UN MILITARY PROTECTION OPERATIONS Table 4.11 Linear probability model

Note: ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

Two conditions stand out as particularly relevant. When UN troops match the perpetrators of violence, and all other conditions are kept constant, there is a 56 percent increase in the chance of a seeing a positive outcome compared to when they do not. When UN troops seek to pre-empt attacks on civilians, there is a 30 percent increase in the chance of seeing a good protection outcome. Both these relationships are statistically significant, as there is a slim chance that this is the result of random occurrences in the data. Keep in mind the standard error—in parentheses in Table 4.11—which indicates that the true estimate is situated somewhere between the coefficient and the +/- standard error value. Conversely, an increase in troop-numbers—in both absolute numbers and better troop-to-population ratios—does not increase the chances of success. None of these conditions portrays a significant relationship with the outcome. The analysis of TCC’s willingness to accept risk is more interesting, while it also echoes the findings from the cross-tabulations. The coefficient suggests that protection operations performed by a mix of troops from risk-averse TCCs and troops that are more willing to accept risk (medium risk) are less likely to lead to a successful outcome than operations performed by risk-averse

MULTIVARIATE LINE AR PROBABILIT Y ANALYSIS

103

troops only. The result is statistically significant. Meanwhile, protection operations that only involved troops that are willing to accept risk (high risk) are also less likely to lead to a successful outcome than operations performed by risk-averse troops only. This relationship does not portray statistical significance. This result suggests a U-shape relationship. Operations involving troops that are risk averse and operations that involved troops more accepting to risk are more successful than operations that performed by a combination of both types of troops. A possible explanation for this counterintuitive result is that high-risk troop contributors select the most difficult and demanding operations. In the cross-tabulations I alluded to this as a possible explanation, but I have not been able to discern such systematic patterns from UNPOCO. A potential future iteration could take into account the selection bias of which troops performs which type of operation. This can be done by controlling for the conflict environment, the intensity of the conflict prior to deployment, and the strength and size of armed groups. The purpose of the analytical steps in this section was to search for each condition’s effect on the outcome while controlling for the simultaneous presence of the other conditions. The results indicate that matching and pre-emption remain the two most interesting conditions, supporting the findings from the cross-tabulations. Troop numbers do not emerge as particularly relevant, while willingness to accept risk portrays interesting patterns, which will be studied more closely in the following chapters. While interesting statistical relationships were identified, the analysis indicates a phenomenon marked by causal complexity. There is a need for other methods that will better handle analyses of complex social phenomena, exploring different combinations of conditions leading to successful outcomes across time and place.

5 Discovering causal pathways to successful protection outcomes “We need to think differently about how we use military force in contemporary operations. Get rid of the idea that force is only about defeating an enemy to defend state boundaries or to maintain alliance cohesion. While these tasks are always important, we keep failing to protect people wherever we deploy, which has had disastrous effects on the outcome of our operations from the Balkans, via Afghanistan, to South Sudan. Protecting people from those who harm them is also a moral imperative we can never escape as military professionals.”¹ —Colonel (R) Petter H. F. Lindqvist Norwegian Armed Forces Former Military Chief of Staff in the UN mission in South Sudan

Moving toward insights that respond to Colonel Lindqvist’s call for thinking anew about the utility of force in wars amongst the people, this chapter explores combinations of conditions—often called causal pathways or recipes—that may systematically explain variations in outcomes of UN military protection operations across time and UN missions. I apply fuzzy set Qualitative Comparative Analysis (fsQCA) to the same sub-set of 126 cases from the UNPOCO dataset as was used for the statistical analyses in Chapter 4. However, each case is now operationalized and scored according to standard principles for fsQCA (Schneider and Wagemann 2012). We are now moving from statistical to set-theoretical approaches, which are driven by a different logic (Goertz and Mahoney 2012; Ragin 2013). Before getting started on the operationalization and analysis of data, I will provide the necessary building blocks along the way to help us understand how fsQCA is designed and how I employ the method on UNPOCO data. The cases are studied along the same four conditions frequently identified as causally relevant in existing literature: deterrent presence, willingness to accept risk, pre-emption, and matching. In addition, each case is coded ¹ Interview with Colonel Petter H.F. Lindqvist, Oslo, April 22, 2022.

Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0005

QUALITATIVE COMPARATIVE ANALYSIS

105

on the outcome (see also Chapter 3 for more details on how outcomes are assessed). A quick reminder: the main criteria for case selection have simply been sufficient depth and quality of information. Although this approach to case selection is not ideal, it remains the only available option if we lack openly available systematic reporting of all potential cases of interest. A few words of caution are still necessary. The selection bias toward events that contain sufficient depth of reported information possibly favors “larger” events, i.e., those that are most likely to make headlines due to many casualties or a significant participation of UN troops. My selection method potentially skews the analysis toward the more violent end of the spectrum. Certainly, UN missions are often judged by how they fare in the most extreme of cases (Annan 1999a; Carlsson, Han, and Kupolati 1999; Center for Civilians in Conflict 2016). Some of the day-to-day operations that aim to protect civilians are likely left out of official reporting, since there is scant information about how they are conducted and little analysis by UN missions on what effect they have had on civilian security. My inability to study every case of interest may systematically weaken the overall insights, as interesting causal pathways toward successful outcomes at the lower end of the spectrum of violence may remain undisclosed. However, with the help of UNPOCO, I have captured many cases that have not been studied systematically in the past, hopefully providing valuable new insights.

Qualitative Comparative Analysis Rather than isolating the additive effect of independent variables on a dependent variable—often the aim in quantitative approaches—QCA uses Boolean algebra to discover potential causal pathways—recipes—that are combinations of causal conditions either necessary or sufficient for an outcome. To arrive at such recipes, QCA portrays “each case as a combination of causal and outcome conditions. These combinations can be compared with each other and then logically simplified through a bottom-up process of paired comparison” (Ragin, Drass, and Davey 2006). The logical comparison is performed with the help of software, allowing more cases to be compared than is traditionally done in qualitative comparative research designs (George and Bennett 2005; Schneider and Wagemann 2012). The results of these paired comparisons are displayed by the software as a “truth table,” where each row “denotes a qualitatively different combination of conditions, i.e. […] a difference in kind rather than difference in degree” (Schneider and Wagemann 2012, 92).

106

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

In practical terms, I score the relevant cases’ membership in sets. Membership scores come in two forms: either “crisp” (either 0.0 or 1.0), or “fuzzy” (somewhere along the scale between 0.0 and 1.0). The fuzziness does not imply lack of clarity. It means that it “permits [set] membership in the interval between 0 and 1 while retaining the two qualitative states of full membership and full non-membership” (Ragin, Drass, and Davey 2006). In contrast, a traditional dichotomous—crisp—set can only portray a two-value variable (either “in” or “out” of a set, scored 1 or 0). Crisp and fuzzy set values can be used interchangeably in QCA research designs, but it is generally recommended to use fuzzy sets when existing data allows for meaningful nuancing. All scores—the qualitative anchors—must be based on existing theory and in-depth knowledge about the cases. This is where the qualitative aspect of QCA comes to the fore. The most critical qualitative anchor is the cut-off point (0.5), which decides whether a case is more in than out of a set. When all cases have been scored on all conditions and the outcome, QCA-software facilitates the investigation of necessary conditions and conditions that in combination may be sufficient for the outcome. A condition is necessary if the outcome cannot occur without it. The outcome is therefore a sub-set of the condition. However, a necessary condition alone is most often not enough to produce the outcome. A condition is sufficient if the outcome always occurs when it is present. In addition, there may be other sufficient conditions that can also produce the outcome. The condition is therefore a sub-set of the outcome (Legewie 2013). As shown in Figure 5.1, Venn diagrams can portray the logic of sufficiency and necessity.

Condition 1 Outcome

Condition

Necessary condition

Condition 2

Outcome

Sufficient conditions

Figure 5.1 Venn diagrams portraying the logic of sufficiency and necessity.

THE OUTCOME

107

In this chapter, I use QCA to investigate whether any of the four causal condition candidates—alone or in some combination—emerge as necessary and/or sufficient for the outcomes. Although the analyses provide interesting insights—with the combination of matching and pre-emption seemingly able to explain about half of the successful outcomes—I am not able to systematically explain successful outcomes for the remaining half. While this is a quite decent result when seeking to explain such a complex phenomenon across time and place, there is a need to dig deeper in particular cases to search for other conditions that might explain how and when protection operations succeed and fail. In the two following chapters, I do just that, adding additional insights. QCA demands a high degree of transparency about how conditions and outcomes are defined and operationalized. Fundamentally, transparency facilitates critique, potential replication, and modification of future research designs. QCA puts particular emphasis on this step of the research process, even more so than most qualitative research strategies. Being fully transparent about how conditions are coded and scored can also help specify and refine theories. In the following, I explain the scores of the outcome and the four causal conditions, before presenting the analysis and results of the analyses.

The outcome The outcome variable—or just the outcome in QCA-language—estimates degrees of success of UN military protection operations. The cases are now operationalized as a fuzzy set, rather than a dichotomous—crisp—set, as used in the statistical analysis. The outcomes include cases at the extreme ends of the scale as well as two “fuzzy” variations in between. The extreme ends—full successes and complete failures—are quite straightforward to determine. No civilians were killed or harmed in the first category, and no one was protected in the latter. Partial successes and failures are more challenging to measure and code. I have applied a combination of counterfactual reasoning, my understanding of the modus operandi of the specific perpetrator in each case, the threat scenario categories developed by Beadle (see Chapter 2), and casespecific knowledge to score variation in the outcomes. Still, it is only possible to broadly estimate how outcomes vary, and my findings should be read in that light. I have asked of each case: what is likely to have happened to civilians under threat without the military intervention of the UN troops? A brief revisit of the

108

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

basic idea underpinning counterfactual reasoning tells us that “the meaning of causal claims can be explained in terms of counterfactual conditionals of the form ‘If A had not occurred, C would not have occurred’” (Menzies 2014). Counterfactual reasoning rests on alternative outcomes—possible worlds— that must resemble the actual event as closely as possible. When introducing a counterfactual condition, it should make minimal change to the real situation, and must therefore be close in time to where the real world and the counterfactual possible world branched off (Mahoney 2021). I try to establish a counterfactual possible world where the UN did not intervene, to be able to analyze what effect the intervention had in real life. I try to minimize the changes to the real world, by only removing one counterfactual condition, i.e., the UN’s military effort to protect. I then compare the possible world with the actual outcome following a UN intervention, leading to an analysis of whether few or many civilians were protected in each case. I do not attempt to explore longer-term effects of the protection operation, which would potentially undermine the value of the counterfactual reasoning by introducing second-order effects. Ideally, the number of civilian casualties should have been compared with the total number of civilians under threat in each case. However, this information is not readily available. More commonly, the available information provides fatality figures, although this is also often missing from official UN reporting. A partial protection success could therefore see 10 civilians killed if we can plausibly expect that the perpetrators commonly kill and harm civilians in the dozens or hundreds. In many other situations, 10 civilians killed would be partial or full failure. Outcomes are scored as “everyone protected” (1.0) in cases where UN peacekeepers protected all potential victims in a specific area at a certain time. One way to achieve this is to deny a perpetrator access to a contested area, by using military force. For example, the former UN/African Union (AU) hybrid operation in Darfur (UNAMID) has on several occasions successfully intervened to prevent armed militias entering camps for internally displaced persons (United Nations 2014d, 2016a, 2017d). In these cases, the militias were deterred, withdrew, and no one was harmed. Armed communal militias entering camps where unarmed civilians from the opposing side reside represent a direct and imminent threat to civilians. Based on the modus operandi of the communal militias in this conflict, it is highly likely that this situation could have led to many civilian casualties, if left alone. Outcomes are coded and scored as “many protected” (0.75) when UN troops used force to protect quite effectively, although some civilians were still

THE OUTCOME

109

killed and/or harmed. For example, on July 6, 2013, more than 30 armed and predatory Mai-Mai elements in civilian clothes attacked M23 elements in Kanyaruchinya, close to Goma in the Democratic Republic of the Congo (DRC), while also firing at the local population, killing one person. In response, MONUSCO engaged the Mai-Mai elements, killing one, injuring two, and arresting another (United Nations 2013e). This situation, if left alone, could easily have led to more than one civilian casualty, both because of the indirect threat to civilians from a possible firefight between the Mai-Mai and the M23, but also due to the sheer number of predatory armed fighters in the vicinity of civilians, on whom they often prey for profit and survival. Outcomes are scored as “few protected” (0.25) when UN troops used force to protect, but many civilians were still killed or harmed. For example, in July 2005, to protect civilians, the UN mission in the DRC (MONUC) and the national armed forces (FARDC) ran a series of joint military operations in South Kivu to obstruct the movement of the Democratic Forces for the Liberation of Rwanda (FDLR). After warning the FDLR combatants to leave, MONUC and FARDC destroyed several FDLR camps (United Nations 2005). The rationale was to force FDLR fighters to relocate to areas where they would pose less of a threat to the civilian population. Despite robust joint operations, unidentified armed elements attacked the village of Ntulamamba—west of Bukavu in South Kivu—close to where these joint UN/FARDC operations were being conducted. A verification mission by MONUC found that some 47 persons, mostly women and children, had been killed ((United Nations 2005). The FDLR denied involvement. In this case, the joint operations against FDLR bases were likely to have triggered revenge attacks inflicting many casualties, even when compared to the typical modus operandi of FDLR. This case was thus coded as a situation where few were protected. Outcomes are assessed as “no one protected” (0.0) in cases where UN forces have failed to protect victims in a specific area at a certain time, despite having intervened militarily. In many cases, such failures occur because UN forces are unable to respond in time or use tactics that have no effect on civilian security. For example, in 2010, “at least 387 civilians, including 300 women, 23 men, 55 girls and 9 boys, were raped by a coalition of combatants from the Democratic Forces for the Liberation of Rwanda and the Mai-Mai Sheka, as well by residual elements of Lieutenant Colonel Emmanuel Nsengiyumva” (MONUSCO 2011, 1). In addition, almost a thousand houses and several shops were looted, and more than one hundred civilians were subjected to forced labor ((MONUSCO 2011, 1). The UN troops stationed nearby responded only after two days of attacks ((MONUSCO 2011, 1). Even then,

110

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

they were not able to identify the ongoing attacks, as most of the population had fled to find shelter in the bush. The UN troops did not link the deserted villages to the presence of violent perpetrators. The presence of Blue Helmets in this situation did nothing to improve civilian security. Table 5.1 shows that 21 out of 126 operations were scored as “everyone protected,” 49 operations were scored as “many protected,” 46 operations were scored as “few protected” and 10 operations were scored as “no-one protected.” Although UN troops had more positive outcomes (70) than negative (56) according to these numbers, it is not possible to suggest that protection operations have been successful more often than not. The cases are selected based on the quality of information. Table 5.1 Estimated outcomes of 126 UN military protection operations in Africa, 1999–2017

Total

Number of operations

Everyone protected

Many protected

Few protected

No-one protected

126 (100 %)

21 (16.7 %)

49 (38.9 %)

46 (36.5 %)

10 (7.9 %)

Table 5.2 lists the cases in the set according to scored outcome, including their fuzzy scores, their qualitative description, the number of cases, and the case IDs from the UNPOCO dataset. For more details on each case, refer to the UNPOCO dataset (Kjeksrud 2019).

Causal condition candidates In this section, I operationalize the four causal conditions candidates according to standard QCA procedures. I have provided each condition with a short form in parentheses to ease identification in the datasets and tables, since the QCA software used for the analysis only allows lower letters and no space between words.

Deterrent presence (deter) Troop-to-population ratios are used as a proxy measurement of the deterrent presence of the uniformed component of UN peace operations. They are measured using monthly data on uniformed UN deployments, in combination with data on the national population size from the UN and the World Bank (United Nations 2021b; United Nations Department of Economic and Social

Table 5.2 Outcome calibrations with fuzzy scores, qualitative descriptions, number of cases, and case IDs from UNPOCO Fuzzy score

Description

No. of cases

Case IDs from UNPOCO

1.0

Everyone protected

21

UNMIL Liberia4

UNISFA Abyei2 Abyei6 Abyei8 Abyei11 Abyei12 Abyei20

UNMISS SouthSudan3

MINUSMA Mali2 Mali5 Mali6

MONUSCO DRC106

MINUSCA CAR1 CAR6

UNMIL Liberia1 Liberia3 Liberia5 Liberia6 Liberia8 Liberia15

UNISFA Abyei1 Abyei3 Abyei4 Abyei5 Abyei7 Abyei9 Abyei18

UNMISS SouthSudan1 SouthSudan13

0.75

Many protected

49

UNAMID Darfur10 Darfur11 Darfur13 Darfur14 Darfur15 Darfur16 Darfur17

UNAMID Darfur1 Darfur5

Continued

Table 5.2 Continued Fuzzy score

0.25

Description

Few protected

No. of cases

46

Case IDs from UNPOCO MINUSMA Mali4 Mali9

MONUC DRC1 DRC5 DRC8 DRC12 DRC14 DRC15

UNMIS Sudan1 Sudan2

UNOCI IvoryCoast1 IvoryCoast2 IvoryCoast3

UNMIL Liberia7 Liberia9 Liberia10

UNAMSIL SierraLeone1

DRC16 DRC17 DRC21 DRC22 DRC24 DRC26

MONUSCO DRC28 DRC45 DRC47 DRC51 DRC53 DRC54

UNMISS SouthSudan5 SouthSudan6 SouthSudan11 SouthSudan12

DRC55 DRC56 DRC61 DRC64 DRC69 DRC85

MINUSCA CAR7

UNAMID Darfur2 Darfur4 Darfur6 Darfur7 Darfur9

MINUSMA Mali1

MONUC DRC2 DRC3 DRC4 DRC10 DRC13

DRC19 DRC23 DRC25

MONUSCO DRC44 DRC49 DRC57 DRC63 DRC66 DRC68 DRC70 DRC71 DRC72 DRC74 DRC76

UNOCI IvoryCoast6 0.0

No-one protected

10

UNISFA Abyei10 Abyei13 Abyei14 Abyei15 Abyei16 Abyei19 Abyei21

MONUSCO DRC27

UNMISS SouthSudan10 SouthSudan15

DRC77 DRC79 DRC81 DRC82 DRC83 DRC88 DRC91 DRC99 DRC102 DRC103

MINUSCA CAR4 CAR8

114

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

Affairs 2016; World Bank 2018). The data on uniformed UN deployments are identical to the data used by recent studies to investigate how the size of UN operations correlates with civilian casualty figures (Hultman, Kathman, and Shannon 2019). Some of the data on population sizes rely on forecasts of population growth, as accurate censuses do not exist for all mission areas (see Table A2 for references). Actual deployment numbers in specified operational areas and local population numbers would have captured this causal condition candidate more precisely. Such data is not easily available across all cases, and my analysis remains a somewhat crude measurement. Some promising research efforts “go local” by capturing a more refined picture of peacekeeping deployments, but they remain limited in the number of cases studied (Autesserre 2014; Ruggeri, Dorussen, and Gizelis 2017, 2018). The anchor-points for membership scores in this set are calibrated according to existing theories on troop-to-population ratios, adapted to a UN setting. Although the idea of determining ratios to explain success or failure in peace and stabilization operations has been criticized, it is included here for two reasons. First, the most compelling research on the positive effect of UN peacekeeping operations suggests that a decrease in civilian targeting is correlated to the number of uniformed peacekeepers deployed (Hultman, Kathman, and Shannon 2019). My statistical analysis failed to find a significant relationship between the size of uniformed components and effective protection operations at the operational and tactical levels, while good troop-to-population ratios did show a significant relationship with the outcomes. In the QCA analysis, I aim to investigate this aspect further, adding more nuance by investigating this aspect in combination with other conditions with the help of set-theory. Second, since UN reform efforts also aim to increase the number of troops available for deployment, there is a need to explore whether there are thresholds that troop-to-population ratios should reach to be likely to have a systematic positive effect on the outcome of protection operations. Table 5.3 captures the scores of my operationalization of cases. Full membership of this set—“fully in” (1.0)—is assigned to cases with a troop-to population ratio better than 1:100, which is half the amount of troops needed for successful outcomes as ascribed by Quinlivan (1:50) (Quinlivan 1995). The reason for this modification is that UN operations are almost never set up for combat operations. They operate with the consent of host authorities and do so impartially. Consequently, peacekeeping should require fewer troops than counterinsurgency operations. The UN mission in Abyei—UNISFA—is the only example from UNPOCO where this ratio has been achieved.

Table 5.3 Troop-to-population ratio calibrations with fuzzy scores, descriptions, number of cases, ratio thresholds, and case IDs from UNPOCO Fuzzy score 1.0

Term Fully in

No. 20

Threshold 1:100 1:500 1:1,000

UNMISS SouthSudan5-6 UNMIL Liberia1 Liberia15 MONUC DRC1-5 DRC8 DRC10 DRC12 DRC13-17 DRC19 DRC21-26

UNMISS SouthSudan1

UNMIL Liberia3-10

UNAMID Darfur1-2 Darfur4-7 Darfur9-11 Darfur13-17

CAR CAR4 CAR6-8

UNMISS SouthSudan3 SouthSudan10-13 SouthSudan15 MONUSCO DRC74 DRC27-28 DRC76-77 DRC44-45 DRC79 DRC47 DRC81-83 DRC49 DRC85 DRC51 DRC 88 DRC53-57 DRC91 DRC61 DRC99 DRC63-64 DRC102-103 DRC66 DRC106 DRC68-72

UNOCI IvoryCoast6

CAR CAR1

UNOCI IvoryCoast1-3

UNMIS Sudan1-2

MINUSMA Mali1-2 Mali4-6 Mali9

116

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

Partial membership of this set is assigned to cases that fall between the 1:100 troop-to-population ratio and the cut-off point at 1:500. They receive the “mostly in” score of 0.75. The cut-off point is determined by the ratio suggested by Goode (1:357), although slightly increased to better reflect the fact that UN forces mostly operate in non-combat environments, guided by the bedrock principles of peacekeeping (Goode 2009). The cases captured in UNPOCO with this ratio are found in Liberia, Sierra Leone, Darfur, and CAR. The “mostly out” score (0.25) is given to cases with a troop ratio between the cut-off point of 1:500 and the “fully out” score (0.0), which has been set at 1:1000. I assume that there are few positive protection effects at the operational and tactical levels from the sheer presence of UN troops when UN missions deploy troops in ratios that see 500 or 1,000 inhabitants per uniformed peacekeeper. With more than 1,000 people per peacekeeper—about three times as many as the ratio presented by Goode—it is considered unlikely that the troop-to-population ratio positively influences the outcome. Although some UN missions reach the ratios suggested in the existing literature, many remain out of the scope of what is deemed causally relevant. In fact, the UNPOCO dataset captures cases with ratios up to eight times as high as those introduced by Goode. However, my calibration should not be interpreted as discounting the potentially positive conflict-reducing effects of the presence of uniformed peacekeepers, despite being few in numbers compared to the population size. Rather, I want to investigate whether particularly “good” ratios can be part of the explanation for protection successes at the operational and tactical levels. By applying the rather strict ratios from existing theories, we should be able to discern whether this condition systematically appears together with other conditions in causal pathways leading to successful outcomes across cases and time.

Willingness to accept risk (risk) UN military protection operations may involve using force to deter, coerce, or even to destroy a perpetrator’s will and ability to use violence against civilians. Forceful military actions challenge the principles of UN peace operations as well as many troop contributors’ acceptance of risk. For these and many other reasons, force is seldom used to protect in UN peace operations (United Nations Office of Internal Oversight Services 2014). However, on at least 200 occasions between 1999 and 2017, UN troops did apply force—in ways defined and captured in this book—to protect.

CAUSAL CONDITION CANDIDATES

117

Different troop contributors operate in different ways. This can be a significant challenge in UN peace operations, to which more than 120 countries contribute uniformed personnel (United Nations 2021a). Differences between troop contributors can include the type of units deployed, their level and quality of training, the type and quality of equipment, command and control arrangements, the history, culture, and traditions of armed units, all the way down to individual traits in each soldier and officer in each contingent. These differences can also influence the results of military protection operations. It is not a viable research strategy to investigate all potential causal candidates linked to differences in various TCCs. Many of them are difficult to measure. In addition, QCA cannot be used to analyze more than a handful of conditions thought to influence a particular social phenomenon. Instead, I have selected one macro condition carrying particular importance when protecting civilians through military means: a troop-contributing country’s general policies on and level of acceptance of the use of force in UN peace operations. Some troop contributors are—for many good reasons—reluctant to risk the lives and well-being of their own forces for any purpose. Others are on principle against expanding the agenda of UN peace operations, which includes the use of force to protect civilians from violence (Bellamy and Williams 2013). Yet others are more “forward leaning”—accepting a higher degree of risk to protect a third party from harm—and on principle more open to perform robust military operations under the UN flag. I investigate whether the official stance of UN troop contributors is reflected in actual operations and systematically affects the outcomes of operations across time and UN missions. Accordingly, different TCCs are ascribed different memberships—either “in” (1.0) or “out” (0.0)—in a crisp set that captures their willingness to use force to protect civilians in the context of UN peace operations (see Table 5.4). The underlying hypothesis is that troops coming from countries that are more willing to use force to protect civilians from violence—accepting higher risks for their own forces—will systematically perform better than troops deployed by TCCs that are more hesitant. Somewhat surprisingly, the statistical analysis indicated an absence of a significant relationship between the willingness to accept risk and the outcomes. Still, it is valuable to test this condition further with the help of other methods. Membership scores are based on the existing literature, national policies, statements in the UN General Assembly, and expert opinions on how the TCCs operate on the ground (Bellamy and Williams 2013; Chesterman 2004; Government of Rwanda 2015; Providing for Peacekeeping 2020; United

118

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

Table 5.4 Calibration of TCCs’ willingness to use force to protect Crisp score

Description

# TCCs

TCCs

1.0

Willing

25

0.0

Hesitant

12

Ethiopia Benin Burkina Faso Chad Gabon Gambia Guatemala Guinea Ireland Bangladesh Cambodia China Egypt

Malawi Mauritania Mongolia Nepal Netherlands Niger Nigeria Portugal Rwanda Ghana India Indonesia Jordan

Senegal South Africa Sweden Tanzania Togo Ukraine Uruguay Kenya Morocco Pakistan Philippines

Nations 2010a, 2015b) (see also Table A3 for more references). At first, I attempted to develop a fuzzy score for this set—assigning degrees of willingness and hesitancy to different TCCs—but I found that it was challenging to do so systematically across all countries contributing troops. Instead, I chose the crisp set approach, which enabled a rougher—but more responsible—coding of TCCs’ degree of willingness to use force to protect. This has been the most challenging and possibly the most controversial condition to operationalize. To be clear, the aim is not to point a judgmental finger at particular troop contributors. However, there is a common perception that some TCCs are not doing enough or that they are doing things wrongly. Therefore, to avoid general finger pointing, there is a need to explore systematically whether the country of origin of UN troops has anything to do with the outcomes of protection operations, when force is applied to protect. As far as I am aware, this represents a first attempt at a more systematic analysis of this aspect across time and UN missions. In many cases, more than one TCC has been involved in an operation. To reflect this qualitative difference between cases, I have added a third score for the QCA analysis. If one or more of the TCCs involved in a case come from a country coded as “willing,” these cases receive the membership score 0.75 (“fairly willing”). The underlying hypothesis is that the presence of at least one “willing” troop contributor will have some positive effect on the outcome, but not as much as would have been expected if every TCC involved was from a “willing” country. Table 5.5 portrays how the scores are distributed across the 126 cases studied in the QCA analysis.

CAUSAL CONDITION CANDIDATES

119

Table 5.5 Calibration of TCCs’ willingness to use force to protect, including constellations of willing/hesitant TCCs, fuzzy scores, description, number of cases and case IDs from UNPOCO Fuzzy Description No. Case IDs from UNPOCO score 1.0

0.75

0.0

Willing

Fairly willing

Hesitant

54

32

40

MONUC DRC2

MONUSCO DRC54 DRC56-57 DRC61 DRC70-71 DRC81

UNOCI IvoryCoast2-3

UNISFA Abyei1-16 Abyei18-21

MINUSMA Mali2 Mali4-6

MINUSCA CAR6

MONUC DRC1 DRC10 DRC19

MONUSCO DRC49 DRC55 DRC64 DRC66 DRC68-69 DRC72

UNOCI IvoryCoast6

MINUSCA CAR1 CAR8

MINUSMA Mali1 Mali9

UNAMSIL SierraLeone1

MONUC DRC3-5 DRC8 DRC12-17 DRC21-26

MONUSCO DRC27-28 DRC44-45 DRC47

UNMIS Sudan1-2

UNOCI IvoryCoast1

MINUSCA CAR4 CAR7

DRC85 DRC88 DRC91 DRC99

UNAMID Darfur1-2 Darfur4-7 Darfur9-11 Darfur13-17 UNMIL Liberia7

DRC74 DRC76-77 DRC79 DRC82-83 DRC102-103 DRC106

UNMISS SouthSudan5 SouthSudan10 SouthSudan13 SouthSudan15

UNMIL Liberia1 Liberia6 Liberia15

DRC51 DRC53 DRC63

UNMISS SouthSudan1 SouthSudan3 SouthSudan6 SouthSudan11-12 UNMIL Liberia3-5 Liberia8-10

120

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

There is generally a lack of openly available and systematic reporting on which country’s troops have been involved in specific operations. Therefore, when specific information is missing, I determined the participants in each case based on the deployment maps provided by the Secretary-General’s reporting to the UNSC. In some cases, the main responsibility is ascribed to the TCC with overall responsibility for the geographical area where the operation took place, and/or the TCCs deployed to the nearest UN camp. UN missions are usually quite static in their deployment patterns, although dynamic deployments have occurred more frequently in recent years. It is still reasonable to assume that the dominant TCC in a particular area is most likely to have been critically involved in the operation. However, it is not completely certain that all participants have been captured for all cases, and some might have been ascribed as participating when they have in fact not been part of the operations. I also searched for complementary information online—from news articles and blogs—which sometimes name the TCCs involved in particular cases. Finally, you should note that all the troop contributors involved in operations captured in UNPOCO portray some level of willingness to expose their troops to risk and use force to protect. First, all operations have been performed under Chapter VII mandates, expecting UN troops to use deadly force to protect civilians under threat when necessary. Second, UNPOCO only captures cases where force has been used. TCCs that often are seen as reluctant—operating under caveats—carried out many operations, both successfully and unsuccessfully.

Pre-emption (preempt) UN peace operations are guided by three bedrock principles: impartiality, consent of the host state and main parties to the conflict, and non-use of force except in self-defense and defense of the mandate (United Nations 2008). These principles—although significantly modified over time—have been part of armed UN peacekeeping since its inception in the mid-1950s. Historically, UN troops were not to use force to fulfill the mandate, but only in self-defense. UN forces were not to pose a threat to the parties of the former conflict, usually fought between states. UN troops were positioned as a buffer between the armies of former warring parties, and their main jobs were to patrol, monitor, and report transgressions of peace agreements and ceasefires. There was no intention of using armed force to make former belligerents comply. For decades, the dominant UN mode of operations has therefore been defensive, only reacting to situations when UN troops were under threat.

