Resisting War

In civil conflicts around the world, unarmed civilians take enormous

risks to protect themselves and stand up to heavily armed combatants.
This is not just counterintuitive – it is extraordinary. In this book,
Oliver Kaplan explores cases from Colombia, with extensions to
Afghanistan, Pakistan, Syria, and the Philippines, to show how and
why civilians are able to influence armed actors and limit violence.
Based on original fieldwork as well as statistical analysis, the book
explains how local social organization and cohesion enables both covert
and overt nonviolent strategies, including avoidance, cultures of
peace, dispute resolution, deception, protest, and negotiation. These
“autonomy” strategies help communities to both retain civilian status
and avoid retaliation by limiting the inroads of armed groups. Contrary
to conventional views that civilians are helpless victims, this book
highlights their creative initiative to maintain decision-making power
over outcomes for their communities.

oliver kaplan is an assistant professor at the Josef Korbel School of

International Studies at the University of Denver. He was previously a
postdoctoral Research Associate at Princeton University in the Wood-
row Wilson School and at Stanford University as an affiliate of the
Empirical Studies of Conflict project. His research for Resisting War
received the Diskin Dissertation award honorable mention from the
Latin American Studies Association.
Resisting War
How Communities Protect Themselves

OLIVER KAPLAN
University of Denver
University Printing House, Cambridge cb2 8bs, United Kingdom
One Liberty Plaza, 20th Floor, New York, ny 10006, USA
477 Williamstown Road, Port Melbourne, vic 3207, Australia
4843/24, 2nd Floor, Ansari Road, Daryaganj, Delhi – 110002, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781107159808
10.1017/9781316671887
© Oliver R. Kaplan 2017
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2017
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
names: Kaplan, Oliver Ross, author.
title: Resisting war : how communities protect themselves / Oliver Ross Kaplan.
description: Cambridge, United Kingdom ; New York, NY : Cambridge University Press, 2017.
identifiers: lccn 2016044944 | isbn 9781107159808 (Hardback)
subjects: lcsh: Community organization–Colombia. | Civilians in war–Colombia. |
Civil war–Protection of civilians–Colombia. | Nonviolence–Colombia. | Colombia–History–
1946–1974. | Colombia–History–1974–
classification: lcc hn303.5 .k37 2017 | ddc 303.6/109861–dc23
LC record available at https://lccn.loc.gov/2016044944
isbn 978-1-107-15980-8 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy
of URLs for external or third-party Internet Web sites referred to in this publication
and does not guarantee that any content on such Web sites is, or will remain,
accurate or appropriate.
To the campesinos of Colombia: the true experts
on making peace.
To Ben.
Contents

List of Figures page xi

List of Tables xiii
Preface and Acknowledgments xv
List of Abbreviations xix

1 Introduction: Civilian Autonomy in Civil War 1

The Argument 9
Current Perspectives on Civilians and Civil Wars 18
Research Design and Methods 24
Plan of the Book 24
2 A Theory of Civilian Decision-Making in Civil War 33
The Advantages of Community Cohesion and Organization 35
Civilians’ Choices and Pathways to Autonomy 42
Explaining Violence: Conditions for De Facto Autonomy 54
Summary 60
3 The History of Conflict and Local Autonomy in Colombia 62
Colombia’s History of Conflict 63
The History of Civil Society Responses to Armed Conflict 70
Peace Communities and Formal Autonomy Organizations 71
Junta Councils and Their Relevance for Violence 79
Summary 83
4 Living to Tell About It: Research in Conflict Settings 85
Why the Case of Colombia? 86
The Framework for Subnational Analysis 88
The Case Studies and Fieldwork: “To the Villages!” 93
Summary 108

vii
viii Table of Contents

5 How Civilian Organizations Affect Civil War Violence 110

The Junta Councils’ Impact on Violence 112
Additional Social Organizations 130
Learning from an Outlier: The Case of Belén de Los Andaquies 132
Conclusions 133
Appendix 135
6 Why Some Communities Are More Organized than Others 155
Explaining the Formation of Junta Councils 156
The Role of La Violencia and Reverse Causality 158
Selecting Cases Under Constraints in Multimethod Projects 163
Selecting Cases from Cundinamarca 169
The ATCC Cases in Santander 173
Conclusions 173
Appendix 175
7 The Institution of the ATCC: Protection through Conciliation 183
The ATCC in Context and Trends in Violence 186
Evaluating Explanations for Violence: The Balance of Control 189
Civilian Institutions as an Explanation: The Process of the ATCC 194
Analysis of Threats and Conciliations, 1987–2007 195
Conditions for the Maintenance of Local Order in Wartime 206
Explaining a Resurgence of Violence, 2000–2007 212
Conclusions 216
8 Discovering Civilian Autonomy in Cundinamarca 219
Manifestations and Impacts of La Violencia 224
The Juntas de Acción Comunal and Preexisting Social Capital 226
The Nature and Severity of Armed Conflict 242
Juntas during the Conflict and Mechanisms for Autonomy 246
An Intervening Explanation: Clientelism 258
Levels of Violence (vs. Reported Violence) 261
Conclusions 266
9 Civilian Autonomy around the World 270
Civilian Autonomy in FARClandia 271
Civilian Autonomy and the Peace Zones in the Philippines 276
Civilian Autonomy in Afghanistan and Pakistan 286
Civilian Autonomy in Syria 291
Conclusions 299
10 Conclusions and Policy Implications 300
The Nuances of Civilian Autonomy 302
The Historical and State-Building Foundations of Civilian Autonomy 304
Methodological Contributions 305
Policy Implications 306
Table of Contents ix

A Research Agenda on Civilians in War 312

Final Thoughts 314

Appendix A: Archives Consulted 317

Appendix B: Supplementary Documentation on the ATCC 319
Glossary 333
References 339
Index 359
Figures

1.1 Map of formal civilian autonomy organizations in Colombia page 6

1.2 Map of civilian autonomy in civil war around the world 7
3.1 The Colombian national homicide rate, 1946–2005 64
3.2 The growth of guerrilla fronts, 1978–1996 65
3.3 Map of FARC activity, 1999–2005 66
3.4 Intensity of the conflict and level of peace mobilization,
1978–2003 72
5.1 Political homicides by presumed perpetrator, 1990–2005 114
5.2 Map of mean annual political homicide rates by municipio,
1990–2005 115
5.3 Map of juntas per communities by municipio 119
5.4 Estimated mean municipal political homicide rates
by levels of junta councils, 1990–2005 121
5.5 The geography of La Violencia 124
5.6 Participation in community organizations by ethnicity, 2005 131
5.7 Distribution of juntas in Colombian municipios 137
5.8 The conditional effect of juntas on the homicide rate by
levels of conflict 148
5.9 The effect of juntas on violence over time 149
6.1 Comparison of La Violencia and junta coverage 160
6.2 Comparisons of selected cases 169
6.3 Map of case study sites 170
6.4 Armed group actions in Cundinamarca case towns 172
6.5 Distribution of junta-treated and control cases by
propensity-score quartile 179
6.6 Propensities for junta “treatment” of Cundinamarca towns 179
6.7 Predicted junta values of Cundinamarca towns 182

xi
xii List of Figures

6.8 Predicted vs. actual homicide rates in Cundinamarca towns 182

7.1 Armed actions in ATCC municipios, 1990–2006 192
7.2 Human rights violations in ATCC municipios, 1990–2006 193
7.3 Outcomes of threats according to whether the victim was
believed to have collaborated 197
7.4 Reasons for threats (and killings) by armed groups 199
7.5 Information channels: how the ATCC learned of threats 200
7.6 Information and appeals used by the ATCC to investigate
threats 200
7.7 Distribution of threats and killings among armed groups by
ATCC villages, 1991–2007 201
7.8 The balance of control among armed groups across ATCC
villages, 1987–2007 202
7.9 Threat and killing rates in the ATCC zone vs. neighboring
areas 204
7.10 Maps of actual and counterfactual violence in ATCC and
neighboring regions, 1987–2007 205
7.11 Estimates of coca production in ATCC villages, 2003 214
8.1 Villages of the Cundinamarca municipios 222
8.2 Estimates of violence from quantitative and qualitative sources 263
9.1 The geography of Peace Zone villages in the Philippines 279
9.2 Frequencies of Peace Zone strategies in the Philippines. 280
B.1 Communication from the paramilitaries to the ATCC 325
B.2 Communication from the FARC to the ATCC 326
B.3 Warning from the FARC to the ATCC 327
B.4 ATCC authorization to seek dialogues with the FARC 328
B.5A ATCC Membership Document circa 2007 330
B.5B ATCC Membership Document 331
Tables

1.1 Cross-national examples of civilian autonomy in civil wars page 27

2.1 Examples of civilian organizational strategies and tactics by
contentiousness 53
3.1 Formation of junta councils in Colombia, 1960–1993 81
4.1 Characteristics of ex-combatant interview subjects 104
5.1 Summary statistics 140
5.2 Models of CINEP political homicide rates, 1990–2005,
by municipios 142
5.3 Models of police homicide rates, 1999–2005, by municipios 144
5.4 Models of additional organizations, 1990–2005, by municipios 146
5.5 Models of killings according to armed group perpetrators 150
5.6 A matching model of the effect of juntas 152
5.7 Pair-wise correlations of conflict variables and juntas 153
6.1 Models of juntas per “communities” in 1985, by municipios 176
6.2 Models of juntas per “communities” in 1985, by municipios 177
6.3 Junta case selection models 178
6.4 Characteristics of Cundinamarca towns 180
7.1 Key variables in the ATCC region, 1975–2007 188
7.2 Predictions of the balance of control theory of violence 190
8.1 Qualitative indicators of social cohesion and organization 227
8.2 Qualitative indicators of violence 263
B.1 Real and counterfactual subcases 321

xiii
Preface and Acknowledgments

This project began with a question and a curiosity. How can we tell if civilians
in conflict settings can protect themselves through social movements? The
reigning theories all said this was unlikely, and yet there were cases that
suggested, shouted, that protection was possible. I wanted to know why, when,
and how these kinds of efforts succeed or fail. I was motivated by the idea that
research could help answer these questions and even contribute to the protec-
tion of people living in the direst of situations. The choice of Colombia as a
research site was dictated by the question, then, rather than the reverse.
I had lived in Central America but had never been to Colombia and knew
little about the country, except that it was probably not an easy location in
which to do research. I started learning all I could, became drawn to the place,
and realized research could be quite feasible and enjoyable, as long as one uses
common sense, or does not dar papaya (literally to “give papaya”). I found a
beautiful country with warm, humble, thoughtful, determined people and,
thankfully, great coffee and rich chocolate. What started as my doctoral
research became a journey of discovery that led to the findings presented in
this book. It was a true education, full of new experiences, treasured memories,
and fast friends. It would also hold my first experience with tear gas and
encounters with unbearable heat and unimaginable carsickness, among other
pleasantries.
I realized early on in my research that I would have to keep a list of all the
people that helped me along the way because I was racking up and continued to
take on many, many debts of gratitude. In this sense, this book is the result of a
true and broad collaboration. However, any remaining errors are, lamentably,
my own.
I first express mis sinceras gracias to all those who shared their precious time,
knowledge, and histories with me and invited me into their homes or stores for
a tinto, aguapanela, or arepita. Not only did they offer friendship and

xv
xvi Preface and Acknowledgments

fascinating, hilarious, and somber tales and insights, but they also genuinely
looked out for my well-being and helped me get to where I needed to go. This
book would be nothing without them. I was also received warmly during my
shorter time in the Philippines, which was similarly eye-opening.
I was fortunate at Stanford to have an amazing group of advisors. My
dissertation committee of David Laitin, Terry Karl, Ken Schultz, and Jeremy
Weinstein gave me a combination of healthy skepticism, constructive criticism,
and frequent encouragement that pushed me to do better. I could not have
overcome many of the research hurdles without them. Other faculty also gave
instrumental guidance. Karen Jusko was incredibly helpful with breaking
down research design issues and I also had many enlightening conversations
with Alberto Díaz-Cayeros, Beatriz Magaloni, Jim Fearon, Josh Ober, Steve
Krasner, and Mike Tomz. I am especially grateful to Eliana Vásquez, for our
conversations and her unwavering enthusiasm and support, and to her family.
I also thank my undergraduate advisors Barbara Walter and David Lake at UC
San Diego for helping me first get started in political science research.
My peer group of fellow Stanford Ph.D. students were good friends and
commiserators throughout. I thank Claire Adida, Mike Albertus, Leo Arriola,
Rik Bhavnani, Thomas Brambor, Matt Carnes, Darah Cohen, Luke Condra,
Roy Elis, Alejandro Feged, Joe Felter, Brodie Ferguson, Desha Girod, Bethany
Lacina, Natan Sachs, Jake Shapiro, and Jessica Weeks for suffering discussions
with me, serving as sounding boards, and giving great feedback. My colleagues
at the Josef Korbel School at the University of Denver also provided encourage-
ment and helpful comments on various later drafts: Debbi Avant, Erica Che-
noweth, Rachel Epstein, Cullen Hendrix, Danny Postel, and Aaron Schneider.
Thanks also go to all my other friends and colleagues at Stanford, Princeton,
the University of Denver, and elsewhere for their support.
I thank Robert Dreesen, my editor at Cambridge University Press, for
believing in this project and his guidance in the publication process. I also
thank Brianda Reyes, Sarah Lambert, Anand Shanmugam, and Julia TerMaat
for their editorial assistance and the three anonymous peer reviewers of the
manuscript, who provided valuable criticism and insights. I thank Marcela
Vega Vargas, a talented Colombian artist, for masterfully channelling my
vision in her wonderful illustration of the cover image.
I am indebted to Mauricio García Durán and Teófilo Vásquez of the Centro
de Investigación y Educación Popular (CINEP), Fabio Sánchez Torres of the
Universidad de Los Andes, Jeff Villaveces at the United Nations (OCHA), Sol
Santos, Nathan Cruz, Zakia Shiraz, and Grant Miller for their facilitation of
data, sage advice, and friendship. I also thank Padre Mauricio Uribe at the
Universidad Sergio Arboleda for his encouragement, wonderment, and polite
driving.
A number of individuals both in Colombia and in the United States were
also extremely supportive and helpful. These include Gloria Inés Restrepo,
Esperanza Hernández Delgado, Hernán Molina, Luis Emiro Valencia, Camilo
Preface and Acknowledgments xvii

Echandía, Marcela Palacios Garzón, Leslie Wirpsa, Mariana Puerto, Fredy

Barrero, Andrés Casas, Janice Gallagher, Adan Griego, Julie Sweetkind-Singer,
Julia Simarra, Estella Duque, Zephyr Frank, Jon Bellish, Santiago Dávila,
Christian Caryl, and Dani Powell.
This book was greatly enhanced by excellent research assistance by Liz
Carolina Garzón, Marie Claire Vásquez Duzán, Juan Jurado, Daniela Uribe,
Anjali Menon, Jenna Rodrigues, Kate Castenson, and Natalie Southwick.
A number of organizations also provided great insight and assistance during
my research. These include the Asociación Campesina de Antioquia, the staff
at FOR-Colombia, the Corporación Júridica Libertad, and the Taller de Vida.
I also received invaluable assistance and support from various offices in the
Government of Colombia. These include the High Advisory for Reintegration,
the Colombian Vice-presidency’s Human Rights Observatory, and the Ministry
of the Interior. In the Philippines, I am grateful for assistance from Catholic
Relief Services and The Asia Foundation.
During my research I was honored to present my project as it evolved at the
Instituto Fedesarrollo, the CEDE seminar at the Universidad de Los Andes, and
CERAC, all in Bogotá, as well as the Department of Political Science at the
University of Maryland – College Park, the American Political Science Associ-
ation Annual Conference, and Stanford’s Workshops in International Relations
and Comparative Politics. The feedback I received from these presentations
greatly steered my thinking.
I am grateful for funding support from the Dwight D. Eisenhower/Clifford
Roberts Fellowship, the Smith Richardson Foundation World Politics and
Statecraft Fellowship, the Stanford Goldsmith Writing Award in Dispute Reso-
lution, the Stanford Diversity Dissertation Research Opportunity grant, the
Stanford CICN Research Grant, the Stanford Graduate Research Opportu-
nity award, Stanford’s FSI O’Bie Schultz Dissertation Travel Fellowship, and
Princeton’s Bradley Foundation Research Fellowship. I received support as a
Postdoctoral Fellow at Stanford and Princeton through the Empirical Studies of
Conflict program (ESOC), for which I am also grateful.
Lastly, I am thankful to my family. I am thankful to my parents, Sue and
Ron, for instilling in me the good sense to be able to complete this work with
sensitivity and insight and without harm. I also know that my “Colombian
journey of discovery” caused them at least a few sleepless nights and for that,
I am sorry. I am appreciative of my brother, Ben. Through the highs and lows,
you were always there with a smile on your face, maybe not always following
whatever I might have been droning on about as a struggling student, but
always nudging me to keep going. Thanks. You are an inspiration.
Abbreviations

ACIA Integral Peasants Association of the Atrato Region

ACR High Advisory for Reintegration/Colombian Agency
for Reintegration
ACVC Peasant Farmer Association of the Cimitarra River Valley
AFP Armed Forces of the Philippines
ANA Afghan National Army
ANUC National Peasant Association
ARMM Autonomous Region in Muslim Mindanao
ASOPROA Association of Small- and Medium-Scale Producers of
Eastern Antioquia
ATCC Peasant Workers Association of the Carare River
AUC United Self-Defense Forces of Colombia
CAFGU Citizen Armed Force Geographical Units
CDF Civilian Defense Forces
CINEP Center for Investigation and Popular Education
CNAC National Confederation of Communal Action
CNRR National Commission on Reparation and Reconciliation
COCOMACIA Community Council of the Peasant Association of the
Atrato Region
CPP Communist People’s Party
CPR Communities of Populations in Resistance
CRIC Cauca Regional Indigenous Council
DANE National Administrative Department of Statistics
DAS Administrative Department of Security
ELN National Liberation Army
EPL Popular Liberation Army
EZLN Zapatista Army of National Liberation
FARC Revolutionary Armed Forces of Colombia

xix
xx List of Abbreviations

FATA Federally Administered Tribal Areas

FMLN Farabundo Martí National Liberation Front
FSA Free Syrian Army
IGO Intergovernmental organization
ISAF International Security Assistance Force
ISIS Islamic State of Iraq and Syria
JAC Community Action Board
LCC Local Coordination Committee
LRA Lord’s Resistance Army
M-19 Movement of April 19
MAQL Quintín Lamé Armed Movement
MAS Death to Kidnappers
MBNC Bolivarian Movement for a New Colombia
MILF Moro Islamic Liberation Front
MNLF Moro National Liberation Front
NATO North Atlantic Treaty Organization
NDF National Democratic Front
NGO Nongovernmental organization
NPA New People’s Army
PCCC Colombian Clandestine Communist Party
RUF Revolutionary United Front
SDA Special Development Area
UP Patriotic Union Party
ZOP Zone of Peace
1

Introduction
Civilian Autonomy in Civil War

They’ve got me pissed with so much gosh-darn questioning

About what color my flag is, if I’m Conservative or Liberal.
They’ve got me fired-up with so much gosh-darn finding-out
About whether I’m an ELN-er, EPL-er, support the AUC or if I’m FARC.
They’ve got me worn-out with so much gosh-darn interrogating
About whether I’ve been opening my gate for the army and
giving them water from my well . . .
I’m a hard-working campesino, poor and very honorable,
I live happily but they’ve got me wound-up like a vine . . .
Well look misters, I’ll answer you all,
I want this to be clear:
I ain’t on nobody’s side, I do what’s right, not what’s wrong . . .
So that’s why I beg you, and ask you: questions – no more
Don’t screw with me anymore!
–Colombian folk song, “El Campesino Embejucao”1
by Oscar Humberto Gómez
Me tienen arrecho con tanta juepuerca preguntadera
que qué color tiene mi bandera que si soy Godo o soy Liberal
Me tienen verraco con tanta juepuerca averiguadera
que si soy Eleno, Epelo o siquiera apoyo a las AUC o soy de las FARC
Me tienen mamao con tanta juepuerca interrogadera
que si yo a la tropa le abro la cerca y si le doy el agua de mi manantial

1
Translated to English by the author. A campesino is a farmer or peasant. Embejucao is deri-
vative of the Spanish word bejuco, or vine, and is taken to mean wound-up like a vine, enraged
or worked-up. The ELN, EPL, and FARC are guerrilla groups. The AUC are right-wing
paramilitaries.

1
2 Introduction: Civilian Autonomy in Civil War

Yo soy campesino trabajador, pobre y muy honrao

vivía muy alegre pero me tienen embejucao . . .
Pues miren señores a todos ustedes yo les contesto
y quero que quede muy claro esto:
yo no soy de naide, hago el bien, no el mal . . .
así que les ruego, suplico y pido: ¡ya no más preguntas,
no me jodan más!

One day in the early 1970s in the village of La India in central Colombia,
residents were warming up for a soccer match on a field that was not much
more than a clearing in the forest. As one of the referees that day recalled, a
ragtag group of guerrillas in campesino garb and boots appeared out of the
jungle and asked if they could join.2 Short on players with few people living in
the area, the villagers welcomed them. It was not long before an army patrol
arrived. None the wiser that there were guerrillas in the mix, the troops asked
if they could also join the game and they all ended up playing a friendly match
(of course, the villagers eventually won). At the end of the game, one group of
“campesinos” said goodbye rather quickly and left through the jungle. For
several years, the villagers continued to play the occasional game against the
army, who remained none the wiser.
Historically, communities like La India have been intertwined with various
armed actors with close, often benevolent relationships and information flows.
This is possible because insurgents sometimes fit the mold of the idealized
benevolent guerrilla hailed by classical theorists like Mao Zedong (1961) and
Che Guevara (1961): noble, disciplined fighters defending the people and
pursuing justice.3 Governments, as counterinsurgents, are similarly advised to
protect the population (Galula 1965). But this is not always the case. At some
point in La India, things began to change. Conflict intensified and armed actors
became more hostile and violent toward the civilian population.
There has been a similar turn of events in the rest of Colombia and in many
other conflicts around the world. Beginning in the second half of the twenti-
eth century, civil wars have been prevalent, claiming the lives of an estimated

2
ATCC#1, La India, 10/2007. Interviews were conducted anonymously and are designated by
community, number of participant, location, and date. In this study, “armed groups” or “armed
actors” are terms used interchangeably to refer to any macro-level army in the armed conflict,
including “leftist” guerrillas, “rightist” paramilitaries, or the public forces of the government
(army, police, etc.). “Towns,” “counties,” and “municipios” are also used interchangeably to
refer to Colombian localities. Interviewee names, some place names, and other potential identify-
ing information have been changed to protect individuals.
3
According to Guevara (1961, 39), “The peasant must always be helped technically, economically,
morally, and culturally. The guerrilla fighter will be a sort of guiding angel who has fallen into the
zone, helping the poor always and bothering the rich as little as possible in the first phases of the
war.” He continues, “The line should be soft and hard at the same time: soft and with a
spontaneous cooperation for all those who honestly sympathize with the revolutionary move-
ment; hard upon those who are attacking it outright, fomenting dissentions, or simply communi-
cating important information to the enemy army” (81).
Introduction: Civilian Autonomy in Civil War 3

16.2 million people (Fearon and Laitin 2003). Indeed, most of the world’s
killings and human rights violations occur in conflict settings and most victims
are noncombatants (Sivard 1993), with civilians comprising four out of every
five of Colombia’s war victims (GMH 2013). Rebels can be abusive (Weinstein
2006), states use mass violence (Valentino 2004), and, with changing conflict
conditions, civilians come to face the predicament of the campesino embejucao of
the song: caught in the crossfire, “entre la espada y la pared,” or “between a sword
and a wall.” They can be stigmatized in the “fog of war” and accused of collabor-
ating with the enemy (Kalyvas 2006). In Colombia, they must additionally deal
with the problems of coca, youth recruitment, and displacement from their lands,
among other maladies. And, like the campesino embejucao, most civilians facing
less-than-benevolent armed groups just want to be left alone.4 What are people in
this predicament to do and what chances do they have? Are they helpless, inactive,
and consigned to a fate of abuse, as many accounts describe?
I argue to the contrary that civilians are not necessarily passive or powerless.
They are actors with agency whose ability to respond to the dangers of conflict
derives from social cooperation. Villages with different social relations deal with
increasing pressures and violence differently. This is illustrated by returning to
the village of La India as the conflict worsened and contrasting its experience
with that of the nearby village of San Tropel, just to the west over some low
hills. In 1998, paramilitary forces that had by then arisen in the region brutally
executed twelve woodcutters in San Tropel and dumped their bodies in the
Carare River (El Tiempo 2009).5 Just a short time later, in 2001, this same
group was preparing to kill eleven residents of La India, but did not because a
community organization that had been formed there to deal with the problems
created by the conflict, the Peasant Workers Association (ATCC in Spanish),
came to the civilians’ defense and advocated on their behalf.6 The eleven people
lived. Though this is but a single episode, it is emblematic of many similar events
in this community (I explore these communities in detail in Chapter 7).
This raises a puzzle: given similar pressures, why were residents of La India
able to act but not those of San Tropel? And why were the people killed in
San Tropel, but not in La India? More broadly, how common are these kinds
of actions? How do they affect armed groups and how can we tell if they affect
levels of violence? It is not obvious that unarmed civilians in civil wars can
protect themselves against heavily armed combatants, and yet some civilians
do. The attention of both journalists and scholars has concentrated on the
many victims of civil conflicts – “If it bleeds, it leads.” Yet in line with a

4
Popkin (1979), Kriger (1992), and Kalyvas (2006) suggest that macro-actor goals are not civilians’
primary concern. According to Nordstrom (1992) on Sri Lanka and Mozambique, civilians are
often not even familiar with macro-actors’ goals, “Civilians often had difficulty distinguishing
sides, especially according to ideological considerations of just and unjust. Indeed, many of the
victims of war – torn from comfort and community, family, and home, too often wounded or
bereaved – do not know what the conflict is about or who the contenders are” (265).
5 6
Heard of by ATCC#2, La India, 10/2007. ATCC#3,4, La India, 10/2008.
4 Introduction: Civilian Autonomy in Civil War

corollary that “If it’s nonviolent, it’s silent,” few accounts examine how the
people who are not victims survive. Millions of people have been displaced
from the countryside in Colombia, but many have been able to remain as well.
The topic of this book is how civilians can retain their autonomy, or self-
rule, in the face of armed groups and protect themselves.7 Civilians may flee
violence or seek protection from an armed group, but these options can be
dangerous, unavailable, or unappealing, since many would prefer to stay in
their communities. Facing this dilemma, their alternative in autonomy is to
actively avoid participating in the conflict between the contending armies to
avoid its damaging effects and gain even a small degree of certainty in their
daily lives.8 However, in changing, complex environments and with but one life
to live, following this course is difficult for most individual civilians. Even when
many civilians might share such preferences and together be more effective in
gaining protection, fear creates collective action problems in confronting com-
batants. Some residents may receive benefits from relationships with armed
groups, or armed groups may seek to penetrate and control communities – to
divide and rule. Social cooperation and organization is therefore the key to help
civilians overcome fear, manage their own communities, and deal with armed
group pressure in an enduring manner. The narrower question of autonomy
and protection from armed groups therefore links to the broader question
of what capacity civilians have for social cohesion and cooperation in war
settings – the question of social capital (Putnam 1993; Buonanno et al. 2009
related to crime). While civilian strategies for autonomy can and do arise
specifically as a response to deal with armed conflict, preexisting bases of social
cooperation are a helpful catalyst.
In Colombia and other countries there are many notable examples of local
organizations for autonomy and organizational actions for protection in war-
time. A review of these cases reveals they are more common than one might
believe. To give an initial sense of their breadth in Colombia, according to one
survey, more than 500 local officials (mayors and governors) held dialogues

7
The concept of autonomy has been previously introduced in UNDP 2003, Sandoval 2004, and
Tarrow 2007. According to the UN report, “Autonomy of citizen movements vis-à-vis the armed
organizations and indeed the State itself . . . but not neutrality has allowed them to keep them-
selves apart from the armed confrontation. They have avoided taking sides in favor of one or the
other band, but always show themselves to be on the side of the population. Since armed groups
attempt to involve them in the conflict, they have claimed their right to survive. That way any
initiative from a group is replicated in another, and any position is communicated equally to all.
They have also earned themselves a degree of autonomy with respect to central government. As
one activist put it, ‘What’s on the line are our lives, not the government’s life.’ Therefore they look
for alternative ways of handling the conflict, beyond the desires and recipes of the national
government. They know that they cannot sit around waiting to act until the government has
organized its grandiose ‘negotiations’ with armed groups.”
8
I make no normative judgment about the righteousness of either participating in “liberation”
struggles or defending the establishment, “la patria.”
Introduction: Civilian Autonomy in Civil War 5

with armed groups during the peak years of conflict (El Tiempo 2001a).9
To more precisely illustrate the prevalence of civilian autonomy organiza-
tions and actions, I classified community cases based on the criteria of being
local and grassroots-based (i.e., at the community level), “apolitical,” based on
social cohesion and organization (not individual), and nonviolently seeking
protection from violence (from one or more groups).10 Across Colombia, more
than fifty locations formally organized for self-protection from armed groups
since the early 1990s.11 Figure 1.1 shows a map of these civilian organizations
by their municipio, or town. These cases are found in many regions of the
country and vary according to types of strategies implemented and the number
of people, villages, and areas involved.12
Civilian autonomy in conflict settings is also found more broadly around the
world than might be expected. Figure 1.2 shows a map of cross-national protect-
ive actions by civilian organizations in conflict conditions that were classified

9
However, few public officials wish to publicly admit to such dealings. Interactions varied from
mere intimidation and conversations under pressure to communication, small-scale humanitar-
ian accords, peace communities, constituent assemblies, and voting. As Gilberto Toro of the
Federation of Municipios observed, “While the state is impotent in guaranteeing local govern-
ability [stopping abuses of the civilian population] we are going to have desperate mayors
turning to new ideas.”
10
These cases may include idiosyncratic actions and social processes as well as formal organiza-
tions created to promote local peace. A case is considered “apolitical” if, in its beginnings, it does
not have apparent, formal, sustained relationships with macro-political actors, such as the public
forces or armed groups. However, some communities are eventually co-opted by or integrated
with the state. Some but not all of these cases confront conflict environments with multiple
macro-actors. The communities that are identified are not believed to have formal relationships
with any macro-political actors, but from afar it is difficult to tell by which group(s) they may feel
threatened. Inferences about the number of armed groups these communities face are found in
Table 1.1.
11
I classified cases from secondary sources through a broad and admittedly nonsystematic search
since there is not usually standard language to describe these experiences (i.e., many more
“silent” cases might exist). These codings are therefore not comprehensive and possibly omit
many actions. In some cases the limited information and context available in the reporting also
present challenges in assessing the nature of events and social cooperation, possibly producing
classifications that are inaccurate. The resulting collection of cases are examples of collective
actions but are not precise analyses of the effects of civilian strategies, which would require much
more labor-intensive measurements – such as those found elsewhere in this text.
12
While the autonomy movements do not encompass every village within each municipio where
they are located, the number represents roughly 10 percent of municipios with the presence of at
least one armed group at the peak of the conflict or about 20 percent of the smaller set of
municipios that were affected by conflict for extended periods during the 1990s and 2000s. A
number of these organizations are profiled in the United Nations Development Program’s Good
Practices website database, a catalog of over 250 community experiences intended to disseminate
examples of and lessons from how different civil society actions and organizations have “over-
come” the armed conflict. Some of these are user-submitted. Though the database is still surely
missing many experiences and movements, 51 of these profiles explicitly mention the aim of
“autonomy.” Greater detail on the diversity of actions taken by communities is contained in the
history and case study chapters ahead.
6 Introduction: Civilian Autonomy in Civil War

figure 1.1 Map of formal civilian autonomy organizations in Colombia

similarly to the Colombian cases (Table 1.1 at the end of this chapter contains
further details on these examples).13 These fourteen countries with civilian

13
Many but not all of the countries where these cases are found fit the accepted national level
definition of civil war of at least 100 annual battle deaths for each party to a conflict and at least
Introduction: Civilian Autonomy in Civil War 7

figure 1.2 Map of civilian autonomy in civil war around the world.
Source: Author’s coding.

autonomy actions comprise around one-quarter of countries that experienced

civil wars since 1980 (including only some communities from these coun-
tries). Instances of protective actions and organizations are found in Asia,
Africa, the Middle East, and Latin America. In some countries, there are only
a few press reports on one or two isolated communities, such as the profile of
a single village leader in Dagestan, Russia, who negotiated a path for his
community between Russian counterinsurgents and Islamist militants (Greene
2010). In other countries, there are cases that are more deeply profiled in the
budding anthropological literature on this topic and involve many commu-
nities. Studies of the Peace Zones in the Philippines identify at least ninety-
one communities that organized with the help of the Catholic Church to opt
out of the conflicts between the Philippine military and communist and
Muslim rebel groups (Santos 2005, Hancock and Mitchell 2007). There are
cases as diverse as the religiously motivated and superstition-based Naprama
movement and Jehovah’s Witness communities in Mozambique (Wilson
1992) as well as strategies of avoidance as found in Guatemala’s Commu-
nities of Populations in Resistance (CPRs, Falla 1994).
Examples of armed resistance for autonomy are also included in Table 1.1,
although they are not the emphasis of this study.14 Cases of local armed

1,000 total annual battle deaths among all sides (Fearon and Laitin 2003). Cases in countries
falling short of this standard were included because their locales still suffered what would
reasonably be considered civil war conditions and contestation.
14
There are additional cases of apparent armed resistance in Colombia and other countries but
these are excluded because their origins are not entirely bottom-up. Rather, they are political and
“pro-government militias” allied with the state. Examples from Colombia in this category
include the Convivir self-defense forces, the village-based peasant soldiers program (soldados
campesinos), and paramilitary groups.
8 Introduction: Civilian Autonomy in Civil War

resistance range from the Rondas Campesinas studied by Starn (1999) and
Fumerton (2001) in different regions of Peru to Iraq’s Anbar awakening of
Sunni tribes (at least in their beginnings) to Muana’s (1997) description of
the origins of the Civilian Defense Forces (CDF) in Sierra Leone. In Chapter 9,
I more closely review instances of civilian autonomy from the conflicts in
Afghanistan, Pakistan, the Philippines, and Syria.
Many of these cases are highly organized and have been publicized, but there
may also be more subtle, underreported kinds of civilian social cooperation and
strategies. Given this variety, it can be hard to tell whether and which social
organizations and collective strategies explain differences in resilience to vio-
lence across communities. This task is further complicated by possible reverse
causality and selection biases since, if conflict harms social cooperation, then
social relations and civilian strategies may be solely derivative of the powerful
armed groups, with no independent success at suppressing violence. Any
observed existence of civilian collective action or impact would then be epi-
phenomenal, or merely due to existing in peaceful places. Alternatively, many
civilian self-protection processes largely exist in stateless, conflict-ridden areas,
which could also make them more predisposed to suffer violence. Given these
research challenges, I address three interrelated research questions:
(1) Where and why do local social organizations arise?
(2) What strategies do such organizations permit communities to use to deal
with civil war violence?
(3) Why and under what conditions do armed groups change their behavior
toward (organized) civilians?
To study the question of civilian autonomy in civil war, I use multiple social
science methods, from statistical tests to interviews and case studies from the
Colombian conflict. This process involves the careful construction of counter-
factual scenarios of what would have happened, what the armed groups would
have done, had civilians not used a given strategy or taken a given action. I find
that some kinds of civilian social arrangements, organizations, and strategies
can reduce civil war violence, suggesting that civilian autonomy occurs more
broadly than originally believed. At the same time, and with reason, there are
also limitations on where civilian organizational processes succeed – they are
not a panacea. By exploring these conditions, I find that, along with successes,
communities experience challenges and failures. While some organizational
processes appear to affect certain kinds of violence and conflict dynamics,
organizations are also more likely to buckle under extreme levels of combat
and overall have few discernable effects on the intensity of the fighting between
belligerents themselves.
In the rest of the chapter, I first present my main argument about the
importance of civilians’ social capital, organizations, and strategies for limiting
civil war violence and delimit its scope. I also briefly summarize the research
findings in support of this argument. I then situate this study in existing
The Argument 9

literatures and indicate how they have so far only obliquely addressed the
questions surrounding civilian autonomy. I then preview the research design
and methods. Lastly, I preview the chapters to come.

the argument
My main argument is that there are conditions where civilians can use social
processes to reduce violence perpetrated against them. The field of conflict
resolution usually pays most attention to state actors, nonstate armed groups,
and national-level peace negotiations. By contrast, this argument is grounded in
civil society (sometimes referred to as “Track II”) in considering why civilian
bystanders succeed or fail in organizing opposition to state oppression or to a
potentially harmful insurgent movement. Since armed actor coercion of (and
violence against) civilians has been theorized to stem from divisions within
civilian communities (Kalyvas 2006), I argue that social cohesion affords
civilians greater chances to overcome fear, break the “law of silence” and
revive communication, and implement collective strategies for protection.
In situations where communities face multiple armed actors or even a single
abusive group, compliance and alliance do not guarantee protection. It is here
that cohesion and collective strategies can help communities achieve autonomy,
or maintain democratic decision-making power over outcomes for the commu-
nity within the community, without influence from outside armed groups.15
Violence can be reduced through institutional solutions to avoid participation
in the conflict, manage the internal order of communities, limit the inroads of
armed groups, and demand accountability from these groups.
With three outcomes to explain in this book, I develop a three-part “civilian
autonomy theory” that links civilian organizations to strategies and then to
security. First, variation in the social and demographic landscape and technical
assistance from external actors (be they the government, churches, or NGOs)
propels some communities to organize more easily than others. Second, cohe-
sive, organized communities can make collective decisions about how to best
deal with the various dangers of civil war conflict. Although civilians may
commonly align with dominant armed groups or displace, in addition to these
standard strategies, I pay special attention to strategies to retain autonomy in
the midst of multiple armed groups. These can include what can be termed
“weapons of the weak” (Scott 1985) for conflict settings to deal with conflicts
and divisions within civilian communities as well as “weapons of the not-
so-weak,” such as overt protest and actions by nonviolent community guards.
The selection of strategies is shaped by an interaction between civilian

15
These arguments were first published as my doctoral dissertation (Kaplan 2010). My main
interest and assumption is that civilian responses to violence are usually instrumental, inten-
tional, and with strategic forethought. However, they can certainly also be emotional, cathartic
processes or be born out of frustration.
10 Introduction: Civilian Autonomy in Civil War

preferences, social cohesion, and the past and prospective threat environment,
and some strategies are more assertive and thus potentially more effective than
others. Third, the strength of civilian organization, selected strategies, and
armed group incentives jointly determine substantive outcomes such as levels
of violence. In sum, it is the unity of civilian centrists that helps impede and
isolate violent “extremists.” This theory is stated in general terms to pertain
both to the case of Colombia and be adapted to explain patterns in other
conflicts.
A main task of this theorizing about autonomy is to specify civilian mechan-
isms that generate protection. I show how different civilian actions reduce
violence by affecting armed actors’ behavior, capabilities, or ways of thinking.
I take a moment here to foreshadow the strategies and mechanisms that
I identified through reading and fieldwork and elaborate on them in greater
depth in Chapter 2.16 The multiple kinds of violent threats civilians face call for
multiple solutions. Subject to the constraints of available ideas and imperfect
information about levels of danger, civilians may thus select different kinds of
autonomy strategies in different places as they organize and adapt responses to
different types of violence.
These autonomy mechanisms can be grouped according to their level of
formalization and depth of cooperation. First, there are cohesion and solidarity
mechanisms, which more resemble ad hoc coordination in that some are less
premeditated or enduring. These can include preexisting social harmony, which
means fewer conflicts among neighbors to exploit, or the common knowledge
among residents that allows them to collectively and spontaneously protest
aggression and resist armed groups’ attempts at domination.
Second, there are formal organizations and mechanisms that are based on
deep and sustained cooperation and intentionally oriented for protection.
Civilians can actively promote ideational norms among residents against aiding
armed actors (a so-called culture of peace), develop local conflict resolution
processes (so civilians do not seek policing by outside actors), develop early
warning systems to avoid combat, dialogue with armed groups and investigate
suspected enemy collaborators for them, and link with external nongovern-
mental organizations (NGOs) and international governmental organizations
(IGOs) to “go public” to protest aggression and shame armed actors. As one
man told me in reference to these kinds of strategies, “Creating peace is an
everyday process.”17 Some strategies are more contentious toward armed
actors than others, and some formally organized communities that use these

16
The terms “strategy” and “mechanism” are interchangeably used to convey processes that affect
violence. However, a mechanism is a causal process whereas a strategy is a plan that is chosen by
an agent. Some mechanisms that affect violence, such as preexisting cooperation, are not
strategies that civilian agents can actively choose, but are still mechanisms that affect armed
group violence.
17
ATCC#3, La India, 10/2007.
The Argument 11

strategies may publicly opt out of the conflict and declare their territory and
population off limits to combatants.
With this articulation of strategies, I argue for a more nuanced view of how
civilian communities act. Formal, public protests have a role to play, but peace
and stability do not always require banning or expelling armed groups from
territory and may not depend on singular, publicized events. There may also be
other, less visible strategies. A main observable implication from both the covert
and overt kinds of mechanisms is that greater organizational capacity should
predict less violence after accounting for other factors that predict violence.
Even as this work shifts focus to civilians and their agency, civilians’ actions
cannot be divorced from armed groups, whose motives and incentives, prefer-
ences and choices are key for explaining violence visited upon communities.
Armed groups coerce or abuse in part because they suffer little consequence,
have poor information, or have not considered other ways to achieve their
goals.18 In theory, by making killing more difficult or even costly and by
reducing mistakes in targeting and costs of governance, civilian cohesion and
the strategies it enables should reduce violence. It becomes more difficult for
competing armed groups to seed violence through social divisions or for com-
manders to tolerate organizational abuses. Civilians may induce armed actors
to change behavior because of what I refer to as the “sensitivity” to their
reputations and legitimacy, by incorporating new ideas, or by affecting internal
group politics.
These armed actor motives provide insight into when civilians’ efforts are
likely to succeed, suggesting a broad but delimited set of conflict conditions and
armed group preference profiles. Communities should be more likely to affect
violence and civilian livelihoods in areas where armed groups might have
incentives to commit violence but hold preferences that are relatively more
flexible and, with the imposition of only small costs, can be persuaded to
change their behavior. This may occur, for instance, when armed groups have
what I call “live and let live” preferences and even in some cases “abusive-
coercive” preferences. They may not be Guevara’s “guiding angels” but are
more concerned about limiting civilians’ defections than winning their full-
fledged support. Groups may also be so inclined when they do not completely
depend on civilians for their resource bases. In contrast, civilians should be less
able to avoid violence in cases where groups are highly resolute in killing or
winning, such as cases of genocide (or “draining the sea” to get to the insurgent
“fish”).19 These scope conditions for autonomy add a dose of realism to the

18
Violence can also be irrational and random. My focus is on systematizing how civilians may on
average influence the “rational” behaviors of armed groups.
19
Armed resistance groups during the Holocaust and other genocides are noteworthy exceptions
(Semelin 1993, Tec 1993). In Colombia, examples of armed civilians fighting off armed groups
include the towns of Don Matías in Antioquia (Ivan García 1994) and Rionegro in Santander
(El Tiempo 1996).
12 Introduction: Civilian Autonomy in Civil War

discussion by not postulating civilian success as inevitable. They are found to

varying degrees within Colombia as well as other countries.
A related question to what impact civilians’ organizational efforts have
on violence is where their organizations come from. Some forms of social
cooperation and organization can be stimulated as a response to violence and
serve as bases for mobilization. Others can exist prior to the onset of conflict.
I draw on collective action theories and literatures on protest and resistance
to theorize about when civilians will coordinate or organize spontaneously
“now out of never” (Kuran 1991) to cope with violence. I explore the social
conditions associated with the prevalence of organizations across towns and
take organizations to reflect a degree of cooperation.20
The sources of organization are an interesting topic in its own right and are
also helpful for validating a causal relationship between organization and
violence. This is because, in an alternative process to civilian autonomy, armed
conflict may damage social relations instead of spurring civilian responses.
With organizations surviving and existing in peaceful areas, researchers could
be susceptible to bias in the selection of cases for study, making an observed
negative relationship with levels of violence spurious. Since I argue that civilians
have independent effects on violence, I additionally argue that conflict does
not on average disrupt social relations or censor the selection of autonomy
strategies.
To the extent that civilian organizations and their strategies are not merely
derivative of (epiphenomenal to) conflict dynamics, they merit consideration as
an additional explanatory variable. Indeed, by accounting for prior levels of
conflict through a historical statistical analysis, I find that conflict does not
necessarily disrupt social relations and can even reinforce them. The case
studies in Chapter 8 reveal examples of exogenous stimuli for organizations
in towns with histories of conflict. Some organizations may also exist in unfav-
orable conflict dynamics and take on lives of their own, persisting even as
dynamics change. However, consistent with the success of civilians’ strategies
being conditioned by armed group actions, cohesive communities and strong
organizations can protect but, under pressure, their resilience will be put to test.