CAUSAL CONDITION CANDIDATES

121

Contemporary UN peace operations, however, are deployed to armed conflicts within states where there is residual violence, despite the presence of peace agreements (Ki-moon 2014). Many armed groups continue to fight, even after the deployment of UN troops. UN forces are now increasingly expected to use force proactively to protect civilians from violence. This development peaked in 2013, when the UN deployed the Force Intervention Brigade of 3,000 troops to the Eastern DRC to neutralize various armed groups by force (Karlsrud 2015; Kjeksrud and Vermeij 2017; United Nations 2013f). The UN mission in Mali—MINUSMA—is also mandated to use force to deter armed groups from moving into population centers (United Nations 2013g). Most recently, the UN Security Council mandated a regional protection force of 4,000 troops to deploy to Juba—the capital of South Sudan—to protect civilians and if needed force parties to comply with the peace agreement (United Nations 2016c). Increasingly, throughout the 2010s, the Security Council seemed to rely on military force to establish some form of stability to ensure a more permissive environment for longer-term peace building. At the time of writing however, in mid-2022, a few years have passed since the Security Council mandated new multidimensional and integrated UN missions. The current war in Ukraine—likely to deteriorate working relationships in the Council—may also impact the future of peace operations and protection of civilians in armed conflict. Based on what we know so far, I seek to explore whether UN protection operations are more effective in reactive or pre-emptive mode. Accordingly, I have coded all 126 cases according to the type of operation UN forces have conducted to protect (see Table 5.6). Of the 126 cases, 43 are coded as preemptive (termed “preempt” (scored 1.0), while the remaining 83 are coded as “reactive” (scoring 0.0). This is also a crisp set, where cases are either “in” or “out.” It may be possible to provide a more finely tuned categorization to enable a fuzzy score analysis of this set, but I found no theoretical qualitative anchors that would allow me to do so responsibly. However, I do acknowledge that there are qualitative differences among cordon-and-search operations, static camp-defense operations, and joint offensive operations with host-state forces. It follows that pre-emptive operations are those cases where UN forces have tried to intervene militarily before attacks against civilians materialized. These can include cordon-and-search operations, as well as direct military confrontations alone or jointly with host-state forces. Conversely, “reactive” operations respond to situations where violent attacks on civilians are already underway. This category includes situations where UN troops sought to provide a deterrent presence, but where force was used as a last resort, defense of

122

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

Table 5.6 Calibration of the pre-emptive/reactive character of UN military protection operations, including fuzzy scores, description, number of cases, and case IDs from UNPOCO Crisp Description No. Case-IDs from UNPOCO score 1.0 Pre-emptive 43 UNAMSIL UNMIL SierraLeone1 Liberia4

0.0

Reactive

83

UNMIS Sudan1

MONUC DRC5 DRC8 DRC10 DRC12 DRC14-17 DRC19 DRC21

MONUSCO DRC28 DRC45 DRC55 DRC61 DRC64 DRC68-69 DRC88 DRC106

UNOCI IvoryCoast2-3

UNISFA Abyei2 Abyei4 Abyei6-7 Abyei10 Abyei12 Abyei20

UNMISS SouthSudan3

MINUSMA Mali2 Mali5-6

UNAMID Darfur13-17

MINUSCA CAR1 CAR6-7 UNAMID Darfur1-2 Darfur4-7 Darfur9-11

UNMIL Liberia1 Liberia3 Liberia5-10 Liberia15 MONUC DRC1-4 DRC13 DRC22-27

MONUSCO DRC44 DRC47 DRC49 DRC51 DRC53-54 DRC56-57 DRC63 DRC66 DRC70-72

MINUSMA Mali1 Mali4 Mali9

UNOCI IvoryCoast1 IvoryCoast6

MINUSCA CAR4 CAR8

UNMIS Sudan2

DRC74 DRC76-77 DRC79 DRC81-83 DRC85 DRC91 DRC99 DRC102-103

UNISFA Abyei1 Abyei3 Abyei5 Abyei8-9 Abyei11 Abyei13-16 Abyei18-19 Abyei21 UNMISS SouthSudan1 SouthSudan5-6 SouthSudan10-13 SouthSudan15

CAUSAL CONDITION CANDIDATES

123

their own or internally displaced person (IDP) camps from attack, and pursuit of perpetrators after attacks against civilians had taken place.

Matching (match) According to the only existing theory on how to maximize the utility of force to protect civilians, military protectors must match the perpetrator’s violence against civilians (Beadle 2011, 2014, 2015; Beadle and Kjeksrud 2018). Only then will protectors be able to effectively influence the will and ability of the perpetrator to attack. The matching theory is described in detail in Chapter 2. Table 5.7 is just a reminder of the functions of force/violence that form the underpinnings for this theory and visualizes protectors’ functions of force on the left and types of violence committed by the perpetrator on the right. Table 5.7 Functions of force (protector) versus types of violence (perpetrator) Functions of force (to protect)

Description

Types of violence (against civilians)

Description

Amelioration

Delivering aid, support humanitarian assistance, observe ceasefires, etc.

Impairment

Fostering insecurity, threatening civilian life without physically targeting them

Containment

Preventing arms, planes, and troops from spreading or passing through a barrier

Incitement

Using violence against civilians to spread fear and insecurity

Deterrence/ coercion

Posing or carrying out a threat, in order to change the opposition’s intentions

Deterrence/ coercion

Using violence to change civilian behavior

Destruction

Attacking the opposing force in order to destroy its ability to prevent the achievement of the political purpose

Destruction

Using violence to directly destroy civilians (or civilian installations)

This condition has been operationalized by first ascribing one or more functions of force to the protector in each case and then assessing the type of violence committed against civilians by the perpetrator, before comparing the two to evaluate whether the use of force matches the violence by the perpetrator. Through a crisp-set approach, a “match” is scored 1.0 (“in” the set) while a “mismatch” scores 0.0 (“out” of the set). Table 5.8 shows that in 99 cases, the

124

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

Table 5.8 Calibration of UN troops’ ability to match the perpetrators by force, including fuzzy scores, description, number of cases, and case IDs from UNPOCO Crisp Description No. Case IDs from UNPOCO score 1.0 Match 99 UNAMSIL MONUC SierraLeone1 DRC1 DRC4-5 DRC8 DRC10 DRC12 DRC14-17 DRC19 DRC21-24 DRC26 DRC28

0.0

Mismatch

27

MONUSCO DRC45 DRC47 DRC49 DRC51 DRC53-57 DRC61 DRC64 DRC68-72

DRC74 DRC76-77 DRC79 DRC81-83 DRC85 DRC88 DRC102-103 DRC106

UNMIL Liberia1 Liberia3-10 Lieria15

UNOCI IvoryCoast1-3

UNAMID Darfur1 Darfur5 Darfur10-11 Darfur13-17

UNMIS Sudan1

UNMISS SouthSudan1 SouthSudan3 SouthSudan10

UNISFA Abyei1-9 Abyei11-16 Abyei18-20

MINUSMA Mali2 Mali4-6 Mali9 UNISFA Abyei10 Abyei21

MINUSCA CAR1 CAR6-8 MONUC DRC2-3 DRC13 DRC25

MONUSCO DRC27 DRC44 DRC63 DRC66 DRC91 DRC99

UNAMID Darfur2 Darfur4 Darfur6-7 Darfur 9

UNMIS Sudan2

UNMISS SouthSudan5–6 SouthSudan11–13 SouthSudan15

MINUSMA Mali1

UNOCI IvoryCoast6

MINUSCA CAR4

NECESSARY CONDITIONS FOR SUCCESSFUL OUTCOMES

125

protectors have matched the perpetrators, while the remaining 27 are coded as a mismatch.

Necessary conditions for successful outcomes In this section, I present the results of the fsQCA analysis, which is performed in two steps. The first step is an analysis of necessary conditions and the second is an analysis of potential causal recipes, conditions that in combination may be sufficient for the outcome. Both analyses rest on the calibrations presented above. Together, the 126 cases now form a so-called QCA matrix, where the membership scores of all cases in all condition sets and the outcome set are compiled (see Table A1). I use QCA software developed by Charles Ragin to perform the analysis (Ragin, Drass, and Davey 2006). Although the mapping of 200 cases captured and coded in UNPOCO covers the period from 1999 to 2017, there were no cases identified from 1999, 2001, and 2002 and this is also the case for the QCA matrix. All 10 UN missions represented in UNPOCO appear in the QCA matrix. The first analytical step in QCA is to control for the presence of necessary conditions. A condition is necessary “if, whenever the outcome Y is present, the condition is also present. In other words, Y cannot be achieved without X” (Schneider and Wagemann 2012, 69). A necessary condition is a super-set of the outcome. Table 5.9 portrays the results of the analysis. Table 5.9 Analysis of necessary conditions for the presence of positive outcomes Condition

Consistency

Coverage

Deterrent presence (deter) Willingness to accept risk (risk) Pre-emption (preempt) Matching (match)

0.411552 0.657040 0.480144 0.902527

0.644068 0.583333 0.773256 0.631313

The software performs two analyses to determine whether a particular condition is necessary for the outcome. First, a consistency analysis—assessing “how far the outcome can be considered a subset of the condition”—and, second, a coverage analysis—measuring “the relevance of a necessary condition” (Schneider and Wagemann 2012, 143, 147). According to common QCA standards, the consistency threshold should be set at least at 0.9 (Ragin 2006; Schneider and Wagemann 2012, 143). Among the four conditions analyzed

126

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

here, only one—matching (match)—portrays a value that fulfills the consistency threshold for a necessary condition. It breaches, with a rather small margin, the 0.90 threshold, with a consistency score of 0.902. From the QCA matrix (Table A1), we can see that 70 cases of military UN protection operations have either fully successful outcomes (21) or partially successful outcomes (49). Matching occurred in 68 of these 70 cases. This explains the high consistency score, although the condition is not always present when a fully successful or partially successful outcome is present. Deterrent presence, willingness to accept risk and pre-emption all score well below the suggested 0.90 threshold. It follows that neither good troop-topopulation ratios, willingness to accept risk, nor pre-emption, is necessary to achieve successful outcomes. The high consistency score of matching requires further examination to determine whether this is a trivial or non-trivial relationship of necessity (Schneider and Wagemann 2012, 139–50). This is done with the help of the coverage analysis. Again, we turn to Venn-diagrams in Figure 5.2 to portray the difference between trivial and non-trivial relationships. Both Venn diagrams in Figure 5.2 portray the fundamental logic of a necessary condition, in that the conditions are super-sets of the outcome. However, they also show different degrees in this relationship. Venn diagram 1 depicts a trivial relationship, while diagram 2 depicts a relevant, non-trivial relationship. Schneider and Wagemann provide a fictitious example to explain the difference, relayed here with slight modifications (Schneider and Wagemann 2012, 145).

Outcome Y

Outcome Y

Condition A Condition B

Figure 5.2 Venn diagrams portraying the logic of a trivial (1) and non-trivial (2) necessary condition.

Imagine that the set Outcome Y is speeches in a country’s parliament when parliamentarians curse. Condition A is the set of parliamentarians born in the country, while Condition B is the set of male parliamentarians. Although both conditions are necessary for the outcome, being a male member of parliament is much more relevant (non-trivial) to explain the cursing phenomenon. Does

CAUSAL RECIPES TOWARD SUCCESSFUL OUTCOMES

127

the coverage score for matching of 0.631 indicate a relevant—non-trivial—set relationship? There are no standard thresholds provided by the literature on QCA. However, according to an example provided by Schneider and Rohlfing, a coverage score of 0.65 indicates a non-trivial relationship (Schneider and Rohlfing 2013, 565). The 0.631 result from my coverage analysis therefore seems to support the relevance of matching. However, although matching is almost always present when outcomes are successful, I also find that UN troops matched the perpetrators of violence in 31 out of 56 cases where few or no civilians were protected. These negative outcomes explain the results of the coverage analysis. Where does this leave the necessity analysis? Matching is relevant for almost all positive outcomes, coming close to being a necessary condition. However, matching also appears quite often when UN troops fail to protect. The results also indicate that matching must appear in combination with other conditions to become relevant. The following section investigates whether matching occurs in combination with other conditions in causal pathways toward successful outcomes.

Causal recipes toward successful outcomes The second analytical step in QCA is to search for causal recipes, or conditions that in combination may be sufficient for the outcome. Again, the QCA software is used to perform the analysis. Now, the method introduces a truth table, which sorts all cases into combinations of sufficient conditions leading toward the outcome in different rows (see Table 5.10). Before using the analytical tools provided by the software, we must decide the consistency cut-off point for relevant solutions, which determines which rows of combinations will be part of the analysis. This is critical, as the cut-off point will influence the causal recipes’ consistency and coverage scores. As can be seen from Table 5.10, I had to decide on a cut-off point between 0.79 and 0.81 (marked in bold). Table 5.10 Relevant rows from the truth table deduced from the analysis of the QCA matrix deter

risk

preempt

match

number

raw consist.

PRI consist.

SYM consist.

1 1 0 0

0 1 0 1

1 1 1 1

1 1 1 1

2 13 12 15

1.000000 0.913043 0.810345 0.796610

1.000000 0.906977 0.717949 0.750000

1.000000 0.951220 1.000000 0.818182

128

DISCOVERING CAUSAL PATHWAYS TO SUCCESSFUL PROTECTION

A cut-off point at 0.81 only captures 27 cases, possibly increasing solution consistency scores, but certainly decreasing their coverage. A cut-off point at 0.79 will add 15 more cases to the analysis, risking a lower solution consistency score, but increasing the chances for a higher coverage score. I opted for the lower cut-off point at 0.79, which yielded the results as shown in Table 5.11. Table 5.11 Intermediate solution analysis of the truth table deduced from the QCA matrix Combinations

Raw coverage

Unique coverage

Consistency

Solution coverage

Solution consistency

Match∗ preempt

0.480144

0.480144

0.791667

0.480144

0.791667

The analysis only proposes one causal pathway based on these choices and calibrations, the combination of pre-emptive operations that also match the perpetrators of violence (match ∗ preempt). Although this combination of conditions scores reasonably well on solution consistency (0.79), it covers slightly less than half of the outcome set (0.48). I also performed a robustness check, analyzing the truth table with a higher cut-off point (0.81). Now, a slightly more consistent solution appeared. In addition, matching and pre-emption was now joined by deterrent presence, i.e., good troop-to-population ratios (match ∗ preempt ∗ deter), yielding a consistency score of 0.85. However, as expected, this solution only covered about a third of the outcome set (0.33). What can we learn from this analysis? The most interesting insight is perhaps the results that do not appear. First, favorable troop-to-population ratios are not able to explain positive protection outcomes across operations at the tactical and operational levels, either alone or together with other conditions. Although I found this condition to have a significant relation to the outcomes in the statistical analysis, it does not hold in the QCA analysis. Again, it seems that being present in large numbers is just not enough to protect civilians from violence from imminent threats. Troop-to-population ratios do seem to be part of the explanation in about one-third of positive outcomes, but that also indicates that this condition demands the presence of other explanatory factors to become relevant. This provides nuance to our existing knowledge about the overall conflict-reducing effect of large peacekeeping operations. Although the presence of many troops reduces the severity of conflict, it does not necessarily explain how UN troops fare in protecting civilians from different types of perpetrators who continue to attack civilians in the presence of UN troops. We

CAUSAL RECIPES TOWARD SUCCESSFUL OUTCOMES

129

should still keep in mind that the country-level data underpinning my analysis is a crude measurement of the potential deterrent effect at the local level. Second, the willingness of Blue Helmets to accept risk does not appear to be part of the answer to how they fare in protecting civilians from imminent threats across conflicts and time. This finding supports the indications from the statistical analysis. Keep in mind that I do not capture the events where civilians were under threat without a military UN intervention. Hence, a pool of troop contributors that were more willing to accept risk might have improved the UN’s results overall. The bottom line is that, from what we can gather from openly accessible reporting, the country of origin of troops is not able to explain outcomes across cases systematically. Combined with the first insight, it also underlines another main point: It seems to matter more what UN troops do rather than where they are from. Less surprising is the finding that pre-emptive protection operations tailored to particular threats—matching the perpetrators of violence—are important parts of the causal pathways toward successful outcomes. We already had indications of their relevance from the statistical analysis. The analysis in this chapter further strengthened the relevance of matching. Now we know that matching and pre-emption combined provide the most consistent solution across almost half of the cases. Although the QCA analysis does provide interesting insights, the results remain inconclusive. While we should not expect to fully explain all cases of a complex phenomenon with the help of scientific methods and analyses, it remains a fact that most outcomes must be explained by other conditions. Therefore, in the next chapters, I perform one more analytical step through deeper qualitative analysis of two particularly interesting cases. Combining QCA with deeper qualitative case studies and process tracing is seen as an emerging good research standard (Beach and Rohlfing 2018; Schneider and Rohlfing 2013). I will continue to investigate the four conditions in local contexts, but equally importantly I seek to identify other causal conditions that may be relevant for successful outcomes.

6 Protecting Civilians from the M23 UN troops in the Force Intervention Brigade had the right training, capabilities, and mindset. –Divisional General (R.) Carlos Alberto dos Santos Cruz, former Force Commander of the United Nations Organization Stabilization Mission in the Democratic Republic of the Congo (2018)

Military force can play a prominent and successful role in protecting civilians from violence. Recent UN peacekeeping history can boast several successful cases, many of which are captured in the UNPOCO dataset. Among the better-known successes are the 2013 joint operations conducted by the Forces armées de la république démocratique du Congo (FARDC) and the UN’s Force Intervention Brigade (FIB) against the armed insurgent group Mouvement du 23 Mars (M23) in the eastern parts of the Democratic Republic of the Congo. These events are interesting for our purposes for several reasons. The first is the exceptionally strong mandate language linking the use of force to protection of civilians. In Security Council resolution 2098 from late March 2013, the Council tasked the United Nations Organization Stabilization Mission in the Democratic Republic of the Congo (MONUSCO) to use military force to “carry out targeted offensive operations through the Intervention Brigade […] to prevent the expansion of all armed groups, neutralize these groups, and to disarm them” (United Nations 2013f, para. 12 b). The rationale underpinning this marching order was to “reduce the threat posed by armed groups to state authority and civilian security […] and to make space for stabilization activities” (United Nations 2013f, para. 12 b). In addition, the mandate provided “default” protection language, instructing MONUSCO and the FIB to use all necessary means to ensure the effective protection of civilians from physical violence (United Nations 2013f, para. a i). This strong language—hitherto unseen in UNSC mandates—paved the way for forceful operations unlike any other contemporary UN military protection effort, with only a possible precedent in the kinetic actions taken by the original UN operation in the Congo (ONUC) in the early 1960s, although its mandate was to stop Katanga from seceding rather than to protect civilians (O’Brien 1962).

Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0006

PROTECTING CIVILIANS FROM THE M23

131

The second reason is the FIB’s robust military capabilities, significantly outweighing regular UN brigades of the same size (Kjeksrud and Vermeij 2017). The FIB consisted of about 3,000 troops, including “three infantry battalions, one artillery and one Special Force and Reconnaissance company with headquarters in Goma, under direct command of the MONUSCO Force Commander […]” (United Nations 2013f, para. 9). The FIB also had access to Rooivalk attack helicopters from South Africa and Ukrainian-piloted Hind attack helicopters, which would prove critical in the fight against the M23 (Defense Update 2013; Olivier 2013b). Artillery fire, Special Forces operations, and regular infantry tactics were also part of the approach (Martin 2013; The Peacekeeping and Stability Operations Institute 2013). The third point of interest is the unexpectedly effective military partnership between the FARDC and the FIB. Although many tend to give credit to the FIB for the win against M23, the FARDC carried out much of the heavy lifting on the frontlines. Undoubtedly, the FIB was a crucial enabler for the joint operations and significantly increased the prospects of success. In the past, the FARDC was—and remains—a notoriously difficult partner for the UN, not least due to their appalling human rights record (United Nations Joint Human Rights Office 2013, paras. 18–24, 2018). Some observers credit the new and talented commander Major General Bahuma Ambamba of North Kivu’s 8th Military Region for this positive development (African Defence Review 2013). General Ambamba also brought other military talents under his command, including Colonel Amadou Ndala, who was a key FARDC commander on the ground. You can catch a few glimpses of Colonel Ndala and his troops in combat in the watch-worthy documentary “This is Congo” (McCabe 2017). To many, Ndala embodied the FARDC’s much improved operational conduct, moral, and discipline following its humiliating defeat against M23 in the fall of 2012. A few months after the successful operations against the M23, the much-appraised Congolese colonel was killed in an ambush near Beni, north of Goma (BBC News 2014). Fourthly is the uncommon regional and international political support for the robust application of force under UN auspices. The humiliating fall of Goma—when M23 with relative ease overran MONUSCO-supported FARDC units—created a rare incentive for forceful military action. Importantly, the International Conference on the Great Lakes Region (ICGLR)—supported by the African Union (AU) and the South African Development Community (SADC)—first introduced the idea of a regionally based intervention force (Kok 2013; United Nations 2012d, 2013f). Swiftly supporting and

132

PROTECTING CIVILIANS FROM THE M23

subsequently adopting the idea, the UN Security Council soon launched the FIB with capable regional troops from South Africa, Tanzania, and Malawi. The final reason is the global powers’ significant pressure on Uganda and Rwanda, both of which supported the M23 by supplying personnel, training, and equipment (Human Rights Watch 2013; United Nations 2012c, 2014b, para. A4). Although it took a while, the United States finally blocked military aid to Rwanda, citing concern over the possible recruitment of child soldiers and the M23’s troublesome human rights record (Smith 2013). Consequently, Rwanda halted its support to the M23—significantly weakening its fighting ability—and thus dealt the FIB and FARDC a better hand in the military confrontation that was to follow. For these reasons alone, a case study of the FARDC-FIB operations against the M23 would probably provide relevant insights on the role and utility of force in protecting civilians from violence. Moreover, there are parallel reasons for case selection flowing from the mixed-methods research design guiding the analyses in this book. The qualitative case studies in this chapter and the next provide further insights to the main findings from the cross-case fsQCA analysis, representing a version of what is known as a condition-centered design (Beach and Rohlfing 2018, 7). This design is concerned with further exploring conditions that—combined—are sufficient to produce the outcome of interest (successful military protection operations). I am interested in gaining a better understanding of the causal role of the two most promising conditions that emerged from the cross-case analysis: matching and pre-emption, while also holding the door open for the two other conditions—which thus far have shown little explanatory promise—troop numbers and willingness to accept risk. Qualitative case studies—in general—also provide opportunities to look for other possible explanations, which may be generalizable to other cases. The main selection criterion for this case is its score on matching the perpetrators of violence. Matching emerged as the only condition to fulfill the criteria for a necessary condition for successful protection outcomes, according to the analyses in Chapter 5. To recap, this implies that matching must occur to achieve successful outcome, while we also know that matching has occurred when UN troops failed to protect. The FARDC-FIB operations clearly matched the perpetrators of violence—with deterrent, coercive, and even destructive force—influencing the M23’s willingness and ability to attack civilians. Moreover, these operations contained an element of pre-emption to stop an impending M23 attack on Goma. From my previous analyses, preemption appears to be highly relevant in combination with matching, and I use this case to illustrate how and why this combination was effective in this

MATCHING THE M23

133

specific situation. Further, my cross-case analyses failed to find systematic positive effects of the deterrent presence of large numbers of UN troops. However, these numbers were limited to the country level, and did not capture variations in particular cases. I therefore use this case to explore an additional—and perhaps more relevant—factor, which is the troop-to-perpetrator ratio, i.e., the potential effect of UN troop numbers relative to the numbers of perpetrators. The joint FARDC-FIB operations clearly had local numerical superiority relative to the M23. In addition, the analyses in earlier chapters found no systematic effect on positive protection outcomes of UN troops’ willingness to accept risk. Nonetheless, I am not confident that this condition is irrelevant in this case. The FIB certainly consisted of highly motivated and risk-tolerant troops, deployed with an “expectation to fight the M23,” as underlined by then Force Commander Santos Cruz.¹ Finally, this case provides particularly rich empirical material—a rare occurrence when studying UN military efforts on the ground—which facilitates a search for alternative explanations yet to be discovered in the existing literature.

Matching the M23 Matching the perpetrators of violence—by now the most promising condition for successful military protection outcomes—simply implies that the use of force must be tailored to particular threats to be effective. A quick reminder about this theory’s logical underpinnings is warranted (see also Chapter 2). Building on Rupert Smith’s four functions of force and his own typology— four correlating perpetrator functions of violence—Beadle explains that if a perpetrator aims to destroy an ethnic group, the protector will not find utility of force by ameliorating the situation by merely supporting the delivery of humanitarian aid. In this situation, greater utility of force is found in matching the perpetrator by destroying his ability to conduct mass killings. Conversely, if a perpetrator uses incitement or impairment against civilians to undermine the legitimacy of a government, using coercive or destructive force against them is likely to lead to stronger incentives for perpetrators to scale up attacks against civilians. In addition, if the most violent functions of force are applied, they risk causing more harm during operations than would otherwise occur in these less violent situations, thus challenging well-established ¹ Interview with Lt. General Carlos Alberto dos Santos Cruz, Oslo, August 31, 2018.

134

PROTECTING CIVILIANS FROM THE M23

rules of proportionality in customary International Humanitarian Law (IHL). Instead, containment and amelioration are better suited to protect civilians in such scenarios. Consequently, to maximize the utility of force, protectors must match the perpetrator’s violence against civilians. Only then will protectors be able to effectively influence the willingness and ability of the perpetrator to attack civilians. Table 6.1 illustrates how military forces can ideally match these four ways in which perpetrators use violence against civilians to provide more effective protection. Table 6.1 Perpetrator’s use of violence vs. protector’s use of military force to protect Perpetrator violence against civilians

Protector use of military force to protect

Impairment (e.g. presence of armed actors and constant threat of armed clashes) Incitement (e.g. indiscriminate attacks by insurgents in government-held areas) Deterrence or coercion of civilians (e.g. threats or retaliatory attacks against civilians associated with the enemy, or demonstrative violence to make people flee) Destruction of civilian life or property (e.g. massacres or scorched earth policies)

Amelioration (e.g. presence of military observers reporting human rights violations) Containment (e.g. creation of weapon-free zones, counter-IED operations) Deterrence or coercion of the perpetrators (e.g. threats or actual use of force to alter the willingness to target civilians through robust show of force or punishing attacks) Destruction of perpetrator capabilities (e.g. neutralization of rebel armed forces)

These functions of force and violence are broad analytical categories, each encompassing a wide range of potential military efforts. However, each category seeks to describe the main purpose of using force (violence) to protect (target civilians), which provides logic to the analysis of how UN troops used force to influence perpetrators of violence. In this first of two case studies in the book, I evolve the matching theory by unpacking what functions of force and violence might mean in practical terms and dig deeper for more fine-grained explanations for protection successes and failures. To evaluate how and to what degree the use of force matched the perpetrators of violence in this case, I assign functions of force and violence to the FARDC-FIB and the M23, respectively, before investigating variations in the outcome. Most of the information underpinning the analyses was captured from two reports by the UN Group of Experts, supported by online sources, and interviews that I conducted with practitioners during fieldwork (United Nations 2013d, 2014a). I treat the joint FARDC-FIB operations against the

MATCHING THE M23

135

M23 as one event in my analyses. It is nonetheless important to convey that it consisted of a series of events beginning with the establishment of the FIB on March 28, 2013, and UNSC resolution 2098 and ending with the defeat of the M23 in November later that year. To trace how and to what extent the joint FARDC-FIB operations matched the M23, I divide the series of operations into five different phases, as shown in Table 6.2. Table 6.2 Five phases of FARDC-FIB operations against the M23 Phase 1

Phase 2

Phase 3

Phase 4

Phase 5

Mar.–Apr.

May

June–July 29

July 30–Aug.

Sep.–Oct.

Resolution 2098 and M23 threats

First Mutaho Hills battle

Second Mutaho Hills battle

Goma red line and the Kibati battle

Three-pronged attack defeating the M23

FIB units were only directly involved in phases four and five. Nevertheless, the establishment and gradual deployment of the FIB in phases one and two also influenced the conduct of M23. Finally, note that the matching theory describes how perpetrators use violence against civilians. Although the violence applied by the M23 during these events did target civilians, they were mostly concerned with combatting the FARDC and the FIB at this time. This underlines one of the key observations made in Chapter 2, that the protection of civilians is seldom the only task of military forces. It also shows that external intervention can alter the behavior of the perpetrator of violence, shifting their efforts toward countering the intervening force.

Phase 1 (March–April 2013): Resolution 2098 and M23 threats On March 28, the Security Council established the Force Intervention Brigade (United Nations 2013f). The FIB did not deploy troops in this first phase, but its planned entry into the fray certainly triggered reactions from the M23. In two open letters from early April, the M23 addressed the parliaments and people of Tanzania and South Africa, respectively, threatening violent retaliation against their troops’ future presence in the DRC (Congo DRC News 2013a, 2013b). At this point, it was evident to the M23 that the UN was in fact preparing to respond by force, threatening their grip on eastern DRC. Both letters warn of dire consequences for FIB personnel if they engage in battle on behalf of the Congolese government. Evidently, the letters did not prompt any of the

136

PROTECTING CIVILIANS FROM THE M23

troop contributors to back down, and preparations for their deployment continued unabated. However, the situation underlines that, even before the FIB deployed, its formal existence influenced the M23 into attempting to deter the most potent troop contributors from taking part. As we shall see later, M23 continued to seek opportunities to throw the FIB off balance before it became operational. Table 6.3 lays out each actor’s actions and functions of violence/force in this phase. Table 6.3 Phase 1 (March–April 2013): Resolution 2098 and M23 threats Date (2013)

M23 actions

M23 function(s) of violence

FARDC/FIB actions

FARDC/FIB function(s) of force

Outcome

MAR– APR

M23 threatens Tanzania and South Africa in letters in April.

Impairment

Establishment of FIB.

Containment/ deterrence

No direct targeting of civilians in Goma.

M23 redeployed troops and heavy weapons toward Goma. M23 fighters train for urban warfare and guerilla tactics

FARDC deployed in defensive positions in and around Goma

Insecurity for civilians remained in areas occupied by M23

The sheer presence of a most capable armed group near Goma—the most populous city in Eastern DRC—fulfills the criteria laid out by the matching theory for impairing the security situation for civilians. For matching to occur, military protectors must improve security for civilians under latent threat, without the application of forceful means. At this point, UN troops were not yet present in the country. The FARDC, however, were deployed in defensive positions in and around Goma, seeking to contain the threat from M23, and to deter them from attacking and entering Goma. In this phase, matching seems to have occurred. Since the M23 never attacked Goma during this period, the approach was to some degree effective. With that said, deterrence is only effective until it fails, and, as will become clear in the next phase, the M23 had not stood down, instead continuing to reposition its troops toward Goma. Simultaneously, the M23 took part in peace talks in Kampala, which had commenced back in December 2012. According to Force Commander Carlos

MATCHING THE M23

137

Alberto Santos Cruz, “much of M23’s actions were tailored to improve their position at the Kampala talks.”² At this point, the M23 still received support from both Rwanda and Uganda and occupied a narrow strip of territory inside the DRC bordering Rwanda. This provided both effective logistical support and a potent staging area for attacks.

Phase 2 (May): First Mutaho Hills battle On May 20, the M23 attacked FARDC positions in the Mutaho Hills, just 7 km north from Goma (Reuters 2013; United Nations 2013d, 11–12). The FARDC held its positions successfully. Part of the successful defense can be ascribed to the fact that the M23 ran out of ammunition and failed to receive “promised troop and ammunition support from within Rwanda” (United Nations 2013b, 12). The first troop detachments from Tanzania had arrived only a week earlier—the FIB still being months away from reaching full operational capacity—and were not able to support the FARDC during this second phase. During the clashes with the FARDC, the M23 suffered more than 70 casualties—including 40 dead—leading them to announce a unilateral ceasefire. Interestingly, the M23 launched the attacks just three days ahead of the UN Secretary-General’s visit to Goma. While the M23 had amassed much of its firepower and personnel in preparation for these attacks, it remains uncertain whether the attacks were seeking to pre-empt future FIB actions, to retake Goma before the FIB reached full operational capacity, or rather to improve M23’s position at the peace talks in Kampala (United Nations 2013b, 12). Table 6.4 portrays the actors’ actions and functions of violence/force schematically. To match a perpetrator who uses force coercively, military protectors must use force to influence the perpetrator into disengaging from a particular strategy of violence or implement direct military action that reduces the perpetrator’s ability to continue its attacks on civilians, without causing more civilian harm in the process. In this phase, the M23 was more concerned with attacking the FARDC—possibly simultaneously testing the FIB’s preparedness—than directly targeting the civilian population. It is plausible that the M23 sought to re-enter Goma at this point, which would have significantly increased the threat to civilians (United Nations 2013d, 11–12). Worryingly, the M23 fired artillery shells toward the Mugungu III IDP camp, causing civilian injuries and forcing many civilians and IDPs to flee toward Sake and Goma (UNHCR ² Ibid.

138

PROTECTING CIVILIANS FROM THE M23

Table 6.4 Phase 2 (May): First Mutaho Hills battle Date M23 actions (2013)

M23 FARDC/FIB function(s) actions of violence

FARDC/ FIB function(s) of force

Outcome

MAY

Coercion

Coercion

FARDC remained in position at Mutaho Hills, defending Goma.

On May 20, three days before the UN Secretary General’s visit to Goma, the M23 attacked FARDC at Mutaho Hills (7 km NW of Goma). On May 21 and 22, five artillery shells (122 mm) fired by M23 landed close to Mugunga III refugee camp and populated areas of Goma

FARDC forcefully defended their positions at Mutaho Hills. First FIB deployment of Tanzanian and South African troops arrived in mid-May. No FIB units involved in operations at Mutaho Hills at this time

M23 suffered at least 70 casualties, incl. 40 killed. M23 declared unilateral ceasefire and most fighters retreated to initial positions. Thousands of IDPs and locals fled to Goma and Sake. A few civilian injuries as a result of M23 shelling

2013). It remains unclear whether this was part of the plan or rather flawed targeting. Since the FARDC held its ground at Mutaho Hills—coercing the M23 to retreat to its initial positions—matching also occurred during this phase, with a largely successful outcome.

Phase 3 (June–July 29): Second Mutaho Hills battle June and early July saw a lull in the fighting between the M23 and the FARDC. On July 14, however, the M23 once again tried to overrun the FARDC position at Mutaho Hills. Three days of intense fighting ensued. As shown in Table 6.5, the FARDC defended its positions successfully. Moreover, the national army managed to push M23 northward, capturing additional territory in the process (United Nations 2014a, 5). During Phase 3, the FARDC also employed

MATCHING THE M23

139

Table 6.5 Phase 3 (June–July 29): Second Mutaho Hills battle Date M23 actions (2013)

M23 function(s) of violence

FARDC/FIB actions

FARDCFIB function(s) of force

Outcome

June– July 29

Coercion

FARDC again held their position at Mutaho Hills, and pushed the M23 further north, gaining additional territory.

Coercion

Some M23 artillery pushed out of range of Goma

Relatively limited fighting in June and early July. M23 maintained troops in forward positions close to Goma. From July 14 to 16, M23 again attacked FARDC at Mutaho Hill and the Kanyarucinya area, north of Goma

FARDC employed attack helicopters. On July 24, a FARDC attack helicopter caused unintended injuries of at least 7 civilians. FIB continued deployment in Sake and Goma

attack helicopters to target M23 positions. On July 24, rounds fired by one of the helicopters hit a civilian area near Rumangabo, 5 km north of Goma, injuring seven civilians (United Nations 2014a, 5). Similar to the situation in Phase 2, the FARDC matched the M23 by coercing them into retreat, although this time with unintended harm to civilians. Importantly, pushing the M23 back reduced the threat from their long-range artillery, which held a reach of 18.5 km.

Phase 4 (July 30–August): Goma red line and the Kibati battle Toward the end of July, it became clear that the M23 had retained significant military capabilities within range of Goma, continuing to fire toward populated areas and UN installations. As a result, on July 30, Force Commander Santos Cruz established a security zone around Goma (United Nations News Centre 2013; YouTube 2013). Anyone carrying arms within this zone— bar the Congolese army—was to disarm within 48 hours, or MONUSCO

140

PROTECTING CIVILIANS FROM THE M23

would use all necessary means to disarm them. According to the Force Commander, the purpose of the security zone was twofold: First, we needed to push the M23 long-range artillery out of reach of the civilian population in Goma. Second, I needed a tripwire that would justify forthcoming operations against the M23. I knew [the M23] would not uphold the security zone. In fact, we [MONUSCO/FIB] were not anywhere near capable of disarming everyone within the security zone. The security zone was mostly a feint to lure the M23 into overstepping the red line.³

The ruse worked. Toward the end of August, the M23 again fired close to UN installations in Goma. The artillery shells made impact close to the Force Commander’s position, who subsequently ordered the Tanzanian artillery component of the FIB to fire back: After the first grenade from M23 landed close to the UN camp, we [the FIB] retaliated with five artillery grenades to adjust our fire. But the M23 would not cease fire. When they continued to fire towards the airport, I ordered the Tanzanians to return fire with 50 artillery grenades.⁴

The forceful UN response marked the first direct engagement of the FIB in the fight. The first artillery salvos destroyed an M23 tank, and over the next three days, FIB artillery and attack helicopters continued to strike the M23 alongside operations conducted by the FARDC. On August 28, joint FARDCFIB operations used “tanks, armored personnel carriers, attack helicopters, mortars and ground troops in attacks on M23 positions along the Kibati front line” (United Nations 2014a, para. 10). Yet again, the M23 declared a unilateral ceasefire. At this point, the M23 was no longer a direct threat to Goma and its inhabitants. All their long-range artillery was pushed back or destroyed. At least 17 M23 officers were killed because of the fighting (United Nations 2014a, para. 10). Table 6.6 shows how the parties acted in this phase, including the types of violence/force employed. The establishment of the security zone was a final attempt at deterring the M23 from attacking Goma. However, we now know that the Force Commander did not expect a deterrent effect from this move. When the M23 expectedly breached the red line, the joint FARDC-FIB operations employed coercive and destructive functions of force. The M23 heavily resisted these operations, ³ Ibid. ⁴ Ibid.

MATCHING THE M23

141

Table 6.6 Phase 4 (July 30–August): Goma red line and the Kibati battle Date M23 actions (2013)

M23 function(s) of violence

FARDC/FIB actions

FARDC/ FIB function(s) of force

Outcome

July 30— Aug.

Coercion

On July 30, MONUSCO Force Commander established a security perimeter around Goma.

Deterrence/ coercion/ destruction

M23 pushed back from their remaining positions around Goma and its Northern suburbs.

On July 30, M23 shelled Goma. M23 strongly resisted FARDC-FIB joint operations, using its full arsenal, including anti-tank weapons

On August 21, fighting restarted at Kibati (15 km N of Goma). From August 22 to 24, FIB artillery and attack helicopters targeted M23 in Kibati in support of FARDC. On August 28, joint FARDC-FIB operations used tanks, armored personnel carriers, attack helicopters, mortars, and ground troops in attacks on M23 positions along the Kibati front line. On August 30, FARDC took “Three Towers” at Kibati

M23 artillery no longer within range of Goma. At least 17 M23 officers killed and much military equipment destroyed. M23 declared unilateral ceasefire towards the end of the last week of August

inflicting many casualties on the FARDC, while also suffering many casualties among their own ranks. Two Tanzanian FIB officers were also killed. The M23 also employed coercive and destructive violence. In this phase, the situation resembled a “traditional” military standoff between two warring parties, with civilians caught in the line of fire. As such, it is challenging to determine if the M23 employed a particular function of violence against civilians during this phase. However, we know that M23 directly threatened MONUSCO’s leadership with launching attacks on Goma and did fire some artillery shells into the city during the battle. As such, the M23 remained an imminent threat to

142

PROTECTING CIVILIANS FROM THE M23

civilians, and protection remained the main rationale for joint FARDC-FIB operations. In Phase 4, the protectors matched the coercive attempts of the perpetrators of violence, denying them the ability to continue to launch attacks on Goma. In addition, the joint FARDC-FIB operations toward the end of August included some destructive functions of force, by significantly reducing the M23’s capability to launch further attacks against civilians elsewhere. Some 17 M23 officers were killed, and much of their military equipment was destroyed. Arguably, the military response was too forceful for pure protection purposes, instead using overwhelming force to defeat the M23. In fact, the end of phase four was the first sign that the FARDC-FIB joint operations would not merely coerce the M23 to retreat from Goma, but rather sought to destroy their overall willingness and ability to continue fighting (which was the FIB’s mandate in the first place). From a protection perspective, the outcome was positive, since the most destructive weapons of the M23 were removed out of range of Goma or destroyed. More worryingly, it remains unclear whether the operations led to unintended harm to the civilian population.

Phase 5 (September–November): Three-pronged attack defeating the M23 While the battle at Kibati in August clearly weakened the M23, the unilateral ceasefire and the continued peace talks in Kampala allowed them to regroup once more before their last stand against the joint operations of the FIB and the FARDC at the end of October (United Nations 2014a). At this time, Malawi had joined the FIB, finally making the brigade fully operational. From October 25 to 28—following another breakdown of the peace talks in Kampala on October 21—joint FARDC-FIB operations dislodged the M23 from their remaining positions along the Kibati–Rutshuru axis, leading to their formal surrender on November 5 (United Nations 2014a), para. 16). The details of the operations have been well described by others—such as the UN Group of Experts on the DRC and Darren Olivier writing for African Defence Review—providing the backbone of the recapitulation of events used here (Olivier 2013a; United Nations 2014a). The FARDC-FIB joint operations were performed along three fronts: northern, western, and southern (Olivier 2013a). The UN brigade consisted of three

MATCHING THE M23

143

task forces, each supporting a task force from the FARDC. Each UN/FARDCpairing held the responsibility for one front. The main role of the FIB was to block the M23 from using the only functional north–south road from Kibati to Rutshuru, while the FARDC undertook most of the offensive efforts to attack M23 positions. Most FIB support came in the form of artillery and mortar fire, as well as air support from attack helicopters, although some FIB ground units were also involved in direct combat with the M23. According to Force Commander Santos Cruz, “the FIB fired some 500 rockets from helicopters, 500 artillery shells and 500 mortars” during just one week of operations in October.⁵ During the fight on the northern front in Kiwanja, the FIB even utilized South African and Tanzanian Special Forces to initiate the operations from the town center, pushing the M23 outward toward FARDC positions on the outskirts of town. According to Force Commander Cruz, MONUSCO feared close quarter combat in Kiwanja—potentially risking the lives of many civilians still living there—leading FARDC-FIB to plan for this somewhat unorthodox move.⁶ Regular UN flights transported the Special Forces to Kiwanja. A lieutenant from Tanzania was killed during the fight there. After four days of intense fighting, the M23 had fled all but one of its positions. The armed rebel group made a final stand at Bunagana, a small village on the border with Uganda. This was a heavily fortified position that contained much of the M23’s remaining equipment. The FARDC attempted to overrun the position but failed on two separate occasions. Eventually, UN attack helicopters from South Africa were employed to raid the position, leading to its destruction. After attempting to torch and burn the equipment and ammunition, the remaining M23 fighters fled to Uganda (United Nations 2014a, 9). In the aftermath, the UN Group of Experts found “numerous cases of M23 troops looting, assaulting, abducting and arbitrarily arresting local people in their shrinking area of control,” based on interviews with residents in Kiwanja and United Nations sources (United Nations 2014a, 7). At this point of the fight, however, the M23 was mostly concerned with surviving the confrontation with the FARDC-FIB operations. Like Phase 4, the M23 used coercive and destructive force, but mostly as a response to these operations rather than to target civilians (see Table 6.7). However, on November 4, the M23 intentionally shelled civilian areas of Bunagana (United Nations 2014a, 9). Civilians thus remained under imminent threat from the M23, the very day before their ⁵ Ibid. ⁶ Ibid.

144

PROTECTING CIVILIANS FROM THE M23

Table 6.7 Phase 5 (September–November): Three-pronged attack defeating the M23 Date (2013)

M23 actions

M23 function(s) of violence

FARDC/FIB actions

FARDCFIB function(s) of force

Outcome

Sep.– Nov.

M23 redeployed heavy weapons during lull in fighting.

Coercion

FIB at full operational capacity (incl. Malawi).

Destruction

M23 plea for ceasefire.

Small-scale clashes during the last half of September. M23’s final confrontation with FARDC-FIB in late October

FARDC-FIB launch joint operations (October 25–28)

Physical defeat of M23. M23 surrenders Surviving M23 fighters flee to Uganda and Rwanda

surrender. Joint FARDC-FIB operations finally destroyed the perpetrators’ willingness and capability to launch destructive attacks, and the M23 ceased to be a threat to civilians on November 5.

Pre-empting the M23 The operations against the M23 were not predominantly pre-emptive, as the armed group had been targeting civilians since April 2012, long before the FIB had been deployed (United Nations Joint Human Rights Office 2014). In fact, the FARDC and MONUSCO had been on the defensive from the M23’s inception—most devastatingly portrayed with the fall of Goma in November 2012—and usually reacted to M23 attacks against them. The main shift came in August 2013 when FARDC-FIB operations turned pro-active to push M23 artillery out of range of Goma. The idea underpinning the security zone is certainly interesting, since it points to a pre-emptive logic at the tactical level, with significant influence on the strategic outcome. The idea underpinning the security zone was to lure the M23 into crossing the red line, which would facilitate a more robust joint response to reduce the threat to civilians from the most destructive of M23’s weapons. When the M23 failed to adhere to the

TROOP-TO-PERPETRATOR RATIOS

145

conditions of the security zone, the FIB and FARDC could more justifiably go on the offensive, pre-empting further attacks on Goma. However, it remains uncertain whether the M23 were in fact preparing to launch an all-out attack Goma. It could be that their troop movements toward the city, attacks on FARDC positions, and the limited shelling of Goma, were in fact signaling to gain a stronger position at the peace talks. It is also plausible that the FIB and the FARDC would have confronted the M23 in any event, without the establishment of a security zone. As such, the preemptive effect created by the security zone may not be causally relevant to the eventual outcome of the joint operations. However, the establishment of the security zone did impact the dynamics of the battle by adding a justification for confronting the M23 pre-emptively before a further potential attack on Goma. Overall, the UN response to the threat the M23 posed to civilians was not primarily pre-emptive. However, when protection operations were finally underway, pre-emptive logic defines one of the most significant events of the armed clash, which shifted the balance from reactive responses to M23 attacks to pro-active operations that finally led to the M23’s demise.

Troop-to-perpetrator ratios In the cross-case analyses described in earlier chapters, I analyzed the potential deterrent effect of the sheer numbers of uniformed personnel in-country, finding no systematic correlation with positive protection outcomes. Troopto-population ratios, however, did show promising signs. The fsQCA analysis in Chapter 5 indicated that good troop-to-population ratios were indeed part of a causal pathway that explained about a third of the successful outcomes of cases captured in the UNPOCO dataset. As such, troop-to-population ratios may have some explanatory merit, although most cases cannot lean on this explanation. Moreover, MONUSCO scores quite poorly on the troopto-population ratio. The DRC is a populous country covering a vast area. In 2013, the DRC had a population of more than 70 million people in an area of 2.3 million square kilometers (about a quarter of the US landmass). We tend to describe MONUSCO as the largest UN mission, but this description fails to convey that there are still disappearingly few UN troops per square kilometer. It nonetheless seems plausible that the number of troops involved in particular events could have an effect on the outcome. As such, I add another element by examining the potential explanatory power of troop-to-perpetrator

146

PROTECTING CIVILIANS FROM THE M23

ratios, which remains unexplored in the context of UN military protection operations. In most cases, I have not been able to find data portraying local variations in troop vs. perpetrator numbers, but in the case of the FARDC-FIB against the M23, we know a bit more. During the joint operations against the M23 in the fall of 2013, the FARDC had “at least 6,000 troops during the final operations” (United Nations 2014a, 9). The full strength of the FIB was 3,000 troops, while few of these were involved in the actual operations. The UN Group of Experts indicates that the FIB used some 400 troops on the ground during the final phases of the operations (United Nations 2014a, 9). It follows from this—depending on how you count—that the UN-DRC government side commanded and controlled somewhere between 6,400 and 9,000 troops. It is challenging to determine an exact number of M23 fighters, although figures in the range of 1,500–2,500 have been suggested (Pflanz 2012; Stearns 2012a, 40). Even if we maximize the number of M23 fighters to about 2,500 and assume the lowest number of FARDC-FIB troops—about 6,400 uniformed personnel—the M23 was clearly outnumbered. Combined, the FARDC-FIB constellation was able to establish three joint task forces to put simultaneous pressure on the M23 on three fronts. As such, troop-to-perpetrator ratios seem to be more relevant to explain the outcomes of protection operations, than troop-to-population ratios. This adds to existing knowledge about the conflict reducing effects of large operations, pointing to the need to ask “compared to what,” when troop numbers are presented as key to reducing violence against civilians (Hultman, Kathman, and Shannon 2019). We also know that numbers mattered to the M23. During the fighting in the fall of 2013, the M23 suffered heavy casualties. Estimates conducted by MONUSCO during and after the conflict indicate more than 400 M23 killed and over 1,000 injured.⁷ UN and FARDC troops on the ground collated these estimates through local populations providing information in the aftermath of operations, reconnaissance with drones and helicopters, and physical observations from the peacekeepers themselves. At the same time, the M23 continuously recruited new personnel from Rwanda and Uganda (United Nations 2014a, 10–12). A baffled MONUSCO Force Commander observed that M23 positions quickly filled up with new personnel after their positions had been pounded with heavy fire from the joint FARDC-FIB operations.⁸ Consequently, we simply do not know how many fighters the M23 employed ⁷ Interview with General (ret.) Carlos Alberto dos Santos Cruz, Force Commander MONUSCO (2013–2015). Oslo, August 31, 2018. ⁸ Ibid.

WILLINGNESS TO ACCEPT RISK

147

during the confrontations. The M23 also received direct support from neighboring countries Rwanda and Uganda during the fighting, mostly in the form of recruitment of new personnel, but also through direct fire support (United Nations 2014a, 10–12). This support might have prolonged the fighting but did not influence the outcome. Another aspect of troop numbers was revealed in the interview with General Santos Cruz. When asked whether he would ideally have had more personnel in the FIB, or if he could have managed with fewer, his response was that the size of the brigade was well-suited to the joint operations north of Goma. The main reason for this was that the fighting occurred on a narrow strip of land just a few kilometers in width bordering Rwanda, with only one north–south functional road. He would not have had room to maneuver a larger force in support of the FARDC, as all belligerents were somewhat dependent on that road. This is also a reason why the UN attack helicopters became so important since they were able to move freely and swiftly above the narrow area of operations. Although it remains unclear exactly how the Security Council ended up with a brigade of this size, and with these capabilities, it matched the needs on the ground in this situation relatively well.

Willingness to accept risk South Africa, Tanzania, and Malawi all provided troops to the FIB. In what was a rare moment for UN peace operations, the troop contributors arrived with an expectation and acceptance of significant risk, matching the strategic intention of the Security Council mandate. Most of the time, FIB forces supported the FARDC during offensive operations. However, during the final battles in October, FIB units were also engaged in unilateral ground combat with the M23 (African Defence Review 2013; The Peacekeeping and Stability Operations Institute 2013). The Blue Helmets respected the UN chain of command, with no need to confer with capitals before acting and were willing to deploy into areas where there was a significant risk of retaliation from the M23. It seems clear in this case that having risk-accepting UN troops can form part of the explanation for the successful outcome. What is puzzling, however, is that the FIB operated alongside less risk accepting MONUSCO troops, also known as the framework brigades. With the same mandate to protect civilians and under unified command with the FIB, the framework brigades did not participate in the robust operations against the M23. This led to some quite bizarre situations. In Kiwanja, for example, MONUSCO framework brigades patrolled in the presence of the

148

PROTECTING CIVILIANS FROM THE M23

M23 while the FIB and the FARDC prepared to fight them.⁹ As became evident in October and early November, the FIB was able to provide sufficient support to the FARDC to win the fight, but MONUSCO was not able to utilize its full potential, mostly because of significant aversion to risk and caveats conveyed by the framework brigades. Interestingly, after the defeat of the M23, later FIB contingents were less accepting of risk than their predecessors were. According to General Santos Cruz, there was “less appetite for robust operations as time went on.”¹⁰ My cross-case analyses found no significant relationship between willingness to accept risk and the outcomes of protection operations across cases. The findings from the DRC, however, highlight that variation in willingness to accept risk clearly does matter in individual cases. Without the presence of South Africa, Tanzania, and Malawi—or troops with a similar mindset— MONUSCO would not have been able to support the FARDC in a way that led to a successful outcome.

Additional explanations for the successful outcome While studying these operations, other potential explanatory conditions emerge as relevant to explain the outcome than those identified in the literature. Some of these might also be relevant to explain variations in outcomes across a larger number of cases, but I have not been able to study them systematically. Future studies might benefit from keeping these conditions in mind.

Operational art The establishment of the security zone around Goma shifted the dynamics of the armed conflict in a positive direction in favor of the FARDC-FIB side. The MONUSCO Force Commander established the security zone based on his personal initiative, without conferring with headquarters or with the Kampala delegates: I had my instructions from New York already. My orders were to neutralize the M23. M23’s failure to adhere to the security zone was just one event in the ⁹ Ibid. ¹⁰ Ibid.

ADDITIONAL E XPL ANATIONS FOR THE SUCCESSFUL OUTCOME

149

line of many where they had displayed complete arrogance in relation to the FIB’s role and potential involvement in the fight. I just used that arrogance to our advantage, attempting to shift the balance of the conflict by drawing a line, being prepared to immediately escalate when that line was crossed.¹¹

At first glance, the decision to establish the security zone is well within the room for maneuver delegated to force commanders in general. Yet General Santos Cruz received a great deal of criticism from headquarters in New York, as well as from the international delegates taking part in the peace talks in Uganda. According to the Force Commander, he merely employed a tactic based on his reading of how the M23 would react, using the breach of the security zone as a stepping-stone to an effective counterattack. The mandate was clear: to neutralize the M23 (United Nations 2013f). For those seeking a political solution in Kampala, however, the establishment of the security zone seemed a reckless move that endangered the volatile peace talks. It remains unclear to what degree the M23 really was willing to engage in sincere negotiations, as they broke the deals made on numerous occasions by continuing combat on the ground, thereby endangering the lives of civilians. In this case, it seems that the decisions made on the ground and operational art had a better effect on civilian security than the political efforts in Uganda ever achieved. This highlights the dilemma of how to strike a balance between conflict management from afar and letting those on the ground make decisions based on their best judgments. Arguably, mission leadership should be free to make decisions without always conferring with headquarters. UN missions do have significant autonomy. In other situations, however, this may not be straightforward. Ideally, the political and military aspects of any forceful intervention should work in concert, not least to avoid stepping on each other’s toes. The UN has not yet mastered how to combine the use of force with political peace efforts at the strategic level. To understand more about the influence of personal initiatives and operational art, deeper case studies of particular operations would be necessary. I have only briefly lifted the veil here, displaying the dynamics in play in one specific situation when UN troops engaged in forceful operations against a perpetrator of violence against civilians.

¹¹ Ibid.

150

PROTECTING CIVILIANS FROM THE M23

Operational readiness The 2013 spring events at Mutaho Hills—just outside Goma—underline the need to deploy UN troops that are ready to engage in operations immediately, as opposing forces may look for any opportunity to discredit them. If the FARDC positions at Mutaho Hills had been overrun by the M23 MONUSCO and the not-yet-operational FIB units would have been faced with a significant challenge, very similar to that of November 2012 when the M23 occupied Goma. The UN Group of Experts found that the M23 leadership had decided to maintain troops in forward positions close to Goma even after their first defeat at Mutaho Hills, indicating plans for a renewed attack on Goma (United Nations 2013d, 11–12). Indeed, the M23 did launch a new attack in midJuly, again before the FIB was fully operational. This aspect is perhaps mostly important when a force is deployed with an expectation to engage in combat, as was clearly the case with the FIB. Luckily, their efforts were not needed at this point, as the FARDC was able to hold their positions.

Escalation When deterrence fails, UN troops must be prepared to escalate the use of force to protect civilians. This might seem obvious, but my impression from studying UN military efforts over time is that deterrence equals uniformed presence, failing to take into account that a credible threat of actual application of force is a core component of effective deterrence. In this case, escalation played a significant role in pushing the M23 back from Goma and eventually in their defeat. The M23 was not easily deterred. The UN Group of Experts points to the fact that the armed group continued to redeploy troops and move equipment toward Goma, even after their significant setback at Mutaho hills in May. In September and October, after having been pounded by the combined arsenal of the FARDC and the FIB for days, the M23 nonetheless regrouped to continue the fight. If the FIB and FARDC had not been able to respond with force to coerce the M23 into retreat, it seems likely that the threat to civilians in Goma and its surroundings would have remained high. This is not to say that escalation is always the right approach when civilians are under threat. However, when faced with a perpetrator threatening to use coercive force with heavy weapons against a densely populated area, the UN and its national partners must pose a credible deterrent threat. Credible deterrence can only be achieved if the opposing party believes that you will retaliate, and that you do so, when deterrence fails.

ADDITIONAL E XPL ANATIONS FOR THE SUCCESSFUL OUTCOME

151

Force mobility and projection The FIB and FARDC may have been able to defend Goma solely with the use of ground forces, artillery, and mortars.¹² However, the attack helicopters from Ukraine, South Africa, and the DRC were critical in the defeat the M23. The attack helicopters were used for surveillance and to obtain information, as well as for direct targeting of M23 positions. Being independent of the eastern DRC’s severely weak road infrastructure, the highly mobile helicopters were able to project force swiftly beyond the more slowly moving FIB and FARDC ground units. In another example of the effect of troop mobility, the UN Special Forces supporting the FARDC in Kiwanja in October were lifted there to catch the M23 off guard. Certainly, the Special Forces were few, but the principle of being able to move quickly to where you are most needed, and to project force with mobile military platforms, is sometimes a critical enabler of effective military protection operations. Force mobility and projection are also linked to credible deterrence. With that said, not all UN missions with a protection mandate would be able to employ attack helicopters, both because they are expensive and controversial (Novosseloff 2017).

Understanding the perpetrators of violence The joint operations against the M23 were surprisingly effective and one of the reasons why the operations succeeded so quickly was how the FARDCFIB operations understood the modus operandi of the M23. Although we now know that the M23 trained in guerilla warfare before the deployment of the FIB, their eventual response—operating in structured formations and in dugin static positions—served the FIB and the FARDC well. Beadle’s deliberations on how to find the utility of force to protect center around a deeper understanding of the perpetrators’ use of violence, before tailoring a response to that threat (Beadle 2014, 14–21). Others have written excellent analyses of the M23 and other Tutsi-based armed groups in the eastern DRC (Beswick 2009; Stearns 2012a). I will not provide a full analysis of the M23 here. Instead, I will highlight a point that seems to have been critical to their defeat in late 2013. The M23 received significant support from two state actors; Rwanda and Uganda (United Nations 2012d, 2014a; United Nations Joint Human Rights Office 2014, para. 112). Due to this external support, the M23 was the most capable non-state armed group in the area at that time, and the fight against ¹² Ibid.

152

PROTECTING CIVILIANS FROM THE M23

them probably dragged out—and became bloodier—because of it. However, the external support also influenced the M23’s modus operandi, as they received a steady flow of weapons from their partners abroad, facilitating a continued reliance on heavy weapons and shelling with artillery. These types of weapons are static, demanding fortified positions to optimize their effectiveness. When the FIB and FARDC targeted these positions, the M23 did not regroup and reconsider their strategy. Instead, they replenished the bombarded positions with new personnel and military material, continuing with more of the same.¹³ Counterintuitively, the exact type of support that made the M23 a powerful actor also strongly contributed to its demise. In this case, the capabilities of the FARDC and FIB were well-suited to weakening the threat posed by the M23. However, later FIB-supported operations against the ADF and other armed groups in the east have proven more difficult. Hence, an intervention brigade of this type and with these capabilities will not always be the right response. As different perpetrators display different modus operandi, so also must the protectors.

Conclusion The operations against the M23 provide examples of what matching the perpetrator might mean in practice, as well as providing an opportunity to study the different functions of force to protect, applied in different phases for different purposes and with different outcomes. This event enables a study of a “typical” case, and one in which matching was clearly part of the causal pathway toward a successful outcome. Since the operations protected every potential future victim from that M23, this can be considered a successful— albeit unusual—outcome for UN military protection operations. In addition, military force was undoubtedly the predominant tool to achieve this outcome since few other UN means were involved in the immediate actions to protect civilians. A typical case does not expose a theory to a rigorous test. However, it is valuable to examine whether the matching theory passes the first test of explaining the most clear-cut cases, where force quite clearly has been used with the utility to protect. It should be noted that this is one of few cases where Blue Helmets—together with government troops—directly engaged a warring party. As such, we should be cautious with the generalizability of the findings to other cases. ¹³ Ibid.

CONCLUSION

153

Although the operations against the M23 were not purely pre-emptive, I also found that one of the events that shifted the dynamics toward a successful outcome—the establishment of the security zone—was based on pre-emptive logic, which supports the findings of the cross-case analyses. Other explanations for the successful outcome include troop numbers and willingness to accept risk. Troop numbers mattered mostly in relation to the number of perpetrators. Willingness to accept risk was certainly part of the explanation for the successful outcome, countering the results from the cross-case comparisons. This condition thus remains relevant, but it is important to continue studying UN military protection operations in different contexts to understand why, how, and when it is relevant. Furthermore, the case study identifies other causal condition candidates that may be part of the explanation for successful outcomes: operational art, operational preparedness, the willingness to escalate when deterrence fails, force mobility and projection, and understanding the modus operandi of the perpetrators of violence. Some of these conditions might be specific to this case, but there is good reason to systematically analyze them in future studies to evaluate the degree to which they are valid across a larger number of cases. These operations represent an outlying case in the empirical universe of UN protection operations, as they completely removed one of the most imminent and potent threats to civilians in that specific time and place, including the killing of many M23 fighters. Although I consider the overall outcome to be a success, some might argue that such short-term victories are not enough to favor using force to protect in UN peace operations. It is true that military UN operations often fail to facilitate longer-term stabilization and are seldom adequately linked to political processes leading toward peace (Berdal 2016, n.d.; High-Level Independent Panel on Peace Operations 2015). This is also true of the operations against the M23. Despite the initial success of defeating the M23, later FARDC-FIB operations have not been able to reduce the overall threat posed by the plethora of armed groups attacking civilians in eastern DRC, which remains the main task of the FIB. However, since the main rationale for using force in UN operations is to protect civilians from imminent violence, MONUSCO and FARDC were undoubtedly successful in their defeat of the M23.

7 Overwhelmed by the White Army Preventing inter-communal violence, deterring it, and protecting civilians from it was at the heart of our mandate. But I soon discovered that our military capabilities were wholly inadequate. —Hilde Frafjord Johnson Special Representative of the Secretary General in South Sudan 2011–2014 (Johnson 2016, 104)

On December 23, 2011—after weeks of mobilization that overtly threatened their long-time enemy—the Nuer White Army marched en masse to punish the Murle. The White Army was determined to avenge previous attacks against themselves and disregarded high-level mediation attempts urging them to stand down (Johnson 2016, 110). Early reports indicated that the White Army moved south with as many as 12,000 fighters.¹ Actual numbers were more likely somewhere between 6,000 and 8,000 (UNMISS Human Rights Division 2012). Still—for a loosely organized community-based armed group—the White Army was massive, constituting an imminent physical threat to the entire Murle community in Pibor County. Organized in columns—each with hundreds of fighters—the White Army razed some 20 villages throughout Pibor County, killing Murle civilians wherever they could be found. In support of the Sudan Peoples’ Liberation Army (SPLA),² the United Nations Mission in South Sudan (UNMISS) deployed more than half of its infantry forces to Jonglei County to defend the Murle from the oncoming attackers. Still, the SPLA and UNMISS were severely outnumbered, doing little to stop the White Army’s violent march before it reached Pibor Town, the main Murle population center in Jonglei.³ This was where the SPLA and UNMISS concentrated the main military protection efforts. During

¹ Interview with high-level UNMISS official involved in the military planning process connected to the violent attacks in 2011–2012. Oslo 2018. ² Now the South Sudan People’s Defence Forces (SSPDF). ³ Pibor town is now part of Boma, after the creation of new South Sudanese states in 2015. I will refer to Jonglei throughout, as this was the name of the province at the time of the events studied here.

Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0007

OVERWHELMED BY THE WHITE ARMY

155

the New Year’s weekend of 2011, the joint UNMISS–SPLA operations succeeded in deflecting an all-out attack on the town center. These efforts saved the lives of many potential victims. However, immediately after being denied access to Pibor town, the White Army crossed Pibor River and proceeded to attack the Murle living south and east of the town. The Murle living in these areas had not been warned of the oncoming threat, leading to several hundred more being killed. Neither UNMISS nor the SPLA was able to respond effectively to these continued White Army attacks. In addition, when Murle militias retaliated against Lou Nuer and Bor Dinka communities—while attacks against themselves were still underway—neither the SPLA nor UNMISS had the will or capability to respond to protect civilians, leading to many more dead, wounded, and displaced. Although UNMISS struggled to find a relevant military response to protect, the mission still saved many lives. The mission provided early warning to many communities living in harm’s way. These efforts—largely delivered by the civil affairs component—ameliorated the situation by facilitating the flight—and ultimately the survival—of many potential civilian victims. The threat to Murle communities came as a surprise to no-one. The years preceding 2011 witnessed a violent tit-for-tat cycle of attacks between Nuer and Murle, with casualties counting several hundred (UNMISS Human Rights Division 2012, 5–9). The immediate trigger for the events treated here was the August 2011 Murle attack against the Nuer communities in Pieri Payam, leading to 600 people killed (Johnson 2016, 101; UNMISS Human Rights Division 2012, 11). Economic incentives and group dynamics among particular age groups are also part of the explanations for these seemingly unending cycles of violence, where cattle raids often have led to large-scale atrocities (International Crisis Group 2009; Lacey 2013; UNMISS Human Rights Division 2012, 5–12; Young 2007). While ethnic animosities and revenge for former attacks may explain much of the violence at this moment, armed conflicts in South Sudan also occur within ethnic groups (Breidlid and Arensen 2014). This case is by far the most violent event captured in the UNPOCO dataset. The attacks “targeted entire communities, including women and children, and possibly aimed at destroying their livelihoods and social and economic infrastructure. Hate speech and incitement to violence based on ethnicity— crimes under domestic law and violations of international human rights law—contributed to the violence” (UNMISS Human Rights Division 2012, i). The investigation performed by the UNMISS Human Rights Division recorded 612 Murle killed, including at least 88 women and 88 children, another 294 unconfirmed deaths, and 370 persons unaccounted for at the time

156

OVERWHELMED BY THE WHITE ARMY

of the report, some of whom were possibly abducted (UNMISS Human Rights Division 2012, 12). The main selection criterion for this case is linked to the UN mission’s inability to match the perpetrators of violence. While the M23-case (see Chapter 6) was a typical case—in that the FIB clearly matched the perpetrators of violence—the case study of the White Army vs. Murle, however, clearly deviates on this condition (Beach and Rohlfing 2018). UNMISS and the SPLA clearly failed to match the White Army, apart from the defense of Pibor town, where the outcome also was more positive. While the responsibility for the violence rests with the White Army, UNMISS and SPLA’s inability to match the perpetrators elsewhere in the county ultimately led to many civilian casualties. The case also allows for an assessment of the three other causal condition candidates that appear throughout my analyses: pre-emption, deterrent presence, and the willingness to accept risk. In this case, it was not the military component who attempted to pre-empt attacks, but rather UNMISS’ civilian leadership and civil affairs component providing early warning to civilians before threats materialized. These efforts clearly saved many lives, although they temporarily aided the perpetrators’ cause in removing much of the targeted population. Moreover, the troop-to-perpetrator ratio was more or less inversed in comparison with the M23 case, as the SPLA and UNMISS military were severely outnumbered by the White Army. Finally, the troop contributors taking part in the operations all score among the more unwilling to accept risk to protect civilians from harm. However, they did indeed take significant risk in deploying forces when met with the massive threat from the White Army. Like the case portrayed in Chapter 6, this case also provides rich empirical material, facilitating the exploration of alternative causal conditions and mechanisms. Most of the descriptions of the events are drawn from the only comprehensive investigation into the events undertaken by the Human Rights Division in UMISS (UNMISS Human Rights Division 2012). Since my initial findings are unable to explain all cases of interest, an important aim of the qualitative case studies is therefore to identify other conditions that had an impact on the outcome. The communal conflict in South Sudan triggered systematic investigations and analysis, both by the UN mission and by outside observers.

Failing to match the White Army To trace how and to what extent the SPLA–UNMISS operations matched the White Army, I divide the series of operations into four different phases, as

FAILING TO MATCH THE WHITE ARMY

157

Table 7.1 Four phases of SPLA–UNMISS operations against the White Army in 2011–2012 Phase 1

Phase 2

Phase 3

Phase 4

Dec. 5–12

Dec. 13–22

Dec. 23–30

Dec. 31–Jan. 4, 2012

White Army mobilization, early warning, and UN deployment

Deterrence fails as the White Army moves south

Multiple White Army attacks

White Army Pibor attack, SPLA/UNMISS defense, and circumnavigation

shown in Table 7.1. I have not added a systematic analysis of how UNMISS matched or mismatched Murle revenge attacks since few attempts were made to reduce the threat to civilians emerging from the avenging Murle militias. However, we need to keep in mind that two protection crises unfolded simultaneously from December 27, when the Murle launched retaliatory attacks against the Nuer. These attacks lasted until February 4, 2012, killing close to 300 civilians. In a separate section, I therefore discuss the absence of relevant responses to the Murle revenge attacks.

Phase 1 (December 5–12): White Army mobilization, early warning, and UN deployment UNMISS was aware of potential retaliatory attacks between the Murle and the Nuer long before the events treated here escalated toward the end of 2011 (Johnson 2016, 108). Since August that year, the mission had deployed infantry patrols, integrated teams, and aerial reconnaissance helicopter patrols to observe—and if possible avert—brewing violence (Johnson 2016, 108). However, in the first week of December, UNMISS officially alerted the South Sudan Government of potential imminent violence in Jonglei, based on their information about the mobilization of Nuer youth in response to recent Murle attacks (Johnson 2016, 110; UNMISS Human Rights Division 2012, 13). UNMISS confirmed these observations yet again on December 9 with the help of aerial reconnaissance flights and UN personnel undertaking field missions to several locations throughout Jonglei state. Still, the Government did not recognize these early-warning efforts of potential large-scale violence against civilians until December 12, when they reported back to UNMISS that peace conferences in Jonglei were suspended due to the forewarned threat to the Murle population.

158

OVERWHELMED BY THE WHITE ARMY

UNMISS begun deploying troops throughout Jonglei according to an existing rapid response plan, seeking to warn those civilians under imminent threat, and preparing a deterrent presence in high-risk areas. The SPLA, however, “dragged its feet,” and the Government made no attempt to pre-empt a potential attack at that time (Johnson 2016, 109). The sheer existence of an armed group of this size involved in a communal conflict fulfills the criteria laid out in the matching theory of impairing the security for civilians (see Chapter 2). To match a perpetrator of violence impairing civilian security, UN military efforts must improve security for civilians under latent threat, without the application of forceful means. With the help of aerial reconnaissance flights, rapid deployment of field missions to hot spots, and other early-warning efforts, this is exactly what UNMISS did during this phase. Table 7.2 provides another reminder of the logic underpinning the matching theory, and it is also worth revisiting Chapter 2 to recall the mechanisms of the matching theory. While it remains difficult to determine exactly how many potential victims were saved by UNMISS’s early-warning efforts during this phase of the event, early warning was most likely a critical condition for reducing the violent impact on the Murle communities. Nevertheless, facilitating the flight of a population faced with an imminent threat is also controversial, as it can assist the perpetrator in fulfilling his aim of removing the opposing community from certain areas. While this motivation is more commonly present in ethnic cleansing campaigns, perpetrators in communal conflicts often display a similar logic, although communal militias seldom possess the means

Table 7.2 Perpetrator’s use of violence vs. protector’s use of military force to protect Perpetrator violence against civilians

Protector use of military force to protect

Impairment (e.g. presence of armed actors and constant threat of armed clashes) Incitement (e.g. indiscriminate attacks by insurgents in government-held areas) Deterrence or coercion of civilians (e.g. threats or retaliatory attacks against civilians associated with the enemy, or demonstrative violence to make people flee) Destruction of civilian life or property (e.g. massacres or scorched earth policies)

Amelioration (e.g. presence of observers reporting human rights violations) Containment (e.g. creation of weapon-free zones, counter-IED operations) Deterrence or coercion of the perpetrators (e.g. threats or actual use of force to alter the willingness to target civilians through robust show of force or punishing attacks) Destruction of perpetrator capabilities (e.g. neutralization of rebel armed forces)

FAILING TO MATCH THE WHITE ARMY

159

needed to do so (Beadle 2014). Certainly, the White Army campaign effectively drove tens of thousands of Murle from their homes, killing hundreds in the process, and destroying much infrastructure and homes. However, the Nuer youth never sought to hold the areas they attacked, and the Murle were able to move back home after the attacks had ended (although much of the housings also were destroyed). Their temporary removal from the White Army’s destructive path was possibly a lesser problem at that time. One question remains: was there any opportunity to deter the White Army militarily before its violent march on the Murle began? I will return to this question toward the end of the analysis. Table 7.3 provides a schematic overview of events in Phase 1, including the parties’ actions, functions of violence/force, and the outcome. Table 7.3 Phase 1 (December 5–12): White Army mobilization, early warning, and UN deployment Date

White Army actions

White Army function(s) of violence

UNMISS/SPLA actions

UNMISS/ SPLA functions of force

Outcome

Dec. 5–12, 2011

White Army mobilization

Impairment

UNMISS alerted the South Sudan Government to mobilization of Nuer youth on December 5 and 9.

Amelioration

Continued mobilization of the White Army

UNMISS deployed aerial reconnaissance missions and field visits to confirm the mobilization UNMISS started deploying troops to key population centers in Jonglei UNMISS provided early warning to civilians under threat of attack

Suspension of peace conferences in Jonglei

160

OVERWHELMED BY THE WHITE ARMY

Phase 2 (December 13–22, 2011): Deterrence fails On December 13, with the help of aerial reconnaissance, UNMISS observed a forward column of White Army fighters north of Pibor County (UNMISS Human Rights Division 2012, 14). On that day, high-level efforts were made by UNMISS to muster an immediate response to protect civilians from the expected violence. Hilde Frafjord Johnson—then the Special Representative of the UN Secretary General and head of UNMISS—warned the SPLA Chief of Staff, urging the Government to take urgent action to protect civilians. No immediate efforts were made by the SPLA at that time. Nevertheless, on December 19, Vice-President Riek Machar started to meet with leaders of both communities in Juba, the capital of South Sudan, to head off the expected violence. He achieved little, as the White Army continued to mobilize and march closer to areas inhabited by the Murle. UNMISS continued its deployment of infantry companies to vulnerable areas for attacks in Pibor County, as well as aerial reconnaissance flights to observe the White Army’s movements, reporting about their progress toward the south. Like Phase 1, the White Army presence close to Murle areas impaired the security for civilians, and many had already started to flee the impending attacks (see Table 7.4). UNMISS continued its early-warning efforts, matching the threat posed by the perpetrators by ameliorating the situation for civilians by facilitating for their early flight away from the threat. On December 21, the SPLA had received orders to “be vigilantly alert and prepared to protect Lou Nuer and Murle civilians if they were attacked” (UNMISS Human Rights Division 2012, 14). Although faced with a massive group of armed fighters—which did little to hide their violent intentions— neither the SPLA nor UNMISS was able or willing to muster a military response to deter the White Army from proceeding.

Phase 3 (December 23–30, 2011): Multiple White Army attacks On December 23, the first physical attack against civilians was committed by the White Army, when they targeted Wuno, close to Likuangole (see Table 7.5). UNMISS aerial patrols observed the attacks in the immediate aftermath. During the following three days, the White Army broke into smaller columns to attack villages along the Nanaam River valley and other areas close to Likuangole (UNMISS Human Rights Division 2012, 14). More than 175 Murle

FAILING TO MATCH THE WHITE ARMY

161

Table 7.4 Phase 2 (December 13–22, 2011): Deterrence fails as the White Army moves south Date

White Army actions

White Army function(s) of violence

UNMISS/SPLA actions

UN/SPLA functions of force

Outcome

Dec. On Impairment On Dec. 13, Amelioration Continued 13–22, December UNMISS police mobilization 2011 13, advance patrol confirmed the of White White Army presence of Nuer Army column youth in Northern First columns present in Pibor of fighters northern On Dec. 13, move south Pibor County UNMISS SRSG toward Pibor Continued urged the County from White Army Government and Nuer areas mobilization SPLA to take urgent Deterrence measures to halt the fails violence On Dec. 19, vice-president meets with Nuer community to hold off attacks On Dec. 21, SPLA orders troops to protect civilians In Jonglei UNMISS continued deployment to key population centers

were killed in these multiple attacks, according to UNMISS’s investigations. The investigation also questions whether the people living in these areas had received warning of the attacks. On December 27, the White Army proceeded to attack Likuangole town, using eight columns of fighters. Since early warning efforts by UNMISS had led most civilians to flee the town, relatively few were harmed during these attacks. Nevertheless, the attacks led to much material damage, and most of the buildings were destroyed. Some 100 civilians sought shelter in the SPLA barracks situated 3 kilometers outside the town, which were spared from attacks. The White Army continued to attack civilian settlements south of Likuangole

Table 7.5 Phase 3 (December 23–30, 2011): Multiple White Army attacks Date

White Army actions

White Army function(s) of force

UNMISS/SPLA actions

UNMISS/SPLA functions of force

Outcome

Dec. 23–30, 2011

On Dec. 23, White Army attacked Wuno, where they established a base

Destruction

On Dec. 23, UNMISS air reconnaissance observed White Army attack on Wuno (Likuangole)

Amelioration/ deterrence

At least 175 persons killed Dec. 23–6

Between Dec. 24 and 26, White Army broke into smaller groups to attack a number of smaller villages in Likuangole along the Nanaam river On Dec. 27, White Army attacked Likuangole town with eight columns of fighters Between Dec. 27 and 30, White Army attacked surrounding villages east and south of Likuangole On Dec. 30, White Army proceeded southward toward Pibor Town, continuing attacks on villages along the way

On Dec. 25, more than 50% of UNMISS troops (8 out of 15 companies) deployed to Jonglei No efforts made by SPLA or UNMISS to deter White Army from attacking Likuangole On Dec. 27, SPLA sheltered 100 civilians in barracks outside Likuangole On DEC 28, the Vice-president met with the White Army in Likuangole to urge them not to attack Pibor UNMISS aerial reconnaissance observed White Army toward Pibor

On Dec. 27, 90 tukuls (huts) burned in Likuangole, and hard wall structures destroyed Relatively few people killed in Likuangole town, as most had received early warning, fleeing before arrival of White Army

FAILING TO MATCH THE WHITE ARMY

163

in the following days. At this point, UNMISS infantry deployments were still underway. By December 25, two days before the attacks on Likuangole, UNMISS had deployed four platoons in Likuangole, four platoons in Bor, two platoons in Gumuruk, three platoons and three APCs in Pibor, and one platoon to the north of Walgak. SPLA had 512 troops stationed in the barracks outside Likuangole at the time of the attacks on December 27. The battalion deployed there had recently received an additional 100 troops but made no active efforts to intervene to deter the White Army. In fact, according to an interview with the SPLA commander at Likuangole—again undertaken by UNMISS Human Rights Division—he had been ordered only to return fire in self-defense. UNMISS troops evacuated 32 civilians from Likuangole, but were also heavily outnumbered, leading to no military response to deter the perpetrators. The function of violence employed by the White Army fulfills the criteria of destruction (see Chapter 2). To match a perpetrator seeking to destroy civilian life and infrastructure, military efforts to protect must destroy or significantly reduce the perpetrator’s will and capability to launch such destructive attacks against civilians. It seems clear that the response mustered by both UNMISS and the SPLA during this phase mismatched the ways and means by which the White Army targeted civilians. Undoubtedly, the early-warning efforts had a positive effect—not least in Likuangole—where most civilian lives were spared. However, since early warnings did not reach all potential victims, the absence of a relevant military response led to many casualties outside Likuangole in this phase.

Phase 4 (December 31, 2011–January 4, 2012): White Army Pibor attack, SPLA/UNMISS defense, and circumnavigation On December 31, the White Army columns had reached the outskirts of Pibor town, where UNMISS and the SPLA had established defensive positions around the town perimeter. The White Army managed to destroy some infrastructure in the southern parts of town but was denied access to the town center. Some 90,000 civilians had already fled south along the Kangen River and east toward Pochalla (Johnson 2016, 113). Over the following days, the White Army positioned itself on the eastern side of Pibor River, which runs east of the town. On January 2, some 150–200 White Army fighters crossed the river in a new attempt to attack the town. SPLA responded with force, killing five

164

OVERWHELMED BY THE WHITE ARMY

of the attackers. UNMISS APCs also repositioned to deter further advances from the White Army from the east. While presenting a significant threat to Pibor town during these days, the White Army also continued attacks against Murle settlements and IDPs south and east of the town. According to one of the interviewees in the UNMISS investigation, many Murle were killed in the continuing attacks south and east of Pibor “mainly because the Murle had all believed […] that the armed Lou Nuer youth would not proceed beyond Pibor” (UNMISS Human Rights Division 2012, 19). Many reported that they had received no warning of the attacks, according to the same investigation. During the attacks on Pibor, the White Army applied coercive violence to gain access to the Murle’s main population center in Jonglei. To match coercive functions of violence, the protectors must influence the perpetrator into disengaging from a particular strategy of violence or engage in direct military action that reduces the perpetrator’s will and ability to continue its attacks on civilians, without causing more harm in the process. The SPLA and UNMISS matched the White Army at the standoff at Pibor Town, successfully protecting those that had remained there. However, little was done to protect those fleeing south and east, or to warn those that already inhabited those areas. According to the SRSG Hilde Johnson, “the continued attacks headed deep into Murle land, where there were no roads, very few helicopter landing sites, and thus no way our forces could pursue [the White Army] through the bush” (Johnson 2016, 112). It seems that the Murle also misread the situation, believing that the White Army would not proceed beyond Pibor town. This was a highly deadly combination for the Murle, leading to hundreds of fatalities. Table 7.6 provides a schematic overview of events in Phase 4.

The absence of relevant responses to Murle revenge attacks Between December 27, 2011 and February 4, 2012, the Murle launched 44 retaliatory attacks into Nuer and Dinka territories, killing 276 persons, abducting 25, and stealing more than 60,000 cattle (UNMISS Human Rights Division 2012, 20). Few attempts were made by UNMISS or the SPLA to protect Nuer and Dinka civilians from these attacks, many of which occurred deep into Nuer and Dinka territories. Attacking civilian settlements with small bands of armed fighters, the Murle perpetrators were less visible from the air. However, the attacks were highly deadly and destructive, constituting almost one third of the combined civilian death toll of these events and leading to much material destruction. Understandably, UNMISS and the SPLA had their

Table 7.6 Phase 4 (December 31, 2011–January 4, 2012): White Army Pibor attack, SPLA/UNMISS defense, and circumnavigation Date

White Army actions

White Army function(s) of force

UNMISS/SPLA actions

UNMISS/SPLA function(s) of force

Outcome

Dec. 31—Jan. 4

On Dec. 31, White Army arrived at Pibor Town

Coercion/ destruction

On Dec. 31, UNMISS aerial reconnaissance saw two columns of fighters approx. 5–10 km outside Pibor

Deterrence/ coercion

Many Murle fled Pibor along Kangen River (south-east of Pibor), toward Pochalla (east of Pibor), before the arrival of the White Army

On Dec. 31, White Army burned tukuls, church, and NGO offices at Pibor’s southern perimeter From Dec. 31, most of the White Army fighters were present east of Pibor town, across the Pibor river

SPLA and UNMISS had established dug-in positions around the town perimeter On Jan. 2, SPLA fired on the advancing fighters, killing five and wounding two

On Jan. 2, 150–200 fighters advanced to cross Pibor River in an attempt to attack Pibor town from the east

UNMISS moved two armored personnel carriers (APCs) to eastern side of town to help deter attacks

Between Dec. 31 and Jan. 3, smaller groups of hundreds of White Army fighters attacked villages east and south of Pibor Town

Few efforts from UNMISS/SPLA to warn IDPs south and east of Pibor of oncoming attacks

On Jan. 3 and 4, White Army retreats

SPLA and UNMISS effectively denied White Army access to Pibor Town, protecting many Murle seeking shelter there Several hundred Murle killed by White Army south and east of Pibor town during and after the failed attacks on Pibor on Dec. 31 White Army stole tens of thousands of cattle and abducted a significant number of Murle

166

OVERWHELMED BY THE WHITE ARMY

hands full with preparing to defend Pibor town at the time when the revenge attacks commenced. Still, the Murle revenge attacks continued for a full month after the White Army had returned home, questioning why no further efforts were made to at least warn communities under threat. UNMISS was aware of the attacks and had received the first reports of Murle attacks on the very same day they started (Johnson 2016, 113). Furthermore, on January 8, UNMISS helped evacuate wounded civilians from Yuai, which had been attacked by Murle (UNMISS Human Rights Division 2012, 21). UNMISS also did send additional forces to Akobo, where the threat of Murle attacks was deemed most severe (Johnson 2016, 115). However, due to the absence of any relevant military response to these revenge attacks, there is no foundation for assessing the degree of matching the perpetrators of violence.

Pre-emption Pre-emption occurs when UN troops attempt to stop attacks on civilians before they occur. This case does not fulfill the criteria for pre-emption—as scored in the UNPOCO dataset—from a military point of view. There are still certain characteristics pointing toward pre-emptive logic that are worth highlighting. Although there were no pre-emptive military operations to deter or coerce the White Army from marching on the Murle—despite the obvious violent potential of their mobilization—UNMISS early-warning efforts effectively removed many potential targets from harm’s way. This approach is also controversial, as it is extremely challenging to protect civilians on the run, and it may assist the perpetrators in reaching their goals. In addition, the early-warning efforts seem to have failed to reach many communities, both west of Likuangole and south and east of Pibor. Undoubtedly, UNMISS sought to influence the SPLA into pre-emptive operations, but the national army’s response was sluggish. Only on December 31, over a week after the first attacks on the Murle, did the President order 3,400 SPLA soldiers to reinforce the troops already stationed in Jonglei, to little effect on civilian security (UNMISS Human Rights Division 2012, 26–7).

Troop-to-perpetrator ratios In UNMISS’s investigation of these events, the lack of sufficient troop numbers appears as one of the main explanations for the absence of a more relevant

TROOP-TO-PERPETRATOR RATIOS

167

military response to the threat posed by the White Army (UNMISS Human Rights Division 2012, 17, 25–6, 31). Although no exact figures exist for the number of White Army fighters, the estimates range from 6–8,000 up toward 12,000 (UNMISS Human Rights Division 2012, i). In comparison, during the attack on Likuangole on December 27, about 500 SPLA soldiers were present in the barracks a few kilometers outside town. In addition, four UNMISS platoons were present in Likuangole at that time, constituting some 120 Blue Helmets (UNMISS Human Rights Division 2012, 14). Severely outnumbered, neither SPLA nor UNMISS made any effort to intervene at that point. However, from December 31 to January 2, an almost similar number of troops were present effectively to defend Pibor town. At that point, 542 SPLA soldiers were present alongside almost 100 UN troops. These forces combined were not able to protect civilians outside the town perimeter, where most of the violent acts were committed. Various estimations exist of the total number of UN troops available during these operations. On paper, UNMISS had about 3,600 infantry troops available in December 2011(Johnson 2016, 99). This consisted mainly of 2,201 Indian troops and 1,359 troops from Bangladesh (United Nations 2011). However, Kenya also had a battalion deployed, counting 691 contingent troops (United Nations 2011). The investigation performed by UNMISS reports that, by December 25, UNMISS had deployed “more than 50 percent of its infantry to hot spots expecting attacks,” amounting to “eight of its 15 companies” (UNMISS Human Rights Division 2012, 14). Further, the investigation reports that “four platoons were deployed to Likuangole, four platoons to Bor, two platoons to Gumuruk, three platoons and three armored personnel carriers to Pibor, and one platoon to the north of Walgak” (UNMISS Human Rights Division 2012, 14). However, only those dispatched to Likuangole and Pibor had any opportunity to influence the White Army.⁴ Furthermore, on December 31, “the President ordered 3,400 SPLA infantry troops to be deployed to Pibor and 800 armed SSPS be deployed by road from Bor to Pibor and Gumuruk” (UNMISS Human Rights Division 2012, 27). These arrived too late to make any difference to the overall casualty figures. One could indeed question the wisdom of spreading out UNMISS military capacity so thinly—as each of these deployments would be completely outnumbered by the White Army—rather than amassing a larger force that may provide some deterrent effect. However, UNMISS’s logistics capacity was completely stretched, and amassing any larger force was near impossible.⁵ In ⁴ Interview with (ret.) Colonel Ebbe Deraas, UNMISS Force Chief of Staff. Kjeller, 10.17.2018. ⁵ Ibid.

168

OVERWHELMED BY THE WHITE ARMY

addition, the civilian airplanes used to deploy troops did not allow for transporting more than personal equipment. For example, for those UN troops flown into Pibor, no additional ammunition was available, besides what they carried on their bodies. The only possibility for obtaining more troops was to hope for SPLA reinforcements. When these troops first were deployed, they arrived too late to make any impact on the overall outcome. Thus, more troops were indeed available, possibly enough to constitute a significant deterrent force against the White Army. One of the reasons for the absence of any significant response from the SPLA was possibly the ethnic composition of the army. The SPLA “included several Nuer units, which made any attempt to counter their own kin severely challenging.”⁶ To sum up, although troop numbers clearly mattered, the events in Jonglei indicate that it also matters what the troops do to protect civilians. Significantly outnumbered in relation to the number of perpetrators, the SPLA and UNMISS managed to defend Pibor town—the main Murle population center in Jonglei state—“only” killing five White Army fighters. The White Army looks to have been more easily deterred than one would have expected, based on their sheer numbers. Nevertheless, the marginal number of troops mustered by UNMISS and SPLA, compared to the massive number of White Army fighters, is most likely part of the explanation why a more robust, and possibly more effective, response, never materialized.

Willingness to accept risk India and Bangladesh provided most of the infantry forces—respectively providing 2,201 and 1,359 troops—while Kenya provided 691 contingent troops. In a previous analysis in this book, all three contributors were coded amongst the more hesitant, often deploying with significant caveats as to how and where their troops should operate (Providing for Peacekeeping 2020)). In this case, however, all three TCCs were willing to take significant risk to protect civilians. Even before the attacks started in late December, UNMISS had deployed troops to volatile areas in Jonglei state. Troops were stationed in Bor, Pibor, Likuangole, and Akobo. As we have seen already, the troops at Likuangole were involved in the evacuation of some civilians before the attacks on December 27. However, it was not until the attack on Pibor that the risk to UN troops ⁶ Ibid.

WILLINGNESS TO ACCEPT RISK

169

increased significantly. India, Bangladesh, and Kenya all deployed to Pibor to support the SPLA in defending the town from the massive threat. Keep in mind that on the same day of the arrival of the perpetrators, the SPLA counted some 550 troops, while UNMISS only mustered about 100 troops. To defend Pibor, UNMISS also redeployed a lightly armed Indian platoon from Bor. Helicopters transported the personnel, while the three APCs were driven from Bor to Pibor, about 200 kilometers away. They arrived on the evening of December 31, the same day the White Army arrived on the outskirts of town (Johnson 2016, 111). A Bangladeshi platoon was flown in from Juba, and a platoon from Kenya was flown in from Western Bahr-el Ghazal. The UN troops were concentrated in dug-in positions around the WFP storehouse. UNMISS leadership and troop contributors took a leap of faith by positioning troops between the perpetrators and their potential victims. According to Ebbe Deraas, a Norwegian Army Colonel involved in the preparation, UNMISS “expected heavy UN casualties if the White Army decided to direct their rage toward us. However, we did not perceive that the White Army saw the UN as a threat and decided that it mattered more to make a robust stance at Pibor town. But it was a very difficult decision, as we knew we were completely outnumbered.”⁷ The SPLA provided the main defense of Pibor, and UNMISS troops fired no shots. Still, the supporting role of UNMISS troops was probably part of the explanation for why the White Army never attempted further attacks on Pibor from the east. Would more willing troop contributors have made a difference to the outcome in this situation? Again, according to Colonel Deraas, “more riskaccepting troops may have been more eager to deploy, but we really did not have the capacity to send more troops. And luckily, we did not have to engage the White Army directly, as we had no capacity to evacuate wounded troops or sustain them logistically in the case of a prolonged operation.”⁸ Indeed, the troops involved in the operations showed incredible courage by departing for Pibor at that time. Moreover, when the concept of operations was defending key population centers, not pro-active attempts at deterring the perpetrators before they launched their attacks, country of origin mattered little in this context. Rather, it shows that risk averse TCCs also sometimes put their troops in harm’s way leading to positive (local) protection outcomes.

⁷ Ibid. ⁸ Ibid.

170

OVERWHELMED BY THE WHITE ARMY

Additional explanations for the unsuccessful outcome Following the attempt to trace the effects of the four most relevant causal candidates, I introduce three other causal condition candidates that have appeared during the analysis: force mobility, host nation support, and the ability to understand the perpetrators of violence.

Force mobility and projection Jonglei is the largest of South Sudan’s states (122,581 sq. km, or slightly smaller than Bangladesh), with a population of about 1.3 million (Wikipedia 2018). It is also the least developed, “virtually without functional roads” (Johnson 2016, 101). Adding to this, one of the world’s biggest wetlands—the Sudd— covers much of Jonglei’s territory, making it inaccessible by road for almost eight months of the year (Johnson 2016, 101–12). Due to the size of the operational area and the inaccessibility of many remote civilian settlements, UNMISS struggled to deploy beyond the main population centers, and, even then, helicopters were critical to move troops and equipment. Making matters worse, UNMISS lost four of its military helicopters in November 2011, when Russia pulled out of the mission. The remaining air assets remained critical throughout the events. Most of the troops taking part in the defense of Pibor town were flown in. The APCs that relied on transport by road only arrived the same day as the White Army approached the town, leaving little time for preparations. Finally, the early-warning efforts provided by UNMISS in the early phases of the event, possibly the most effective protection tool in this situation, were largely facilitated by air mobility and teams of civilian UN staff. Air assets are expensive and there will probably never be enough of them provided to cover the needs of UN peace operations (Novosseloff 2017). Nevertheless, as protection of civilians remains a core activity for UN missions, helicopters remain a core enabler for early warning, force mobility, and sometimes the use of force.

Host-nation support One of the conditions explaining the feeble military response to the threat posed by the White Army is the reluctant approach of the SPLA to addressing communal conflict and to using military forces to protect civilians (Johnson 2016, 98). Clearly, the SPLA was the strongest military actor in South Sudan

ADDITIONAL E XPL ANATIONS FOR THE UNSUCCESSFUL OUTCOME

171

at the time of the events. If the Government had responded more swiftly to the pleading of UNMISS’s leadership to deploy more troops to defend the Murle, the SPLA would most likely have been able to deter or coerce the White Army from marching south, saving hundreds of lives, although perhaps risking the lives of others. According to Special Representative of the Secretary-General: “the SPLA never warmed to the idea that protection of civilians was central to its role: this was a job for the police. Besides, protecting civilians could be seen as the favoring of one community against another; they preferred to stay away […]” (Johnson 2016, 105). When the President finally decided to reinforce the SPLA in Jonglei, it was largely too late to save civilian lives, except for defending Pibor town. With limited host-nation support, there was little UNMISS could do besides its significant efforts to warn civilians from the impending threat and to deploy its limited number of forces to key population centers. One of the reasons for the lukewarm attempts to protect their own citizens was possibly linked to the ethnic composition of the SPLA and the White Army, and there was most likely never much hope of mustering a relevant military response in that particular context.