Colombian Social Organizations as Platforms for Civilian Autonomy

One Saturday morning in August of 2008, I saw the central role of organiza-
tions in vivid relief. I was sitting in a plastic chair in front of a store on a dusty
lane in the village of La India, idly chatting with a new acquaintance from the

20
Communities face challenges to mobilization because they may have few resources to offer
selective benefits, have difficulty using the law to enforce contracts given state weakness, and
cannot use coercion since they espouse nonviolence. Many organizations persist and maintain
membership because they provide social benefits and opportunities for social interaction and
exchange. In some cases this may be because the organization is the only “game in town.”
The Argument 13

community. Suddenly, a woman ran out into the street bawling at the top of her
lungs. Wailing with sorrow, she dropped to her knees, hands covering her face,
and shrieked, “They killed him, they killed him!”
In an instant, seemingly the entire village came out onto the street and
surrounded her, including the man I was chatting with, who ran to join the
growing crowd. These friends and neighbors encircled the crying woman
five people deep, practically giving her a big group hug, asking about what
happened and showing concern. Feeling like an outsider and challenged to
interpret the scene, when the man eventually returned I asked, “Who? Who’s
been shot? What’s happening?”
The woman had received a cell phone call from her son, who was working in
the fields nearby. He had been shot, apparently by a local criminal gang
(pandilla). He said, “Mom, I’ve been shot, and I’m dying.”21
What happened next was even more remarkable, as I saw the village swing
into action and saw the community process of the local farmer association – the
ATCC – that I had heard about play out before my eyes. With all the commo-
tion, the vice-president of the association quickly arrived at the scene. When he
found out what was going on, he immediately jumped on a motorcycle and
sped off to make the two-hour trip to the nearest police station to get help, report
the crime, and push the authorities to capture the assailants. As he passed
me, he turned his head and we briefly made eye contact. Behind his stoic
expression – like the one often worn by campesinos – I felt there was also a
message, as if he meant to say, “Sometimes, this is what we have to do here. It’s
not happy, but we do it.”
The collective spirit to console the woman and protect the community made
an impression on me as I realized this is what communities like La India have
done for years. When transporting my argument to Colombia, it applies first
and foremost to formal “peace” organizations, such as the ATCC organization
in La India. But a key insight from the argument is that civilians also take
important actions outside of formal, well-known organizations that were
designed or named specifically to promote peace, such as “peace communities.”
Taking a step back, it is apparent that many of the peace organizations are
built on the foundations of smaller, more basic community organizations
and social unity. I therefore consider several additional measures of rural
community organization. These include the highly organized Indigenous and
Afro-Colombian populations, economic cooperative associations, and land
reform councils (ANUC), though the principal source of variation I study are
local village or neighborhood councils of campesinos (mestizo peasants) known
as juntas de acción comunal (Community Action Boards) across Colombian
municipios (towns).

21
I was not able to determine which group was responsible (whether it was a neo-paramilitary
criminal band) or what the outcome was for the young man.
14 Introduction: Civilian Autonomy in Civil War

The most common form of rural organization in Colombia, the juntas are
the main forum through which residents coordinate to solve local problems and
provide public goods. They were formed beginning in the late 1950s to bolster
the countryside after a bloody conflict known as La Violencia (The Violence)
but were later largely left to their own devices. The juntas were intended to
foster reconciliation and “coexistence” but were not specifically designed to
undertake protective strategies during war. Nevertheless, I theorize that juntas
affect violence because they embody the high levels of coordination and social
capital necessary for a community to implement more complex autonomy
(or other) strategies and procedures to preserve the community in the face of
conflict.22 Examining the presence and functioning of juntas presents a conveni-
ent way of tackling challenging research design issues because they are more
simply measurable and comparable than larger, irregular autonomy organiza-
tions, vary across many geographical units, and predate the recent period of
armed conflict.
The analysis of the junta councils yields several notable research findings.
After controlling for combat and contention among armed groups, a variable
representing juntas has a negative effect on violence. On average, if a town were
to move from no juntas to the 75th percentile in junta coverage, it would
produce an approximate 25 percent reduction in selective killings, but mainly
when the conflict intensity remains moderate. Ethnic minority populations also
appear to suffer less violence, but the evidence on other social organizations
suggests they are less helpful. The juntas are found, if anything, to be more
prevalent in historically conflicted areas, including those areas that were
affected by the brutal La Violencia conflict during the 1950s.
The case studies in Chapters 7 and 8 show that juntas helped cement social
relations and later were catalysts for communities to maintain social unity and
stand up to armed groups. In the ATCC zone, juntas were key subunits of an
information-gathering, coordinating, and pacifism-promoting network. In
other towns in Cundinamarca, the juntas appear to have played more subtle
but still important roles for dealing with armed conflict. Histories from these
towns also show that while the juntas were effective at buffering communities
they were also weakened prior to the onset of conflict by factors such as
clientelism. This suggests a prominent role for the legacies of juntas in addition
to the juntas themselves.
Since directly measuring autonomy and predicting strategies prior to their
occurrence is complicated, my analysis focuses on the prior characteristic of
organization. Organization enables a variety of strategies that could account
for differences in levels of violence across towns, of which autonomy is only

22
The relevant “social capital” of local organizations such as juntas can take the forms of shared
preferences and reduced intracommunity conflict, participation, and information networks,
which can affect outcomes of violence through several mechanisms. These organizations may
also reflect a stronger collective identity.
The Argument 15

one. Organization may have general benefits for reducing violence above and
beyond what individuals are capable of, regardless of conflict conditions. But
organization can also stand in for collective choices to flee or ally, in addition to
enabling autonomy. To isolate the link between organization and autonomy, I
also analyze the effects of community organization under the conditions of
threats by multiple armed groups, where autonomy is most expected.

Scope of the Argument

My argument is delimited by several bounds of scope that clarify the concept of
civilian autonomy and the conditions for when cohesion, organization, and
autonomy strategies are likely to help avoid violence. To begin, I focus my
attention on collective, localized, ongoing processes at the community level in
conditions of civil war. Collective processes are prioritized because individuals
have limited choice sets facing violence. Localized processes are prioritized
because they are principally aimed at dealing with the specific, immediate
problems a community faces and are more geographically delimited than
general, broad-based peace “movements.”23 Ongoing community processes
are elevated over specific actions because actions must generally be sustained
over time to continue to keep violence from visiting the community since armed
groups can return. In this framework, protest events are but one of various
strategies that an organized community may employ.
The study also focuses on the rural sector. In rural areas the state tends to be
less present and the physical and membership boundaries of communities are
blurrier, making autonomy strategies both more necessary and likely. Collect-
ive strategies and movements for protection and autonomy can and do occur in
urban slums, but urban areas have clearer lines of armed group control, better
information flows, and the possibility of more forceful responses by state forces
if conflict spikes, and armed groups operate more clandestinely.
I acknowledge but do not emphasize the role for civilians’ preferences over
political affiliations or particular strategies. When civilian communities hold
strong preexisting preferences for or against a given armed actor these prefer-
ences can affect their choices. However, evidence suggests these ideological
preferences tend to be weak relative to the preference for doing what it
takes to survive when under threat – they have no “dog in the fight.” Though
I discuss and measure the preferences of residents of different communities,
I assume civilians’ preferences in violent situations are not static ideologies but
instead tend to be influenced by armed groups (similar to Kalyvas 2006). I see

23
Civil society peace actions such as region- or nation-wide marches, symbolic votes, protests, etc.
can certainly still play important roles for resolving civil conflicts and reducing violence (for
instance, general arguments by Sharp 1973 and Sharp and Paulson 2005, anti-ETA protests in
Spain by Funes 1998, and nonviolent protest in Colombia by Cante and Ortíz 2005 and García
Durán 2006).
16 Introduction: Civilian Autonomy in Civil War

the main limiting factors for civilians to respond to armed actor aggression as
organizational capacity, cohesion, and coordination, rather than preferences to
do so. To the extent that preferences influence civilians’ prospects, I argue it is
the variance in residents’ preferences that matters – either social unity or
disunity, which can influence civilians’ capacity to work together and keep
armed groups out.
I refine my argument by identifying certain types of organizations as more
effective at reducing violence through autonomy than others. A critical organ-
izational characteristic is whether organizations attempt to remain apolitical
relative to the interests of macro-armed and -political actors. If civilian organ-
izations or communities hold political stances, they can be stigmatized or
targeted for perceived alliances. Indeed, this alternative depiction of organiza-
tions has found support in other notable cases such as El Salvador (e.g., land
cooperatives were targeted by the army, see Wood 2003). Given the risk of
stigmatization, I assume communities or organizations that have delimited
aims, act for self-defense, and do not pursue larger political projects (including
state takeover) will have the best chances for protection through autonomy
strategies, since their claims to political neutrality will be more credible.24 For
instance, I find that the relatively apolitical juntas reduce violence more than
legacies of land reform councils, which may sign a community as espousing
leftist platforms (Chapter 5). A community may be targeted if it is seen as a rival
or contender for power (in the case of land reform councils, by right-wing
paramilitary groups).
I concentrate primarily on nonviolent civilian strategies to deal with armed
groups. Social movements (Tarrow 1994) and nonviolent resistance (Schock
2005, Chenoweth and Stephan 2011) have been shown to be effective at
pressuring states to respect rights and overthrow autocratic regimes, but rela-
tively little attention has been paid to how these tools can be used during armed
conflict or to pressure nonstate actors. Yet civilians’ nonviolent strategies
during war merit attention, first, because they are more prevalent than armed
local civilian resistance strategies in Colombia (though there certainly are
noteworthy examples of effective armed resistance for self-defense and auton-
omy, including the indigenous Movimiento Armado Quintín Lame; Peñaranda
et al. 2006). These civilian groups are differentiated from the armed actors of a
conflict by acting for self-defense, with no wider goal of state takeover. In the

24
Communities and organizations that aim to be apolitical often still advocate for community
development. Advocacy for development can blur the line of a community’s “political” position
when rural peasants incorporate anti-neoliberal platforms and rhetoric into their organizing, as
they sometimes do, since they resemble the political platforms of many insurgent groups. This may
increase risks since armed actors may associate these positions of residents with their stances in
other policy domains including, for example, whether they have explicit links with or sympathies
for other macro-political actors. However, when simply staying on one’s land becomes political
and grounds for being targeted, the situation begins to approximate genocidal conditions as there
may be little a community can do given an armed actor’s extreme preferences.
The Argument 17

case of armed strategies for autonomy, this criteria effectively excludes political
paramilitary projects that either from their inception or in due course aid the
state with counterinsurgency.25 Second, focusing on nonviolence avoids the
ambiguity and ethical dilemma of whether armed civilians are still “civilians.”
Armed resistance can certainly be an important strategy for maintaining auto-
nomy, though by directly participating in hostilities and forgoing the status
of “noncombatant,” civilians can become legitimate targets of armed groups
(permissible targets under International Humanitarian Law).
My main interest is understanding the ability of civilians to reduce violence
against their communities and protect human rights. Yet civilian social
processes may also fulfill many other functions during conflicts. These can
include increased cooperation and participation, improved governance, eco-
nomic development, and less tangible results such as the psychological benefits
of inspiration and empowerment (within the community and as a symbol for
other communities), and a sense of security, hope for the future, belonging,
identity, community, etc. I set these benefits aside for others to study because,
even if they are realized, their value and the value of civilian autonomy strat-
egies could ultimately be questioned if they fail to reduce violence. While there
may be many senses in which civilian movements promote peace, here peace is
conceived of as the extent to which violence is reduced.
I mainly study the effects of social processes on violence against civilian
community members and do not deeply explore outcomes for the broader
armed conflict or for armed groups. It would be helpful to know whether and
when local initiatives affect larger dynamics of conflict such as onset, termin-
ation, the spread (or displacement) of violence to other communities, or the
intensity of fighting between armed groups. However, since some conflict
dynamics are largely a byproduct of interactions between the armed groups
themselves, only certain civilian mechanisms might have influence, by affecting
armed groups’ resources or capacities, for example. There are also meas-
urement challenges to detect these effects with the available granularity of
measurements at the local level. Even so, some parts of the study have impli-
cations for the broader conflict and I later discuss some pertinent examples,
such as how autonomous communities can assist with the demobilization of
combatants.26
Lastly, and to reiterate, civilians’ success at protecting themselves also
depends on the conflict conditions and armed group preferences. Cohesion

25
Some armed actors build their own civilian organizations or cadres as support networks, such as
the FARC’s Movimiento Bolivariano por la Nueva Colombia (MBNC) militias and Unión
Patriótica (UP) political party. With the criterion of being apolitical, I distinguish forms of
organization where initiative is rooted in civil society.
26
The ATCC has provided guarantees for local troop demobilizations (ATCC#3, La India, 10/
2007). The Nasa Indians have also used nonviolent methods to stop FARC attacks on police
posts (El Tiempo 2001b).
18 Introduction: Civilian Autonomy in Civil War

and community processes are expected to be able to deal with up to moderate

levels of contestation and at least moderately amenable armed group types.
Communities are more likely to be overrun in the midst of heavy fighting or
when facing highly resolute or ruthless groups.

current perspectives on civilians and civil wars

This study engages and bridges several academic literatures. I first review
studies of violence in civil war and how civilian behavior has been portrayed
in these studies. I then review the contributions of studies on civilians in other
kinds of conflict settings short of civil war yet where the state is absent. Lastly,
I contrast these studies with the anthropological literature on community resist-
ance to armed groups (originating mainly on Colombia). I highlight how the
study of both violence and social movements can be advanced if these litera-
tures better engage with each other and how this study works toward that goal.

The Literature on Civil Wars

The turn toward micro-level analysis of violence in the recent literature on
civil wars provides a strong foundation for analyzing the effects of civilian
institutions and autonomy. Rather than focusing on the normative and legal
aspects of the human rights regime and rights violations, these works have
espoused positivist arguments to account for what actually happens in conflicts.
For instance, Valentino (2004) argues that massive violence in the form of
“draining the sea,” or targeting civilians to deprive rebels of support, is a
common strategy in counterinsurgency. Weinstein (2006) argues that rebel
organizational structures can enable or limit abusive behavior toward civilians,
even in zones of complete control.27 Kalyvas (2006) argues that, as a result
of the “identity” problem of outing enemy supporters in the “fog” of irregular
wars, armed actors use “selective” violence to coerce support. There have been
several valuable quantitative studies on the patterns of conflict and violence in
Colombia by authors such as Restrepo et al. (2006) and Sánchez Torres (2007).
These studies all share the common thread of placing primacy on macro-actors
in armed conflicts and how their decisions and behaviors affect war onset, rebel
recruitment, violence, and war termination. The insights from these works
mean it is now essential to control for these factors in any analysis of civilian-
based explanations of armed conflict. Yet these existing explanations leave
unexplained variation in violence – how does violence vary within the same
conflict conditions and among the same armed groups? How does violence vary
within areas where states are weak?

27
Azam (2006) provides another explanation for violence in zones of control, arguing that
warlords may also have incentives to victimize their own civilians, first for plunder and second
to suppress wages and lower the opportunity cost of recruitment.
Current Perspectives on Civilians and Civil Wars 19

The role of civilians in this literature is eclectic, though civilians are seen as
primarily integrated with armed groups, with limited independence. Little is
said about how civilians deal with and diminish violence. Many works have
studied the challenge of rebel recruitment in civil wars including Popkin (1979),
Taylor (1988), Lichbach (1994), and Humphreys and Weinstein (2008). These
works analyze the collective action problem rebel groups face in inducing
civilians to join them and become combatants – to mobilize for violence.
Explanations for participation range from ideology to selective enticements to
coercion. Even if civilians are not recruited directly into the rebel ranks, they
may still provide aid to armed groups in the form of resources (Wood 2003) or
information (Kalyvas 2006).
Studies on civilian participation in conflicts share some similarities with this
study but also exhibit important differences. Similar to my arguments, Wood’s
study argues civilians’ risky choice to support rebels is also a form of agency. In
her study, based in El Salvador, the civilians’ aid to guerrillas was rooted in the
guerrillas’ benevolence toward the peasants (by pursuing the revolutionary
goals of liberation and political autonomy in relation to landholders) and
largely observed in areas of complete guerrilla control. With these differing
conditions from those that I focus on, Wood arrives at a distinct form of civilian
agency that is neither autonomous nor protective.28 In another study, Petersen
(2001) emphasizes the role of strong communities in how people are “pulled
into rebellion” to form resistance movements. By contrast, I consider how
communities avoid rebellion (e.g., how to stay at what he calls the “zero”
position when it is the safest option). On Colombia, Salazar and Castillo’s
(2001) theoretical models describe when civilians will either aid a dominant
armed group or displace, but nothing more.
In the context of civil wars, Kalyvas is one of the main scholars to systematic-
ally incorporate civilian processes into the production of violence. He highlights
and emphasizes the mechanism of denunciations by individual civilians (for
personal or ideological reasons) to explain selective violence by political actors
as a “joint process” over any civilian collective action aimed at opposing such
divisions.29 His theory does make a place for what he calls civilian “local
committees,” which he sees as “small information processing groups” that
can screen denunciations by individuals and veto selective killings. However,
he does not deeply theorize about or measure this process.30

28
Wood also studies the demilitarized community of Tenancingo, which might be considered a
case of civilian autonomy though it is not clear whether this effort was organized from the
bottom up.
29
However, coercive violence is not always well-targeted, “There is evidence that political actors
are successful in generating deterrence via selective violence in spite of killing many innocent
people” (Kalyvas 2006, 109).
30
According to Kalyvas, “Local, usually village-based committees handle and screen information
for armed actors. . . . In exchange for monitoring and information, local agents obtain a valued
immanent good: the power to rule over their communities” (Kalyvas 2006, 110).
20 Introduction: Civilian Autonomy in Civil War

The details are vague concerning where these committees arise, how they
collect information, the nature of the principal–agent relationship with armed
actors, and whether they are necessarily relegated to the role of “agents.”31
Kalyvas only argues that these committees’ effectiveness is determined by
territorial control – they persist only in areas of contested control through the
fear of “double defection” (with his few examples limited to completely con-
tested areas).32 The committees are rendered unnecessary in these zones
though, since armed groups are unable to use selective violence as they
cannot sufficiently protect informants to encourage denunciations. Implicit in
his discussion is that committees often cooperate (collaborate) with only one
side in the conflict, with little chance for enduring autonomy, neutrality, or
“fence-sitting” as conditions change (or what he calls “hedging,” or “double-
dealing”). In this book, I point to a greater diversity of civilian-combatant
relationships.
To summarize, although the field has examined why civilians join armed
groups and why they support them with manpower, resources, or information,
explanations for violence tend to omit civilians as autonomous actors. Their
organizational processes to avoid conflicts or ability to protect themselves are
discounted. They are generally assumed to be powerless and usually for good
reason: they are unarmed and are subject to coercion by often ruthless armed
combatants (Kalyvas 2006, Weinstein 2006). A realist perspective would there-
fore assert that civilian nonviolent tactics and even armed resistance are futile.
By describing civilians as “caught in the crossfire,” many accounts equate them
with helplessness and subject to various forms of suffering. If help comes from
the international community, it is often too little or too late, and sometimes

31
As one village committee’s behavior is explained, “The village’s solidaristic reputation was
endogenous to its relatively peaceful behavior during the war and that the fear of mutual
denunciations led the otherwise contentious villagers on a path to cooperation” (Kalyvas
2006, 294). However, this is an incomplete account of which villages can manage committees.
He asserts that, “There is no significant variation in local practices and institutions of factional
accommodation or types of factional and individual interaction in the villages of the Argolid”
(297). But, it is not clear how these are measured. His litigiousness measure could represent well-
functioning courts as well as discord and there are no other measures of institutions or inter-
actions within villages. While he notes “avoiding a vicious cycle was a key concern of local
leaders” (295), some of whom were “credited . . . for managing the village’s fortunes success-
fully” (296), the mechanism is simply described as “Diplomacy” (297). This relationship
between committees and conflict could also be unique to Greece’s social landscape.
32
In Kalyvas’s anecdotal cross-national examples of committees, the three zones of contested
control are blended together even though they have different predictions for violence and civilian
capabilities. While he acknowledges challenges in determining control (236) from secondary
sources, the committees are nevertheless lumped in zone 3, which could be inaccurate (242).
Further, in the examples of committees in Greece there is no explanation of variation in village
committees within zone 3. Eleven villages are coded with examples of mutual deterrence (293) of
at least 20 villages coded as existing in zone 3 in at least some time period.
Current Perspectives on Civilians and Civil Wars 21

even counterproductive (Luttwak 1999, Kuperman 2000).33 It is puzzling for

these theories that in some instances, civilians are not mechanistically part
of the landscape of battle, but rather are calculating agents that seek to foil
armed actors.