Understanding the modus operandi of perpetrators of violence While UNMISS clearly understood the violent potential of the White Army long before the attacks were underway in late December 2011, the mission misread the ultimate motivation of the perpetrators, which went far beyond a “traditional” retaliatory attack. The White Army attacks rather sought to target “entire communities, including women and children, […] possibly aimed at destroying their livelihoods and social and economic infrastructure. Hate speech and incitement to violence based on ethnicity […] contributed to the violence” (UNMISS Human Rights Division 2012, i). In addition, while UNMISS and SPLA thought Pibor town would be the main objective for the Nuer youth—consequently amassing its defensive forces in and around the town—the White Army rather sought to target the Murle wherever they could be found. Civilians were the center of gravity for the White Army, not a particular town. Although UNMISS and SPLA saved many civilians in Pibor, most of the victims were killed elsewhere, where there was no UN or SPLA presence. Indeed, UNMISS was aware of the destructive tactics of the White Army long before they reached Pibor, as UN troops deployed in

172

OVERWHELMED BY THE WHITE ARMY

Likuangole had assisted in the evacuation of vulnerable civilians before the White Army arrived. Still, some of the attacks—for example along the Nanaam River north-west of Likuangole—only became known to UNMISS through their own investigation in the aftermath (Johnson 2016, 112). Similarly, the violent reprisal attacks committed by the Murle, largely went unnoticed by UNMISS and the SPLA, and few efforts were made to soften the impact of these attacks on Nuer and Dinka communities. Certainly, the modus operandi of the Murle made detection and early warning challenging, but the revenge attacks lasted over a month, with fatality figures close to 300. These observations underline that understanding the perpetrator of violence is a key element in any military effort to protect civilians from violence, supporting the thesis introduced by Beadle on how utility of force to protect can be found (Beadle 2015). Although UNMISS made a remarkable effort to warn and report on the threat emerging from the White Army and made a brave choice in defending Pibor town alongside the SPLA although vastly outnumbered, the mission failed to comprehend the full extent of threats to civilians generated by this situation.

Conclusion This case provided a tougher test for the four causal condition candidates. The violent events from Jonglei facilitate a study of a “deviant” case, where the condition of most interest—matching—were much less pronounced than in the typical case from the DRC. Although unable to prevent the White Army from marching on the Murle in the first place, tactically matching the perpetrator still seems to have led to positive outcomes in crucial sub-phases of the attacks, such as in Pibor town. However, the attacks (from both parties) led to hundreds of civilian deaths combined, many more injuries, and much material destruction. The events are clearly the deadliest of all the 200 cases captured in the UNPOCO dataset. The tactical protection victories neither influenced the overall motivation and capability of the White Army to continue its attacks nor deterred the Murle’s immediate acts of revenge. To trace the effects of the four conditions deemed causally relevant, I have found that UNMISS matched the perpetrators of violence during the mobilization phase, by providing early warning to those under threat. However, as the perpetrators shifted to destructive tactics, the ameliorating efforts of UNMISS were not nearly enough to prevent violence against civilians. One exception was the defense of Pibor town alongside the SPLA, where they

CONCLUSION

173

matched the White Army’s coercive efforts to breach the defensive perimeter, effectively protecting those that had sought refuge there. Neither SPLA nor UNMISS was able to protect those outside the town center, where most of the killing occurred. Early-warning efforts somewhat fulfilled the criteria for pre-emption, but in the absence of a more robust pre-emptive response before attacks commenced, hundreds of civilians were still killed. Judged by the limited number of SPLA and UNMISS troops involved in the operations—and the fact that the White Army largely fulfilled its aim of severely punishing the Murle—there was little deterrent effect from the number of protective forces deployed. Still, although few in numbers, they managed to deter and coerce the White Army from taking Pibor town. This indicates that it matters what military protectors do to protect, more than being present in large enough numbers. Furthermore, although all TCCs involved in the operations are considered risk-averse, they did in fact take considerable risk in deploying to Pibor town, which supported the overall positive outcome at this time. Nevertheless, it remains unknown if more risk-accepting troops would have had a better impact on civilian security. Although the UNMISS response ameliorated the security situation for many civilians under threat, one critical question remains unanswered: Were there any windows of opportunity to deter or coerce the White Army military before its march commenced? To respond, I return to similar events that unfolded in Abyei in March 2014, where UNISFA successfully managed to intervene to deter a column of some 4–5,000 armed Misseriya fighters from proceeding to Abyei town, protecting all potential victims (United Nations 2014c). In that case, the Misseriya were easily deterred by the presence of armed UN troops. Certainly, Abyei is a much smaller theatre of operations than is Jonglei. In addition, the White Army was intent on punishing the Murle for earlier actions against themselves. They were highly motivated to go through with their violent campaign. Most importantly, the only actor that could potentially pose a threat to the White Army—the SPLA—responded too slowly and half-heartedly. However, the White Army was deterred at Pibor town, which perhaps indicates that an early joint forceful response by UNMISS and the SPLA could have influenced their decisions to march, potentially protecting all civilians under threat from the White Army mobilization.

8 Increasing the utility of force to protect “The UN needs a better framework for assessing perpetrator threats, their severity and intent, and the likely courses of action needed to prevent or mitigate that harm. Without that analysis in the initial (and on-going) mission design, as part of the political strategy, and as the basis for preparing the leadership, missions will lack understanding of the choices they may face or the political and materiel support they need.” —Victoria K. Holt (2019) Then Managing Director at the Henry L. Stimson Center (Gorur et al. 2019, 30)

Victoria Holt—one of the most influential voices in the international discourse and politics on protection of civilians over the past two decades—points to several remaining gaps in UN peace operations’ approach to protecting civilians from violence. Among the most severe is our lack of understanding of “what works,” in other words which courses of action are most effective in different situations. In short, Holt observes that we still do not understand the anatomy of military protection successes across time and operations. Hopefully, this book provides new knowledge to increase such understanding. Building on a new dataset—bridging the gap of missing systematic event data—it captures the core characteristics of 200 military protection operations at the tactical and operational levels in ten UN missions across Africa from 1999 to 2017 (Kjeksrud 2019). Besides describing the core characteristics of an understudied phenomenon, the empirical mapping also facilitates an evaluation of variations in outcomes of military efforts to protect across time and locations. Somewhat surprisingly, Blue Helmets perform better than expected when they apply force to protect, although there is certainly room for improvement. Even more, my analysis provides stepping-stones for a future theory of the utility of military force to protect civilians from violence. Through a mixed-methods approach I identify a combination of two conditions— a causal recipe—that seem generalizable across many cases, although many cases remain unexplained. In the following, I provide a short summary of the

Using Force to Protect Civilians. Stian Kjeksrud, Oxford University Press. © Stian Kjeksrud (2023). DOI: 10.1093/oso/9780192857101.003.0008

SUMMARY OF MAIN FINDINGS

175

main findings and show how this book is relevant for research, policy, and practice of using military force to protect civilians in UN peace operations.

Summary of main findings I set out to answer two questions different, but interrelated, questions. First, I asked to what degree UN military troops have provided protection to civilians under imminent physical threat in Africa between 1999 and 2017. My estimations indicate that UN troops in Africa have used force to protect civilians effectively on 76 occasions, alongside an almost similar number of failures. This finding is quite remarkable, as it underlines that UN troops are certainly able to protect civilians from violence. It challenges deep-rooted perceptions that the UN is unfit to wield force to improve security for those under imminent threat of violence. Still, although successes do happen, UN troops often struggle to muster effective military responses to protect civilians from perpetrators that harm them as part of their warfare. This confirms what we already knew about existing challenges and indicates that there still is some potential to improve UN troops’ ability to protect by force. However, we should also acknowledge that providing protection to civilians under threat is challenging, and that failures will also occur in the future. These findings do not come without caveats. More than 20 percent of the cases collected for the analyses in this book have unknown outcomes. Furthermore, many cases were most likely never reported via the Secretary-General’s reporting to the Security Council. I cannot therefore be certain that the distribution of successes and failures is representative of the entire universe of cases where UN troops have used force to protect. In addition, due to limited depth and consistency of openly available information, I have only been able to broadly estimate variations in outcomes across cases based on counterfactual reasoning, the modus operandi of perpetrators of violence, and case specific knowledge. The findings should be read in that light. Perhaps most troubling, I found that UN forces only addressed a small fraction of the violence perpetrated against civilians in the conflict areas where they deploy. Others had already found that UN troops seldom apply force to protect (United Nations Office of Internal Oversight Services 2014). However, compared with comprehensive datasets on violence against civilians, my compilation of 200 reported protection operations starkly underlines that the demand for protection significantly outweighs what Blue Helmets can provide.

176

INCRE ASING THE UTILIT Y OF FORCE TO PROTECT

Second, I asked what determines UN military troops’ ability to protect civilians from physical violence. Rather than looking exclusively at failures, I sought to understand cases with more fortunate outcomes: military operations in which UN troops did protect civilians from imminent threats. I applied a mixed-methods research approach to explore this second question, including statistical analyses, counterfactual reasoning, fuzzy set Qualitative Comparative Analysis, and qualitative case studies. I found four promising causal condition candidates from existing literature—(i) deterrent presence, (ii) willingness to accept risk, (iii) pre-emption, and (iv) matching—and systematically pursued these throughout the analyses. I found that once UN peacekeepers decide to intervene militarily, the ability to understand perpetrators’ motives and modus operandi for attacking civilians—matching particular threats with tailored and timely (pre-emptive) military responses—often determine the extent to which UN troops achieve successful protection outcomes. Although we know that large, uniformed components decrease conflict intensity and civilian targeting across operations, I did not find this effect reflected in the outcomes of protection operations at the tactical and operational levels. Somewhat unexpectedly, I also found that risk-averse troops did just as well, or even better, as those troops coming from countries that are more willing to accept risk. The findings indicate that riskaverse troops sometimes also wield force effectively to protect civilians from violence. Although pre-emption and matching are often part of successful outcomes, this causal pathway is still not able to explain most positive outcomes across missions and time. We should keep in mind that finding such “covering laws” for any social phenomenon is quite unlikely. Therefore, in a final analytical step, I sought to trace hypothesized observable implications from the causal mechanisms underpinning matching and pre-emption in two qualitative case studies, with emphasis on substantiating the main findings from the cross-case fsQCA analysis. I also broadened the scope to search for new and omitted causal condition candidates. The first case was from the DRC in 2013—where the Force Intervention Brigade (MONUSCO) defeated the M23 alongside the FARDC—and the second case was from South Sudan in 2011–2012, where UNMISS, alongside the SPLA, sought to protect the Murle against the Nuer White Army. In the first case (DRC), I found that matching the perpetrators of violence, pre-emption, and willingness to accept risk were all part of the explanation for the successful outcome. Troop numbers also mattered. These findings reintroduced the potential causal relevance of troop numbers and risk-acceptance. However,

SUMMARY OF MAIN FINDINGS

177

while troop-to-population ratios remained low in this case, it seems like troopto-perpetrator ratios were a highly influential condition, as the protectors significantly outnumbered the M23 fighters. The findings also supported the relevance of pre-emption and matching the perpetrators. New potential causal conditions emerged, including the will and ability to escalate when deterrence fails, force mobility and projection, operational art, operational readiness, and understanding the modus operandi of the perpetrators of violence before tailoring a military protection response. The second case (South Sudan) indicated that matching the perpetrators of violence resulted in the protection of civilians, although this case also provided an example where mismatching the perpetrators ultimately led to hundreds of civilians being killed. Pre-emptive logic was present in UNMISS’s efforts to protect, but came in the form of early warning, removing many civilians under threat. Although this saved many lives, it became evident that it was extremely challenging to protect those on the run. Furthermore, insufficient troop numbers clearly influenced the choices made by UNMISS and SPLA, which also impacted the overall outcome. Interestingly, UNMISS and the SPLA managed to deter and coerce the White Army from entering Pibor town, although severely outnumbered. The combined force was not able to protect civilians outside the town, however, where most of the killing occurred. Other causal condition candidates also emerged in this case. Like the DRC case, force mobility and understanding the perpetrators of violence stand out as conditions that influence the outcome. In this case, the lack of host-nation support was perhaps the single most defining condition, which led to many civilian casualties and much material damage. To date, most research has highlighted the UN’s inability to provide effective physical protection, stressing problems such as troop-contributing countries’ reluctance to use military force when needed, insufficient pre-deployment training, and a general lack of military guidance on how to protect most effectively. In contrast, I find that military UN peacekeepers can use force to protect civilians effectively from imminent threats. The findings point to the need to tailor operational concepts and military protection practices based on a better understanding of the threat civilians are facing. Moreover, more than being present in large enough numbers, UN troops must sometimes engage militarily with armed groups who deliberately attack civilians, and they must do so in ways that match the perpetrator’s aims and actions. In sum, the book provides new ideas to improve future military protection practices, as well as theoretical contributions to an understudied field of inquiry.

178

INCRE ASING THE UTILIT Y OF FORCE TO PROTECT

Implications for theory, policy, and practice of UN military protection operations How can future research efforts and the policy and practice of military protection operations benefit from the insights provided in this book? Underpinned by data captured in UNPOCO, information collected during fieldwork in the DR Congo and South Sudan, and interviews I have conducted with key individuals, my findings support, challenge, and refine existing knowledge on the use of force to protect civilians in UN peace operations.

Contributions to existing knowledge Perhaps the most significant finding of this book is that to protect civilians effectively from violence, UN troops must match the perpetrators of violence, tailoring their responses to influence the will and ability of perpetrators to harm civilians. Put simply, it matters what UN troops do to protect civilians from different types of threats. Military forces usually develop doctrines, concepts, and practices to counter an opposing party’s violence against themselves, and to avoid harming civilians in the process. Protection by force as understood in UN peace operations challenges this notion, as the violence they are meant to counter is primarily targeting those that do not partake in the armed conflict. Victoria Holt et al.’s benchmark study on the UN’s ability to protect civilians highlighted the importance of tailored operations based on a proper understanding of perpetrators of violence (Holt, Taylor, and Kelly 2009, 196–7, 227–9). The only existing theory on the utility of force to protect civilians developed by Alexander Beadle rests mainly on this same idea (Beadle 2014). However, until now these claims have remained untested. This book provides the first empirical test of Beadle’s matching theory across operations and time. Based on my findings, matching the perpetrators of violence emerges as the most promising causal condition to explain successful outcomes across cases. Moreover, although a highly relevant condition, matching the perpetrators is not enough to protect successfully, as this condition relies on other conditions to form a sufficient explanation for successful outcomes. From the cross-cases analysis, matching and pre-emption combined were able to explain about half of the successful outcomes. The UN’s POC policy has already recognized pre-emption as an important factor to success (United Nations 2019b). However, the policy says little about how successful preemption can be achieved, and there is little systematic empirical evidence

IMPLICATIONS FOR THEORY, POLICY, AND PRACTICE

179

supporting this policy guidance. This book provides empirical evidence supporting the potential positive effects of pre-emptive operations and adds further nuance. I found that pre-emptive operations that also matched the perpetrators of violence were highly relevant to explain many successful outcomes. Put simply, it is not enough to be in the right place at the right time to intervene before perpetrators attack: the use of force must simultaneously be tailored to the threats civilians are facing. The two qualitative case studies pointed toward more proximate causal condition candidates that facilitated matching in these cases. Some of them are already well-known explanatory conditions, and I can only re-emphasize their importance. First, we already know that air assets are critical in remote and inaccessible areas where UN forces commonly deploy (Dorn 2014; Novosseloff 2017). My case studies indicated that helicopters were essential to project force beyond weak road infrastructure, collect critical information, and provide early warning to those under threat. Second, I found that engagement from the host nation is critical to increase the effect of UN efforts to protect. Much has been written about the consent of host countries—one of the bedrock principles of peacekeeping—and how its presence or absence can influence the effectiveness of UN peace operations (Howard 2008; Tardy 2011; United Nations 2008). Both cases indicated that UN troops played a supporting role to government forces in protecting civilians from violence, as they should, but that the level of engagement displayed by the host countries varied significantly, with substantial influence on the outcome of operations. In the DRC, the FARDC bore the brunt of the robust operations to defeat the M23, effectively supported by the FIB. In South Sudan, UNMISS urged the SPLA to take early action to protect. Failing to do so, however, led to hundreds of civilian fatalities, as the White Army was never deterred effectively from its violent march on the Murle. It also exposed that UNMISS was ill suited to protect civilians unilaterally. Only when the SPLA and UNMISS united to defend civilians seeking shelter in Pibor town did they manage to deter and coerce the White Army from attacking the town. However, they were not able to project force beyond the town, where most of the killing was perpetrated. The two case studies also re-emphasized the need to understand the perpetrators of violence before designing a specific protection response, supporting existing knowledge. I also found a potential causal condition candidate that is less pronounced in existing research on UN peace operations, namely the influence of operational art and personal initiatives. Both cases show that it matters how key individuals think and act when force is used to protect. In the DRC, Force Commander Santos Cruz established the security zone

180

INCRE ASING THE UTILIT Y OF FORCE TO PROTECT

around Goma without any intention to uphold it and without consulting headquarters or the delegates at the peace talks in Kampala. Rather, he sought to lure the M23 into overstepping the red line, which facilitated a more robust response, eventually leading to the armed group’s demise. In South Sudan, SRSG Johnson and others in the leadership urged the government to respond in time, but when they failed to do so, she used the mission to its maximum capacity to warn civilians ahead of the attacks, which saved many lives. This book does not provide a deep analysis of this aspect, but I believe such microdynamics can provide highly valuable insights about causes and effects of UN military-protection operations. Furthermore, the book adds to existing knowledge about the effect of sufficient troop numbers to protect civilians from violence. From quantitative studies on UN peacekeeping and civilian victimization in civil wars, we knew that large uniformed components do have significant conflict-reducing effects across conflicts and time (Hultman, Kathman, and Shannon 2019)). However, this effect is not reflected in the outcomes of protection operations at the tactical and operational levels. While my measurements are crude, there are indications that it matters more what UN troops do, rather than that they are merely present in large enough numbers. My observations do not conflict with the importance of large deployments of uniformed peacekeepers, as observed by Lisa Hultman and her colleagues. Rather, I highlight that both phenomena can occur simultaneously. This effect has also been observed by other scholars. While the presence of UN troops may reduce the overall conflict intensity, many missions remain ill-suited to actively address the surplus of violence remaining in the conflict areas where UN troops deploy (Costalli 2014, 358–9). While I found that large, uniformed components were not part of the causal explanations for successful outcomes across the macro-analyses of 126 cases taken from UNPOCO, troop numbers emerged as a significant factor to explain the outcomes in both qualitative case studies. In both cases, the UN deployed few troops in relation to the total population number in each country. However, troop-to-perpetrator ratios differed significantly. In the DRC, the FIB and FARDC considerably outnumbered the M23, while in South Sudan, UNMISS and the SPLA were completely outnumbered by the White Army. The first case was highly successful, with UN and government troops completely removing one of the main threats to civilians from the DRC. The second case highlighted that UNMISS and SPLA struggled to muster a relevant military response due to being too few to influence the motivation of the White Army, leading to hundreds of fatalities, many injuries, and tens of thousands of displaced people. While force ratios are part of the literature

IMPLICATIONS FOR THEORY, POLICY, AND PRACTICE

181

on counterinsurgency, stabilization operations, and “traditional” inter-state warfare, I am not aware of any efforts to analyze systematically the effect of different troop-to-perpetrator ratios when Blue Helmets use force to protect. My findings indicate that there is value in further pursuing this condition. Similarly, acceptance of risk influenced the outcomes in both qualitative case studies. The troops taking part in the FIB, supporting the FARDC in defeating the M23, deployed with no caveats and were prepared to take significant risks to combat the M23. This condition facilitated the robust response that eventually removed the M23 as a threat to civilians in the eastern DRC. In South Sudan, all troop contributors that took part are deemed risk-averse, but still took significant risk to protect civilians in Pibor. It took some effort to convince the contingents to deploy, but this particular situation would probably generate similar reactions among more risk accepting TCCs, as the troops were completely outnumbered. My findings indicate that risk-averse troops can protect civilians from violence under the right conditions. Finally, the UNPOCO dataset may be a valuable contribution to existing research. To the best of my knowledge, this is the first attempt at capturing the entire universe of cases where UN troops have used force to protect civilians from violence. UNPOCO provides a systematic compilation of cases reported to the UN Security Council from the inception of protection as a mandated task for UN troops in 1999, covering 18 years of UN military protection operations. Although many cases remain unreported, I believe this dataset provides a reasonably comprehensive overview of this understudied phenomenon. Besides providing rich descriptive statistics of military UN protection operations—a gap in this field until now—UNPOCO will potentially facilitate other comparative studies across time and UN missions.

Stepping-stones for future research While I set out to trace the effects of UN military protection operations and to seek explanations for their successes and failures across time and UN missions, I also was somewhat restrained by the macro approach and the choice of methods. QCA is best suited to explore macro conditions, which made sense when comparing 200 cases across time and locations, but it cannot easily handle more than a handful of causal conditions. It struggles to capture fully the multiple causal variations in particular operations. Although this analysis helped me unearth important explanatory conditions for variations in outcomes of operations, I was only able to explain about half of the positive outcomes.

182

INCRE ASING THE UTILIT Y OF FORCE TO PROTECT

Furthermore, while I found other potential conditions with the help of the case studies, I certainly did not exhaust the potential list of other explanatory conditions. Therefore, future studies would benefit from applying comparative qualitative case study designs to discover causal condition candidates at the micro levels of analysis, as also identified by other observers (Autesserre 2014). The two cases studied in this book are interesting for many reasons, but they are also outliers in the universe of UN military protection operations. I hope that the UNPOCO dataset can facilitate and inspire future studies of cases that are less known. Most critically, I believe more knowledge could be obtained by exploring military protection efforts performed by MONUSCO, the FIB, and the FARDC after the fall of M23. Operations against the ADF and the Patriotic Resistance Front of Ituri are perhaps of most interest, as they portray different outcomes from those achieved against the M23. Furthermore, operations performed by the UN mission in Abyei remain understudied, but probably contain useful insights about how to use force to protect civilians in large-scale communal conflicts, as UNISFA has done successfully on quite a few occasions (Osterrieder, Lehne, and Kmec 2015). In the Central African Republic, there are still many lessons to be learned from MINUSCA’s simultaneous efforts to counter predatory violence and communal conflict. Valuable studies of the conflict and UN mission in CAR already exist, but there is a need to update that knowledge and look more closely at the military protection efforts over the past years (Karlsrud 2015; Øen 2014). The study of UN military protection operations is still immature and suffers from poor theory development. This is understandable. The use of military force to protect is a phenomenon characterized by causal complexity, there is a lack of openly available datasets, and it is challenging to collect information systematically in conflict areas where UN troops deploy. I have not been able to form an encompassing theory of the utility of force to protect civilians in UN peace operations, as too many of the cases with successful outcomes remain unexplained. I am certain that we must systematically dig deeper into proximate causal conditions to come closer to a more holistic understanding of what works when. Nevertheless, I hope future studies will be able to build on some of my findings for this purpose. The matching theory remains our best stepping-stone to explain how military forces can succeed in protecting civilians from different types of threats. Nevertheless, it remains incomplete as a way of explaining what matching looks like in practice at the tactical and operational levels in UN peace operations. The case studies provided in this book add some insights, but we need

IMPLICATIONS FOR THEORY, POLICY, AND PRACTICE

183

more cases to build on to investigate whether some of the case-specific conditions, such as force mobility and the ability to escalate, can help explain outcomes across cases. Future studies could also prove valuable in explaining how pre-emptive operations come about, and if there are particular situations which allow for UN troops to intervene before perpetrators attack. Both the M23 and the White Army were quite easy to understand. Other perpetrators, however, will be near impossible to counter before they launch attacks. The LRA and the ADF are possibly two of the most notorious examples from contemporary conflicts, where deadly attacks on civilians are mostly performed by smaller groups of fighters moving in inhospitable terrain that provide effective cover from any counter efforts from the air. The Murle militias described in Chapter 7 also portray some of the same characteristics. Finally, considering that matching is close to being a necessary condition, and this study has put weight on understanding cases where matching—to various degrees— has occurred, a future case selection strategy would be to analyze cases where matching did not lead to the expected outcome. UNPOCO identifies 31 cases in which matching did occur, but with an unsuccessful protection outcome.

Implications for policy and practice Some of the findings provided in this book may be relevant for the policy and practice of UN military protection efforts. To match and pre-empt the perpetrators of violence, it is essential to understand how, why, and with what means perpetrators attack civilians. Although the UN system is able to collect considerable relevant information about the conflict dynamics in areas to which they deploy—including the characteristics of the perpetrators of violence— the organization has thus far not been able to develop useful POC-specific threat-assessment methods for those sent to protect by force. Some attempts exist, but they remain rather generic, failing to take into account the various motivations of different perpetrators to attack civilians, the wealth of information and research we already have about different armed groups that target civilians, and lessons learned from military efforts to protect (UN Integrated Training Service 2018). Threat assessment methods should differentiate systematically between various types of perpetrators; what type of actor is responsible for the majority of violence? What is his rationale for targeting civilians? What strategies and tactics are used for attacking civilians? What are the relevant military capabilities needed to target civilians? What is the expected outcome in terms of harm to

184

INCRE ASING THE UTILIT Y OF FORCE TO PROTECT

civilians if the perpetrator succeeds? Based on such information, it is possible broadly to designate different threats to different categories or threat scenarios, which all demand different military responses. This book provides empirical evidence to inform the development of such threat-assessment methods. Although I find that a more pro-active approach to protection is likely to yield better results, it is not certain that perpetrators always should be pre-empted. If UN troops become too forward-leaning, always seeking to intervene before civilians are attacked, the organization may be perceived as suppressive, potentially worsening the overall situation. Second, all attacks can obviously not be pre-empted, and one therefore needs to be careful using preemption as a blueprint solution to the problem. As such, more value lies in better understanding what facilitates successful matching, as this condition also cover situations where pre-emption is not warranted or feasible. Some of the conditions identified in the case studies could be used as stepping-stones to further explore such facilitating conditions for matching. Troop contributing countries also need better pre-deployment and inmission scenario training that rests on systematic knowledge both about different perpetrators of violence and about what military efforts have worked in the past to protect civilians against them. This knowledge is equally important in order to know when force is likely to have little impact on civilian security, or even increase the threat to civilians. I have not been able to discover all potential causal conditions in play when UN troops succeed in protecting civilians. There is great value in systematically exploring more cases of this phenomenon to identify causal conditions and mechanisms that are similar across comparable situations. UNPOCO provides one stepping-stone for identifying cases from which to learn, but this dataset is still limited to cases reported to the Security Council. The UN has access to a wealth of additional information that can significantly enrich this knowledge base and is well positioned to facilitate systematic studies of its own operations. I am aware that the capacity to do that is not always available in-house. It is also challenging for UN staff to be too critical of their own organization. Clearly, outside experts can help them in this process. Independent commissioned research groups have worked well in the past, such as the one led by Victoria Holt in the late 2000s. Cedric de Coning at the Norwegian Institute of International Affairs has established a global network of research institutes, universities, and think tanks that seek to systematically research the effectiveness of peace operations. Knowledge from this network should also inform future developments of training material for UN troop contributors.

IMPLICATIONS FOR THEORY, POLICY, AND PRACTICE

185

Future efforts would greatly benefit if systematic information about UN efforts to protect civilians were made openly available. This has been a continuous challenge throughout the process of working with this book. Reporting practices have certainly improved over time, but there is still scant analysis of the causes and effects of UN military efforts to protect civilians from violence in the openly available material. I am well aware that some information must remain undisclosed. However, POC is the key task for almost every UN troop deployed, and given UN missions’ uneven record of accomplishment, there is great potential value in providing more and better data to a larger group of observers. In order for Blue Helmets to plan and implement tailored operations that match the perpetrators of violence, they need an enhanced UN intelligence system. In 2017, the UN published its first comprehensive intelligence policy (United Nations 2017b). While this is a significant step forward, much work remains before UN troops on the ground are provided with actionable intelligence to facilitate more effective protection operations. To pre-empt attacks on civilians, UN missions also need better mobility to move troops and personnel to the right place at the right time. Both cases studies show that air mobility was key to protect civilians from violence. Unfortunately, air assets are expensive and there will probably always be too few planes and helicopters to cover all the needs of a UN mission. Therefore, there is a need to start discussing other forms of organizing and deploying troops that take part in UN peace operations, where mobility and flexible military logistic systems could be a core issue. This book indicates that deploying many uniformed personnel does not necessarily lead to better protection outcomes at the tactical and operational levels. It matters more what the troops do in particular situations. Moreover, I have found that contributors that are deemed risk-averse are also able to protect civilians from violence. Since such contributors most likely will continue to provide the majority of UN troops in the future, the UN should engage more deeply with them, rather than expect an influx of high numbers of riskaccepting troops. Some of the findings in this book might be helpful in this regard. First, using force to protect civilians has seemingly not made peacekeeping more dangerous. Most of the perpetrators UN troops and civilians are facing are loosely armed groups with only marginal military capabilities. For professional armies, they seldom represent a significant risk, although they may be highly deadly for civilians. Being able to project a more robust stance, which may include matching and pre-emption, can in fact increase the security of the UN troops.

186

INCRE ASING THE UTILIT Y OF FORCE TO PROTECT

Finally, as long as protection remains the primary task for almost every Blue Helmet, there is a need to think harder about what the UN can and cannot do to protect civilians from physical threats. While this book concerned with outcomes of tactical protection operations, it also portrays how the UN Security Council wields military force strategically as part of its overarching mandate to maintain international peace and security. My findings underline that even more knowledge is needed to better understand the utility of using force to protect civilians. UN peace operations—despite all their limitations and weaknesses—remain the Security Council’s most important tool to address threats to international peace and security (Coleman and Williams 2021). We also know that they are quite effective, significantly reducing the negative impact of armed conflict on civilian life. Still, Blue Helmets can do better. I am still surprised how little academic attention is given to understanding conditions leading to successful outcomes. I hope this book provides insights to inspire more knowledgeable debates to improve the utility of force to protect civilians. While Blue Helmets cannot save humanity from hell, they can indeed protect more people—under the right conditions—wherever they deploy.

Appendix Table A1 QCA-matrix From left to right, the columns contain: case ID (corresponding to case ID in UNPOCO), deterrent presence (deter), troop contributors’ willingness to accept (risk), pre-emptive/ reactive operations (preempt), the ability to match perpetrators by force (match), and the outcome variable (outcome). This matrix combines fuzzy and crisp scores. No

case ID

deter

risk

preempt

match

outcome

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

SierraLeone1 DRC1 (MONUC) Liberia1 DRC2 (MONUC) DRC3 (MONUC) Liberia3 Liberia4 Liberia5 Liberia6 DRC4 (MONUC) DRC5 (MONUC) DRC8 (MONUC) DRC10 (MONUC) DRC12 (MONUC) DRC13 (MONUC) DRC14 (MONUC) DRC15 (MONUC) DRC16 (MONUC) IvoryCoast1 Liberia7 DRC17 (MONUC) DRC19 (MONUC) Liberia8 DRC21 (MONUC) DRC22 (MONUC)

0.75 0.0 0.25 0.0 0.0 0.75 0.75 0.75 0.75 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.75 0.0 0.0 0.75 0.0 0.0

0.75 0.75 0.75 1.0 0.0 0.0 0.0 0.0 0.75 0.0 0.0 0.0 0.75 0.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.75 0.0 0.0 0.0

1.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 1.0 1.0 1.0 1.0 0.0 1.0 1.0 1.0 0.0 0.0 1.0 1.0 0.0 1.0 0.0

1.0 1.0 1.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

0.25 0.75 0.75 0.25 0.25 0.75 1.0 0.75 0.75 0.25 0.75 0.75 0.25 0.75 0.25 0.75 0.75 0.75 0.75 0.25 0.75 0.25 0.75 0.75 0.75

Continued

188

APPENDIX

Table A1 Continued No

case ID

deter

risk

preempt

match

outcome

26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

DRC23 (MONUC) Darfur1 DRC24 (MONUC) DRC25 (MONUC) Sudan1 Sudan2 Liberia9 Liberia10 DRC26 (MONUC) DRC27 (MONUSCO) DRC28 (MONUSCO) Darfur2 SouthSudan1 IvoryCoast2 IvoryCoast3 Abyei1 Abyei2 Abyei3 Abyei4 Abyei5 Abyei6 DRC44 (MONUSCO) DRC45 (MONUSCO) DRC47 (MONUSCO) DRC49 (MONUSCO) Darfur4 Darfur5 SouthSudan3 SouthSudan5 SouthSudan6 Abyei7 Abyei8 Abyei9 Abyei10 Abyei11 DRC51 (MONUSCO) DRC53 (MONUSCO) DRC54 (MONUSCO) DRC55 (MONUSCO) DRC56 (MONUSCO)

0.0 0.75 0.0 0.0 0.0 0.0 0.75 0.75 0.0 0.0 0.0 0.75 0.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 0.0 0.0 0.0 0.75 0.75 0.25 0.75 0.75 1.0 1.0 1.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0

0.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 0.0 0.0 0.75 1.0 1.0 0.0 0.75 0.0 1.0 1.0 1.0 1.0 1.0 0.0 0.0 1.0 0.75 1.0

0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 1.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 1.0 0.0

1.0 1.0 1.0 0.0 1.0 0.0 1.0 1.0 1.0 0.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 1.0 1.0 0.0 1.0 1.0 0.0 0.0 1.0 1.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0

0.25 0.75 0.75 0.25 0.75 0.75 0.25 0.25 0.75 0.0 0.75 0.25 0.75 0.75 0.75 0.75 1.0 0.75 0.75 0.75 1.0 0.25 0.75 0.75 0.25 0.25 0.75 1.0 0.25 0.25 0.75 1.0 0.75 0.0 1.0 0.75 0.75 0.75 0.75 0.75

66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110

DRC57 (MONUSCO) Darfur6 Darfur7 Darfur9 Mali1 SouthSudan10 SouthSudan11 SouthSudan12 SouthSudan13 Abyei12 Abyei13 Abyei14 Abyei15 DRC61 (MONUSCO) DRC63 (MONUSCO) DRC64 (MONUSCO) DRC66 (MONUSCO) Darfur10 Darfur11 Mali2 Mali4 Mali5 Mali6 DRC68 (MONUSCO) DRC69 (MONUSCO) DRC70 (MONUSCO) DRC71 (MONUSCO) DRC72 (MONUSCO) DRC74 (MONUSCO) DRC76 (MONUSCO) DRC77 (MONUSCO) DRC79 (MONUSCO) DRC81 (MONUSCO) MALI9 DRC82 (MONUSCO) DRC83 (MONUSCO) DRC85 (MONUSCO) DRC88 (MONUSCO) DRC91 (MONUSCO) DRC99 (MONUSCO) Darfur13 Darfur14 Darfur15 DRC102 (MONUSCO) DRC103 (MONUSCO)

0.0 0.75 0.75 0.75 0.0 0.25 0.25 0.25 0.25 1.0 1.0 1.0 1.0 0.0 0.0 0.0 0.0 0.75 0.75 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.75 0.75 0.75 0.0 0.0

1.0 1.0 1.0 1.0 0.75 0.75 0.0 0.0 0.75 1.0 1.0 1.0 1.0 1.0 0.0 0.75 0.75 1.0 1.0 1.0 1.0 1.0 1.0 0.75 0.75 1.0 1.0 0.75 0.75 0.75 0.75 0.75 1.0 0.75 0.75 0.75 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.75 0.75

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 1.0 0.0 1.0 0.0 0.0 0.0 1.0 0.0 1.0 1.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 1.0 1.0 0.0 0.0

APPENDIX

189

1.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0

0.25 0.25 0.25 0.25 0.25 0.0 0.25 0.25 0.75 1.0 0.0 0.0 0.0 0.75 0.25 0.75 0.25 1.0 1.0 1.0 0.75 1.0 1.0 0.25 0.75 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.75 0.25 0.25 0.75 0.25 0.25 0.25 1.0 1.0 1.0 0.25 0.25

Continued

190

APPENDIX

Table A1 Continued No

case ID

deter

risk

preempt

match

outcome

111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126

DRC106 (MONUSCO) Darfur16 Darfur17 Abyei16 Abyei18 Abyei19 Abyei20 Abyei21 Liberia15 IvoryCoast6 SouthSudan15 CAR1 CAR4 CAR6 CAR7 CAR8

0.0 0.75 0.75 1.0 1.0 1.0 1.0 1.0 0.25 0.25 0.25 0.25 0.75 0.75 0.75 0.75

0.75 1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.75 0.75 0.75 0.75 0.0 1.0 0.0 0.75

1.0 1.0 1.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 1.0 0.0 1.0 1.0 0.0

1.0 1.0 1.0 1.0 1.0 1.0 1.0 0.0 1.0 0.0 0.0 1.0 0.0 1.0 1.0 1.0

1.0 1.0 1.0 0.0 0.75 0.0 1.0 0.0 0.75 0.25 0.0 1.0 0.25 1.0 0.75 0.25

Table A2 Uniformed peacekeeping deployments No

Case ID

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Sierra Leone1 DRC1 (MONUC) Liberia1 DRC2 (MONUC) DRC3 (MONUC) Liberia3 Liberia4 Liberia5 Liberia6 DRC4 (MONUC) DRC5 (MONUC) DRC8 (MONUC) DRC10 (MONUC) DRC12 (MONUC) DRC13 (MONUC) DRC14 (MONUC) DRC15 (MONUC) DRC16 (MONUC) IvoryCoast1 Liberia7

Inhabitants per uniformed peacekeeper

Population

Troop numbers

366 5,649 694 4,950 4,194 219 216 217 203 3,925 3,925 3,387 3,338 3,368 3,368 3,375 3,306 3,260 2,939 207

4,564,297 51,390,033 3,116,233 53,034,217 53,034,217 3,176,414 3,176,414 3,176,414 3,176,414 54,751,476 54,751,476 54,751,476 54,751,476 54,751,476 54,751,476 54,751,476 54,751,476 54,751,476 18,336,303 3,261,230

12,481 9,098 4,487 10,715 12,646 14,496 14,739 14,616 15,632 13,950 13,950 16,163 16,402 16,258 16,258 16,221 16,561 16,791 6,237 15,775

Fuzzy score

Term

Sources

0.75 0.0 0.25 0.0 0.0 0.75 0.75 0.75 0.75 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.75

Mostly in Fully out Mostly out Fully out Fully out Mostly in Mostly in Mostly in Mostly in Fully out Fully out Fully out Fully out Fully out Fully out Fully out Fully out Fully out Fully out Mostly in

World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data

Continued

Table A2 Continued No

Case ID

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

DRC17 (MONUC) DRC19 (MONUC) Liberia8 DRC21 (MONUC) DRC22 (MONUC) DRC23 (MONUC) Darfur1 DRC24 (MONUC) DRC25 (MONUC) Sudan1 Sudan2 Liberia9 Liberia10 DRC26 (MONUC) DRC27 (MONUSCO) DRC28 (MONUSCO) Darfur2

Inhabitants per uniformed peacekeeper

Population

Troop numbers

3,365 3,113 214 3,182 3,197 3,183 463 3,173 3,043 3,250 3,251 334 332 2,955 3,264 3,320 321

56,543,011 56,543,011 3,375,838 58,417,562 58,417,562 58,417,562 7,000,000 62,409,435 62,409,435 32,154,490 32,154,490 3,811,528 3,811,528 64,523,263 64,523,263 64,523,263 7,000,000

16,803 17,519 15,808 18,357 18,275 18,352 15,136 19,670 20,509 9,894 9,891 11,406 11,471 20,819 19,685 19,437 21,816

Fuzzy score

Term

Sources

0.0 0.0 0.75 0.0 0.0 0.0 0.75 0.0 0.0 0.0 0.0 0.75 0.75 0.0 0.0 0.0 0.75

Fully out Fully out Mostly in Fully out Fully out Fully out Mostly in Fully out Fully out Fully out Fully out Mostly in Mostly in Mostly in Fully out Fully out Mostly in

World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data 5th Sudan Population and Housing Census—2008 World Bank/UN peacekeeping data World Bank/UN peacekeeping data 5th Sudan Population and Housing Census—2008 5th Sudan Population and Housing Census—2008 World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data 5th Sudan Population and Housing Census—2008

38

SouthSudan1

1,649

9,000,000

5,457

0.0

Fully out

39 40 41

IvoryCoast2 IvoryCoast3 Abyei1

2,213 2,213 26

20,895,311 20,895,311 100,000

9,444 9,444 3,796

0.0 0.0 1.0

Fully out Fully out Fully in

42

Abyei2

26

100,000

3,800

1.0

Fully in

43

Abyei3

26

100,000

3,920

1.0

Fully in

44

Abyei4

25

100,000

3,966

1.0

Fully in

45

Abyei5

25

100,000

3,966

1.0

Fully in

46

Abyei6

25

100,000

3,966

1.0

Fully in

47 48 49 50

DRC44 (MONUSCO) DRC45 (MONUSCO) DRC47 (MONUSCO) DRC49 (MONUSCO)

3,617 3,619 3,607 3,601

68,978,682 68,978,682 68,978,682 68,978,682

19,070 19,060 19,122 19,154

0.0 0.0 0.0 0.0

Fully out Fully out Fully out Fully out

Population estimate, based on the 2008 Sudan Population and Housing Census and Statistical Yearbook for Southern Sudan (2010) World Bank/UN peacekeeping data World Bank/UN peacekeeping data Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data

Continued

Table A2 Continued No

Case ID

Inhabitants per uniformed peacekeeper

Population

Troop numbers

Fuzzy score

Term

Sources

51 52 53 54 55 56

Darfur4 Darfur5 SouthSudan3 SouthSudan5 SouthSudan6 Abyei7

317 325 1,003 943 935 25

7,000,000 7,000,000 7,185,100 7,185,100 7,185,100 100,000

22,106 21,510 7,161 7,616 7,684 3,977

0.75 0.75 0.25 0.75 0.75 1.0

Mostly in Mostly in Mostly out Mostly in Mostly in Fully in

100,000

3,983

1.0

Fully in

25

100,000

3,952

1.0

Fully in

Abyei10

25

100,000

3,952

1.0

Fully in

Abyei11

25

100,000

3,952

1.0

Fully in

5th Sudan Population and Housing Census—2008 5th Sudan Population and Housing Census—2008 World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000)

57

Abyei8

25

58

Abyei9

59

60

61 62 63 64 65 66 67 68 69 70

DRC51 (MONUSCO) DRC53 (MONUSCO) DRC54 (MONUSCO) DRC55 (MONUSCO) DRC56 (MONUSCO) DRC57 (MONUSCO) Darfur6 Darfur7 Darfur9 Mali1

71 72 73 74 75

SouthSudan10 SouthSudan11 SouthSudan12 SouthSudan13 Abyei12

76

Abyei13

3,727 3,476 3,476 3,447 3,319 3,364 336 355 362 2,400

71,316,033 71,316,033 71,316,033 71,316,033 71,316,033 71,316,033 7,000,000 7,000,000 7,000,000 19,148,219

19,134 20,519 20,519 20,688 21,485 21,198 20,852 19,703 19,327 7,980

0.0 0.0 0.0 0.0 0.0 0.0 0.75 0.75 0.75 0.0

Fully out Fully out Fully out Fully out Fully out Fully out Mostly in Mostly in Mostly in Fully out

958 807 789 789 24

9,000,000 9,000,000 9,000,000 9,000,000 100,000

9,387 11,151 11,405 11,405 4,124

0.25 0.25 0.25 0.25 1.0

Mostly out Mostly out Mostly out Mostly out Fully in

24

100,000

4,124

1.0

Fully in

World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data 5th Sudan Population and Housing Census—2008 5th Sudan Population and Housing Census—2008 5th Sudan Population and Housing Census—2008 United Nations Population Division/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000)

Continued

Table A2 Continued No

Case ID

Inhabitants per uniformed peacekeeper

Population

Troop numbers

Fuzzy score

Term

Sources

77

Abyei14

24

100,000

4,109

1.0

Fully in

100,000

4,070

1.0

Fully in

3,479 3,474 3,480 3,505 379 379 1,791

73,722,860 73,722,860 73,722,860 73,722,860 7,000,000 7,000,000 17,467,905

21,189 21,219 21,187 21,033 18,472 18,472 9,754

0.0 0.0 0.0 0.0 0.75 0.75 0.0

Fully out Fully out Fully out Fully out Mostly in Mostly in Fully out

Mali4

1,791

17,467,905

9,754

0.0

Fully out

87

Mali5

1,516

17,467,905

11,511

0.0

Fully out

88

Mali6

1,516

17,467,905

11,511

0.0

Fully out

Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data 5th Sudan Population and Housing Census—2008 5th Sudan Population and Housing Census—2008 United Nations Population Division/UN peacekeeping data United Nations Population Division/UN peacekeeping data United Nations Population Division/UN peacekeeping data United Nations Population Division/UN peacekeeping data

78

Abyei15

25

79 80 81 82 83 84 85

DRC61 (MONUSCO) DRC63 (MONUSCO) DRC64 (MONUSCO) DRC66 (MONUSCO) Darfur10 Darfur11 Mali2

86

89 90 91 92 93 94 95 96 97 98 99

DRC68 (MONUSCO) DRC69 (MONUSCO) DRC70 (MONUSCO) DRC71 (MONUSCO) DRC72 (MONUSCO) DRC74 (MONUSCO) DRC76 (MONUSCO) DRC77 (MONUSCO) DRC79 (MONUSCO) DRC81 (MONUSCO) Mali9

3,618 3,618 3,624 3,851 3,851 3,851 4,076 4,085 4,069 4,225 1,539

76,196,619 76,196, 619 76,196,619 76,196,619 76,196,619 76,196,619 76,196,619 76,196,619 76,196,619 78,736,153 17,994,837

21,060 21,060 21,023 19,784 19,784 19,784 18,695 18,653 18,727 18,634 11,692

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Fully out Fully out Fully out Fully out Fully out Fully out Fully out Fully out Fully out Fully out Fully out

100 101 102 103 104 105 106 107

DRC82 (MONUSCO) DRC83 (MONUSCO) DRC85 (MONUSCO) DRC88 (MONUSCO) DRC91 (MONUSCO) DRC99 (MONUSCO) Darfur13 Darfur14

4,176 4,179 4,185 4,229 4,182 4,352 410 411

78,736,153 78,736,153 78,736,153 78,736,153 78,736,153 81,339,988 7,000,000 7,000,000

18,855 18,841 18,814 18,620 18,826 18,692 17,063 17,023

0.0 0.0 0.0 0.0 0.0 0.0 0.75 0.75

Fully out Fully out Fully out Fully out Fully out Fully out Mostly in Mostly in

World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data United Nations Population Division/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data 5th Sudan Population and Housing Census—2008 5th Sudan Population and Housing Census—2008

Continued

Table A2 Continued No

Case ID

Inhabitants per uniformed peacekeeper

Population

Troop numbers

108 109 110 111 112 113 114

Darfur15 DRC102 (MONUSCO) DRC103 (MONUSCO) DRC106 (MONUSCO) Darfur16 Darfur17 Abyei16

115

116

Fuzzy score

Term

Sources

410 4,331 4,380 4,512 420 420 24

7,000,000 81,339,988 81,339,988 81,339,988 7,000,000 7,000,000 100,000

17,093 18,780 18,571 18,029 16,673 16,673 4,090

0.75 0.0 0.0 0.0 0.75 0.75 1.0

Mostly in Fully out Fully out Fully out Mostly in Mostly in Fully in

100,000

4,525

1.0

Fully in

100,000

4,546

1.0

Fully in

5th Sudan Population and Housing Census—2008 World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data 5th Sudan Population and Housing Census—2008 5th Sudan Population and Housing Census—2008 Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000)

Abyei18

22

Abyei19

22

117

Abyei20

22

100,000

4,491

1.0

Fully in

118

Abyei21

22

1,000,000

4,538

1.0

Fully in

119 120 121 122 123 124 125 126

Liberia15 IvoryCoast6 SouthSudan15 CAR1 CAR4 CAR6 CAR7 CAR8

754 863 842 689 372 383 381 382

4,499,621 4,613,823 11,000,000 4,500,000 4,500,000 4,650,000 4,650,000 4,650,000

5,962 5,347 13,072 6,528 12,092 12,158 12,208 12,159

0.25 0.25 0.25 0.25 0.75 0.75 0.75 0.75

Mostly out Mostly out Mostly out Mostly out Mostly in Mostly in Mostly in Mostly in

Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) Approximate population number based on the 2008 Sudan population and housing census (54,000) and a 2014 estimate (124,000) World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data World Bank/UN peacekeeping data

Table A3 Troop contributors’ willingness to accept risk TCC

Crisp score

Description

Source I

Source II

Bangladesh

0.0

Hesitant

Bellamy & Williams (2013). Providing Peacekeepers. p. 185. Oxford University Press

India

0.0

Hesitant

Pakistan

0.0

Hesitant

China

0.0

Hesitant

Ghana

0.0

Hesitant

Philippines

0.0

Hesitant

Providing for Peacekeeping: http://providingforpeacekeeping.org/2014/04/03/ contributor-profile-bangladesh/ Providing for Peacekeeping: http://ipi-pdf-documentstore.s3-website-us-east-1.amazonaws.com/ppp-profiles/ asia/ipi-pub-ppp-India.pdf Providing for Peacekeeping: http://providingforpeacekeeping.org/2014/04/03/ contributor-profile-pakistan/ Providing for Peacekeeping: http://www.providingforpeacekeeping.org/wp-content/ uploads/2017/05/China-Chin-Hao-27Apr2017_FINAL. pdf Providing for Peacekeeping: http://www.providingforpeacekeeping.org/2014/04/03/ contributor-profile-ghana/ Providing for Peacekeeping: http://www.providingforpeacekeeping.org/2014/04/03/ contributor-profile-philippines/

Bellamy & Williams (2013). Providing Peacekeepers, p. 228. Oxford University Press Bellamy & Williams (2013). Providing Peacekeepers, p. 223. Oxford University Press

Bellamy & Williams (2013). Providing Peacekeepers, p. 269. Oxford University Press

Indonesia

0.0

Hesitant

Jordan

0.0

Hesitant

Morocco

0.0

Hesitant

Kenya

0.0

Hesitant

Cambodia

0.0

Hesitant

Egypt

0.0

Hesitant

Mongolia

1.0

Willing

Mauritania

1.0

Willing

Providing for Peacekeeping: http://providingforpeacekeeping.org/wp-content/ uploads/2016/02/ipi-pub-ppp-Indonesia.pdf Report of the Special Committee on Peacekeeping Operations 2010 substantive session (February 22—March 19, 2010) Morocco’s Statement in UNSC debate (January 21 2013): Le Maintien de la paix multidimensionnel. Providing for Peacekeeping: http://providingforpeacekeeping.org/2015/03/23/ contributor-profile-kenya/ Providing for Peacekeeping: http://ipi-pdf-document-store.s3-website-us-east-1. amazonaws.com/ppp-profiles/asia/ipi-pub-pppCambodia.pdf Egypt and peacekeeping: https://www.cairopeacekeeping. org/en/egypt-and-peacekeeping Providing for Peacekeeping: http://providingforpeacekeeping.org/2014/04/03/ contributor-profile-mongolia/ Mauritania in MINUSMA: http://northafricapost.com/ 4169-mauritania-sets-minusma-participation.html

Mauritania part of robust Sahel G5 initiative. https:// www.diplomatie.gouv.fr/en/french-foreign-policy/ defence-security/crisis-and-conflicts/g5-sahel-jointforce-and-the-sahel-alliance/

Continued

Table A3 Continued TCC

Crisp score

Description

Source I

Senegal

1.0

Willing

Ethiopia

1.0

Willing

Rwanda

1.0

Willing

South Africa

1.0

Willing

Uruguay

1.0

Willing

Nepal

1.0

Willing

Ireland

1.0

Willing

Sweden

1.0

Willing

http://reliefweb.int/report/world/opinions-divided-overprotection-civilians-fourth-committee-concludes-generaldebate Providing for Peacekeeping: http://providingforpeacekeeping.org/wp-content/ uploads/2017/11/Ethiopia-profile.pdf Providing for Peacekeeping: http://providingforpeacekeeping.org/2015/03/30/ peacekeeping-contributor-profile-rwanda/ Providing for Peacekeeping: http://providingforpeacekeeping.org/2014/04/03/ contributor-profile-south-africa/ Providing for Peacekeeping: http://providingforpeacekeeping.org/2014/04/03/ contributor-profile-uruguay/ Statement to UNGA (Nov 5 2015). Ambassador to UN Durga Prasad Bhattarai Providing for Peacekeeping: http://www.providingforpeacekeeping.org/wp-content/ uploads/2014/04/Ireland-Murphy-18April2017.pdf Providing for Peacekeeping: http://ipi-pdf-document-store.s3-website-us-east-1. amazonaws.com/ppp-profiles/europe/ipi-pub-pppSweden.pdf

Source II

http://civilianprotection.rw/wp-content/uploads/ 2015/09/REPORT_PoC_conference_Long-version. pdf Bellamy & Williams (2013). Providing Peacekeepers, p.376. Oxford University Press Bellamy & Williams (2013). Providing Peacekeepers, p. 312. Oxford University Press Bellamy & Williams (2013) Providing Peacekeepers, p. 291. Oxford University Press.

Gambia

1.0

Willing

Guatemala

1.0

Willing

Ukraine

1.0

Willing

Togo

1.0

Willing

Nigeria

1.0

Willing

Niger

1.0

Willing

Burkina Faso

1.0

Willing

Statement to UNGA by President Adama Barrow (Sep 25 2018) (https://www.un.org/africarenewal/sites/www.un. org.africarenewal/files/gm_en_0.pdf Providing for Peacekeeping: http://www.providingforpeacekeeping.org/2016/06/21/ country-profile-guatemala/ Statement by the Ukrainian Delegation at the general debate of the Special Committee on Peacekeeping operations: https://mfa.gov.ua/en/news-feeds/foreignoffices-news/18416-vistup-delegaciji-ukrajini-nazagalynih-debatah-speckomitetu-oon-z-pitanymirotvorchoji-dijalynosti Signatory to Kigali principles: http://www.globalr2p.org/ media/files/kp-signatories-31-july-2018.pdf Providing for Peacekeeping: https://s3.amazonaws.com/ipi-pdf-document-store/pppprofiles/africa/ipi-pub-ppp-Nigeria.pdf

Part of Sahel G-5 force. https://www.diplomatie.gouv.fr/ en/french-foreign-policy/defence-security/crisis-andconflicts/g5-sahel-joint-force-and-the-sahel-alliance/ Part of Sahel G-5 force. https://www.diplomatie.gouv.fr/ en/french-foreign-policy/defence-security/crisis-andconflicts/g5-sahel-joint-force-and-the-sahel-alliance/

Opinions divided over protection of civilians as Fourth Committee concludes general debate on peacekeeping matters http://reliefweb.int/report/ world/opinions-divided-over-protection-civiliansfourth-committee-concludes-general-debate Signatory to Kigali principles: http://www.globalr2p. org/media/files/kp-principles-17-november-2016. pdf Signatory to Kigali principles: http://www.globalr2p. org/media/files/kp-principles-17-november-2016. pdf

Continued

Table A3 Continued TCC

Crisp score

Description

Source I

Guinea

1.0

Willing

Chad

1.0

Willing

Signatory to Kigali principles: http://www.globalr2p.org/ media/files/kp-principles-17-november-2016.pdf Part of Sahel G-5 force. https://www.diplomatie.gouv.fr/ en/french-foreign-policy/defence-security/crisis-andconflicts/g5-sahel-joint-force-and-the-sahel-alliance/

Benin

1.0

Willing

Malawi

1.0

Willing

Tanzania

1.0

Willing

Netherlands

1.0

Willing

Gabon

1.0

Willing

Portugal

1.0

Willing

Benin Statement in UNGA (Oct 1 2018). https://gadebate. un.org/en/73/benin Providing for Peacekeeping: https://s3.amazonaws.com/ipi-pdf-document-store/pppprofiles/africa/ipi-pub-ppp-Malawi.pdf Providing for Peacekeeping: http://www.providingforpeacekeeping.org/wp-content/ uploads/2017/07/Tanzania-Minde-10July2017-FINAL.pdf Providing for Peacekeeping: http://ipi-pdf-document-store.s3-website-us-east-1. amazonaws.com/ppp-profiles/europe/ipi-pub-pppNetherlands.pdf Gabon statement in UNGA (Sep. 23 2010). https://www. un.org/en/ga/65/meetings/generaldebate/View/ SpeechView/tabid/85/smid/411/ArticleID/149/reftab/ 227/t/Gabon/Default.html Providing for Peacekeeping: https://s3.amazonaws.com/ipi-pdf-document-store/pppprofiles/europe/ipi-pub-ppp-Portugal.pdf

Source II

Karlsrud (2015). “The UN at war: Examining the consequences of peace-enforcement mandates for the UN peacekeeping operations in the CAR, the DRC and Mali.” Third World Quarterly, 36/1. 40–54

APPENDIX

205

Table A4 Cross tabulations, Chi-square tests, and multivariate linear probability model Case ID Sierra Leone1 DRC1 (MONUC) Liberia1 DRC2 (MONUC) DRC3 (MONUC) Liberia3 Liberia4 Liberia5 Liberia6 DRC4 (MONUC) DRC5 (MONUC) DRC8 (MONUC) DRC10 (MONUC) DRC12 (MONUC) DRC13 (MONUC) DRC14 (MONUC) DRC15 (MONUC) DRC16 (MONUC) IvoryCoast1 Liberia7 DRC17 (MONUC) DRC19 (MONUC) Liberia8 DRC21 (MONUC) DRC22 (MONUC) DRC23 (MONUC) Darfur1 DRC24 (MONUC) DRC25 (MONUC) Sudan1 Sudan2 Liberia9 Liberia10 DRC26 (MONUC) DRC27 (MONUSCO) DRC28 (MONUSCO) Darfur2 SouthSudan1 IvoryCoast2 IvoryCoast3 Abyei1

Ratios Troop Matching RiskPreOutcome numbers willingness emptive 366 5,649 694 4,950 4,194 219 216 217 203 3,925 3,925 3,387 3,338 3,368 3,368 3,375 3,306 3,260 2,939 207 3,365 3,113 214 3,182 3,197 3,183 463 3,173 3,043 3,250 3,251 334 332 2,955 3,264 3,320 321 1,649 2,213 2,213 26

12,481 9,098 4,487 10,715 12,646 14,496 14,739 14,616 15,632 13,950 13,950 16,163 16,402 16,258 16,258 16,221 16,561 16,791 6,237 15,775 16,803 17,519 15,808 18,357 18,275 18,352 15,136 19,670 20,509 9,894 9,891 11,406 11,471 20,819 19,685 19,437 21,816 5,457 9,444 9,444 3,796

1 1 1 0 0 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1 1 1 0 1 0 1 1 1 1

0,5 0,5 0,5 1 0 0 0 0 0,5 0 0 0 0,5 0 0 0 0 0 0 1 0 0,5 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 1 1 1

1 0 0 0 0 0 1 0 0 0 1 1 1 1 0 1 1 1 0 0 1 1 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 1 1 0

0 1 1 0 0 1 1 1 1 0 1 1 0 1 0 1 1 1 1 0 1 0 1 1 1 0 1 1 0 1 1 0 0 1 0 1 0 1 1 1 1

Continued

206

APPENDIX

Table A4 Continued Case ID Abyei2 Abyei3 Abyei4 Abyei5 Abyei6 DRC44 (MONUSCO) DRC45 (MONUSCO) DRC47 (MONUSCO) DRC49 (MONUSCO) Darfur4 Darfur5 SouthSudan3 SouthSudan5 SouthSudan6 Abyei7 Abyei8 Abyei9 Abyei10 Abyei11 DRC51 (MONUSCO) DRC53 (MONUSCO) DRC54 (MONUSCO) DRC55 (MONUSCO) DRC56 (MONUSCO) DRC57 (MONUSCO) Darfur6 Darfur7 Darfur9 Mali1 SouthSudan10 SouthSudan11 SouthSudan12 SouthSudan13 Abyei12 Abyei13 Abyei14 Abyei15 DRC61 (MONUSCO) DRC63 (MONUSCO) DRC64 (MONUSCO) DRC66 (MONUSCO)

Ratios Troop Matching RiskPreOutcome numbers willingness emptive 26 26 25 25 25 3,617 3,619 3,607 3,601 317 325 1,003 943 935 25 25 25 25 25 3,727 3,476 3,476 3,447 3,319 3,364 336 355 362 2,400 958 807 789 789 24 24 24 25 3,479 3,474 3,480 3,505

3,800 3,920 3,966 3,966 3,966 19,070 19,060 19,122 19,154 22,106 21,510 7,161 7,616 7,684 3,977 3,983 3,952 3,952 3,952 19,134 20,519 20,519 20,688 21,485 21,198 20,852 19,703 19,327 7,980 9,387 11,151 11,405 11,405 4,124 4,124 4,109 4,070 21,189 21,219 21,187 21,033

1 1 1 1 1 0 1 1 1 0 1 1 0 0 1 1 1 0 1 1 1 1 1 1 1 0 0 0 0 1 0 0 0 1 1 1 1 1 0 1 0

1 1 1 1 1 0 0 0 0,5 1 1 0 0,5 0 1 1 1 1 1 0 0 1 0,5 1 1 1 1 1 0,5 0,5 0 0 0 1 1 1 1 1 0 0,5 0,5

1 0 1 0 1 0 1 0 0 0 0 1 0 0 1 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 0

1 1 1 1 1 0 1 1 0 0 1 1 0 0 1 1 1 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 0 0 0 1 0 1 0

APPENDIX

Darfur10 Darfur11 Mali2 Mali4 Mali5 Mali6 DRC68 (MONUSCO) DRC69 (MONUSCO) DRC70 (MONUSCO) DRC71 (MONUSCO) DRC72 (MONUSCO) DRC74 (MONUSCO) DRC76 (MONUSCO) DRC77 (MONUSCO) DRC79 (MONUSCO) DRC81 (MONUSCO) Mali9 DRC82 (MONUSCO) DRC83 (MONUSCO) DRC85 (MONUSCO) DRC88 (MONUSCO) DRC91 (MONUSCO) DRC99 (MONUSCO) Darfur13 Darfur14 Darfur15 DRC102 (MONUSCO) DRC103 (MONUSCO) DRC106 (MONUSCO) Darfur16 Darfur17 Abyei16 Abyei18 Abyei19 Abyei20 Abyei21 Liberia15 IvoryCoast6 SouthSudan15 CAR1 CAR4 CAR6 CAR7 CAR8

379 379 1,791 1,791 1,516 1,516 3,618 3,618 3,624 3,851 3,851 3,851 4,076 4,085 4,069 4,225 1,539 4,176 4,179 4,185 4,229 4,182 4,352 410 411 410 4,331 4,380 4,512 420 420 24 22 22 22 22 754 863 842 689 372 383 381 382

18,472 18,472 9,754 9,754 11,511 11,511 21,060 21,060 21,023 19,784 19,784 19,784 18,695 18,653 18,727 18,634 11,692 18,855 18,841 18,814 18,620 18,826 18,692 17,063 17,023 17,093 18,780 18,571 18,029 16,673 16,673 4,090 4,525 4,546 4,491 4,538 5,962 5,347 13,072 6,528 12,092 12,158 12,208 12,159

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 1 0 1 1 1

1 1 1 1 1 1 0,5 0,5 1 1 0,5 0,5 0,5 0,5 0,5 1 0,5 0,5 0,5 1 1 1 1 1 1 1 0,5 0,5 0,5 1 1 1 1 1 1 1 0,5 0,5 0,5 0,5 0 1 0 0,5

0 0 1 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 1 1 0 0 1 1 1 0 0 0 1 0 0 0 0 1 0 1 1 0

207

1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 1 1 1 0 0 1 1 1 0 1 0 1 0 1 0 0 1 0 1 1 0

References ACLED. 2020. “Armed Conflict Location and Event Dataset (ACLED).” http://www. acleddata.com/ (January 2, 2017). ACLED. 2022. “Armed Conflict Location and Event Data Project—Curated Data for Africa.” https://acleddata.com/africa/. African Defence Review. 2013. “How M23 Was Rolled Back.” http://www.africandefence. net/analysis-how-m23-was-rolled-back/ (June 9, 2014). Al Jazeera. 2019. “‘Why Are They Here?’ Anti-UN Rallies Spread in East DRC.” https:// www.aljazeera.com/news/2019/11/27/protests-spread-in-east-drc-as-fury-against-unpeacekeepers-rises (July 5, 2021). Allansson, Marie, Erik Melander, and Lotta Themnér. 2017. “Organized Violence, 1989– 2016.” Journal of Peace Research 54(4): 574–87. Allen, Tim, and Koen Vlassenroot, eds. 2010. The Lord’s Resistance Army: Myth and Reality. London; New York: New York: Zed; Distributed in the USA exclusively by Palgrave Macmillan. Annan, Kofi. 1999a. The Fall of Srebrenica: The Report of the Secretary-General. New York: United Nations. Annan, Kofi. 1999b. “Two Concepts of Sovereignty.” The Economist. (September 8, 1999). Annan, Kofi. 1999c. “Walking the International Tightrope.” New York Times. http://www. nytimes.com/1999/01/19/opinion/walking-the-international-tightrope.html (June 2, 2015). Annan, Kofi. 2000. We the Peoples: The Role of the United Nations in the 21st Century. New York: United Nations. Antonini, Blanca. 2009. Security Council Resolutions under Chapter VII: Design, Implementation, and Accountabilities, The Cases of Afghanistan, Côte d’Ivoire, Kosovo and Sierra Leone. Madrid: FRIDE. Appleget, Jeff, Robert Burks, and Fred Cameron. 2020. The Craft of Wargaming: A Detailed Planning Guide for Defense Planners and Analysts. Annapolis, MD: Naval Institute Press. Autesserre, Séverine. 2014. “Going Micro: Emerging and Future Peacekeeping Research.” International Peacekeeping 21(4): 492–500. BBC News. 2011. “Hundreds Die in South Sudan Raids.” BBC News. https://www.bbc.com/ news/world-africa-14595368 (April 1, 2020). BBC News 2012. “DR Congo’s M23 Rebels Threaten to March to Kinshasa.” http://www. bbc.com/news/world-africa-20427682 (November 19, 2014). BBC News 2014. “DR Congo Commander Ndala ‘Killed in Ambush.’ ” BBC News. https:// www.bbc.com/news/world-africa-25579653 (November 5, 2020). Beach, Derek, and Ingo Rohlfing. 2018. “Integrating Cross-Case Analyses and Process Tracing in Set-Theoretic Research: Strategies and Parameters of Debate.” Sociological Methods & Research 47(1): 3–36. Beadle, Alexander William. 2011. Finding the ’Utility of Force to Protect’—Towards a Theory on Protection of Civilians. Kjeller: Norwegian Defence Research Establishment (FFI). Beadle, Alexander William. 2014. Protection of Civilians—Military Planning Scenarios and Implications. Kjeller: Norwegian Defence Research Establishment (FFI).