The Literature on Civilian Institutions

Related to the studies on subnational (dis)order is a growing body of works in
political science that has shifted attention from states and armed actors to
analyze “informal” institutions. These studies document subnational protec-
tion schemes and conflict resolution by civilians and civil society in the contexts
of riots and interethnic conflict, without recourse to the state. Scott (1985,
1992) describes the everyday resistance of peasants against their landlords
and elites in the form of “weapons of the weak” and “hidden transcripts” in
Malaysia. Fearon and Laitin (1996) describe how interethnic cooperation, not
conflict, is actually the norm and highlight in-group policing institutions that
limit interethnic spirals of violence. Varshney’s (2002) study of Indian cities
argues that enduring associations are sturdier than ad hoc relationships and
communication to deal with rumors and quell interethnic riots.34
These informal institutions for resolving disputes and providing order where
the state is absent (Ellickson 1991) play important and prevalent roles across
Latin America (Levitsky and Helmke 2006). This study is therefore certainly
not the first to pay close attention to the savviness of civilians. It departs from
these works, however, with its different scope of analyzing strategies in the
arguably more dangerous context of civil war settings, where communities may
face multiple armed groups and a different set of violence-related problems.

The Literature on “Peace Communities” and Civilian Autonomy

Diverging from the literature on civil war violence, a growing philosophical
and anthropological literature on peace movements has identified “peace
communities” as a local collective strategy to avoid and end conflicts (García
Durán 2005). Many of these strategies to avoid violence involve opting out of
the conflict by carving out a geographically defined area that is off limits to

33
However, this is not to say that interventions never happen or are ineffective. Kuperman shows
that even a small intervention during the Rwandan genocide could likely have protected
thousands of Tutsi civilians. Indeed, the small UN peacekeeping force in Kigali was able to
employ some strategies that were successful in protecting Tutsis threatened by the Interahamwe,
including harboring them in the city’s soccer stadiums (see the film Ghosts of Rwanda).
34
In constrast to Varshney, while “associations” may have explanatory power for understanding
riots, my theory does not predict such organizations are necessarily resilient enough to deal with
armed groups (though for some moderate forms of violence they can be). Indeed, I find coopera-
tives and ANUC land reform councils do not consistently reduce violence and instead posit even
more intense forms of cooperation may be required.
22 Introduction: Civilian Autonomy in Civil War

armed actor hostility (Hancock and Mitchell 2007; Anderson and Wallace
2012).35 These studies point to some apparently successful responses to vio-
lence and displacement, though with origins in anthropological and advocacy
traditions, research has tended to be either normative or descriptive. Ethno-
graphic and advocacy works on different communities include Nordstrom
(1997), Amnesty International (2000), Hernández Delgado (2004), Sanford
(2003, 2004), Bouvier (2006), and Hancock and Mitchell (2007), while Leder-
ach (1997, 2003) adds a religious and spiritual perspective on these movements.
Much of the discourse on these movements draws on the concept of
“peacebuilding,” which has origins in post-conflict rebuilding of communities
(Lederach 1997 and 2005, Pearce 1997, Bouvier 2009, USAID 2009). The
application of this approach to communities in the midst of conflict has empha-
sized education and promoting a “culture of peace” and the common Colom-
bian refrain of “convivencia,” or “coexistence,” as end goals. These goals are
noble but can also be vague, ostensibly suggesting a linear process that may
oversimplify how to arrest complex conflict dynamics.
It is thus a challenge to gauge how prevalent these civilian-led efforts
have actually been or how successful they have been at protecting civilians.
While human rights and peace activists and some scholars embrace civilian
efforts to “resist” and “create peace” in the face of adversity and hail such
bravery, scholars such as Kalyvas (2006) and Luttwak (1999) would argue
that any apparent effectiveness on the part of civilians is largely a result of
armed actor permissiveness and is epiphenomenal, or derivative of conflict
dynamics. No empirical research to date has attempted to rigorously adjudicate
this debate.
Part of the problem is that despite anecdotes of effectiveness research
remains undertheorized. There may be many local movements in the name
of “peace,” but it is not clear how such movements can be compared to
distinguish mere rhetoric from strategies and effects. This leads to concep-
tual questions about how to define a comparable set of organizations and
recognize autonomy organizations. Given these first-order issues, it is also not
surprising that the causal mechanisms civilians may use during conflicts, or
processes by which organized civilian resistance might affect substantively
interesting outcomes, have not been precisely specified.36 There has been an
incomplete consideration of the interests and motives of armed groups and how
civil society might affect their incentives for using violence. As a result, the

35
Sometimes also referred to as “Humanitarian Spaces,” these areas often pursue policies of
neutrality.
36
However, some NGO and International Organization (IO) programs have tried to consolidate
and standardize approaches (European Union, Redepaz, Constituent Assemblies, Middle Mag-
dalena Program for Peace and Development [PDPMM], Suippcol, United Nations Development
Programme, etc.).
Current Perspectives on Civilians and Civil Wars 23

causal “force” of civilians has not been made falsifiable or pitted against the
explanations of violence previously reviewed in this section that are rooted in
the macro-politics of armed actors.
Additional issues with the state of research on civilian autonomy organi-
zations stem from the body of cases that have been studied. First, while the
high-quality anthropological work on various cases offers many lessons, the
cases have not always been methodically or self-consciously chosen, leading
to possible selection biases (i.e., overlooking unorganized, highly violent, or
demolished communities). Second, many of these cases involve nonstandard-
ized geographic units of analysis, which complicates making comparisons and
measurements both across these cases and possible control cases of counter-
factual communities that did not organize for autonomy. Third, cases have
been used to build theories of behavior, but these theories have then rarely been
tested on out-of-sample cases with either qualitative or quantitative methods to
assess generalizability of findings (King et al. 1994). In sum, these works have
not risen to Kalyvas’s challenge that they are rarely independently or lastingly
effective.
In this book I aim to go beyond both the buoyancy of activists and
the vagueness of scholarly pessimism to analyze the mechanisms of civilian
autonomy processes and the evidence relating these processes to outcomes
of violence. I synthesize the best elements of the violence, institutions, and
peace literatures to supply theoretically grounded empirical answers on when
and how civilian autonomy is possible. The rich body of case studies on
“autonomy” inspired my shift in perspective toward civilian agency to argue
that other variables and actors matter in addition to “Weberian” state presence,
armed actor control, and wealth. From the informal institutions literature
I incorporate a theoretical approach for how civilians’ institutions solve prob-
lems in anarchy. From the civil war studies I import the causes of violence,
armed group behavior and preferences, and competitive hypothesis testing. Yet
I focus on the opposite collective action problem from that of rebel recruitment
– civilians seeking to limit violence and end civil wars.
In the end, this study is not a critique of the legal and normative field of
human rights scholarship but rather an extension toward analyzing causal
mechanisms and conditions for success. Going beyond the rhetoric is impor-
tant because political actors may not have incentives to follow normative
prescriptions and rules as talk can be cheap. Like studies of the empirical
effects of human rights treaties (Simmons 2009) and trade regime issue-linkage
(Hafner-Burton 2009) on compliance with human rights norms, I put sub-
national human rights mechanisms to realist scrutiny. By arguing that civilians
and their institutions can independently affect their own livelihoods in the
“crossfire,” but only under certain conditions, I embrace the tension between
collective action and opportunity structure approaches (Taylor 1988 and
Lichbach 1998). I conclude the viewpoints of human rights activists and more
structure-oriented civil war scholars are both partially correct.
24 Introduction: Civilian Autonomy in Civil War

research design and methods

The research pitfalls in the literature point to careful research design as essential
for credibly addressing the controversy of civilian autonomy and responding to
realist critiques of civilian strategies. As I elaborate in Chapter 4, I test my
theories across multiple units of analysis by integrating quantitative techniques
with fieldwork and qualitative analysis, bringing methodological structure and
rigor to the question of civilian agency. Tests are from Colombia and with
emphasis on the years from 1990 through 2005. This is an especially relevant
period since the conflict and civilian peace movements intensified during this
time, but the study is also closely grounded in the historical context of the
preceding years. The design accounts for alternative (structural) hypotheses of
violence to isolate the marginal protective effects of civilian organization.
This study has two quantitative components that fuse many disparate sub-
national datasets. I first use statistical analysis to test for a relationship between
different civilian organizations and violence across the standardized geograph-
ical unit of municipalities and examine the conditions where this relationship
holds. I then analyze the determinants of civilian organizations as proxied by
junta councils, including the prior violence of La Violencia as a causal factor.
This helps shed light on possible selection bias and circular relationships, and
suggests they are not great threats to valid inference.
In the case study component, I compare five “towns” from two different case
sites in different regions to include a diversity of experiences. The case analyses
reflect over 200 interviews with a wide variety of respondents, including ex-
combatants, and contain careful process tracing of history. A first case is the
formal organization from the introductory anecdote of the ATCC in the
department of Santander, which serves as a plausibility probe. I trace how the
ATCC’s investigatory institution deals with the problem of denunciations using
a within-case dataset and compare it with counterfactual unorganized neighbor
communities.
To select additional “out-of-sample” cases, I use statistical matching tech-
niques to settle on the rural municipios of Bituima, Vianí, and Quipile from the
department of Cundinamarca, not far from the capital of Bogotá. These cases
embody a quasi-experiment since they contain variation in their levels of junta
organizations but otherwise share similar characteristics. The qualitative
analysis depicts over sixty years of small-town Colombian history from La
Violencia through the present day. It showcases the texture of campesino life
as well as perspectives from ex-combatants on cross-community differences and
armed group decision-making.

plan of the book

This book is written for readers with general interests in civil conflict and
human rights. However, some chapters contain relatively more social science
Plan of the Book 25

detail about the analysis of civilian autonomy than others. Readers interested in
the general argument and cases of the book should pay most attention to
Chapters 2, 3, 7, 8, 9, and 10. Readers that are especially interested in the
analytical approach may additionally benefit from Chapters 4, 5, and 6.
Chapter 2 further develops civilian autonomy theory to explain the degree
that communities retain self-rule in the face of competition among armed
groups in civil wars. I elaborate on the three-part theory that links civilian
organizations to strategies and then to outcomes of violence, highlighting the
decision-making of both civilians and armed groups.
Chapter 3 provides context for the study by reviewing the historical pro-
cesses and events since the 1950s that shaped social capital and armed conflict
violence in Colombia. I explain the origins and politics of social movements for
peace such as the peace communities and review how the junta village councils
in Colombia came to be central forms of civilian organization in the context of
the conflict.
Chapter 4 describes the integrated, multimethod research design for the next
four empirical chapters and shows how such an approach can both be imple-
mented safely in potentially risky settings and help overcome various threats to
inference. I discuss how the large-n quantitative methods provide an overview
of the impact of community organizations on violence while qualitative case
and interview methods provide additional depth. I also describe my field
research process and preview the data sources I collected during fieldwork.
Chapter 5 contains a quantitative analysis of how civilian communities and
their organizations affect civil war violence. I measure the presence of these
organizations across Colombian municipalities with a unique dataset on the
local junta councils. When tested against extant explanations for violence
including the balance of military control and lootable natural resources, I find
these councils (as well as Indigenous and Afro-Colombian minority group
organizations) have salutary and significant effects on levels of violence.
In Chapter 6 I analyze why junta organizations emerged where they did
across Colombian towns. I address the potential concern that the influence of
the armed conflict on the councils themselves could invalidate the finding that
juntas reduce violence. Historical analysis and data from the La Violencia
conflict of the 1950s show these councils are in fact more likely to be found
in areas with social capital and that experienced past violence. I use statistical
matching procedures to help identify “matched” pairs of neighboring towns
with different levels of junta councils (“nesting” these cases within the larger
statistical analysis). These cases (from the department of Cundinamarca) are
analyzed in Chapter 8.
Chapter 7 analyzes the rural organization known as the Peasant Workers
Association of the Carare River (ATCC) to better understand whether and
how specific autonomy “mechanisms” function to protect civilians. The Carare
civilians constructed a local institutional process to investigate threats against
suspected armed group collaborators to clarify the “fog of war” and reform
26 Introduction: Civilian Autonomy in Civil War

civilian preferences to participate in the conflict. I analyze a unique within-case

database I created on threats and killings and find that the local institution itself
proved to be a critical factor for both explaining and limiting levels of violence.
Neither the creation of the institution nor its effects were merely the results of
the capabilities or choices of armed actors.
In Chapter 8, I qualitatively explore the approaches that communities
quietly innovated (or failed to innovate) to deal with armed conflict violence
in the small, impoverished Cundinamarca coffee-growing towns of Quipile and
its neighbors to the north, Vianí and Bituima. Though only a few hours from
Bogotá, these cases suffered from state absence and guerrilla pressure. The
stronger junta councils and levels of community cohesion and organization in
Bituima and Vianí relative to Quipile allowed their populations to better resist
and cope with the pressures from the FARC guerrillas (and from the army, and
later, paramilitaries). The findings highlight strengths and weaknesses of the
quantitative analysis and point to new hypotheses about how local politics and
clientelism can shape the capabilities of civilian organizations.
In Chapter 9, I explore additional instances of civilian autonomy from out-
of-sample cases from around the globe. I assess support for the theories I
developed in Colombia among additional out-of-sample communities of
Colombia’s contested FARC demilitarized zone of the Macarena region,
among Zones of Peace communities that have pursued autonomy in the Philip-
pines, among communities in Afghanistan and Pakistan seeking to avoid
counterinsurgent–Taliban crossfire, and among communities in Syria caught
between government and rebel forces. These cases show that the book’s main
argument is not culturally bounded and in fact supersedes cultural differences
as an explanation for violence.
The last chapter, Chapter 10, synthesizes the contributions of the book
for the study of human rights, conflict processes, and order in weak states.
I provide an overview of why, how, and when armed groups will change their
behavior in response to civilian actions or cross-community differences. I also
outline a research agenda for studying civilian behavior in conflict settings and
discuss policy implications for protecting human rights. Lastly, I reflect on the
normative implications of this research and how NGOs and external actors can
ensure they “do no harm” in working with embattled civilian populations.
 table 1.1 Cross-national examples of civilian autonomy in civil wars

Country Year(s) Armed or Number of Sources Brief Description Number

Unarmed? Localities of Armed
Actors
Afghanistan 2006 Unarmed 1–3 Gall and Wafa 2006; Some local towns, tribes peacefully Multiple
Suleman and Williams resisted Taliban (and NATO),
2003 Musa Qala district of Helmand
Province; Jaghori District of
Ghazni Province.
Afghanistan 2009 Armed Unknown Gall 2009 Villagers in some [Helmand] Multiple
districts took up arms against
foreign troops to protect their
homes or in anger after losing
relatives in airstrikes.
Afghanistan 2009 Armed 1 Gopal and Rosenberg A village in Nangahar in eastern 1
2009 Afghanistan rose up against the
Taliban, in response both to
Taliban violence and abuses as
well as an entreaty of
development resources from the
Afghan government.
Afghanistan 2009 Armed and 1 Jaffe 2009 U.S. entreaty to insurgents and 1
unarmed village elders in Kamdesh village,
Nurestan, to encourage them to
develop a plan to manage local
security affairs in the midst of a
withdrawal of U.S. forces from
region to more populated areas.
27

(continued)
28
table 1.1 (continued)

Country Year(s) Armed or Number of Sources Brief Description Number

Unarmed? Localities of Armed
Actors
Colombia 1980s Armed Multiple Houghton and Villa Quintín Lame Indigenous self- Multiple
2005 defense group in Cauca, other
sporadic local examples.
Congo-Uganda 2009 Armed 2+ Gettleman and Schmitt After Ugandan Lords Resistance 1
2009; Bavier 2009; Army (LRA) rebels were pushed
Gettleman 2009a into Congo, some terrorized
villages started self-defense
committees armed with shotguns
and slingshots to protect
themselves near Dungu. Also
wider movement in larger town
of Faradje, but also
displacement.
Greece 1944 Unarmed ~10 Kalyvas 2006, citing Local village committees vetoed Multiple
Frangoulis violence from both rebel and
German armies.
Guatemala 1982–1993 Unarmed 2–4 Stoll 1993; Falla 1994 Communities of Population in Multiple
Resistance (CPRs) of displaced
civilians remained in disputed
territories hiding from the
government.
Iraq 2006 Armed 20+ Al-Ansary and Adeeb Tribes in Anbar Province defended 1
2006 villages from Sunni insurgents,
requested government
armaments.
 Kenya 1960s Unarmed Unknown; Barnett and Njama 1966 Mau Mau local councils were said Multiple
Kiambu to veto violence.
Lithuania 1941–1952 Unarmed (?) Multiple Petersen 2001 Some towns tried to remain neutral Multiple
during World War II.
Mexico 1992 Unarmed 1+ Tavanti 2003 In Chiapas, Acteal/ Las Abejas Multiple
religious community’s peaceful
movement against state and
paramilitary oppression;
nonviolent pacificism separated
them from Zapatista National
Liberation Army (EZLN)
guerrillas. Nonviolence in
response to paramilitary
pressure and Acteal massacre.
Mozambique 1989–1993 Unarmed Multiple Wilson 1992 Jehovah’s Witness peace zones; Multiple
Naprama movement against
Renamo rebels.
Pakistan 2008 Armed Multiple Perlez and Shah 2008a; Lashkar militias resisted Taliban Multiple
Gettleman 2007 violence, especially in “lawless”
FATA region; U.S. considered an
“Anbar” tribal strategy for the
Pakistani tribal areas along the
Afghan border.
Pakistan 2008 Armed and 2+ Perlez and Shah 2008b Neutral/ peace zone in town of Multiple
unarmed Buner, near FATA, “The
villagers in Buner say they would
prefer to handle the Taliban on
their own, rather than have the
29

(continued)
30 table 1.1 (continued)

Country Year(s) Armed or Number of Sources Brief Description Number

Unarmed? Localities of Armed
Actors
heavy hand of the army come
and do it for them. . . . A new
peace committee composed of
elders and politicians passed a
resolution declaring Buner a
zone free of both the army and
the Taliban.” Also nonviolent
negotiations by councils in Landi
Kotal.
Pakistan 2009 Armed 1 Masood 2009; Khan Tribal leaders in Bannu, 1
2009 Sertelegram, and Kanju in Swat
who had armed against Taliban
were attacked and killed. Before
Taliban could kill survivors,
villagers came out with guns to
fight them off.
Pakistan 2009 Armed and 1 district, Tavernise and Ashraf 1,000+ villagers in the Dir District Multiple
unarmed many 2009 armed to fight off the Taliban
villages after they attacked a mosque
with suicide bombers, killing
thirty.
Peru 1982–1997 Armed ~4,000 Starn 1999; Fumerton Rondas Campesinas repelled 1
2001 Sendero Luminoso, coopted by
government.
 Peru 1982–1997 Unarmed Unknown Starn 1999; Fumerton Some Rondas Campesinas did not Multiple
2001 arm; they opted out of conflict in
some regions, resisted joining
government counterinsurgency
programs.
Philippines 1988–2004 Unarmed 91+ Coronel-Ferrer 2005; “Peace zones” were declared to Multiple
Santos 2005; peacefully resist first NPA and
Hancock and Mitchell government forces and then
2007 MNLF, MILF, and government
forces in Mindanao and other
regions.
Russia 2010 Unarmed 1 Greene 2010 Village leader in Sulak in Dagestan Multiple
dialogued with army and
Islamist insurgents to avoid
participating in conflict.
Rwanda 1994 Unarmed Multiple Doughty and Ntambara Highly cohesive Muslim 1
2005 communities and other
minorities resisted/did not
participate in genocide; tried to
protect themselves and Tutsis.
Sierra Leone 1990s Armed Multiple Muana 1997; Civilian Defense Forces (CDF) 1
Humphreys and arose in local villages to repel
Weinstein 2006a Revolutionary United Front
(RUF) rebels, later coopted by
the government.
Somalia 2008–2009 Armed Localities in Gettleman 2009b; Sufis around the town of Dusa 1
Dusa Gettleman 2009c; Marreb armed against
Marreb Raghavan 2010 incursions by the extremist
Islamic militant movement, the
Shabab.
31

(continued)
32
table 1.1 (continued)

Country Year(s) Armed or Number of Sources Brief Description Number

Unarmed? Localities of Armed
Actors
Sudan 2007 Unarmed Unknown Gettleman 2007 Some neutral towns mediated Multiple
conflicts between Arab tribes,
“The wali, or governor, of South
Darfur called a peace conference
and urged neutral tribes to
mediate a cease-fire.”
Sudan 2009 Armed Multiple Snapp 2010; Heaton “Arrow Boy” militias against LRA 1
and Fick 2010 incursions.
Zimbabwe 1970s Unarmed Unknown Kriger 1992 Civilians avoided aiding both Multiple
ZANU rebels and government
forces; some committees vetoed
violence.
2

A Theory of Civilian Decision-Making in Civil War

“Many people wrongly think it is [just] the Association that should come to the
defense of each individual, but it should be the opposite.”
– Resident of La India, Santander, 1995 (ATCC Archives)
“The guerrilla respects the thinking of the Association [ATCC]. There’s no reason
why it should disappear. On the contrary, it should be strengthened.”
– FARC Commander, 2001, near La India, Santander
(ATCC Archives)
“The mentality of the armed groups has changed a lot and we respect certain
things.”
– Paramilitary Subcommander, 2001, near La India, Santander
(ATCC Archives)

One Saturday morning in La India, an elder conciliator invited me to join him

at the Adventist mass. Since the church had a strong influence on the ATCC’s
nonviolent approach, I jumped at the chance. From the main street, we walked
a few minutes through a small wooded area and came upon a clearing and the
church. I watched as the congregation greeted each other, then we entered and
sat in a pew in the front row. As the service began, the people sang hymns and
read along from the prayerbook. I quickly lost my place, however, so I quietly
nudged my companion to ask him where we were. So as not to disturb the
songs, he whispered, “I don’t know.” Since he was following along and
skimming the pages with his finger, I quizzically asked, “What do you mean
you don’t know?” He looked over at me sheepishly, shrugged his shoulders,
and said, “I don’t read.” His reply was stunning, and was an epiphany for how
impressive the mobilization of the ATCC and other communities around
Colombia truly was. Indeed, how is it that ordinary people like this elder
conciliator are able to confront violence? The answer is a story of organization.