REFERENCES

209

Beadle, Alexander William. 2015. “Protection of Civilians as a New Objective in Military Operations.” In International Military Operations in the 21st Century: Global Trends and the Future of Intervention (Eds. Tore Nyhamar and Per Martin Norheim-Martinsen). Cass Military Studies, Routledge, 195–214. Beadle, Alexander William, and Stian Kjeksrud. 2018. “The Utility of Military Force to Protect Civilians in UN Peace Operations.” In The Use of Force in UN Peacekeeping, Routledge, 100–23. Bellamy, Alex J., and Paul Williams, eds. 2013. Providing Peacekeepers: The Politics, Challenges, and Future of United Nations Peacekeeping Contributions. 1st ed. Oxford, U.K: Oxford University Press, 100–23. Bellamy, Alex J., Paul Williams, and Stuart Griffin. 2004. Understanding Peacekeeping. Polity Press. Bellamy, Alex J., Paul Williams, and Stuart Griffin, eds. 2010. Understanding Peacekeeping. 2nd ed. Cambridge, UK; Malden, MA: Polity. Benson, Jay. 2016. The UN Intervention Brigade: Extinguishing Conflict or Adding Fuel to the Flames? One Earth Future. http://oefresearch.org/sites/default/files/documents/ publications/uninterventionbrigade.pdf. Berdal, Mats. 2016. “The State of UN Peacekeeping: Lessons from Congo.” Journal of Strategic Studies: 1–30. Berdal, Mats. 2012. “Lessons from a Study in Failure—The Force Intervention Brigade and the United Nations Mission in Congo, 2012–2017.” http://podcasts.ox.ac.uk/lessonsstudy-failure-force-intervention-brigade-and-united-nations-mission-congo-2012. Berdal, Mats, and David H. Ucko. 2015. “The Use of Force in UN Peacekeeping Operations: Problems and Prospects.” The RUSI Journal 160(1): 6–12. Beswick, Danielle. 2009. “The Challenge of Warlordism to Post-Conflict State-Building: The Case of Laurent Nkunda in Eastern Congo.” The Round Table 98(402): 333–46. Bode, Ingvild, and John Karlsrud. 2018. “Implementation in Practice: The Use of Force to Protect Civilians in United Nations Peacekeeping.” European Journal of International Relations: 1–28. Bostad, Eline Knarrum. 2018. Regime Crackdown in Syria (2011–2017)—Unpacking Violence against Civilians. FFI-report. Boutros-Ghali, Boutros. 1992. An Agenda for Peace: Preventive Diplomacy, Peacemaking and Peace-Keeping. New York: United Nations. Brahimi, Lakhdar. 2000. Report of the Panel on United Nations Peace Operations. New York: United Nations. Breidlid, Ingrid Marie, and Michael J. Arensen. 2014. “Anyone Who Can Carry a Gun Can Go”: The Role of the White Army in the Current Conflict in South Sudan. Oslo: PRIO. PRIO Paper. Buotros-Ghali, Boutros. 1995. Supplement to an Agenda for Peace. New York: United Nations. January 3, 1995. Cammaert, Patrick C. 2008. “A Peacekeeping Commander’s Perspective: From Headquarters and the Field.” The RUSI Journal 153(3): 68–71. Cammaert, Patrick C., and Fiona Blyth. 2013. The UN Intervention Brigade in the Democratic Republic of the Congo. International Peace Institute. IPI Issue Brief. Carlsson, Ingvar, Sung-Joo Han, and Rufus M. Kupolati. 1999. Report of the Independent Inquiry into the Actions of the United Nations during the 1994 Genocide in Rwanda. United Nations. Center for Civilians in Conflict. 2015. Within and Beyond the Gates: The Protection of Civilians by the UN Mission in South Sudan.

210

REFERENCES

Center for Civilians in Conflict. 2016. Under Fire: The July 2016 Violence in Juba and UN Response. Center for Civilians in Conflict. Chenoweth, Erica, and Adria Lawrence, eds. 2010. Rethinking Violence: States and NonState Actors in Conflict. Cambridge, MA: MIT Press. Chesterman, Simon. 2004. “The Use of Force in UN Peacekeeping Operations.” External Study for the United Nations Peacekeeping Best Practices. www.useofforceunpko.pdf Cil, Deniz, Hanne Fjelde, Lisa Hultman, and Desirée Nilsson. 2019. “Mapping Blue Helmets: Introducing the Geocoded Peacekeeping Operations (Geo-PKO) Dataset.” Journal of Peace Research 57(2): 1–11. Clayton, Govinda. 2016. “The Known Knowns and Known Unknowns of Peacekeeping Data—Introduction to the Special Data Review Forum.” International Peacekeeping 24(1): 1–62. Coleman, Katharina P., and Paul D. Williams. 2021. “Peace Operations Are What States Make of Them: Why Future Evolution Is More Likely than Extinction.” Contemporary Security Policy 42(2): 241–55. Congo DRC News. 2013a. “M23 Open Letter Sent to The Parliament of South Africa and SA People.” Congo Drc News. https://m23congordc.wordpress.com/2013/04/08/m23-openletter-sent-to-the-parliament-of-south-africa-and-sa-people/ (February 25, 2018). Congo DRC News. 2013b. “M23 Open Letter To People and to the Parliament of Tanzania.” Congo DRC News. https://m23congordc.wordpress.com/2013/04/13/m23-openletter-to-people-and-to-the-parliament-of-tanzania/ (February 25, 2018). Congo Research Group. 2017. Mass Killings in Beni Territory: Political Violence, Cover Ups, and Cooptation. Investigative report. de Coning, Cedric, John Karlsrud, and Chiyuki Aoi. 2017. UN Peacekeeping Doctrine in a New Era Adapting to Stabilisation, Protection & New Threats. Routledge. Costalli, Stefano. 2014. “Does Peacekeeping Work? A Disaggregated Analysis of Deployment and Violence Reduction in the Bosnian War.” British Journal of Political Science 44(2): 357–80. Davies, Shawn, Therése Pettersson, and Magnus Öberg. 2022. “Organized Violence 1989– 2021 and Drone Warfare.” Journal of Peace Research 59(4): 593–610. Defense Update, News. 2013. “Combat Debut for Rooivalk.” Defense Update. https:// defense-update.com/20131107_rooivalk_combat_debut.html (November 5, 2020). Di Salvatore, Jessica, and Andrea Ruggeri. 2017. 1 Effectiveness of Peacekeeping Operations. Oxford University Press. http://politics.oxfordre.com/view/10.1093/acrefore/ 9780190228637.001.0001/acrefore-9780190228637-e-586 (October 22, 2018). Diehl, Paul F., and Daniel Druckman. 2015. “Evaluating Peace Operations.” In The Oxford Handbook of United Nations Peacekeeping Operations, Oxford: Oxford University Press, 93–112. Dorn, A. Walter, ed. 2014. Air Power in UN Operations: Wings for Peace. Farnham, Surrey, UK; Burlington, VT: Ashgate. dos Santos Cruz, Carlos Alberto, William R. Phillips, and Salvator Cusimano. 2017. Improving the Security of United Nations Peacekeepers: We Need to Change the Way We Are Doing Business. United Nations Secretary-General. https://peacekeeping.un.org/sites/ default/files/improving_security_of_united_nations_peacekeepers_report.pdf (August 7, 2018). Doyle, Michael W., and Nicholas Sambanis. 2006. Making War and Building Peace: United Nations Peace Operations. Princeton, N.J: Princeton University Press. Durch, William J. 2006. Twenty-First-Century Peace Operations. United States Institute of Peace (USIP).

REFERENCES

211

Duursma, Allard. 2019. “Obstruction and Intimidation of Peacekeepers: How Armed Actors Undermine Civilian Protection Efforts.” Journal of Peace Research 56(2): 234–48. Duursma, Allard. 2021. “Pinioning the Peacekeepers: Sovereignty, Host-State Resistance against Peacekeeping Missions, and Violence against Civilians.” International Studies Review 23(3): 670–95. Duursma, Allard, and John Karlsrud. 2019. “Predictive Peacekeeping—Strengthening Predictive Analysis in UN Peace Operations. Stability: International Journal of Security and Development, 8(1): 1. DOI: http://doi.org/10.5334/sta.663 Eck, Kristine. 2012. “In Data We Trust? A Comparison of UCDP GED and ACLED Conflict Events Datasets.” Cooperation and Conflict 47(1): 124–41. Elster, Jon. 1978. Logic and Society: Contradictions and Possible Worlds. Chichester; New York: Wiley. Everett, Andrea L. 2017. “Mind the Gap: Civilian Protection and the Politics of Peace Operation Design.” Security Studies 26(2): 213–48. Fahey, Daniel. 2015. “New Insights on Congo’s Islamist Rebels.” The Washington Post. http://www.washingtonpost.com/blogs/monkey-cage/wp/2015/02/19/new-insightson-congos-islamist-rebels/ (April 14, 2015). Fearon, James D. 1991. “Counterfactuals and Hypothesis Testing in Political Science.” World Politics 43(02): 169–95. Findlay, Trevor. 2002. The Use of Force in UN Peace Operations. Solna, Sweden: Oxford; New York: SIPRI; Oxford University Press. Fjelde, Hanne, Lisa Hultman, and Sara Lindberg Bromley. 2016. “Offsetting Losses: Bargaining Power and Rebel Attacks on Peacekeepers.” International Studies Quarterly 60(4): 611–23. Fjelde, Hanne, Lisa Hultman, and Desirée Nilsson. 2019. “Protection through Presence: UN Peacekeeping and the Costs of Targeting Civilians.” International Organization 73(1): 103–31. Fortna, Virginia Page. 2007. Does Peacekeeping Work?: Shaping Belligerents’ Choices after Civil War. Princeton: Princeton University Press. Fortna, Virginia Page, and Lise Morjé Howard. 2008. “Pitfalls and Prospects in the Peacekeeping Literature.” Annual Review of Political Science 11(1): 283–301. France 24. 2016. “Four Killed in CAR Protest against UN Peacekeepers.” France 24. https:// www.france24.com/en/20161024-central-africa-republic-un-peacekeepers-minuscabangui (July 5, 2021). Friedman, Jeffrey A. 2011. “Manpower and Counterinsurgency: Empirical Foundations for Theory and Doctrine.” Security Studies 20(4): 556–91. Friedrichs, Rebecca. 2011. “Côte d’Ivoire: UN Peacekeeping, Impartiality and Protection of Civilians.” The Stimson Centre. http://www.stimson.org/spotlight/cote-divoireun-peacekeeping-impartiality-and-protection-of-civilians/ (June 6, 2014). Frost-Nielsen, Per Marius. 2017. “Conditional Commitments: Why States Use Caveats to Reserve Their Efforts in Military Coalition Operations.” Contemporary Security Policy 38(3): 371–97. Gallagher, Adrian, Blake Lawrinson, and Charles T. Hunt. 2022. “Colliding Norm Clusters: Protection of Civilians, Responsibility to Protect, and Counter-Terrorism in Mali.” Global Responsibility to Protect 1(aop): 1–28. Galula, David. 1964. Counterinsurgency Warfare: Theory and Practice. New York: Praeger. http://thewaywefight.com/Galula%20David%20-%20Counterinsurgency%20Warfare. pdf.

212

REFERENCES

Gates, Scott. 2002. “Recruitment and Allegiance: The Microfoundations of Rebellion.” Journal of Conflict Resolution 46(1): 111–30. Gates, Scott, and Simon Reich, eds. 2010. Child Soldiers in the Age of Fractured States. Pittsburgh, Pa: University of Pittsburgh Press. Gates, Scott, and Håvard Strand. 2004. “Modeling the Duration of Civil Wars: Measurement and Estimation Issues.” In Presentation at the Joint Session of Workshops of the ECPR, Uppsala, http://www.academia.edu/download/39850876/Modeling_the_Duration_of_ Civil_Wars_Meas20151109-22962-1xprpt7.pdf (January 2, 2017). George, Alexander L., and Andrew Bennett. 2005. Case Studies and Theory Development in the Social Sciences. MIT Press. Gettleman, Jeffrey. 2012. “Accounts Emerge in South Sudan of 3,000 Deaths in Ethnic Violence.” The New York Times. https://www.nytimes.com/2012/01/06/world/africa/insouth-sudan-massacre-of-3000-is-reported.html (April 1, 2020). Giffen, Alison. 2010. Addressing the Doctrinal Deficit: Developing Guidance to Prevent and Respond to Widespread or Systematically Attacks Against Civilians. Washington DC: The Henry L. Stimpson Center. https://www.stimson.org/2011/addressing-doctrinal-deficitdeveloping-guidance-prevent-and-respond-widespread-or/. Goertz, Gary, and James Mahoney. 2012. A Tale of Two Cultures: Qualitative and Quantitative Research in the Social Sciences. Princeton, N.J: Princeton University Press. Goldstein, Joshua S. 2012. Winning the War on War: The Decline of Armed Conflict Worldwide. New York: Plume. Goode, Steven M. 2009. “A Historical Basis for Force Requirements in Counterinsurgency.” Parameters 39(4): 45. Gorur, Aditi. 2013. Community Self-Protection Strategies. How Peacekeepers Can Help or Harm. Stimson Centre. Gorur, Aditi, Richard Gowan, Victoria K. Holt, and Lisa Sharland. 2019. Evolution of the Protection of Civilians in UN Peacekeeping. Australian Strategic Policy Institute. Special report. https://s3-ap-southeast-2.amazonaws.com/ad-aspi/2019-07/SR%20140%20 Protection%20of%20civilian%20in%20UN%20peacekeeping.pdf (August 12, 2021). Government of Rwanda. 2015. Report of the High-Level International Conference on the Protection of Civilians. Kigali. Griffith, Samuel B. 1963. Sun Tzu: The Art of War. New York: Oxford University Press. Guterres, António. 2018. “Secretary-General Highlights Dangers, Value of United Nations Peacekeeping as He Honours Fallen ‘Blue Helmets’ at Headquarters Wreath-Laying Ceremony.” un.org. https://www.un.org/press/en/2018/sgsm19061.doc.htm (August 7, 2018). Headquarters Department of the US Army. 2006. Field Manual 3-24. Washington: Headquarters Department of the US Army. Hedstro¨m, Peter, and Petri Ylikoski. 2010. “Causal Mechanisms in the Social Sciences.” Annual Review of Sociology 36(1): 49–67. Hegre, Håvard, Lisa Hultman, and Håvard Mokleiv Nygård. 2010. “Evaluating the ConflictReducing Effect of UN Peace-Keeping Operations.” In National Conference on Peace and Conflict Research, Uppsala, http://cega.berkeley.edu/assets/miscellaneous_files/122_-_ Hegre_Hultman_Nygard_-_PKO_prediction_2015_-_ABCA.pdf (June 30, 2015). Hegre, Håvard, Lisa Hultman, and Håvard Mokleiv Nygård. 2015. “Peacekeeping Works: An Assessment of the Effectiveness of UN Peacekeeping Operations.” Hegre, Håvard, Lisa Hultman, and Håvard Mokleiv Nygård. 2019. “Evaluating the ConflictReducing Effect of UN Peacekeeping Operations.” The Journal of Politics 81(1): 215–32.

REFERENCES

213

Henke, Marina E. 2016. Has UN Peacekeeping Become More Deadly? Analyzing Trends in UN Fatalities. International Peace Institute. Providing for peacekeeping. High-Level Independent Panel on Peace Operations. 2015. Uniting Our Strengths for Peace: Politics, Partnerships and People. Report of the High-Level Independent Panel on Peace Operations. Holt, Victoria K., and Tobias C. Berkman. 2006. The Impossible Mandate?: Military Preparedness, the Responsibility to Protect and Modern Peace Operations. Washington, DC: The Henry L. Stimson Center. Holt, Victoria K., Glyn Taylor, and Max Kelly. 2009. Protecting Civilians in the Context of UN Peacekeeping Operations: Successes, Setbacks and Remaining Challenges. New York: United Nations. Howard, Lise Morjé. 2008. UN Peacekeeping in Civil Wars. Cambridge; New York: Cambridge University Press. Howard, Lise Morjé. 2019. Power in Peacekeeping. Cambridge, United Kingdom; New York, NY: Cambridge University Press. Howard, Lise Morjé, and Anjali Kaushlesh Dayal. 2018. “The Use of Force in UN Peacekeeping.” International Organization 72(01): 71–103. Hultman, Lisa. 2016. “Action for Protection: What Peacekeepers Do to Protect Civilians.” International Peacekeeping Special Data Review Forum. Hultman, Lisa, Jacob D. Kathman, and Megan Shannon. 2019. UN Peacekeeping in the Midst of War. New product edition. New York, NY: Oxford University Press. Hultman, Lisa, Jacob Kathman, and Megan Shannon. 2013a. “UN Peace Operations and Protection of Civilians: Cheap Talk or Norm Implementation?” Journal of Peace Research 50(1): 59–73. Hultman, Lisa, Jacob Kathman, and Megan Shannon. 2013b. “United Nations Peacekeeping and Civilian Protection.” American Journal of Political Science 57(4): 1–17. Human Rights Watch. 2013. DR Congo: M23 Rebels Kill, Rape Civilians. https://www.hrw. org/news/2013/07/22/dr-congo-m23-rebels-kill-rape-civilians (January 19, 2018). Human Rights Watch. 2018. “World Report 2018: Rights Trends in Democratic Republic of Congo.” Human Rights Watch. https://www.hrw.org/world-report/2018/countrychapters/democratic-republic-congo (July 3, 2018). Hunt, Charles T. 2016. “All Necessary Means to What Ends? The Unintended Consequences of the ‘Robust Turn’ in UN Peace Operations.” International Peacekeeping: 1–24. Independent International Commission on Kosovo, ed. 2000. The Kosovo Report: Conflict, International Response, Lessons Learned. Oxford; New York: Oxford University Press. International Committee of the Red Cross. 2004. “What Is IHL?” https://www.icrc.org/en/ doc/assets/files/other/what_is_ihl.pdf (October 9, 2020). International Crisis Group. 2009. Jonglei’s Tribal Conflicts: Countering Insecurity in South Sudan. ICG. Africa Report. https://d2071andvip0wj.cloudfront.net/154-jonglei-s-tribalconflicts-countering-insecurity-in-south-sudan.pdf. Isberg, Jan Gunnar, and Lotta Victor Tillberg. 2011. Med Alla No¨dvendiga Medel: Brigadgeneral Jan-Gunnar Isbergs Erfarenheter Från Tja¨nstgjo¨ring i Kongo 2003-2005. Fo¨rsvarsho¨gskolan. Johnson, Hilde F. 2016. South Sudan: The Untold Story: From Independence to Civil War. London: I.B. Tauris. Jones, Pete, and David Smith. 2012. “Congo Rebels Take Goma with Little Resistance and to Little Cheer.” The Guardian. http://www.theguardian.com/world/2012/nov/20/congorebel-m23-take-goma (May 23, 2014).

214

REFERENCES

Kaldor, Mary. 2007. New & Old Wars—Organized Violence in a Global Era. 2nd edition. Stanford, Calif.: Stanford University Press. Kalyvas, Stathis N. 2006. The Logic of Violence in Civil War. Cambridge; New York: Cambridge University Press. Karlsrud, John. 2015. “The UN at War: Examining the Consequences of Peace-Enforcement Mandates for the UN Peacekeeping Operations in the CAR, the DRC and Mali.” Third World Quarterly 36(1): 40–54. Kathman, Jacob D., and Reed M. Wood. 2014. “Stopping the Killing During the ‘Peace’: Peacekeeping and the Severity of Postconflict Civilian Victimization.” Foreign Policy Analysis 57(2): 149–69. Keating, Colin. 2014. Statement of Former New Zealand Ambassador to the UN. http:// www.responsibilitytoprotect.org/Statement%20of%20Former%20New%20Zealand %20Ambassador%20to%20the%20UN.pdf (October 29, 2020). Keenan, Marla B., and Alexander William Beadle. 2015. “Operationalizing Protection of Civilians in NATO Operations.” Stability: International Journal of Security & Development 4(1). http://www.stabilityjournal.org/articles/10.5334/sta.gr/ (July 5, 2016). Ki-moon, Ban. 2014. “Secretary-General’s Remarks at Security Council Open Debate on Trends in UN Peacekeeping.” http://www.un.org/sg/statements/index.asp?nid=7769 (October 16, 2014). Kingsley, Regeena. 2018. “Military Caveats.” Military caveats. http://militarycaveats.com/ (May 2, 2018). Kjeksrud, Stian. 2014. “The Future of Peacekeeping Operations.” In International Military Operations in the 21st Century: Global Trends and the Future of Intervention, eds. Per Martin Norheim-Martinsen and Tore Nyhamar. Routledge, 116–36. Kjeksrud, Stian. 2019. “Replication Data for: Using Force to Protect Civilians.” DataverseNO. https://doi.org/10.18710/FZAVCN. Kjeksrud, Stian, Alexander William Beadle, and Petter H. F. Lindqvist. 2016. Protecting Civilians from Violence: A Threat-Based Approach to Protection of Civilians in United Nations Peace Operations. Oslo/Kjeller: NODEFIC/FFI. https://www.ffi.no/no/ Publikasjoner/Documents/Protecting-Civilians-from-Violence.pdf. Kjeksrud, Stian, and Jacob Aasland Ravndal. 2011. “Emerging Lessons from the United Nations Mission in the Democratic Republic of Congo: Military Contributions to the Protection of Civilians.” African Security Review 20(2): 3–16. Kjeksrud, Stian, and Lotte Vermeij. 2017. “Protecting Governments from Insurgencies: The DRC and Mali.” In UN Peacekeeping Doctrine in a New Era: Adapting to Stabilisation, Protection & New Threats, Global Institutions, Routledge, 227–46. Kok, Naomi. 2013. “From the International Conference on the Great Lakes Region-Led Negotiation to the Intervention Brigade: Dealing with the Latest Crisis in the Democratic Republic of Congo.” African Security Review 22(3): 175–80. Koops, Joachim, Norrie MacQueen, Thierry Tardy, and Paul D. Williams, eds. 2015. The Oxford Handbook of United Nations Peacekeeping Operations. 1. ed. Oxford: Oxford Univ. Press. Lacey, Elizabeth. 2013. Restive Jonglei: From the Conflict’s Roots to Reconciliation. IJR. Lacina, Bethany, and Nils Petter Gleditsch. 2005. “Monitoring Trends in Global Combat: A New Dataset of Battle Deaths.” European Journal of Population/Revue européenne de Démographie 21(2–3): 145–66. Legewie, Nicolas. 2013. “An Introduction to Applied Data Analysis with Qualitative Comparative Analysis.” Forum Qualitative Sozialforschung/Forum: Qualitative Social Research 14(3). http://www.qualitative-research.net/index.php/fqs/article/view/ 1961 (January 12, 2017).

REFERENCES

215

Lewis, David K. 2001. Counterfactuals. Rev. ed. Malden, Mass: Blackwell Publishers. Macura, Tomas. 2020. “Accountability and Protection of UN Peacekeepers in Light of MONUSCO.” Die Friedens-Warte 88(3/4): 143–56. Mahoney, James. 2021. The Logic of Social Science. Princeton, New Jersey: Princeton University Press. Malone, David, ed. 2004. The UN Security Council: From the Cold War to the 21st Century. Boulder, Colo: Lynne Rienner. Martin, Guy. 2013. “DRC Sniper Revelation Compromising SANDF Troops.” Defence Web. http://www.defenceweb.co.za/index.php?option=com_content&view=article& id=31797:drc-sniper-revelation-compromising-sandf-troops-expert&catid=111:sadefence&Itemid=242 (June 4, 2014). McCabe, Daniel. 2017. This Is Congo. Dogwoof. https://www.thisiscongo.com/home (November 4, 2020). McChrystal, Stanley. 2009. “ISAF Commander’s Counterinsurgency Guidance.” Melander, Erik. 2015. Organized Violence in the World 2015: An Assessment by the Uppsala Conflict Data Program. Headquarters International Security Assistance Force (ISAF), Kabul, Afghanistan. Menzies, Peter. 2014. “Counterfactual Theories of Causation.” Stanford Encyclopedia of Philosophy. https://stanford.library.sydney.edu.au/entries/causation-counterfactual/ (October 15, 2018). Mold, Francesca. 2017. “Mongolian Peacekeepers Awarded UN Medal in South Sudan.” UNMISS. https://unmiss.unmissions.org/mongolian-peacekeepers-awarded-un-medalsouth-sudan (April 30, 2018). MONUSCO. 2011. Final Report of the Fact-Finding Missions of the United Nations Joint Human Rights Office into the Mass-Rapes and Other Human Rights Violations Committed by a Coalition of Armed Groups along the Kibua-Mpofi Axis in Walikale Territory, NorthKivu 30 July to 2 August 2010. DRC. Moore, Riley M. 2013. “Counterinsurgency Force Ratio: Strategic Utility or Nominal Necessity.” Small Wars & Insurgencies 24(5): 857–78. Nadin, Peter. 2018. Use of Force in UN Peacekeeping. S.l.: Routledge. https://www.routledge. com/The-Use-of-Force-in-UN-Peacekeeping/Nadin/p/book/9781138686861. NATO. 2016. “NATO Policy for the Protection of Civilians.” NATO. http://www.nato.int/ cps/en/natohq/official_texts_133945.htm (October 9, 2020). NATO. 2022 Strategic Concept. https://www.nato.int/strategic-concept/. NATO Joint Analysis and Lessons Learned Centre. 2015. “JALLC’s PORA Briefs OPC on Protection of Civilians Project.” https://www.jallc.nato.int/articles/jallcs-pora-briefsopc-protection-civilians-project (April 4, 2022). Ngulube, Matthew, Gustav Hagglund, and Emmanuel A. Erskine. 1994. Report of the Commission of Inquiry on Armed Attacks on UNOSOM II. United Nations. Novosseloff, Alexandra. 2017. Keeping Peace from Above: Air Assets in UN Peace Operations. International Peace Institute. https://www.ipinst.org/wp-content/uploads/2017/ 10/1710_Keeping-Peace-from-Above-1.pdf. O’Brien, Conor Cruise. 1962. To Katanga and Back: A UN Case History. London: Hutchinson. O’Brien, Conor Cruise. 2011. To Katanga and Back: A UN Case History. London: Faber. Øen, Ulrik Hallén. 2014. Protection of Civilians in Practice – Emerging Lessons from the Central African Republic. Kjeller: Norwegian Defence Research Establishment (FFI). FFI-report. http://rapporter.ffi.no/rapporter/2014/01918.pdf (November 19, 2014).

216

REFERENCES

Oksamytna, Kseniya, and John Karlsrud, eds. 2020. “United Nations Peace Operations and International Relations Theory: An Introduction.” In United Nations Peace Operations and International Relations Theory, Manchester University Press. https://www. manchesterhive.com/view/9781526148889/9781526148889.00006.xml (October 7, 2020). Olivier, Darren. 2013a. “How M23 Was Rolled Back.” African Defence Review. https://www. africandefence.net/analysis-how-m23-was-rolled-back/ (September 24, 2018). Olivier, Darren. 2013b. “Rooivalk Attack Helicopters Perform Well in First Combat Action against M23.” African Defence Review. https://www.africandefence.net/after-23-yearsrooivalks-fire-first-shots-in-drc/ (November 5, 2020). Olsson, Ola, and Michele Valsecchi. 2010. “Quantifying Ethnic Cleansing: An Application to Darfur.” https://gupea.ub.gu.se/handle/2077/24045 (November 2, 2015). Osterrieder, Holger, Johannes Lehne, and Vladimir Kmec. 2015. “United Nations Interim Security Force for Abyei (UNISFA).” In The Oxford Handbook of United Nations Peacekeeping Operations, Oxford University Press, 818–29. Patrick, Stewart M. 2015. “President Obama Tackles UN Peacekeeping.” Council on Foreign Relations—The Internationalist. http://blogs.cfr.org/patrick/2015/09/25/presidentobama-tackles-un-peacekeeping/ (September 28, 2015). Paul, Christopher, Colin P. Clarke, and Beth Grill. 2010. Victory Has a Thousand Fathers: Sources of Success in Counterinsurgency. RAND Corporation. https://books.google.co. za/books?id=0J_JZbLElKkC. Pettersson, Therése, and Magnus Öberg. 2020. “Organized Violence, 1989–2019.” Journal of Peace Research 57(4): 597–613. Pflanz, Mike. 2012. “DRC Rebels Capture Goma without Firing a Shot.” http://www. telegraph.co.uk/news/worldnews/africaandindianocean/democraticrepublicofcongo/ 9689774/DRC-rebels-capture-Goma-without-firing-a-shot.html (January 25, 2018). Phayal, Anup, and Brandon Prins. 2020. “Deploying to Protect: The Effect of Military Peacekeeping Deployments on Violence against Civilians.” International Peacekeeping 27(2) (March 14, 2020): 311–36. Posen, Barry R. 2017. “Civil Wars & the Structure of World Power.” Daedalus 146(4): 167–79. Powers, Matthew, Bryce W. Reeder, and Ashly Adam Townsen. 2015. “Hot Spot Peacekeeping.” International Studies Review 17(1): 46–66. PRIO. 2016. “Armed Conflict Dataset.” https://www.prio.org/Data/Armed-Conflict/ (January 2, 2017). Providing for Peacekeeping. 2018. “Total Number of Uniformed UN Peacekeepers Deployed by Type.” https://s3.amazonaws.com/providing-for-peacekeeping-docs/ summaries/Monthly+Deployments.png (July 5, 2018). Providing for Peacekeeping. 2020. “Providing for Peacekeeping—Country Profiles.” Providing for Peacekeeping. http://www.providingforpeacekeeping.org/profiles/ (August 19, 2020). Quinlivan, James. 1995. “Force Requirements in Stability Operations.” Parameters 25(4). http://strategicstudiesinstitute.army.mil/pubs/parameters/Articles/1995/quinliv.htm. Ragin, Charles C. 2006. “Set Relations in Social Research: Evaluating Their Consistency and Coverage.” Political Analysis 14(3): 291–310. Ragin, Charles C. 2013. The Comparative Method: Moving beyond Qualitative and Quantitative Strategies: With a New Introduction. http://search.ebscohost.com/login. aspx?direct=true&scope=site&db=nlebk&db=nlabk&AN=784602 (October 5, 2015).

REFERENCES

217

Ragin, Charles C., Kriss A. Drass, and Sean Davey. 2006. “Fuzzy Set/Qualitative Comparative Analysis 2.0.” http://www.u.arizona.edu/~cragin/fsQCA/index.shtml (August 8, 2016). Ravndal, Jacob Aasland. 2015. “Thugs or Terrorists? A Typology of Right-Wing Terrorism and Violence in Western Europe.” Journal for Deradicalization (3). http://journals.sfu. ca/jd/index.php/jd/article/view/16. Razza, Namie Di. 2018. Protecting Civilians in the Context of Violent Extremism: The Dilemmas of UN Peacekeeping in Mali. International Peace Institute. https://www.ipinst.org/ wp-content/uploads/2018/10/1810_-POC-in-the-Context_of_Violent-Extremism.pdf. Reuters. 2013. “Congo Army Clashes with M23 Rebels Close to Eastern City of Goma.” Reuters. https://www.reuters.com/article/us-congo-democratic-m23/congo-armyclashes-with-m23-rebels-close-to-eastern-city-of-goma-idUSBRE96D0AV20130714 (February 25, 2018). Rogers, James I., and Caroline Kennedy. 2014. “Dying for Peace? Fatality Trends for United Nations Peacekeeping Personnel.” International Peacekeeping 21(5): 658–72. Ruggeri, Andrea, Han Dorussen, and Theodora-Ismene Gizelis. 2017. “Winning the Peace Locally: UN Peacekeeping and Local Conflict.” International Organization 71(01): 163–85. Ruggeri, Andrea, Han Dorussen, and Theodora-Ismene Gizelis. 2018. “On the Frontline Every Day? Subnational Deployment of United Nations Peacekeepers.” British Journal of Political Science 48(4): 1005–25. Schmidl, Erwin A. 1997. “Speak Softly and Carry a Big Stick.” In Peacekeeping with Muscle: The Use of Force in International Conflict Resolution., eds. Alex Morrison, Douglas A. Fraser, and James Kiras. The Canadian Peacekeeping Press, 83–90. Schneider, Carsten Q., and Ingo Rohlfing. 2013. “Combining QCA and Process Tracing in Set-Theoretic Multi-Method Research.” Sociological Methods & Research 42(4): 559–97. Schneider, Carsten Q., and Claudius Wagemann. 2012. Set-Theoretic Methods for the Social Sciences: A Guide to Qualitative Comparative Analysis. Cambrigde: Cambridge Univ. Press. Security Council Report. 2020. “Security Council Report.” https://www.securitycouncil report.org/. Security Council Report (SCR). 2015. Protection of Civilians: Cross-Cutting Report. Security Council Report (SCR). https://www.securitycouncilreport.org/atf/cf/%7B65BFCF9B6D27-4E9C-8CD3-CF6E4FF96FF9%7D/cross_cutting_report_1_protection_of_ civilians_2015.pdf. Sewall, Sarah, Dwight Raymond, and Sally Chin. 2010. MARO-Mass Atrocity Response Operations: A Military Planning Handbook. DTIC Document. http://oai.dtic.mil/oai/ oai?verb=getRecord&metadataPrefix=html&identifier=ADA525455 (July 4, 2016). Seybolt, Taylor B., Jay D. Aronson, and Baruch Fischhoff, eds. 2013. Counting Civilian Casualties: An Introduction to Recording and Estimating Nonmilitary Deaths in Conflict. Oxford: Oxford University Press. Sheeran, Scott, and Stephanie Case. 2014. The Intervention Brigade: Legal Issues for the UN in the Democratic Republic of the Congo. http://www.ipinst.org/publication/ policy-papers/detail/442-new-ipi-report-examines-legal-issues-surrounding-unsintervention-brigade.html. Shortland, Neil., and John Bohannon. 2014. “Civilian Casualties in Afghanistan.” Science 345(6198): 723–24. Shortland, Neil, Huseyin Sari, and Elias Nader. 2019. “Recounting the Dead: An Analysis of ISAF Caused Civilian Casualties in Afghanistan.” Armed Forces & Society 45(1): 122–39.

218

REFERENCES

Slim, Hugo. 2007. Killing Civilians: Method, Madness and Morality in War. London: Hurst. Small Arms Survey. 2012. My Neighbor, My Enemy: Inter-Tribal Violence in Jonglei. Human Security Baseline Assessment. Smith, David. 2013. “US Blocks Military Aid to Rwanda over Alleged Backing of M23 Child Soldiers.” The Guardian. http://www.theguardian.com/global-development/2013/oct/ 04/us-military-aid-rwanda-m23-child-soldiers (January 19, 2018). Smith, Rupert. 2008. The Utility of Force: The Art of War in the Modern World. 1st Vintage Books ed. New York: Vintage Books. Stearns, Jason. 2012a. From CNDP to M23: The Evolution of an Armed Movement in Eastern Congo. Rift Valley Institute. Usalama project Stearns, Jason. 2012b. North Kivu: The Background to the Conflict in North Kivu Province of Eastern Congo. Nairobi: Rift Valley Institute. Usalama project. Stearns, Jason. 2013a. Banyamulenge: Insurgency and Exclusion in the Mountains in South Kivu. Nairobi: Rift Valley Institute. Usalama project. Stearns, Jason. 2013b. Mai-Mai Yakutumba. Nairobi: Rift Valley Institute. Usalama project. Stearns, Jason, Judith Verweijen, and Maria Eriksson Baaz. 2013. The National Army and Armed Groups in the Eastern DRC: Untangling the Gordian Knot of Insecurity. Nairobi: Rift Valley Institute. http://www.riftvalley.net/download/file/fid/3072 (February 25, 2014). Sundberg, Ralph. 2020. “UN Peacekeeping and Forced Displacement in South Sudan.” International Peacekeeping 27(2): 210–37. Sundberg, Ralph, and Erik Melander. 2013. “Introducing the UCDP Georeferenced Event Dataset.” Journal of Peace Research 50(4): 523–32. Tardy, Thierry. 2011. “A Critique of Robust Peacekeeping in Contemporary Peace Operations.” International Peacekeeping 18(2). Thakur, Ramesh. 2013. “Protection Gaps for Civilian Victims of Political Violence.” South African Journal of International Affairs 20(3): 321–38. The Guardian. 2021. “Wounded from Anti-UN Protests Flood DR Congo Hospitals.” The Guardian Nigeria News - Nigeria and World News. https://editor.guardian.ng/news/ wounded-from-anti-un-protests-flood-dr-congo-hospitals/ (July 5, 2021). The Peacekeeping and Stability Operations Institute. 2013. UN Force Intervention Brigade against the M23. PKSOI. SOLLIMS—Stability Operations Lessons Learned & Information Management System. Townsen, Ashly Adam, and Bryce W. Reeder. 2014. “Where Do Peacekeepers Go When They Go?” Journal of International Peacekeeping 18(1–2): 69–91. Tull, Denis M. 2016. United Nations Peacekeeping and the Use of Force: The Intervention Brigade in Congo Is No Model for Success. Stiftung Wissenschaft und Politik. https://www. swp-berlin.org/en/publication/united-nations-peacekeeping-and-the-use-of-force/. UN Commission on Human Rights in South Sudan. 2016. “UN Human Rights Experts Says International Community Has an Obligation to Prevent Ethnic Cleansing in South Sudan.” https://www.ohchr.org/EN/HRBodies/HRC/Pages/NewsDetail. aspx?NewsID=20970&LangID=E (July 4, 2018). UN Integrated Training Service. 2018. “Comprehensive Protection of Civilians Training Materials (CPOC) for United Nations Peacekeeping Operations.” //research.un.org/en/ peacekeeping-community/training/CPOC/Intro (October 8, 2018). UN News. 2017. “DR Congo: Over a Dozen UN Peacekeepers Killed in Worst Attack on ‘Blue Helmets’ in Recent History.” https://news.un.org/en/story/2017/12/638812dr-congo-over-dozen-un-peacekeepers-killed-worst-attack-blue-helmets-recent (December 15, 2021).

REFERENCES

219

UNHCR. 2013. “UNHCR Urges Protection for Civilians amid Fresh Fighting in DR Congo.” UNHCR. http://www.unhcr.org/news/latest/2013/5/519ce44b6/unhcr-urgesprotection-civilians-amid-fresh-fighting-dr-congo.html (September 19, 2018). United Nations. 1945. “Charter of the United Nations.” http://www.un.org/en/documents/ charter/ (January 13, 2015). United Nations. 1999a. S/RES/1265. United Nations. 1999b. S/RES/1270. United Nations. 2004. S/2004/972. United Nations. 2005. S/2005/506. United Nations. 2008. United Nations Peacekeeping Operations: Principles and Guidelines. United Nations. 2009. S/RES/1894. United Nations. 2010a. A/64/19. United Nations. 2010b. DPKO/DFS Operational Concept on the Protection of Civilians in United Nations Peacekeeping Operations. United Nations. 2010c. S/2010/164. United Nations. 2010d. S/2010/512. United Nations. 2011. “Troop and Police Contributions to UN Peace Operations— December 2011.” https://peacekeeping.un.org/sites/default/files/dec11_5.pdf. United Nations. 2012a. S/2012/65. United Nations. 2012b. S/2012/68. United Nations. 2012c. S/2012/175. United Nations. 2012d. S/RES/2076. United Nations. 2013a. S/2013/96. United Nations. 2013b. S/2013/110. United Nations. 2013c. S/2013/225. United Nations. 2013d. S/2013/433. United Nations. 2013e. S/2013/581. United Nations. 2013f. S/RES/2098. United Nations. 2013g. S/RES/2100. United Nations. 2014a. S/2014/42. United Nations. 2014b. S/2014/157. United Nations. 2014c. S/2014/336. United Nations. 2014d. S/2014/515. United Nations. 2014e. S/2014/957. United Nations. 2015a. “DPKO/DFS Policy: The Protection of Civilians in United Nations Peacekeeping.” United Nations. 2015b. “GA/SPD/597.” http://www.un.org/press/en/2015/gaspd597.doc. htm. United Nations. 2015c. “Protection of Civilians: Implementing Guidelines for Military Components of United Nations Peacekeeping Operations.” United Nations. 2015d. The Functions and Role of the Office for Peacekeeping Strategic Partnerships (OPSP). United Nations. 2016a. S/2016/812. United Nations. 2016b. S/2016/1130. United Nations. 2016c. S/RES/2304. United Nations. 2017a. “Guidelines on the Use of Force by Military Components in United Nations Peacekeeping Operations.” http://www.iihl.org/wp-content/uploads/2018/ 03/2016.24-Guidelines-on-Use-of-Force-by-Military-Components-in-PeacekeepingOperations.pdf.

220

REFERENCES

United Nations. 2017b. “Peacekeeping Intelligence Policy.” United Nations. 2017c. S/2017/206. United Nations. 2017d. S/2017/746. United Nations. 2018. “Fatalities in United Nations Peacekeeping.” https://peacekeeping. un.org/en/fatalities (March 3, 2018). United Nations. 2019a. Authority, Command and Control in United Nations Peacekeeping Operations. Department of Peace Operations. United Nations. 2019b. “UN POC Policy.” https://www.globalprotectioncluster.org/ wp-content/uploads/DPO-Policy-on-The-Protection-of-Civilians-in-United-NationsPeacekeeping.pdf (May 19, 2020). United Nations. 2020a. S/2020/366. United Nations. 2020b. S/2020/536. United Nations. 2020c. S/2020/919. United Nations. 2020d. “The Protection of Civilians in United Nations Peacekeeping Handbook.” United Nations. 2021a. “Peacekeeping Fact Sheet—January 2021.” https://peacekeeping. un.org/sites/default/files/peacekeeping_facsheet_01_2021_english_1.pdf (February 23, 2022). United Nations. 2021b. “Troop and Police Contributors.” United Nations Peacekeeping. https://peacekeeping.un.org/en/troop-and-police-contributors (August 16, 2021). United Nations Assistance Mission in Afghanistan. 2020. “Afghanistan: Protection of Civilians in Armed Conflict.” https://unama.unmissions.org/sites/default/files/unama_ protection_of_civilians_in_armed_conflict_-_2020_first_quarter_report_english.pdf (October 9, 2020). United Nations Department of Economic and Social Affairs. 2016. “The World Population Prospects.” http://www.un.org/en/development/desa/news/population/2015report.html (August 24, 2016). United Nations Joint Human Rights Office. 2011. Report on the Investigation Missions of the United Nations Joint Human Rights Office into the Mass Rapes and Other Human Rights Violations Committed in the Villages of Bushani and Kalambahiro, in Masisi Territory, North Kivu, on 31 December 2010 and 1 January 2011. United Nations Joint Human Rights Office. 2013. Report of the United Nations Joint Human Rights Office on Human Rights Violations Perpetrated by Soldiers of the Congolese Armed Forces and Combatants of the M23 in Goma and Sake, North Kivu Province, and in and around Minova, South Kivu Province, from 15 November to 2 December 2012. Democratic Republic of the Congo. https://reliefweb.int/sites/reliefweb.int/files/resources/ UNJHRO%20-%20HRVs%20Goma%20and%20Minova%20-%20May%202013.pdf (November 5, 2020). United Nations Joint Human Rights Office. 2014. Report of the UNJHRO on Human Rights Violations Committed by the M23 in North Kivu Province between April 2012 and November 2013. United Nations Joint Human Rights Office. 2018. Détérioration de La Situation Des Droits de l’homme Dans Le Masisi et Le Lubero (NordKivu) et Défis Relatifs à La Protection Des Civils Entre Janvier 2017 et Octobre 2018. https://ohchr.org/Documents/Countries/CD/ Rapport_Masisi_Lubero_19Dec2018.pdf. United Nations News Centre. 2013. “UN Mission Sets up Security Zone in Eastern DR Congo, Gives Rebels 48 Hour Ultimatum.” http://www.un.org/apps/news/story. asp?NewsID=45535#.U5ggnyi5T5M (June 11, 2014).

REFERENCES

221

United Nations Office of Internal Oversight Services. 2014. Evaluation of the Implementation and Results of Protection of Civilians Mandates in United Nations Peacekeeping Operations. United Nations Office of Internal Oversight Services. 2017. Triannual Review. United Nations. https://oios.un.org/inspection-evaluation-reports (March 23, 2020). United Nations Office of Internal Oversight Services. 2018a. Inspection of the Performance of Missions’ Operational Responses to Protection of Civilians (POC) Related Incidents. United Nations Office of Internal Oversight Services. 2018b. POC Performance Study. UNMISS Human Rights Division. 2012. Incidents of Inter-Communal Violence in Jonglei State. UNMISS. Uppsala University. 2020. “UCDP—Uppsala Conflict Data Program.” UCDP—Uppsala Conflict Data Program. http://ucdp.uu.se/ (February 8, 2018). Uppsala University. 2022. “UCDP Georeferenced Event Dataset (GED) Global Version 22.1.” https://ucdp.uu.se/downloads/index.html#ged_global. Våge, Anders Skeibrok, and Alexander William Beadle. 2014. Assessing Protection of Civilians in Military Operations. Kjeller: FFI. FFI-report. van der Lijn, Jaïr, and Timo Smit. 2015. “Peacekeepers under Threat? Fatality Trends in UN Peace Operations.” SIPRI Policy Paper. September 2015. SIPRIPB1509.pdf Verweijen, Judith, and Claude Iguma. 2015. “Understanding Armed Group Proliferation in the Eastern Congo.” Usalama Project Briefing Paper. https://biblio.ugent.be/publication/ 7017636/file/7017665 (December 12, 2016). Weiss, Thomas G., and Sam Daws, eds. 2018. The Oxford Handbook on the United Nations. Second edition. Oxford, United Kingdom: Oxford University Press. White House. 2015. “Declaration of Leaders’ Summit on Peacekeeping.” whitehouse.gov. https://www.whitehouse.gov/the-press-office/2015/09/28/declaration-leaders-summitpeacekeeping (July 6, 2016). Wikipedia. 2018. “Jonglei State.” Wikipedia. https://en.wikipedia.org/w/index.php?title= Jonglei_State&oldid=853786889 (October 2, 2018). Williams, Paul D. 2023. “How Peacekeepers Fight: Assessing Combat Effectiveness in United Nations Peace Operations.” Security Studies. Williams, Paul D., and Alex J. Bellamy. 2021. Understanding Peacekeeping. Third edition. Medford, MA: Polity Press. Willmot, Haidi. 2016. “The Evolution of the UN Collective Security System.” In Protection of Civilians, eds. Haidi Willmot, Ralph Mamiya, Scott Sheeran, and Marc Weller. Oxford University Press, 109–38. https://oxford.universitypressscholarship.com/view/10.1093/ acprof:oso/9780198729266.001.0001/acprof-9780198729266-chapter-6 (October 13, 2020). Willmot, Haidi. 2017. Improving UN Situational Awareness: Enhancing the U.N.’s Ability to Prevent and Respond to Mass Human Suffering and to Ensure the Safety and Security of Its Personnel. Washington, D.C: Stimson. https://www.stimson.org/sites/default/files/ file-attachments/UNSituationalAwareness_FINAL_Web.pdf. World Bank. 2018. “Countries and Economies.” https://data.worldbank.org/country (August 13, 2018). Young, John. 2007. The White Army: An Introduction and Overview. Geneva: Small Arms Survey/Graduate Institute of International Studies. YouTube. 2013. “MONUSCO Force Commander Announces the Establishment of Security Zone for Goma and Its Northern Suburbs.” https://www.youtube.com/ watch?v=4lMeT9uUAxo (May 22, 2014).

Index Abyei 38, 51, 64, 70, 72, 85, 90, 96, 111–13, 114, 115, 119, 122–4, 173, 182; see also UNISFA Misseriya 51, 66, 85–6, 173 act of genocide 17, 71, 83 Afghanistan 3–4, 22, 46, 104 Taliban 4, 46 all necessary means 4, 30, 130, 140 Allied Democratic Forces/National Army for the Liberation of Uganda (ADF/NALU); see Democratic Republic of the Congo (DRC) amelioration 10, 44, 47, 59, 62, 73–4, 123, 134, 158, 159, 161, 162 Armed Conflict Location Events Dataset (ACLED) 18, 63, 75, 78–80 Armed Forces of the Democratic Republic of the Congo; see Democratic Republic of the Congo and Forces armées de la république démocratique du Congo (FARDC) attack helicopters 24, 57, 131, 139, 140, 141–7, 151 Balkans 17, 71, 83, 104 Bangladesh 34, 118, 167–70 bedrock principles 2, 5, 16, 37, 42, 120, 179 Bosnia 6, 15, 29, 45–6, 51, 59, 71 capabilities 4, 8, 20, 23, 24, 40, 47, 48, 50, 55, 56, 57, 59, 70, 78, 130, 131, 134, 139, 147, 152, 153, 158, 183, 185 Capstone doctrine 2, 16, 58 CAR; see Central African Republic case selection 13, 105, 132, 183 casualties 4, 18, 53, 80, 88, 105, 109, 37, 138, 141, 146, 155, 163, 169 civilian casualties 4, 18, 33, 57, 63, 75–81, 83, 84, 108, 156

causal conditions 7, 10, 27, 48, 70, 94, 97, 99, 101, 105, 107, 110, 129, 156, 181–4 causal mechanisms 7, 11, 35–48, 73 causal pathway(s) 27, 104–29 causal recipe(s) 27, 125, 127–9 caveats 33, 41, 59, 78, 91, 120, 148, 168, 175, 181 Central African Republic (CAR) 42, 182 Chi-square test(s) 11, 89, 94–9 coercion 10, 18, 28, 34, 45–8, 59, 62, 73–4, 83, 123, 134, 138–44, 158, 165 collateral damage 3–4 command and control 2, 22, 51, 117 communal conflict 51–5, 66, 70–2, 156, 158, 170, 182 Congrès national pour la défense du people (CNDP), see Democratic Republic of the Congo (DRC) consent 5, 16, 37, 38, 57, 72, 114, 120, 179 containment 10, 45, 47, 59, 62, 73–4, 123, 134, 136, 158 cordon and search 45, 69, 121 counterfactual reasoning 11–12, 27, 48, 85, 93, 107–8, 176 counterinsurgency (COIN) 3–4, 29, 37–8, 56, 114, 181 Darfur 29, 64, 69, 72, 77, 108, 111–12, 115, 116, 119, 122, 124 deadly force 4, 16 defense of the mandate 5, 120 Democratic Republic of the Congo (DRC) 1, 40, 109, 130; see also MONUC/MONUSCO ADF/NALU 66, 69, 152, 182, 183 Congrès national pour la défense du people (CNDP) 66 Force Intervention Brigade (FIB) 24, 45, 58, 66, 121, 130–53

INDE X Forces armées de la république démocratique du Congo (FARDC) 24, 69, 86, 92, 109, 130–53 Forces démocratiques de libération du Rwanda (FDLR) 1, 46, 66, 69, 86, 92–3, 109 Framework brigades 147–8 Goma 24, 109, 131–2, 135–51, 180 Kibati 135, 139–43 Kibua-Mpofi axis 6, 86 Kiwanja 143, 147, 151 Mai-Mai 66, 69, 86, 109 Mai-Mai Sheka 80, 86 Mouvement du 23 March (M23) 23–4, 42, 45–6, 66, 109, 130–53 Mutaho Hills 135–9, 150 North Kivu 1, 69, 131 Sake 137–9 South Kivu 1, 69, 92, 109 destruction 10, 28, 45–7, 51, 59, 62, 73–4, 123, 134, 141, 143–4, 158, 162–5, 172 deterrence 9, 10, 18, 23, 34, 37, 39, 45–7, 52, 59, 62, 73–4, 83, 90, 123, 134, 136, 141, 150–1, 153, 157–8, 160–2, 165, 177; see also deterrent presence deterrent presence 7, 26, 33, 35, 43, 59, 89–90, 94, 96, 100, 104, 110, 121, 125–6, 128, 133, 156, 158, 176 dialogue 42–3 early warning 23, 35, 155–6, 158–9, 161–3, 170, 172, 177, 179 effectiveness 8–9, 13, 15, 20, 33–4, 41–2, 61, 84, 152, 179, 184 escalation 53, 150 ethnic cleansing 15, 17–19, 22, 51, 53, 55, 71–2, 83, 158 fatalities 23, 56, 78–9, 81–2, 164, 180 civilian fatalities 18, 23, 76, 78–9, 84, 179 UN fatalities 19, 81–2, 84 fatality figures 62, 75, 78–9, 84, 108, 172 FDLR; see Democratic Republic of the Congo and Forces démocratiques de libération du Rwanda force commander 40, 130, 133, 136, 139–40, 143, 146, 148–9, 179

223

Force Intervention Brigade (FIB); see Democratic Republic of the Congo (DRC) force mobility 8, 151, 170, 177, 183 forced recruitment 52, 55 freedom of movement 50–1 functions of force 10, 46–7, 58–62, 73–4, 84, 92–3, 123, 133–4, 140, 142, 152 functions of violence 133, 159, 164 fuzzy set Qualitative Comparative Analysis (fsQCA) 11, 27, 104–7, 110, 125, 132, 145, 176; see also Qualitative Comparative Analysis genocide 15, 17, 25, 46, 51, 53, 55, 71, 83 government repression 54, 57, 71 High-level Independent Panel on Peace Operations (HIPPO) 33, 39 host-state consent 5, 37, 38, 57, 72, 114, 120, 179 host-state support 8, 27 Hultman, Lisa 9, 36–7, 61, 180 humanitarian aid 44, 47, 133 human rights 13, 47, 131, 132, 134, 155, 158 impairment 46–8, 123, 133–4, 136, 158–9, 161 impartiality 5, 16, 55, 120; see also bedrock principles incitement 46–7, 123, 133–4, 155, 158, 171 India(n) 34, 45, 118, 167–9 insurgency 50, 70–3 intelligence 35, 42, 89, 185 internally displaced persons (IDPs) 86, 123, 137, 138, 164, 165 International Humanitarian Law (IHL) 3, 47, 134 Ivory Coast 50–1, 57, 64–5, 112–13, 115, 119, 122, 124 Johnson, Hilde Frafjord 154, 160, 164, 180 Jonglei 23, 66, 76, 77, 154, 157–73 Kampala 136–7, 142, 148–9, 180 Kenya 118, 167–9 Kosovo 16, 22, 50, 71

224

INDE X

Law of Armed Conflict (LOAC) 3, 4, 81 Liberia 38, 50, 64, 67, 71, 72, 80, 111–12, 115, 116, 119, 122, 124 Libya 67, 71 Lord’s Resistance Army (LRA) 77, 183 Malawi 118, 132, 142, 144, 147, 148 Mali 19, 21, 38, 40, 45, 46, 50–1, 56, 58, 64, 69, 82, 84, 111–13, 115, 119, 121, 124 mandate(s) 2, 5, 10, 14, 15, 16, 17, 40, 45, 58, 62, 67, 69, 70, 78, 81, 120, 130, 142, 147, 149, 151, 154, 186 mandated 7, 14, 40, 64, 67, 121, 181 mass killings 15, 28, 46, 47, 133 matching, see matching theory matching theory 33, 35, 59, 123, 134, 135, 152, 158, 178, 182 MINUSCA 45, 64, 65, 69, 72, 111–13, 119, 122, 124, 182 MINUSMA 19, 45, 64, 65, 69, 82, 84, 111–13, 115, 119, 121, 122, 124 mob violence 18, 25, 49, 50, 54, 56–7, 67, 70, 71, 72 Mongolia 34, 40, 118 MONUC 40, 66, 69, 70, 90, 109, 112– 13, 115, 119, 122, 124; see also MONUC/MONUSCO MONUC/MONUSCO 18, 64, 66, 69, 70, 72, 83 multivariate linear probability analysis 11, 89, 94 National Army for the Liberation of Uganda (NALU); see Democratic Republic of the Congo and ADF/NALU necessary conditions 106, 125 non-state armed actor/group 4, 5, 17, 57, 66, 151 non-use of force 5, 16, 44; see also bedrock principles North Atlantic Treaty Organization (NATO) 3, 4, 5, 22 operational art 8, 27, 149, 153, 177, 179 Pakistan 34, 45, 118 peace agreement(s) 14, 120, 121 physical protection 4–5, 30, 38, 42–3, 62–3, 83, 177

possible worlds 12, 108 post-conflict revenge 50, 71 predatory violence 18, 24, 50, 52, 70, 71, 72, 83, 182 pre-emption 7, 8, 26, 27, 32, 33, 34, 35, 41–4, 59, 90–4, 98–100 principles, see bedrock principles protection failures 6, 9, 15, 31, 32, 73 protection mandate(s) 5, 16, 58, 69; see also mandate(s) QCA; see Qualitative Comparative Analysis and fuzzy set Qualitative Analysis qualitative case studies 73, 129, 132, 156, 176, 179, 180 Qualitative Comparative Analysis (QCA) 11, 27, 90, 104–27 rationale 11, 20, 22, 48–9, 54, 59, 70–3, 109, 130, 142, 153, 183 retaliatory attacks 47, 134, 157–8, 164, 171 revenge 23, 50–1, 53, 54, 55, 71, 76, 85, 109, 155, 157, 166, 172 Rules of engagement (RoEs) 2, 40, 43, 45 Rwanda 1, 6, 15, 16, 17, 34, 46, 51, 71, 81, 83, 110, 117, 118, 132, 137, 144, 146–7, 151 Santos Cruz, Carlos Alberto dos 130, 133, 137, 139, 143, 146–9, 179 scenarios 12, 18, 47–54, 70–3, 83, 85, 93, 134, 184 second-order effects 12, 108 selection bias 103, 105 self-defense 5, 14, 38, 44, 53, 55, 120, 163 show of force 43, 47, 57, 134, 158 Sierra Leone 38, 50, 64, 65, 116 Somalia 6, 46, 71 South Africa 118, 131–2, 135–6, 138, 143, 147–8, 151 South Sudan 7, 8, 21, 23, 24, 25, 27, 35, 37, 40, 46, 51, 64, 66, 70–3, 76, 91, 104, 121, 153, 156, 159, 160, 170, 176, 177, 178, 179, 180, 181; see also UNMISS Bor Dinka 155 Dinka 164, 172 Likuangole 160–3, 166–8, 172 Lou Nuer White Army 23–4, 154–73, 176, 177, 179, 180, 183

INDE X Murle 23, 51, 76, 154–73, 176, 179, 183 Nanaam 160, 162, 172 Ngok Dinka 51, 66, 85, 86 Nuer 23, 51, 76, 155, 157, 159–61, 164, 168, 171, 172 Sudan People’s Liberation Army (SPLA) 46, 66, 154–73, 176–7, 179, 180 Special Forces 57, 131, 143, 151 Srebrenica 16, 17, 46, 51, 71, 83 stabilization 37, 114, 130, 153 statistical analysis/analyses 13, 27, 89–90, 101, 104, 107, 114, 117, 128, 129, 176 strategies and tactics 48, 51, 183 sufficient conditions 106, 127 Syria 29, 51, 54, 67 Tanzania 118, 132, 135–48 temporary operating bases (TOB) 21, 42 threat-based approach to protection of civilians 26 threats to international peace and security 14, 16, 186 torture 17, 52 troop-contributing country/countries (TCC(s)) 2, 21, 33, 39, 42, 91, 98, 102, 117–20, 168, 169, 173, 177, 181 troop numbers 7, 22, 27, 33, 35–9, 69, 90, 94–7, 100–3 troop-to-perpetrator ratios 8, 27, 39, 133, 145–6, 156, 166–8, 180–1 troop-to-population ratios 37–9, 90–1, 93, 96–7, 100, 102, 110, 114–16, 128, 145–6, 177 truth table 105, 127–8 Uganda 69, 73, 132, 137, 143–4, 146–7, 149, 151 Ukraine 14, 15, 34, 67, 118, 121, 151 UNAMID 29, 64–5, 69, 72, 108, 111, 113, 115, 119, 122, 124 UNAMSIL 64, 65, 67, 112, 115, 119, 122, 124

225

United Nations (UN) Charter 13–14 General Assembly (UNGA) 41, 91, 117 Group of Experts 134, 142, 143, 150 Protection of Civilians Operations dataset (UNPOCO) 7, 10, 11, 13, 17, 18, 19, 27, 39, 52, 57, 61–89, 91, 93, 94, 98, 100, 101, 103, 104, 105, 110, 114, 116, 120, 125, 129, 145, 155, 166, 172, 178, 180, 181, 182, 183, 184 Protection of Civilians (PoC) policy 16, 22, 34, 36, 42, 43, 64, 178 Secretary-General (UNSG) 10, 11, 15, 16, 19, 21, 44, 62, 67, 72, 75, 88, 120, 137, 171, 175 Security Council (UNSC) 2, 14, 15, 16, 17, 30, 62, 72, 75, 80, 88, 121, 130, 132, 135, 147, 175, 181, 184, 186 Special Representative of the SecretaryGeneral (SRSG) 154, 160, 164, 171, 180 UNIFIL 45 UNISFA 38, 64, 65, 66, 69, 72, 85, 86, 96, 111, 113, 114, 115, 119, 122, 124, 173, 182 UNMIL 64, 65, 67, 69, 72, 80, 111, 112, 115, 119, 122, 124 UNMISS 23, 37, 55, 64, 65, 66, 69, 111–15, 119, 122, 124, 154–73, 176, 177, 179, 180; see also South Sudan UNOCI 64, 65, 69, 112–13, 115, 119, 122, 124; see also Ivory Coast Uppsala Conflict Data Program (UCDP) 63, 78, 79, 80 utility of force 1, 9, 17, 20, 24, 25, 26, 28–48, 55, 84, 104, 123, 133, 134, 151, 172, 173–86 willingness to accept risk 7, 22, 26, 27, 32, 33, 34, 35, 39–41, 59, 90, 91, 94, 97, 98, 100, 102, 103, 116–20, 125, 126, 132, 147–8, 153, 156, 168–9, 176