33
34 A Theory of Civilian Decision-Making in Civil War

Civil wars are not fought in social vacuums. They are fought in social
landscapes. These landscapes are often variable, with notable social differ-
ences from one town or village to the next. My central argument is that these
differences shape how civilians cope with civil war conflict and, in turn,
how they are treated by armed actors. In this chapter, I outline a theory of
when, how, and why variation in cross-community characteristics, such as
organization, affect outcomes for civilians in civil war settings.
Civilians tend to be viewed in limited ways, as either collaborators or victims
of armed groups. In contrast, I argue civilians have greater latitude as agents
and explain how differences in civilian organization and cooperation determine
their ability to implement protective strategies to retain autonomy, or self-rule.
I further explain why and how these strategies can be effective under condi-
tions that might normally invite violence, whether these are zones of contested
territorial control or confrontation with armed groups with weak disciplinary
structures. I show how these civilian strategies can affect armed group behavior
and discipline, and also explore their limits to do so.
I explore the full range of civilians’ available choices, but primarily focus on
collective strategies designed to retain autonomy and self-rule in the face of
competition among multiple armed groups in civil wars. The concept of autonomy
is adapted from the goal of communities that have declared themselves “peace
communities” or “resisting” communities. More specifically, autonomy means
maintaining democratic decision-making power over outcomes for the community
within the community, without influence from outside armed groups. I also
subsume under this concept the more tangible results of being able to keep the
community in place and resist forced displacement as well as mitigate violence
against residents. Self-rule can encompass decision-making abilities and the
realization of decisions. In the realm of justice, community autonomy means the
community (and not other actors) retains decision-making power over whether
individuals should live or be sanctioned or killed. I refer to “de facto” autonomy,
then, as the ability of civilians to shield themselves from the effects of external
actors and therefore see their own decisions implemented, enforced, and respected.
The drive for autonomy stems from the great uncertainty civilians face in
their daily lives in conditions of civil war. With only one life to live, one mistake
or one random act can spell doom. Under the peaceful conditions known to
most residents of the developed world, political order is sustained by a hege-
monic macro-political actor – usually the state – that guarantees security, often
in the form of a “social contract” where security is exchanged for support and
taxation (Tilly 1992). Order can break down and residents may face insecurity
when this actor is either repressive or abusive (e.g., Valentino 2004) or under
anarchy when other political actors compete for power. This absence of order
and rule of law most often occurs in “weak” or “failed” states (if the state were
strong, there would be no problem of civil war).
An option to make daily life more certain and increase chances of survival
is to turn to indigenous – meaning local – organizations. Since armed actor
The Advantages of Community Cohesion and Organization 35

coercion of civilians has been theorized to stem from divisions within civilian
societies (Kalyvas 2006), I argue that social cohesion among civilian commu-
nities affords them greater chances to keep armed groups out and implement
collective strategies to retain autonomy than acting individually. I develop a
general three-part civilian autonomy theory that links civilian organizations to
strategies and then to outcomes.
First, variation in the social and demographic landscape and technical
assistance from external actors (be they the government, church, or NGOs)
propel some communities to organize more easily than others. Second, cohe-
sive, organized communities can make collective decisions about how best to
deal with the dangers of civil war conflict. In addition to the standard strategies
of alignment with macro-actors these can include more subtle “tactical”
responses, or “weapons-of-the-weak” (Scott 1985), to deal with discord within
civilian communities as well as more forceful “weapons-of-the-not-so-weak”
tactics, such as overt protest and actions by nonviolent community guards.
Civilian preferences, social capital, and the (past and prospective) threat envir-
onment all interact to determine strategy selection. Third, the strength of
civilian organization, selected strategies, and armed group incentives jointly
determine substantive outcomes such as levels of violence.
The existing literature on micro-studies of civil war raises the possibility of
civilian agency but does not contain answers. Kalyvas suggests in his study of
selective violence that local committees can play an important role in limiting
violence against civilians but admits that, “[W]e know little about how they
actually operate. Perhaps their most important feature is that they often have a
role in determining what violence is visited on the locality in which they
operate, but how this power is wielded varies” (2006: 110; emphasis added).
Weinstein (2006) examines how rebel organization affects governance struc-
tures toward civilian populations, and argues that abusive groups invite resist-
ance or cause civilians to flee. But he says nothing about when these choices will
be made or how civilian “agency” occurs. As this body of scholarship stands, it
is like having a model of democratic participation without a theory of the
participants – the voters. Yet as Kalyvas (2006) has stated, civil wars are
“highly endogenous processes,” opening the possibility that armed actor
choices may partly result from civilian moves as well. In the rest of the chapter,
I explain how civilians come to cooperate, how they act collectively, and when
armed groups will be affected by their actions.

the advantages of community cohesion

and organization
To retain autonomy, I argue that civilians can act strategically and effectively,
even in the harsh conditions of civil war violence. But, for a hope of mounting
protective strategies, communities must first be able to cooperate. Lone civil-
ians, like lone soldiers, are unlikely to be effective in complex and changing
36 A Theory of Civilian Decision-Making in Civil War

environments and against military organizations. An individual may have

trouble, for instance, assessing the conditions of the conflict, committing to
not collaborating if most of his neighbors are collaborating, or credibly convin-
cing an armed group that he is not a collaborator. He or she may also face risks
for blowing the whistle about threats or acts of violence. In contrast, groups of
civilians are more likely to have the capacity to deal with complex and changing
environments. They have more information and options, longer time horizons,
and can pool risk (see Ober 2008 on epistemic communities of Ancient Greece).
In this section, I explain the sources of and variation in communities’ cohesion
and organizational potential.
From an organizational standpoint, communities face the risk that armed
groups may peel off individual community members from a pacifist cause by
offering selective incentives, either in the form of payments or the application of
violence to settle feuds. Peaceful residents of communities seeking to collectively
avoid or minimize the effects of the conflict must face the tyranny of the few
“extremists,” militants, or collaborators from the various sides in the conflict –
those who, by virtue of picking up weapons, exhibit a greater willingness to use
violence. So the question becomes: how do civilian organizations keep their
residents from giving in to the temptations to participate in the conflict and
fracture the community? This challenge faced by civilians is the opposite
collective action problem to that of rebel groups’ recruitment efforts. In sum,
it is the unity of civilian centrists that helps impede and isolate violent
“extremists.”

Challenges to Collective Action

When communities are threatened with violence, most residents are better off if
they can act to stop or reduce it.1 Individuals would generally prefer to protect
their communities and remain in their homes than choose displacement. Yet
there is a reason why organized responses to civil war violence are not ubiqui-
tous: there are considerable challenges to collective action given its risks, the
high costs of retaliation, and short windows of time for coordination.2 These
organizational pitfalls reflect a civilian’s dilemma that is similar to the chal-
lenges of collective action discussed by Olson (1965).
Most individuals want to stay put, but do not have strong incentives to
defend their communities on their own since they are unlikely to succeed and
armed actors might respond with further violence. Participation in community
defense may preclude safely fleeing to refuge, making hesitance on the part of
individuals quite rational. In these situations, fear can be both pervasive and

1
Some individuals may still benefit from or take advantage of the conflict for personal gain.
2
The challenge of rebel recruitment has been cast as a related collective action problem under fear
in civil wars.
The Advantages of Community Cohesion and Organization 37

devastating to cooperation. Under the “law of silence,” residents stop commu-

nicating out of the fear that they will be ratted out to an armed group for even
discussing the situation.
To act, individuals require certainty that they can count on their neighbors.
Despite fear, two types of communities are well positioned to act collectively
and implement strategies to avoid violence. First, communities of residents with
homogenous interests for remaining in the community to protect their liveli-
hoods will mobilize relatively easily (Olson’s “privileged” groups). These kinds
of communities may organize precisely because the residents all highly value
protection and avoiding displacement and because their similarities help them
agree when choosing collective strategies. Many residents of rural communities,
for instance, possess a keen sense of “belonging” (pertenencia), or identification
with their homes and communities, and cannot envision new livelihoods else-
where. Because of common interests, civilians will participate even with little
knowledge about the preferences of their neighbors.3
Second, social explanations of collective action as discussed by Marwell and
Oliver (1993) suggest even some communities with divergent interests are able
to organize. In these communities, although some residents may be risk averse
and require high certainty that a resistance movement will be successful, they
can be spurred to organize insofar as individuals’ choices are interdependent
and influenced by the strength of community leadership and by what other
members do.4 In this case, first movers with the strongest preferences are likely
to form a critical mass that reassures the rest of the community that the project
will succeed.5 The minority with strong preferences for defending the commu-
nity in the face of violence may be unable to successfully organize alone. But
they are likely to move first and successfully mobilize others when they either
(1) are able to persuade their neighbors or compensate them with selective
benefits so they join in or (2) believe their actions alone will inspire confidence
in their neighbors. Chwe (2001) argues that communication and the ability of
the community to generate “common knowledge” (or, in this case, a culture of
resistance) is crucial for reassuring more reluctant residents (see related models

3
Forced displacement can have a homogenizing effect on civilian preferences within a community
if, when civilians consider whether to return to their lands, less resolute residents are separated
from their more resolute neighbors.
4
This could reflect a prisoner’s dilemma, with individual displacement as a form of defection. It
may be converted into a cooperation game depending on “tipping points”: Civilian leaders would
hope to spark a pattern of participation whereas armed actors seek to cauterize participation and
deactivate civilian organizations to achieve displacement on the cheap. The last people to join in
the community effort may determine whether the effort succeeds or fails because their allegiance
can eliminate armed actor inroads into the community and provide information about the
community’s unity.
5
Applying Marwell and Oliver’s terminology, the public good of community defense has an
“accelerating” production function where “successive contributions generate progressively larger
payoffs” (63).
38 A Theory of Civilian Decision-Making in Civil War

by Kuran 1991 and Lohmann 1994).6 The creation of common knowledge and
even greater reassurance can be facilitated by social capital, network links, and
repeated interactions among individuals.

Social Capital and Pathways to Organization

The civilian’s dilemma points to social capital as a key explanation of cooper-
ation. This is because privileged groups with homogenous preferences for
collective action tend to be rare. The importance of civil society as a counter-
balance to formal (state) institutions has been noted by Bobbio (1988) and
Cohen and Arato (1994). The related concept of social capital – the vibrancy of
civil society – was originally explored at the local level by Putnam (1993), and
has been expanded and adapted to many different forms and contexts (Almond
and Verba 1963 studied the related concept of civic political culture; see Rubio
1997, Sudarsky 2007, and Kaplan and Nussio 2015 on social capital in
Colombia). The notion of strong horizontal relationships among residents is
particularly applicable to the context of rural communities in civil war. Social
ties help residents decide what is best for their community, to launch organiza-
tions that survive over time, and to directly affect levels of violence by prevent-
ing armed groups from seeding dissent among them. I later describe how
conflict-relevant civilian social capital in Colombia comes from three main
sources: existing natural bases for networking and reciprocity, preexisting
organizations, and technical assistance for cooperation provided by external
actors. Further, even acts of violence can promote coordination and the hom-
ogenization of preferences, depending on the severity and timing.
Certain demographic and geographic configurations foment strong social
ties and reciprocity within civilian communities better than others (Petersen
2001). Population size, population density, and network links among residents
including natural bases such as shared ethnicity or ties from economic comple-
mentarities may aid the implementation of collective strategies. A village’s
relative isolation may also promote close intracommunity social relationships
(Humphreys and Weinstein 2006a). For instance, in Colombia, colonizers and
homesteaders that migrated to “wild” areas (after experiences with armed
conflict) tend to have long-running reciprocal relationships and especially deep
commitments to stay on their lands (Legrand 1986). In less developed areas,
social interaction is also supported where favorable geography lessens costs
of intracommunity transportation, communication, and exchange through

6
Strong links between individuals may encourage participation, but weak links may speed up
information transmission (Chwe 2001). This information can be transmitted through rituals, the
strength of interpersonal relationships, a common spoken language, strong leaders, communi-
cation technology, and technical support from NGOs including church networks. Once an
organization has been established, institutional arrangements may be created to prevent reneging
on commitments to defend the community. These are likely to include positive inducements since
moral stances against violence often rule out coercion.
The Advantages of Community Cohesion and Organization 39

The former community store and cooperative of the ATCC came to be used as a meeting
space, La India, Santander, Colombia, 2007.
Photograph by Oliver Kaplan.

features such as rivers or footpaths. By contrast, if a community is close to large

cities or international borders, there may be more exit options and therefore
fewer close relationships, likely making local organizations weaker and dis-
placement less costly. Clear property rights and titles to land may also both
promote cooperation by preventing disputes between neighbors over property
boundaries and instilling stronger preferences to organize to avoid
displacement.
Prior experiences of cooperation among residents in other issue domains
may also translate into cooperation in the face of threats and violence when the
conflict comes to town. Civilians may rely upon interpersonal relationships
and authority structures of preexisting local organizations when assessing how
to deal with an armed group and how many individuals might participate
in a collective strategy. Examples of these kinds of organizations and experi-
ences include local junta public goods councils, committees formed to deal with
cattle rustling, indigenous tribes, agricultural cooperatives, or experiences with
political party mobilizations. With these foundations, some communities are
said to possess particularly strong cultures of organization and resistance
(in Colombia, such communities are informally described as “berracos,” or
“toughies”).
40 A Theory of Civilian Decision-Making in Civil War

External actors can also serve as focal points that help communities commu-
nicate and coordinate when facing armed groups. For instance, NGOs and
churches can support communities in their efforts to respond to conflicts by
providing them with “technical assistance” including ideas about how to
organize or what strategies are available. They can also provide accompani-
ment and a forum – a safe physical space – for meetings and communication.7
Finally, the threat environment may stimulate or impede the formation of
social capital and organization (what could be termed the “endogenous”
formation of social capital). An absence of violence may mean there is little
reason for increased civilian organizing. Strong interests among armed actors to
commit high levels of violence may preclude any civilian social capital or
resistance actions. Similarly, under intense conflict conditions, leaders of com-
munity organizations may also be directly targeted with repression in an effort
to weaken these structures. Moderate levels of violence may therefore most
stimulate intensified social relations and a collective response by civilians since
there is both enough reason (“demand”) to unify and tolerable risks for first
movers.8 Moderate violence may also increase the ability of civilians to act
collectively by producing sorting within civilian communities where less reso-
lute civilians leave the community. The result may be increased homogeneity of
preferences among the remaining residents to protect themselves.

The Added Value of Cooperation and Organizations

The essential feature of organizations for the theory of civilian autonomy is that
they embody intense, institutionalized forms of cooperation among individuals.
This helps civilians in two ways. First, cooperation can mean fewer divisions
within a community that armed groups can exploit and act as a mechanism of
protection. Second, as institutions that operate in conditions of subnational
rather than international anarchy, community organizations help civilians
engage in “repeat play” (Keohane 1984).9 Because organizing on the spot in
the midst of a conflict is possible but difficult, preexisting organizations that
have already solved collective action problems empower civilians in several
different ways.
First, well-functioning organizations can withstand the loss of any single indi-
vidual and live on. They therefore have the ability to maintain decision-making

7
NGOs that have provided such assistance include the Inter-Church Justice and Peace Commission
(Justicia y Paz), the Middle Magdalena Peace Program (PDPMM), and the Network of Peace
(Red de Paz).
8
Under such circumstances, civilian responses would be partly though not entirely endogenous to
the moves of armed actors.
9
They play a similar subnational role as international regimes, which are “Sets of implicit or
explicit principles, norms, rules and decision-making procedures around which actors’ expect-
ations converge” (Krasner 1982, 57). These frameworks define liability and decrease costs of
information and transactions.
The Advantages of Community Cohesion and Organization 41

procedures that endure over time and are not ad hoc. Second, organizations can
aggregate information from many people and places to understand security
conditions and bases of civilian support, and they have sufficient lead time to
deliberate and make decisions before armed groups arrive. Third, organizations
can tap collective memory and knowledge and develop best practices for dealing
with particular threats as they arise. Fourth, organizations have bureaucratic
capacity to implement collective strategies to maintain internal order within
communities. Fifth, organizations can act as figureheads for communities and
interface, advocate, and negotiate with external actors and other organizations.
Civilians’ initiatives to bargain or negotiate are frequently attributed to the
strength of a single popular leader instead of organizations. While some individ-
uals may be capable of undertaking effective nonviolent actions, such individ-
uals do not materialize out of thin air. They are usually produced by especially
cooperative and visionary communities. Community leaders are frequently
emboldened to advocate for their community’s interests when they have the
support of a broad section of their members (and leaders themselves may also
shape incentives of residents to support the common good). With broad support,
leaders’ positions are also likely to carry greater weight with armed groups.

A community planning meeting at the 25th (26th) anniversary celebration of the

founding of the ATCC, La India, Santander, Colombia, 2013.
Photograph by Oliver Kaplan.
42 A Theory of Civilian Decision-Making in Civil War

I argue that organizations with three main characteristics are the most
helpful for protecting communities. First, organizations will be effective if
they have decision-making experience, with clear lines of authority and proced-
ures for aggregating preferences and incorporating new information. Second,
broad and legitimate organizations should be more effective. Organizations
that are all-encompassing of a community’s population are more likely to be
recognized and respected as the community’s decision-making body by resi-
dents and as the community’s figurehead by outside actors. They may also be
inclined to act in moderation, as Kant’s liberal theory of “perpetual peace”
would expect (1970). Broad organizations also suffer from fewer “spoilers”
who might seek to undermine the community for selective benefits from armed
groups. Third, organizations that are “apolitical” relative to macro-actors have
greater chances for successful protection. Politically oriented organizations may
be polarizing or be more easily “stigmatized” by armed actors. They may be
seen as threatening for their politics or as tacitly supporting enemies and
therefore quickly become targeted (e.g., labor or land reform movements, or
local arms of political parties). By contrast, organizations that remain local,
represent local interests, and are cautious in their entanglements with larger
political movements are less likely to alienate armed actors.

civilians’ choices and pathways to autonomy

After organizing, what can civilians do to protect themselves? “Resistance” –
“resistencia!” – is a common mantra but it can be conceptually vague. What
does it mean? Resistance against whom, how, toward what goal? In what way
would resistance affect armed groups and provide protection? How can resist-
ance be evaluated empirically at the micro level? Going beyond the resistance
mantra requires a testable theory with clearly specified mechanisms.
From a realist perspective in which material power determines outcomes,
nonviolent tactics and even armed resistance implemented by organized civil-
ians should have little success against more powerful armed actors. This section
argues to the contrary that civilians can still have agency and avenues to affect
their livelihoods and perhaps the strategic situation in a civil war, even when
they are less (militarily) powerful than armed actors.
I first enumerate the general strategies of alignment available to civilian
communities in relating with armed actors and explain how these strategies
are selected.10 I then proceed to explore the special case of the strategy of opting
for “autonomy” from the armed groups in greater depth, including the tactics,
or specific mechanisms, civilians use to deal with threats. I also explain why and

10
These strategies may not reflect a community’s first preference and may instead be chosen out of
necessity (Kalyvas 2006, Mason 1996).
Civilians’ Choices and Pathways to Autonomy 43

when certain “bundles” or sets of tactics are likely to be chosen by different

communities and over different periods of time.

Civilian Selection of Strategic Alignments

Civilians facing the threat of violence in armed conflict are primarily concerned
with survival, although they would also like to protect their property, economic
prospects, standards of living, and social relations. A main choice civilians
make to balance these interests is how they position themselves (politically) in
relation to competing macro-level actors. Hirschman’s (1970) “voice-exit-loy-
alty” framework provides a productive way to think about the variety of
responses available to civilians. Organized civilians have a number of alignment
strategies from which to choose: do nothing or flee (exit); ally with a macro-
actor (loyalty); and autonomy (voice), or actively avoiding alignments, reducing
civilian participation in the conflict, and demanding accountability through
various armed and nonviolent mechanisms.11
Prior scholarship has suggested that civilians’ alignments are primarily
endogenous to, or a result of, armed actor preferences and pressures. I do not
dispute that armed actors’ choices and strategies greatly shape the options open
to civilians.12 It is only logical that armies are advantaged in their ability to
coerce civilian support through force. Nevertheless, I argue that existing
accounts and portrayals of civilians are incomplete, with important variation
in the selection of civilian strategies left to be explained by community-level
factors. Short of sudden attacks against the population, civilians may carefully
deliberate about their options before implementing responses to violence.13
Civilian cultural or political preferences can also influence alignment deci-
sions. Civilian preferences may tip a community toward (or against) allying
with ideologically close (or distant) macro-actors as balances of control shift.
For instance, communities with leftist or socialist preferences rooted in poverty
may be naturally inclined toward guerrillas and their ideological goals.14

11
Sometimes referred to as neutrality. Stoll (1993) as well as Kalyvas talk about allying with both
sides, or “double-dealing,” as a strategy.
12
This parallels how elites can structure available choices for voters in voting models.
13
Forward-looking armed groups would like to quash civilian threats to their hegemony. In
equilibrium there must be some uncertainty on the part of armed actors (rooted in civilians’
collective action problems) about how civilians might respond that conditions the oppressiveness
of their strategy to root out potentially hostile civilian movements before they arise. Or, armed
actors, in trying to organize communities or hold meetings with them to gain control, may “tip
their hands,” giving communities much-needed time to coordinate amongst themselves. Civilian
movements face uncertainty because they do not know what the armed actors do not know
about them and may miscalculate armed group preferences.
14
Despite facing dangers from government forces, the peasants in Wood’s (2003) account of the
civil war in El Salvador were naturally inclined to support the FMLN as a result of their
economic oppression, desire for justice, and contact with Liberation Theology.
44 A Theory of Civilian Decision-Making in Civil War

In contrast, other types of communities, such as those that are either isolated
or founded around ethnicity, may be accustomed to local political autonomy
and have stronger preferences for maintaining such a position even when
armed groups arrive and contest their territory.15 So, depending on the conflict
dynamics, civilians’ preferences may either speed or slow civilian alignments
with macro-actors.
Civilians’ most basic strategy is to “do nothing.” In other words, regardless
of organization, civilians have no collective strategy. While this choice is most
likely observed because of a failure to organize, even organized communities
may decide not to align or implement specific policies to maintain their inde-
pendence if these choices incite greater repression or if they are unfamiliar with
available tactics. Under threat and with no strategy, the displacement of civil-
ians may occur “drop by drop” (gota a gota) as residents flee one by one to
separate destinations. This displacement can further inhibit future attempts
to organize if there is no one left to count on, a reality not lost on preying
armed groups.
Aside from doing nothing, organized communities may be advantaged in
being able to coordinate a collective exodus from dangerous territory in a more
orderly process. Peasants in El Salvador developed the guinda system to prepare
temporary shelters and evade government troops (Todd 2010), similar to the
avoidance strategies of some communities in Guatemala (Falla 1994) and
Kenya (Barnett and Njama 1966). Families can be kept together as they arrive
in a safer location that is ready with supplies and assistance from neighbors,
NGOs, and/or the government, as the ATCC helped absorb and resettle resi-
dents who fled combat in outlying villages (CNRR 2011, 418). This collective
strategy may be used with an eye toward later facilitating a more rapid and
orderly return to the territory if and when conflict conditions improve, as the
communities along the Cacarica river in Chocó eventually did, forming the
CAVIDA network of humanitarian zones (CINEP 2003, Justicia y Paz 2003).
Unfortunately, resorting to the “do nothing,” individual displacement, or
collective displacement options has been all too common in the history of the
Colombian conflict as many people have been killed or displaced from their

15
Shared preferences for autonomy may arise from a variety of sources. First, residents that share
certain socioeconomic characteristics may naturally tip toward autonomy. For instance, recipi-
ents of land parcels may unite for common defense of their properties. Preferences may be
strongest among lower middle-class communities that are not so poor as to look to guerrillas to
promote their grievances or get involved in the coca economy nor so wealthy as to look to
paramilitaries for protection, as large cattle ranchers have (even though levels of organization
may vary within these groups). Second, culture may play a role. Residents’ pervasive religious
beliefs may make them more inclined to try nonviolent strategies and increase the likelihood
that residents share autonomy preferences (for instance Evangelicals and Seventh-Day Advent-
ists). Lastly, activist NGOs and religious institutions such as churches can plant the seeds
for organized resistance by providing technical assistance and acting as focal points for
coordination.
Civilians’ Choices and Pathways to Autonomy 45

homes (perhaps consistent with Colombia’s generally low levels of community

social capital as observed by Sudarsky 2007). Today, there are many ghost
towns in Colombia, as Sepúlveda Roldán (2004) illustrates in his portrayal of
the microcosm town of Saiza in Córdoba.
Civilians may also ally with an armed actor for protection from that actor as
well as from incursions by a nearby enemy army. This is often a tacit bargain or
social contract where support in the form of material provisions or information
is exchanged for protection and the maintenance of public order.16 Where the
presence of the army is strong, most communities side with the government for
protection from insurgent groups. Armed groups also often force peasants into
the “bargain” of paying protection taxes (vacunas or vaccination payments) in
exchange for “protection.” In an illustration from Putumayo, Colombia, the
cocalero movement allied with guerrillas for a time to resist government coun-
terinsurgency pressure and aerial pesticide spraying to continue producing their
coca crops (Ramírez 2001). This alliance strategy can risk future retaliation if
the balance of control later shifts and a new armed actor becomes dominant.
The final alignment strategy communities may select is autonomy from all
armed actors in their region, or self-rule.17 According to Sandoval (2004,
translated from the Spanish), “Autonomy is that which is determined from
within . . . one cannot conceive of peace with justice as a paternalistic gift of two
powerful actors that supposedly agree . . . Autonomy is independence when
facing armed actors” (187). The motive for pursuing autonomy is frequently to
gain independence from the “alliance” bargain described earlier to avoid
complications with enemy groups who would punish real or perceived defec-
tions. Autonomy strategies may therefore be especially common after shifts in
the military balance of control, when civilians are signaled and stigmatized as
supporters of the formerly dominant power (and when that power also remains
close enough to exact revenge). As such, autonomy strategies often involve
formal declarations of centrist political positioning for increased credibility.18
However, while the term “autonomy” has been frequently adopted by commu-
nities and scholars alike, what it means for influencing behaviors has received
scant critical and theoretical attention.

16
This strategy and its incarnations are discussed at length in Mason (1996) and Kalyvas (2006).
17
The language and rhetoric of autonomy can present some confusion. Autonomy is claimed by
many communities because it is softer and less threatening than claiming “neutrality,” which is
more positional and less about civilian decision-making. This is why formal declarations of
autonomy can reflect the implementation of autonomy policies and tactics although the concep-
tual fit may be loose: groups that do not use the language of autonomy may implement its
mechanisms and groups that use the rhetoric may see autonomy as a goal but not implement
many policies to that end.
18
Declarations themselves may be considered a tactic. Declarations of autonomy can be antagon-
izing to armed groups if they fear that other communities view the toleration as a reason to “hold
out” too. However, declarations may serve as a useful tool to gain outside attention or clarify
which community members are “in” and which are “out.”
46 A Theory of Civilian Decision-Making in Civil War

Autonomy in conflict settings has often been seen as synonymous with

“neutrality” or nonalignment. The difference may be semantic, but for concep-
tual clarity I see autonomy as a broader concept since it does not require
positional decisions and can be an ideal that is sought even in the face of a
single abusive group (whereas neutrality is often a relational concept vis-à-vis
other actors). Autonomy can be conceived of as independence in decision-
making and the freedom from violence required to sustain it.
Autonomy is a general strategy as well as an end goal that a community
may try to reach through various “autonomy strategies” or tactics. Autonomy
can include more subtle, internally oriented, community-building strategies
(weapons of the weak) and other cost-imposing strategies on armed groups
(weapons of the not-so-weak) that may not be contingent on being neutral or
declaring neutrality. In other words, neutrality is only one of the various tactics
a community may select to attain autonomy, and there are also strategies to
seek autonomy that do not involve neutral positioning. In the next section, I
explore more deeply the special case of seeking autonomy and the various
tactics employed to maintain autonomy.

Civilians’ Selection of Bundles of Tactics to Reduce Violence

and Maintain Autonomy
Autonomy-seeking strategies are usually adopted by civilians as a response to
multiple armed groups, when each group is worried about losing civilian
support to the enemy and willing to use coercion to halt the erosion. When
multiple armed groups “signal” and stigmatize civilians as collaborators, com-
munities may seek to credibly commit to not having allegiances to enemy
groups and incentivize armies to desist from violence.19 Communities may act
collectively to demand greater accountability when facing a single aggressive
armed group as well.
In some rare cases, communities may be relatively cohesive and well armed,
making deterrence through force an option. In other organized but unarmed
communities, a number of other nonviolent mechanisms can be implemented to
reduce levels of violence. Some are based around loose, spontaneous coordin-
ation of actions and preferences while others are implemented and sustained

19
The “fog of war” and the inability of civilians to commit to not take sides and collectively deter
armed actors are the main pathways to the coercive violence perpetrated against them. These
problems stem from two different principal-agent monitoring problems when armed actors are
working to consolidate control of a given territory. One is at the group level between civilian
groups and either the rebels or the government (or paramilitaries). Without control of an area,
armed actors cannot perfectly monitor what “side” a group of civilians is on – whether certain
residents of a community are aiding their enemies or if civilian leaders are working to stop them.
The other is among civilian group leaders, who may not be able to perfectly monitor and limit
the defections of their constituents to exit the community or aid armed groups.
Civilians’ Choices and Pathways to Autonomy 47

through deeper, ongoing forms of cooperation. Some actions are taken in direct
response to specific incursions by armed groups: a displaced community may
return to its lands, an indigenous group may confront and protest against
armed group kidnappers en masse to rescue victims. Other actions are more
of a daily, routine variety – social unity, managing information about conflict
conditions, dissuasion of youth from participating in armed groups. For either
set of tactics, civilian cooperation is a catalyst and the aim is to manage the
costs and benefits to armed groups of using violence. The tactics work by
dealing with social divisions to make using violence more difficult, providing
armed groups with benefits so they are more judicious in their uses of violence,
or directly imposing costs on armed groups for using violence.
What follows is a description of the logic behind six of the most popular
tactics I identified through site visits, although other tactics may exist that are
not yet publicized. I argue that communities mix and match these tactics to
select bundles of policies for their specific needs. The tactics are additive
(though not necessarily linearly) and increasing in their contentiousness and
their probable effectiveness. I then theorize about the decision process for
selecting various tactics, which involves similar considerations to those already
mentioned for the selection of the broader alignment strategies.

The principles of the Peace Community of San José de Apartadó, Antioquia, Colombia.
Photograph by Oliver Kaplan.
48 A Theory of Civilian Decision-Making in Civil War

First, civilian organizations can promote what has been called a “culture of
peace.” More concretely, they can attempt policies to change the distribution
of preferences among (and incentives that shape) the civilian population of
a community from belligerence to pacifism, making civilians less easily sedu-
cible by armed actors. Armed groups often have incentives to peel off civilian
allies to gain informational or logistical advantages over their opponents.
Unfortunately, when this same pressure exists from multiple groups, civilians
become fearful of being implicated and killings can spawn desires for revenge.
Trust among civilian residents can break down, begetting even more killings.
“Preference-changing,” culture-of-peace policies can lead fewer civilians to
collaborate with armed actors and, when implemented in an even-handed
way,20 can reduce the incidence of “valid” threats and violence against actual
civilian enemy collaborators. Cultures of peace can also limit groups’ available
manpower by impeding recruitment.
Preference-changing, culture-of-peace policies can be implemented in at least
two ways. First, leaders can collectively promote ideational norms among
residents to influence their preferences and persuade them against participating
in the conflict – against seeking selective benefits from aiding armed actors. This
can involve appeals to the inherent value of nonviolence or be rooted in religion
or the good of the community. Appeals may also be made to logic and history.
For instance, educational programs about the dire consequences of spirals of
violence sparked by defections can be enough to help community members
coordinate on nonparticipation for fear of personal harm. Second, civilian
leaders can manipulate the levers of social incentives – pressures and rewards –
to convince residents to forgo the possible gains of selective benefits armed
groups. Communities with elders (e.g., Indigenous elders) or leaders who are
especially revered may have the most power to award special responsibilities,
promote social acceptance, or threaten potentially devious residents with social
ostracism. Such measures along with paths to leadership and recreational
activities can be especially important to persuade youths not to join the ranks
of armed groups.
If successful, these coordinated preference-changing efforts can yield an
added benefit: reassurance among residents in the form of common knowledge
that mitigates possible intracommunal security dilemmas. Similar to the collect-
ive action process described earlier, an individual’s choice to desist from par-
ticipating in the conflict is linked to the choices of neighbors. Individuals do
not want to be the only ones “losing out” if everyone else is dipping into the
benefits provided by armed groups (or alternatively may want protection if
everyone else is denouncing and has an armed actor patron). If civilians see that
they are not losing protective advantages or selective benefits relative to their
peers they will be less interested in collaborating and denouncing.

20
I.e., not biased toward or against any macro-actor.
Civilians’ Choices and Pathways to Autonomy 49

In theory, a culture of peace is minimally “contentious” in that it imposes

few direct costs on armed actors and so should run relatively low risks of
retaliation. In fact, armed actors may not actually be aware that a community
has an intentional policy of this nature. They may only find that it is harder to
find good informants or material aid. Though a positive step toward peace, this
mechanism’s impact is limited without other tactics that clarify information for
armed actors about residents’ behavior because people could still be falsely
accused of collaboration and then violently sanctioned.
Second, strong communities can implement local conflict resolution pro-
cesses to minimize the degree to which civilians go to outside actors to police
problems of local order. In areas of weak state presence where judicial
courts are largely absent, civilians may bring disputes between neighbors over
issues such as property boundaries, debts, or livestock trampling crops to
armed actors to gain an advantage through coercion or violence (such disputes
can occur even when social capital is high). The acts of delinquent youths
or common criminals can similarly balloon into larger problems without a
system of justice. These conflicts give armed actors an excuse to get involved
in a community’s affairs and they often exploit these divisions to gain infor-
mation about and purge their suspected enemy collaborators, even though
accusations by locals are often false and for personal gain (e.g., Kalyvas
2006). If the community can implement procedures to successfully resolve
disputes before they reach armed actors, they can short-circuit cycles of
informants and killings.
Third, communities can establish local investigatory institutions that lever-
age civilian advantages in local information to clarify accusations by armed
actors against suspected enemy collaborators. For a variety of reasons, civilians
may be accused of participation in the conflict and threatened because of the
“fog of war.” For instance, armed groups may misinterpret observed activities
of civilians as providing material aid to their enemies or they may receive false
denunciations for the reasons mentioned in the preceding paragraph. By
“vouching” for falsely accused suspects but not confirmed collaborators, a
civilian transparency process can reduce indiscriminate violence. This is espe-
cially true if it is costly for armed actors to monitor and police civilians and
their reputations or popular support suffer for killing wrongly accused enemy
collaborators (see next section and Chapter 7).21 Transparency about civilian
actions can also de-incentivize participation in the conflict if it increases the
probability that civilian collaboration with an armed group will be identi-
fied and punished. Investigatory procedures are usually established through

21
I resist using the term “innocent” to describe individuals in this situation, since the inverse
(“guilty”) can disregard the contextual risks that war creates. It suggests that other victims are
somehow culpable and that targeting them may be justifiable, when it is instead preferable to
resolve such cases without resorting to violence. However, some communities use this shorthand
in their everyday discourse when referring to rights violations.
50 A Theory of Civilian Decision-Making in Civil War

dialogues with armed groups and convincing them of the benefits of allowing
civilians to bear the costs of (credibly) policing their own communities.22
Fourth, communities can protest and “go public” to denounce aggression
and abuses and shame armed actors. This tactic relies on the dissemination
of messages to affect armed groups’ reputations not only in the eyes of the
government and international actors (Keck and Sikkink 1998) such as the
United States, but also in the eyes of other civilians and communities who
may be influenced about whether or not to support a particular actor (alluded
to in prior works such as Petersen 2001 and Weinstein 2006; see Hafner-
Burton 2008 on the mixed effects of NGO shaming of governments for human
rights abuses). Communities may engage in marches or other symbolic acts
and link with external NGOs and IGOs to help magnify the wrongdoings to a
wider audience.23 One tactic is to call out armed groups on their contradictions
and hypocrisies. For instance, armed groups that at least marginally value
ideology (political, religious, or other) and claim to be the “defenders of the
people” among their goals may be susceptible to communities’ use of rheto-
rical traps to impugn the group’s legitimacy for acts of violence (these rhetorical
traps may additionally serve as a form of moral persuasion; see Guerra Curvelo
2004).24
The threat of publicly protesting an armed group’s misdeeds may also be
used so that protests stay “off the path of play.” The option of protesting may
provide increased leverage in quieter forms of bargaining over protecting
civilians with would-be aggressors by, for example, allowing the group(s) to
save face.25 The choice to actually “go public,” then, may be made under
bargaining failures after more conciliatory overtures toward armed groups
are not reciprocated. Protest can bring important gains in protection to a
community but can be highly contentious and confrontational toward armed
groups. Since protest may anger armed actors and provoke greater repression if
it fails, this strategy requires high degrees of cohesion and commitment among
the civilian population.
A variant of protesting seen in Colombia as well as some other countries
is the declarations of neutrality that set a community’s territory off limits to

22
Dialogues can also mitigate problems of private information about capabilities, increase trust,
and possibly humanize the civilian community in the eyes of armed groups.
23
Churches and NGOs such as Peace Brigades and Fellowship of Reconciliation have engaged in a
practice known as “accompaniment” to maintain enduring observer presence in threatened
communities. They can then immediately report on threats or rights violations to media outlets,
foreign governments, and international organizations and raise the costs of attacking a commu-
nity by introducing the possibility that a foreigner or city-dweller might be harmed. See Mahony
and Eguren 1997.
24
Although verifying the impact of this mechanism requires further study, Ball (1998) finds
support in his study of El Salvador and argues it should not be underestimated.
25
Using the reputational lever may also affect the internal politics of armed groups by empowering
more restrained factions over other factions.
Civilians’ Choices and Pathways to Autonomy 51

armed groups (a “peace zone”). When balances of power are unstable or may
shift, demarcating territory aims to avoid retribution by shunning entangle-
ments with an out-group. This can also help attract broader attention from
outside the community. The act can furthermore be seen as an attempt to
proclaim the unity of the civilians and signal that if the group tries to divide
or coerce the population it will fail.
An additional protest variant is the use of massive nonviolent manpower to
force armed groups to concede to specific actions. Communities may use the
sheer number of unarmed people to confront and overpower a few armed
combatants. They may persuade combatants to concede through moral
appeals. If the combatants instead prefer to fight, they must decide whether to
commit a massacre. The protestors may even be able to physically stop the
troops if they try to attack. On several occasions, the Nasa Indigenous Guard,
consisting of around 500 unarmed community members who carry symbolic
staffs, have appeared at guerrillas’ and kidnappers’ hideouts to pressure these
groups to release residents who were taken captive (Wirpsa 2009).
Fifth, organized civilians may devise early warning systems to help civilians
avoid being caught in the crossfire of combat between armed groups. Infor-
mation and emergency procedures may be set up within the community or
through dialogues with combatants to either convince them to fight outside the
boundaries of civilian settlements or temporarily displace or shelter residents
until fighting has subsided.
Sixth, communities may attempt local-based armed resistance against exter-
nal armed groups to protect residents. This strategy employs a logic of deter-
rence to directly impose (or threaten to impose) military costs on armed groups
for abuses they may commit. This strategy has been implemented in several
example communities both within Colombia and abroad, but many commu-
nities across many conflicts do not arm because it entails great risks. Creating a
local defense force requires high degrees of cohesion and coordination, not to
mention manpower (though civilians may be advantaged in their local know-
ledge of the social and physical landscape). Many rural civilians are farmers,
not soldiers, and have little military experience.26 Furthermore, resistance with
arms can forfeit a community’s legitimate claim to being noncombatants.
Lastly, if civilians miscalculate their capacity relative to existing armed groups,
arming – a form of hostility toward armed groups – may lead to harsher
retributions and unacceptable levels of casualties. Since civilians reside in
stationary villages, they can become sitting ducks if they are outgunned.
The bundles or combinations of tactics that communities might employ can
be ordered according to their differing degrees of costliness, aggressiveness, or
“contentiousness” toward armed actors (the concept of contentiousness is

26
The choice to arm may additionally depend on the availability of weapons and balance of
firepower held by the armed groups, whether there is a tradition of armed self-defense in the
community, and even possibly on gender balances and the role of women in the community.
52 A Theory of Civilian Decision-Making in Civil War

discussed by Tarrow 1994 and McAdam, Tarrow, and Tilly 2001 and applied
by García Durán 2006a to Colombia).27 I argue that civilians look both
internally within their communities and organizations and externally toward
their environments when selecting strategies, reflecting the collective action-
opportunity structure framework (CARP-SPOT) elaborated by Lichbach (1998).
Civilian organizations select tactics of differing levels of “contentiousness” after
consideration of both the organizational potential they have for implementing
them and the risks they face from armed groups should they mount conten-
tious, cost-imposing strategies that fail. The choices can be conceived of as
falling along a continuum of stylized ideal strategies, where civilians calculate
whether to “go public” or pursue less contentious “backroom”-style negoti-
ations and appeasements.28
In addition to any calculations about costs and benefits of various strategies,
the repertoire of strategies available for any given community will in many
cases be determined by the epistemic constraint of the availability of ideas
about different strategies. Communities can only choose and implement strat-
egies they know of and about which they have information to assess prospects
for success. The tactical repertoire may therefore be shaped by brainstorming
and creativity among community members themselves, or may come from
knowledge about which strategies have been used in other places, what their
requirements are, and how they have fared. Indeed, I am only able to list here
the general strategies that I have been able to identify during my research and,
naturally, other effective strategies could be improvised in the future. The choice
of a bundle of multiple tactics may be further constrained if some strategies are
mutually exclusive. For example, if a community decides to arm against outside
groups, this precludes a philosophy of nonviolence and attempting to convince
armed groups of wishes to be left alone as nonparticipants.
Optimally, communities choose the strongest bundle of policies they can
manage to implement for the greatest possible amount of protection without
provoking a more violent response (i.e., where the marginal benefit of the
decline in violence of the next incremental tactic equals zero). In other words,
they decide based on an assessment of expected utility – the probability of
success and the expected benefits if they are successful relative to alternative
outcomes. Although I argue that a calculation process underlies decision-
making, some communities may be predisposed to naturally prefer particular
tactical approaches because of risk aversion or risk tolerance and cultural
factors including ideas about nonviolence or the inherent value of resistance
(what Wood 2003 calls the “pleasure of agency”).

27
This presumes that more contentious strategies are generally more effective at reducing human
rights abuses or the odds of displacement, conditional on a given level of armed actor threat. The
sensitivity to costs imposed by civilians’ strategies may vary from the government forces to illegal
armed groups.
28
Kernell (1986) develops a similar decision framework in the realm of presidential politics.
Civilians’ Choices and Pathways to Autonomy 53

table 2.1 Examples of civilian organizational strategies and tactics

by contentiousness

(by increasing levels of hostility)

Contentiousness Strategy set Examples

of Civilian
Strategy
Low Displace (individually) San Francisco and Angelópolis
(Antioquia), Saiza (Córdoba);
many others
Ally Peasant Soldiers and Informant
Network programs, coca
growers in Colombia, Rondas
Campesinas in Peru, FMLN
supporters in El Salvador
Displace (organized/ Guindas in El Salvador, CPRs in
collectively) Guatemala, ATCC, Cacarica
Early warning systems COCOMACIA, ATCC
Peace culture, norms of ACVC, ASPROBRAS,
nonparticipation in ASOPROA
conflict
Declare autonomy, negotiate ATCC, some Indigenous groups,
with armed groups ACIA (Chocó Afro-Colombians)
Declare autonomy, bar Peace communities of San José de
passage of armed groups Apartadó, San Francisco de Asís,
in territory, shaming Cacarica; COCOMACIA
Declare autonomy, direct Nasa Indigenous Guard
unarmed confrontation
and protest
High Armed resistance Quintín Lame Armed Movement,
Rondas Campesinas (Peru), CDF
(Sierra Leone)

Different strategy bundles are shown in Table 2.1 along a continuum of their
“contentiousness” toward armed actors in terms of the costs they impose
and the benefits they provide. The community examples provided are only
roughly classified based on qualitative knowledge of the types of strategies they
have chosen.
More subtle tactics are commonly attempted first. If they should fail, com-
munities may then vociferously protest or arm as a last resort, when they feel no
other options are available. According to this theory, a community will select
the most contentious tactic bundles when highly organized, when there is
external support, and when there is moderate armed group pressure. Civilians
may also be more inclined to try contentious actions when there is a relative
54 A Theory of Civilian Decision-Making in Civil War

mismatch in capabilities that favor the civilians relative to armed groups

(perhaps if armed groups are scattered) and they have high confidence of
success. By contrast, when armed group pressure is extremely high, contentious
autonomy actions may be attempted but will be difficult to sustain or may lead
to retaliation, and quieter strategies may be more prudent and prevalent. By
their nature, some of these strategies are more visible to the rest of society (and
therefore social science researchers) than others.
Some tactics are also more effective for responding to particular types
of violence than others. Institutional solutions for information management
and intra-communal disputes may work best for problems with denunciations
to armed groups. Advocacy and early warning systems may work best for
extricating a community from the crossfire of combat. By contrast, these
mechanisms may be less relevant for cases of abusive behavior, where armed
groups may instead primarily be influenced by their legitimacy being put in
jeopardy or by more aggressive responses. In the next section, I outline the
conditions under which armed groups are influenced by these strategies.

explaining violence: conditions for

de facto autonomy
This section explains how civilians’ strategies interact with the conflict
environment and armed groups’ strategies to determine outcomes of interest
such as violence. When civilians’ strategies are successful at mitigating violence,
I refer to it as “de facto” autonomy. In addition to avoidance strategies, I
outline the scope conditions for civilians’ ability to affect the behavior of armed
groups and militaries with strategies that address specific causes of violence.
Armed actors are brought back into the discussion by analyzing the variety of
preferences they hold under differing conditions. This leads to several observ-
able implications for empirical testing about when violence is likely.
I argue that for armed groups to be influenced by organized civilian commu-
nities, they must both have incentives to use violence (or civilian resistance
would be moot) and be sufficiently flexible in their motives for violence to be
open to civilian influence.29 I call this a group’s sensitivity to civilians. Existing
theories provide a helpful starting place for the discussion of sensitivity,
although taken individually, they do not provide a complete depiction of when
armed groups might respond to or be impeded by civilians.
I cull three central factors from existing theories about how armed actors
fight for their survival in the face of threats of external pressure and internal
disintegration. First, a group’s founding institutions, ideology, and norms of

29
For instance, armed actors may exhibit a middle range of preferences where they would still
derive benefit from killing civilians but could also be persuaded not to. If they choose not to kill,
they may bear only a small net loss if they can be assured they are not losing civilian support.
Explaining Violence: Conditions for De Facto Autonomy 55

behavior not only serve as recruitment planks but also shape its attitudes
towards violence and the place of civilians (internal group characteristics).
Second, the relative dependence of an armed group on civilians with respect
to its resource base will determine both its discipline and concern for its
reputation (in interactions with other actors). Third, a group’s time horizons
and security situation, which are closely related to territorial control, will affect
its tendency to cooperate for the gains of repeated play with a community
(situational factors).
These three factors can combine and interact to produce hybrid groups and
conflict conditions with various sensitivities to civilians. For armed groups to be
responsive, they cannot solely be interested in maximizing expropriations from
civilians or violence as an end in itself. Groups of this nature will mostly be
influenced when they are met with force. However, this does not mean that less
extreme groups with moderate preferences over these pursuits will necessarily
have few motives for using violence. These factors can further interact with
each other. For instance, a group’s ideology may determine its resource usage,
and resources may degrade group norms, and both of these may determine how
much groups want to expand their territorial reach or contest control.30 After
first reviewing each of these factors, I illustrate how this argument applies to
several archetypal armed groups in the civil war literature by highlighting their
interests and expected sensitivities to civilians.

Internal Armed Group Characteristics

Armed groups vary in their governing institutions (Arjona 2014), and an armed
group’s susceptibility to civilians can vary based on its ideological beginnings
(Weinstein 2006) and its institutions. A group’s early goals and attitudes
toward violence may shape its violent behavior – for instance, whether armed
groups pursue genocide and abuse or place weight on protecting populations.
Members of some groups may come from civilian communities and retain
familial ties and loyalties to these communities, even if they become more
opportunistic and involved in illicit economies. Group norms may guide the
discipline and restraint of members and determine how much weight they give
to civilians’ moral arguments about the use of violence.
These base dispositions and ideologies toward using violence can be sticky
and persist over time, even after resource or conflict-event “shocks.” Group
members’ particular beliefs about violence may endure in the group to reinforce
and revive them at different points in time. Group norms may also shift as
the composition of a group changes. This is increasingly seen as conflicts

30
For most groups, these traits are not as static as they may commonly be viewed. They can change
over time and the salience of different concerns can be activated based on new considerations
or strategic situations. Groups may also not be monolithic and can simultaneously consist of
ideologues, opportunists, abusive criminal elements, etc.
56 A Theory of Civilian Decision-Making in Civil War

internationalize and foreign fighters that hold other ideologies and norms mix
with local fighters with closer ties to communities (Bakke 2014). In some
instances, civilians can exploit these divides to “nudge” armed groups toward
rights-respecting norms (Kaplan 2013b).

Resource Bases and Dependence on Civilians

The mix of an armed group’s resources can affect both its concern for
legitimacy and its discipline and ability to improve its behavior. Civilians can
provide essential goods to armed groups (relative to their opponents) including
food, water, cellular phones, etc. Groups may also use civilians as sources of
information and manpower, for instance as recruits or labor for extractive
resource production. Beyond civilian inputs, groups can support themselves
with resources from external backers, diasporas, cash crops (coca, poppy),
natural resources (oil, diamonds, etc.), kidnapping, and extortion. A greater
balance of voluntary civilian inputs means groups depend more on civilians for
their survival. This can produce a greater concern for their legitimacy and
reputation within a single community or among nearby communities that
may be watching. Even resource-intensive groups may want to avoid bad
reputations if notoriety draws outside intervention or creates political problems
for external patrons.
Groups that are resource-poor tend to emphasize ideology over selective
incentives in their recruitment strategies. Such groups will therefore attract
fewer opportunist recruits who might be rewarded through pillaging and will
develop stronger disciplinary structures, although some bad apples may remain
(Weinstein 2006). Some of these groups may not be averse to policing abusive
troops but do not because they lack the minimum incentives or necessary
organizational structures and capacity. In some cases, civilian pressure can
provide these incentives for reform.

Situational Factors
Sensitivity may also be shaped by the conflict settings armed groups find
themselves in and their preferences for controlling or conquering territory.
Kalyvas (2006) argues groups have incentives to use violence when they are
contesting control of territory (also referred to as the control-collaboration
theory). The intensity of conflict, a group’s level of security, and pressure from
opponents may shape how patient armed groups are in waiting for the gains
of repeated interactions with civilians versus taking the immediate gains of
targeting suspected enemy supporters or extorting communities. According to
ancient Chinese philosopher Sun Tzu (2005), soldiers in hostile territory –
“desperate” or “death” ground – will fight especially hard. Groups in such
situations or those that place high value on victory may be prone to coerce
civilians. But groups may have varying preferences over attaining victory and
Explaining Violence: Conditions for De Facto Autonomy 57

may care relatively more about not losing what ground they hold and avoiding
losing out to enemy groups.31
Armed groups may have incentives to use selective violence, but also may
place less value on the gains they would obtain through the use of violence. For
instance, if a group already has an expansive base area, the marginal strategic
gains of controlling new areas may be low. When armed groups are able to reap
lootable resource profits on the black market, protecting their bases may be
relatively more important than bearing the costs to take new territories. In these
cases, civilians may be able to increase the group’s security on the cheap by
providing information and reassurance that they are not collaborating with the
enemy. In sum, when there are stalemates, there is little interest in winning a
war, or the costs required to win are high, preventing defections to the enemy
can become a compromise measure for armed groups short of winning the full
allegiance of civilians.
Sensitivity to civilians is distinct from the use of violence against civilians and
can be measured prior to acts of violence. It can be thought of as how armed
groups experience costs (either material or moral) for using violence – an
elasticity of how responsive they will be if civilians decide to pressure them.
Groups sometimes use violence because it is cheap, easy, or simply a “default”
behavior and how they are used to operating (Kaplan 2013b). Sensitivity
combined with pressure from civilians can tip the behavior of armed groups in
circumstances when violence is a means to an end rather than an end in itself.
This can be hard to observe, however, because acts of violence (or their absence)
are binary events that either occur or do not occur and therefore usually reveal
little about the preferences of armed groups, such as their resoluteness or the
utility they gain from committing the acts. Some acts of violence may be much
more valuable to them than others. This raises the question of whether armed
groups are just on the tipping point between killing and abstaining from killing,
or they are set to kill no matter what. They may choose to rule by fear because it
is cheap or because they do not have alternative considerations or roadmaps for
how to interact with the population. But they could be induced to act differently.
I now show how some of these factors combine to produce armed groups
with different sensitivities to civilians according to existing categories in the
literature. How frequently groups actually hold these sensitivities is a separate
empirical question.

The Sensitivities to Civilians of Some Archetypal Armed Groups

I outline the sensitivities of groups that are genocidal, economically motiva-
ted, ideological, and hybrid. A first type of armed group may have strong

31
An implicit assumption of Kalyvas’s theory is that armed groups are continuously trying to
expand their reach and gain more territory, for whatever ends.
58 A Theory of Civilian Decision-Making in Civil War

preferences to target or kill the civilian population and derive little benefit from
civilian support. Such groups could be considered “ideologically genocidal”
and will be insensitive (inelastic) to civilian overtures. In general terms, these
groups could be defined as having extreme interests and are capable of and
willing to use a “final solution” to achieve them (Valentino 2004). They
may pursue ends such as ethnic cleansing (identity-based targeting), but also
resort to annihilation of populations to defeat an enemy at all costs or to
obtain economic benefits that are obstructed by civilians (e.g., to gain land
through displacement for illicit crop cultivation, African palm plantations, or
megaprojects such as dams). Since these “genocidal” groups target civilian
organizations, avoidance strategies or direct force may be the approaches with
best hope for influencing they (e.g., Jewish resistance in Nazi-controlled areas in
World War II; Tec 1993).
Strictly economic groups such as cartels or purely economically motivated
rebel groups can be similarly inured from civilians, but for a slightly different
reason. “Opportunistic” groups permit abusive violence through indiscipline as
a selective reward for their members (Weinstein 2006). Groups like drug cartels
do not necessarily harbor animosity toward civilians and are not necessarily
trying to defeat opponents or the state militarily, but neither greatly depend on
civilians for support or their livelihoods. As pure criminals, cartels have no
compunction against killing if civilians get in the way. This is seen in the recent
extreme violence perpetrated by cartels in Mexico. With these groups, avoid-
ance strategies or meeting them with force may be most effective (as the CDF
forces countered the looting RUF rebels in Sierra Leone). It is primarily when
these groups’ support is threatened that they might be responsive to civilians.
On the other end of the spectrum are groups that are strictly “ideological”
and promote the liberation of the people in the traditions of Mao and Guevara.
These groups will measure their use of violence to a greater degree, making
civilians’ organized autonomy strategies less necessary. Where these groups reign,
civilians can use the dominant strategy of allying with an ideological group for
protection to avoid punishment. These groups most commonly meet resistance
from civilians under conditions of shifting control. These less abusive ideological
groups or benevolent state actors will proactively respond to the population and
incite little resistance, and so will not likely be influenced by civilians.
A primary type of group I identify as meeting the sensitivity conditions for
effective civilian autonomy movements is what could be called “hybrid oppor-
tunist.” These groups may partially depend on civilians for resources but also
benefit from other resource bases. They may be shaped by ideological begin-
nings, resource shocks, and interactions with civilians.32 So while they may

32
Weinstein mainly theorizes about governance in zones of complete control, where whether or not
bargains are struck between rebel groups and communities derives primarily from armed group
characteristics. I argue that groups are more malleable and that there can be learning and that
dynamic relationships can be reshaped by civilians.
Explaining Violence: Conditions for De Facto Autonomy 59

have incentives to use violence or permit abuse, they may also be varyingly
susceptible to organized civilian strategies and able to reform. These groups
may be poorly organized and permit their members some selective “pillaging”
benefits but may also be susceptible to having their cover blown, which would
bring the intrusion of authorities. Within hybrid opportunist groups, ideo-
logues may at times overpower opportunists, or vice-versa.
A key insight from the identification of the hybrid opportunist groups is that
even Weinstein’s economic groups can seek a veneer of political legitimacy.
Although Weinstein suggests these groups will be out-competed by more eco-
nomic groups and wither, it is not clear how frequently this occurs in the real
world, and he gives little indication about the actual prevalence of these
different groups. Three important examples may be the cocaine-trafficking
FARC “revolutionaries” and the “anti-subversive” paramilitaries in Colombia,
and even the heroin-trafficking but religious Islamist Taliban in Afghanistan.33
This argument about armed group sensitivity does not make civilians epi-
phenomenal (ineffectual) or make the production of violence a tautology
(i.e., that belligerents commit violence when they have incentives to commit
violence) because armed groups can still have incentives to use violence. The
theory is falsifiable and would be weakened if groups with sensitive prefer-
ences ended up using violence against organized communities as much as
unorganized communities even in moderate conflict conditions. It would also
be undermined if violence were used for some other motives that civilian
organization fails to effectively address.

Civilians’ Levers
I argue that civilian-sensitive armed actors can be induced by organized civil-
ians to reduce acts of violence. Armed groups in civil war often face a strategic
dilemma that creates incentives to commit violence. Even when armed groups
are well-intentioned, uncertainty about civilian allegiances due to the “fog of
war,” private information, and civilians’ collective action problems may lead to
violence, especially if the costs of using violence are low. Furthermore, armed
groups’ predatory tendencies or organizational pathologies can lead to abuse.
Still, despite these incentives to commit violence, I identify two ways in which
civilians’ strategies can affect armed group calculations: inducing cooperation
between enemy armed groups to not kill civilians and marginally but signifi-
cantly raising the costs of killing.
First, civilians may attempt to shift armed groups’ “cooperative” incentives.
This may encompass situations where there is joint interest among the armed
actors to preserve or not directly target a civilian community. This does not

33
Ironically, these groups all frequently ban drug use in their territories but also participate in the
illegal drug trade.
60 A Theory of Civilian Decision-Making in Civil War

mean that an armed actor alone has no interests in committing violence

against civilians. They might, dependent on the choices of their rivals. Rather
it reflects a logic where coercion of civilians is like an arms race spiral. The
question is whether the groups pull back from the brink of more and more
violence. Groups would like to be able to sustain “if you won’t, I won’t” kinds
of agreements, but lack credibility and reassurances. In the face of this kind
of “security dilemma,” civilians can reassure groups and boost trust by, for
example, conducting their own investigations, which decreases the cost of
policing to armed groups. In addition, civilians may provide other joint benefits
through the linkage of other issues to the nonabuse of civilians, including
assistance in negotiating ceasefires, demobilizations, etc.
Second, civilians can attempt to manipulate “uncooperative” or unilateral
incentives. These are incentives that matter for a group to reduce killing
regardless of what their enemy is doing or whether civilians may be perceived
to be defecting. For instance, if abusive armies face reputational costs for
their acts from NGO monitoring, commanders might be willing to invest in
organizational reforms even though they may entail costs (the costs can be
multiplied if there is a prospect of being brought to justice when the conflict
ends). Preventing societal divisions is another way of increasing costs. With a
culture of peace it becomes harder to find informants and more likely that
targeted, denounced people are wrongly accused of collaborating with the
enemy. Finally, and not to be underestimated, there are the malleable ideas
and default attitudes that armed groups may hold about the use of violence.

summary
This chapter has outlined a theory of civilian behavior in the risky conditions
of civil conflict to explain how and when civilians can protect themselves.
It emphasizes the role played by organization and social capital. I speci-
fied civilian mechanisms that require and are facilitated by cooperation for
communities to maintain autonomy when facing shifting constellations of
armed groups. As illustrated, many mechanisms fall short of the risky option
of arming and can alter outcomes for civilians through their influence on
armed groups’ behavior. I also considered the conditions under which armed
actors might be most receptive to or influenced by civilian entreaties or
pressures.
For civilian communities, their level of organization is both a catalyst and a
constraint that explains the puzzle of why we observe some but not all commu-
nities acting for autonomy and succeeding in reducing violence. Cooperation
can arise from many sources both internal and external to a community and
both independent of and endogenously from the armed conflict. Some types
of organization may be more helpful than others and will enable more con-
tentious strategies. But even when civilian communities have the desire to
survive and protect themselves, they may lack the capacity or knowledge
Summary 61

to act. Even where civilian organizations are capable, neither they nor their
attendant strategies guarantee civilian protection.
Organization shows promise for providing additional protection under
certain conditions, armed group preferences and sensitivities, and configur-
ations of conflict. How broadly these conditions exist in any given conflict
is an empirical question that I begin to address in the statistical tests and
case studies in later chapters. In the next chapter on the social history of
Colombia, I connect the theory developed here to specific social organizations
and give an overview of the patterns of the conflict and collective actions to
protect civilians.
3

The History of Conflict and Local Autonomy

in Colombia

Don Apolinar Moscote (Aureliano Buendía’s father-in-law): “The Liberals were

Freemasons, bad people, wanting to hang priests, to institute civil marriage and
divorce, to recognize the rights of illegitimate children as equals to those of
legitimate ones, and to cut the country up into a federal system that would take
power away from the supreme authority . . . The Conservatives, who had received
their power directly from God, proposed the establishment of public order and
family morality. They were the defenders of the faith of Christ, of the principle of
authority, and were not prepared to permit the country to be broken down into
autonomous entities” (104).
Aureliano Buendía: “If I have to be something, I’ll be a Liberal because the
Conservatives are tricky” (106).
– Gabriel García Márquez, One Hundred Years of Solitude

The first response you will get from an average Colombian when inquiring
about the armed conflict is, “It’s complicated.” And it is true. There are regional
dynamics, multiple actors, changes across time, various theories and logics, and
hidden narratives and subtexts. Colombia has been in a state of conflict since
the early 1960s. The fighting between guerrilla, paramilitary, and government
forces, with narco-traffickers thrown in the mix, has had devastating conse-
quences for a large portion of the civilian population. Here I provide an
overview of the conflict and civilian mobilizations for peace from the second
half of the twentieth century through the present.1 This overview illustrates the
breadth, timing, and regionalization of these trends.

1
This is only a cursory overview of the social and conflict history most relevant to this study. The
study of the Colombian conflict is such a large field it has come to be known as Violentology. For
deeper histories by scholars with far more expertise than I see Pizarro and Peñaranda 1991,

62
Colombia’s History of Conflict 63

The history shows how a variety of social mobilizations grew in response to

rising levels of violence. A look back in time shows there have been various
historical iterations of conflict followed by civil society responding and working
to oppose violence and repair the damage. The causes of these cycles lay deeper
in the country’s broader social trends and political history.
Structural factors such as the illicit economy, difficult terrain, and state
weakness have shaped the state’s approach to the conflict and economic
development, but the state’s approach to governance has also been shaped by
politics. A series of inconsistent choices by elites, driven by political contingen-
cies, produced the social landscape that exists today and set the stage for
civilian autonomy. As alluded to in the García Márquez passage and explored
by Hartlyn (1988), the tensions between Liberal and Conservative partisans
produced various bouts of greater rural inclusion and development depending
on which party was in power and how severe governance problems became.
When confronting the consequences of the La Violencia conflict and later social
issues and conflicts, elites – mainly urbanites little affected by the enduring but
localized insurgencies – wanted to do just enough but not too much to address
the needs of the rural sector.
Staying “between legitimacy and violence” (Palacios 2006) perpetuated
inequality, the exclusion of rural classes, and the concentration of power. The
inconsistency among elites combined with narco-trafficking and nascent insur-
gencies was a formula for persistent rural conflict. The vacillation in rural
state-building policies and start-and-stop national-level peace negotiations
(García Durán 2006b) were also triggers that seeded desperation among the
population and capability for local autonomy. With private institutions such as
the Catholic Church also advocating for the needs of peasants, a variegated
landscape arose, with some communities stimulated to cooperate more than
others. As conditions worsened in the 1990s, civilians that had been left to
their own devices for decades began experimenting with different responses
depending on their community capabilities.

colombia’s history of conflict

The history of Colombia’s conflict can be simply summarized by looking at
the national homicide rate over time. The graph in Figure 3.1 shows two
significant waves of violence from 1946–2005, with a relatively calm period
in between.
The assassination of the Liberal politician Jorge Eliécer Gaitán on April 9,
1948, and the ensuing widespread riots in Bogotá, known as the Bogotazo,
contributed to the beginning of the period of civil conflict in Colombian history
known as La Violencia, or The Violence. It affected large swaths of the

Bejarano and Pizarro 2001, Romero 2003, Leal Buitrago 2006, Duncan 2006, García Durán
2006a, Palacios 2006 and Wills et al. 2006.
64 The History of Conflict and Local Autonomy in Colombia

90

80
Homicides per 100,000 people

70

60

50

40

30

20

10

0
19 6
48

19 0
19 2
54

19 6
19 8
19 0
19 2
64

19 6
68

19 0
72

19 4
76

19 8
19 0
82

19 4
86

19 8
90

19 2
94

19 6
98

20 0
20 2
04
4

5
5

5
5
6
6

7
8

0
0
19

19

19

19

19

19

19

19

19

19

19

20
Year
figure 3.1 The Colombian national homicide rate, 1946–2005.
Source: Fabio Sanchez (2007), Colombian National Police.

countryside, resulting in over 200,000 deaths from 1948–1958 (Sánchez and

Meertens 2001; Guzmán et al. 1963). The conflict unfolded in several waves
throughout the subsequent decades. The conflict has been characterized as
neighbor-against-neighbor partisan vendettas between Liberals and Conserva-
tives, which later degenerated into the terrorizing of campesinos by local,
clientelist armed bands tied to landholding bosses (gamonales).2 The violence
finally subsided beginning in 1958 with the establishment of the Frente Nacio-
nal (National Front) pact among national elites, which alternated power
between the Liberal and Conservative parties, and army operations to eliminate
many key bandits (Time Magazine 1964). But the government failed to eradi-
cate emergent guerrilla groups that represented leftist political movements that
were excluded from access to power and representation.
The origins of modern-day guerrilla groups are found in these early peasant
self-defense leagues. These groups were scattered in different regions that
came to be known as the “Independent Republics,” including Marquetalia
(in southern Tolima), the Aríari region in the Eastern Plains, and Sumapaz in
Cundinamarca, southwest of the capital. These areas were targeted by the
Army and Air Force in the early 1960s, but holdouts survived. After the
bombing of Marquetalia in 1964, Manuel Marulanda organized rebel resist-
ance to the state and, in conjunction with the Communist Party, held the First

2
For additional interpretations of La Violencia and its regional variants, see Guzmán et al. (1963),
Henderson (1985), Oquist (1980), Ortiz (1985), Roldán (2002), and Sánchez (2007).
Colombia’s History of Conflict 65

70

60

50
Number of fronts

40

30

20

10

0
1978 19791980 198119821983 19841985 19861987 1988 19891990 1991 1992 1993 19941995 1996
Year
FARC fronts ELN fronts EPL fronts

figure 3.2 The growth of guerrilla fronts, 1978–1996.

Source: Echandía (1999).

Guerrilla Conference in 1965. With Marulanda as its leader, the FARC arose as
an organization that championed the longstanding “agrarian struggle.”
In its early days, the FARC developed focal points for its efforts in the
departments of Tolima, Cauca, Meta, Huila, Caquetá, and Cundinamarca, as
well as the Urabá and Middle Magdalena River regions. It did not have a
national presence until 1982, when the group added the “People’s Army”
(FARC-EP) to its name, renewed its goals of pressing the causes of marginalized
peasants, and doubled the number of fronts. Although its origins were rural, the
FARC later expanded to urban areas. Figure 3.2 shows the growth of the FARC
over the time. The guerrilla group grew from seven fronts and 850 fighters in
1978, to more than 16,000 fighters in 2000, distributed across sixty-six fronts.
Figure 3.3 indicates where FARC attacks were concentrated over the period
from 1999–2005. The FARC formed a political wing in the late 1980s called
the Unión Patriótica (Patriotic Union Party) that ran local- and national-level
political candidates as part of a strategy of pursuing “all forms of struggle.” In
2000, the FARC renewed its political activities by forming citizen militias as
part of the Bolivarian Movement for a New Colombia (MBNC in Spanish) and
the Colombian Clandestine Communist Party (PCCC). These organizations
espoused communist and anti-imperialist teachings and promoted the forma-
tion of cells of spies and infiltrators to support the FARC’s armed wing.
Colombia’s other major guerrilla group, the National Liberation Army
(ELN), formed in 1962, with its first operations in 1965. Modeled after the
Cuban Revolution, the ELN was motivated by Liberation Theology and
66 The History of Conflict and Local Autonomy in Colombia

figure 3.3 Map of FARC activity, 1999–2005

organized around advocating for the poor with the goals of anti-imperialism
and forcing the national oligarchy out of power. The ELN grew from 350
members in 1984 to 4,500 in forty-one fronts by 2000, greatly extending its
reach across the country (Figure 3.2).
Colombia’s History of Conflict 67

As a consequence of the guerrilla escalation of the 1980s, there were several

state-led attempts to negotiate peace. Under president César Gaviria, smaller
insurgent groups demobilized, including the Quintín Lame Armed Movement
(MAQL), the M-19 (Movement of April 19), and parts of the Popular Liber-
ation Army (EPL). The agreements produced a National Constituent Assembly
that led to the approval of a new constitution in 1991, featuring an important
set of institutional reforms that included new rights, political decentralization,
and greater fiscal control for municipal governments.
Despite these peace efforts, the conflict continued as the FARC, ELN, and
factions from the EPL disengaged from the peace process. These different
guerrilla groups expanded beyond their traditional roles as intermediaries for
the peasants in their interactions with the government and also began financing
themselves through illicit activities such as extortion (especially in oil-rich
regions), alliances with drug-trafficking cartels, and kidnapping. They began
to charge protection taxes on illicit crops, cocaine laboratories, and traffickers
(Thoumi 1997). Despite its smaller size, the ELN is responsible for similar
numbers of kidnappings and acts of sabotage as the FARC.
On the other side of the conflict, paramilitary groups emerged in the early
1980s after a failed peace process with the guerrilla groups. Their origins are as
self-defense groups that were sponsored and financed by large landowners in
response to guerrilla transgressions, especially as the guerrillas increasingly
targeted these landowners for kidnapping and extortion. The earliest ones arose
in the Magdalena Medio (Middle Magdalena) region and along the Caribbean
coast, as well as in several other regions of the country. Before long, they
morphed to take on some of the counterinsurgency functions of the state and
fight the guerrillas. In the process they targeted leftist leaders and people with
perceived links to the guerrillas. Thousands of members of the Unión Patriótica
party who were stigmatized as being linked to the guerrillas were murdered or
disappeared in the late 1980s and early 1990s in what has been termed a
politicide, providing the FARC further justification to continue with a militar-
ized strategy (Dudley 2003, Gómez-Suárez 2007).
In 1997, different groups joined together to become the United Self-Defense
Forces of Colombia (AUC in Spanish) under the leadership of Carlos Castaño,
who had previously founded self-defense groups in the northwestern regions of
Córdoba and Urabá. This new group expanded counterinsurgency operations to
retake territory that had been held by the guerrillas. It also became involved in the
illegal drug trade and introduced new, grotesque forms of brutality into the
conflict, including torture, mutilations and quarterings (descuartización), and
sexual violence. Between 1997 and 2002, many massacres were attributed to
groups under the AUC umbrella (Sánchez 2007). When these groups demobilized
beginning in 2003, more than 30,000 fighters across thirty-seven blocs and fronts
turned in their arms (Oficina del Alto Comisionado para la Paz 2006).
State forces proved insufficient and inadequate to defeat these nonstate
actors, but this did not mean they were unimportant actors in the conflict.
The Colombian military has had the capacity to protect civilian populations at
68 The History of Conflict and Local Autonomy in Colombia

various times in various regions. Yet the army has not been omnipresent and
has also been seen by some as “one more actor in the violence” (Ladrón de
Guevara 1998). The public forces have been accused of cases of complicity with
narco-traffickers and with paramilitaries, who shared their counterinsurgent
goals (Richani 2002).3 During the peak years of conflict, the office of the
Procuraduría investigated, disciplined, and dismissed hundreds of military offi-
cials, including generals (Amnesty International 2001). More recently there
have been several thousand incidents of “false positives” (CINEP 2007, Human
Rights Watch 2015). These incidents involved the killing of peasants and urban
poor (predominantly male youths) by state forces and dressing them in fatigues
to frame them as guerrilla casualties and increase guerrilla body counts for pay
and vacation incentives. While the history of the armed forces has given some
residents reason to be fearful, it is a multifaceted institution which has also
undergone important police reforms and human rights training, and in at least
some cases has prosecuted abusers within its ranks and held them accountable.
The trends of growth of these different armed groups meant rising conflict
during the 1990s. The guerrilla groups expanded to have a presence in nearly
two-thirds of Colombian towns at the peak of the conflict in the early 2000s.
President Andrés Pastrana initiated negotiations with the FARC beginning in
1998, ceding to them a demilitarized zone, or zona de despeje, for the talks that
was roughly the size of Switzerland in the southeast Macarena region (civilian
autonomy within this zone is analyzed in Chapter 9). The negotiations failed in
that they did not end the conflict and allowed the guerrillas to regroup and
rearm. But the negotiations also exposed the guerrillas’ violent nature and
generated a greater consensus among the population that a harder line against
them was necessary. By the late 1990s and 2000s, with so many armed groups
and drug traffickers affecting so much of the country and penetrating multiple
levels of government, Colombia began to be discussed as a possible “failed
state” (Bejarano and Pizarro 2001, Kline 2003). As a commonly cited symbol
of the extent of the threat, mortars fired by the FARC fell inside the presidential
palace on the day of President Álvaro Uribe’s inauguration in 2002.
The government implemented the Plan Patriota counterinsurgency program
to respond to the conflict crisis. The Plan Colombia foreign aid package from
the United States came to more than $10 billion was disbursed over more
than a decade for areas including training and equipment for the armed forces,
counternarcotics programs, development programs, and institutional
strengthening. The initiatives helped bring about a gradual de-escalation of
the conflict beginning around 2003 with the gradual repulsion of guerrillas and
the beginning of the demobilization of paramilitary blocs. Still, the guerrillas’
bellicose actions continued and some of the paramilitaries then quietly
remobilized and new criminal bands, known as BACRIM, began to appear

3
According to a 1999 State Department human rights report, “Security forces actively collaborated
with members of paramilitary groups by passing them through roadblocks, sharing intelligence,
and providing them with ammunition” (U.S. Department of State 2000).
Colombia’s History of Conflict 69

Anti-narcotic police back from patrol and seeking respite from the midday sun,
La India, 2013.
Photograph by Oliver Kaplan.

(CNRR 2007). Although the Colombian government once again engaged in

peace negotiations with the FARC in 2012 and signed an agreement in 2016,
the armed conflict has persisted.
The consequences of this long period of conflict in Colombia have been
severe for civilians, marked by widespread violence and terror.4 There have
been massacres of tens and sometimes hundreds of people as well as political
killings of opposition politicians, social organizers, and labor leaders. At least
220,000 people have been killed in the armed conflict, and many victims have
been disappeared and dumped in mass graves (GMH 2013). An estimated 6
million people, representing nearly one-sixth of the rural population, have been
forcibly displaced from their homes and communities since 1985, ranking
Colombia second only to Syria in the number of internally displaced persons
(UNHCR 2010, El Espectador 2015, IDMC 2010, IDMC 2015). The conflict
has also contributed to the recruitment of child soldiers (Human Rights Watch
2003) and persistently high rates of inequality and poverty.

4
According to an accounting from the Justice and Peace Law, 2,719 mass graves had been
exhumed and over 300,000 people registered themselves as victims of the conflict as of 2010
(El Tiempo 2010).
70 The History of Conflict and Local Autonomy in Colombia

An obelisk memorial to ATCC leaders who were killed in 1990 in La India,

Santander, Colombia.
Photograph by Oliver Kaplan.

All of these factors have added up to great suffering and uncertainty in people’s
daily lives. These effects can be explained by the increasing contestation among
armed groups and narco-trafficking (e.g., Sánchez 2007) but also by state absence,
which, in addition to stemming from Colombian politics is also a product of
structural factors such as histories of colonization in far-flung areas and the coun-
try’s mountainous terrain (LeGrand 1986). The expansion of the conflict has, by
contrast, only infrequently been associated with the characteristics of specific
towns as armed groups broadly sweep over differing neighbor communities.

the history of civil society responses

to armed conflict
Colombia’s history has been characterized by much violence, but vibrant
sources of social capital remain and have served as crucial bases for collective
responses. For instance, Colombia’s patterns of colonization have meant some
towns have remained independent for much of their history. Isolated, difficult
to secure, and neglected, such communities have also created tight-knit social
relationships (LeGrand 1986). With these bases of cooperation, some towns
Peace Communities and Formal Autonomy Organizations 71

were organizing as early as La Violencia to maintain their own security by

forming neighborhood watches and self-defense posses to guard against bandits
(see examples in Chapters 6 and 8).
In the late 1980s and 1990s, what has been characterized as a broad,
nationwide peace movement emerged in response to the spiking violence. Social
movements that had previously been centered on social and political protest
shifted toward the goals of survival and pressuring the government and armed
groups to respect the civilian population and enter negotiations. A graph
produced by Mauricio García Durán (2005; see Figure 3.4) shows the number
of recorded collective actions for peace over time in Colombia, including
national protests, local “peace communities,” and other nationwide activities
such as the Civilian Mandate for Peace symbolic vote of 1997. García Durán
tracks these against increases in the intensity of conflict according to the
number of casualty events and kidnappings over time. He finds that since the
1980s there have been many national and local movements of different forms
across many parts of the country. His analysis also shows that the prevalence of
peace actions tends to increase after spikes in conflict, with the 1997 Mandate
for Peace vote as a prime example.

peace communities and formal autonomy

organizations
Like the broader peace movement, the formation of local organizations for
protection and autonomy also tends to occur as conflict intensifies. Though
regional conflict dynamics vary, many communities’ decisions for autonomy
were made around periods when paramilitaries arose and contested guerrilla-
held zones. As guerrilla groups became more coercive and increasingly aban-
doned their roles as protectors of and interlocutors for rural populations,
peasants looked for other options. The paramilitaries issued many ultimatums
for displacement and were feared for their ruthlessness. Various formal organ-
izations formed and adopted strategies in response to displacement through a
sorting process where either less resolute civilians fled or whole communities
displaced with only the resolute returning to reconstitute.
The trend of community peace strategies can be traced to a number of other
contributing factors. First, a variety of social stimuli in the rural sector brought
communities together. These included government rural development policies
beginning in the 1960s (e.g., ANUC land reform councils and the junta
councils, discussed later); Patriotic Union and Communist Party political
organizing in the 1980s; and church organizing, which, with the rise of Liber-
ation Theology, increased outreach in rural areas through programs such as
Pastoral Social and later peace-oriented organizations such as CINEP, the
Middle Magdalena Peace Program (PDPMM), and the Inter-Church Justice
and Peace Commission (Justicia y Paz) that provided accompaniment and
humanitarian aid. Second, policies such as those embodied by the 1973 Pact
72

figure 3.4 Intensity of the conflict and level of peace mobilization, 1978–2003.
Source: García Durán (2005), CINEP.
Peace Communities and Formal Autonomy Organizations 73

of Chicoral, an agreement between the political parties and large landowners

that ended agrarian reform, symbolized the beginning of an era of broader
exclusion of rural residents from the political system and reduced state services.
This left communities that had been organized by previous rural development
programs adrift. Third, the growing connection and communication between
small towns, NGOs, and leaders in cities increased the dissemination of social
organizing lessons, ideas, and technical assistance. Fourth, the Indigenous
rights movement that burgeoned in the 1980s fought for and won legal
political autonomy in the 1991 Constitution, which not only aided Indigenous
groups but also caused the concept of autonomy to drift to other populations
(e.g., Houghton and Villa 2005, Caviedes 2007, and Sánchez and Del Mar
Palau 2006 on municipal budgetary autonomy). Fifth, the failures of national-
level negotiations and the disengagement of the state left many localities aban-
doned and unwilling to hold out hope for a broader resolution to the conflict
(this was true both during as well as prior to the Pastrana administration; see
García Durán 2006a, Bouvier 2009).
The earliest communities to seek autonomy were located in places with both
social unity and where armed conflict arrived earliest. Many were in the Urabá
and Magdalena Medio regions, and the departments of Cauca, Chocó, and
Putumayo. Nascent local movements for protection – not precisely for autonomy –
began with marches in response to early displacements. The protests by residents
of towns like El Pato are examples of movements that are, at least in name,
separate from guerrillas and also against repression by the army (Molano and
Reyes 1978). The Indigenous groups’ history of resistance to Spanish and
colonial institutions and subsequent exclusion from Colombian society, coupled
with Indigenous rights movements during the twentieth century, helped them
become the first population to advocate for autonomy, establishing an ideal for
other communities to emulate. The formation of the Regional Indian Council of
Cauca (CRIC) in 1971 was a key antecedent to the Nasa (Paez) community later
issuing the Declaration of Ámbalo in 1985, demanding respect of their auton-
omy from the guerrillas and other armed groups (Houghton and Villa 2005).
One of the earliest campesino autonomy movements can be traced to the ATCC
of La India in Santander, analyzed in Chapter 7, which formed spontaneously in
1987. Impressively, these communities include some of the more poor, illiterate,
and marginalized populations, with few apparent organizational capabilities.
Some early programs and NGOs supported these communities in their
struggle. Key organizations include the Middle Magdalena Program for Peace
and Development (PDPMM), founded in 1995 by Father Francisco de Roux,
and the Jesuit organization CINEP, which promoted the formation of the first
peace communities along the Atrato River in the department of Chocó in 1998.
These and other NGOs such as Redepaz (Peace Network) have worked to con-
solidate and standardize community-organizing procedures and promote peace
communities more broadly. As Sandoval (2004) shows, formally organized
communities arose in many parts of the country (Figure 1.2 in this volume).
74 The History of Conflict and Local Autonomy in Colombia

Formal autonomy organizations take a variety of forms and actions. They

are commonly broken down by ethnic background due to cultural differences
in forms of organization (many communities have mixed populations which for
present purposes are considered as campesino). Later, I describe some of the
most prominent communities to provide a sense of who they are and how they
behave. The institutional rules and declarations for many formally organized
communities for autonomy are contained in Villarraga (2003).
First, there are the extremely organized and well-known Indigenous group
autonomy movements. Colombia has more than eighty Indigenous groups
comprising roughly 3 percent of the population (approximately 1.4 million
people in 2005). These communities are organized around cabildo councils and
spiritual authorities such as shamans (Kaplan 2013c). Many have been affected
by the conflict entering their territorial reservations and, rooted in their com-
munities’ unique cosmovisions and “life plan” documents, many have organ-
ized in various ways in response. They have carried out many discrete
protective actions with success. Indigenous groups have rescued kidnap victims
from armed groups, resolved internal conflicts through systems of Indigenous
justice, and protested en masse against the taking of towns (up to even holding
guerrilla soldiers and evicting army troops).
The Nasa (Paez) Indians in the department of Cauca are especially well
organized. Their nonviolent Indigenous Guard, composed of thousands of
volunteers who carry symbolic staffs known as bastones, have rescued kidnap
victims from the FARC, including the kidnapped Paez leader and mayor of the
town of Toribío (Forero 2005). Four hundred members of the guard marched
for two weeks over the Andes to the rebel camp where he was being held and
persuaded the FARC to release him. Several Indigenous groups are also known
for their massive protest marches against the conflict and neoliberal economic
policies, and in support of Indigenous culture.
Afro-Colombian communities have also innovated their own movements for
autonomy from the conflict. Afro-Colombians comprise about 26 percent of
the national population (Colombia has the third-largest Afro-descendent popu-
lation in the Americas after the United States and Brazil). Their initiatives have
been mainly concentrated in the departments of Chocó, Valle del Cauca,
Nariño, and Antioquia, where their populations are greatest. The Law of the
Black Communities (Law 70 of 1993) helped cement Afro-Colombians’ access
to their collective territories and formalized their community councils, or con-
sejos comunitarios. This has helped them to manage their own affairs and
provided grounds for organizing. ACIA (2002) is one of the largest Afro-
Colombian autonomy organizations with more than 120 villages with 45,000
residents along the Atrato River in the Pacific department of Chocó. These
villages have procedures to peacefully resolve community conflicts and com-
prise an information and early-warning network, among other policies.
Campesino communities have fewer formal organizations and are more
variable in composition and policies. The organizations commonly have origins
Peace Communities and Formal Autonomy Organizations 75

in farmer associations and village junta councils. Perhaps the best-known

campesino autonomy organization both in Colombia and internationally is
the “peace community” of San José de Apartadó (Hernández 2004, de
Sousa Santos and García 2004). This community of displaced returnees in the
northwestern Urabá region of Antioquia received accompaniment from NGOs
and the Catholic Church, and the Inter-American Court of Human Rights
issued precautionary measures to protect the population. Still, this community
also suffered the murder of more than 200 residents. Rooted in faith and
pacifist ideas, the community’s declaration for peace includes the policies that
residents will not participate in violent activities, not allow firearms, and not
collaborate with armed groups or provide them with information. The commu-
nity also implemented more extreme policies including a “rupture” with state
institutions for lack of responsiveness in investigations of abuses by paramili-
tary and state forces (Recorre 2007). Since there are many studies on this case
and some difficult measurement issues, I do not study it closely here.
The ATCC in Santander (Chapter 7) is a similar association of campesinos
though it has distinct policies and institutions. In addition to its investigation
procedures and network of information-gathering village councils, it has taken
various actions of civil resistance to counter the aggression by armed groups in
recent years. In 2001, 800 residents protested for two days against being
displaced from their homes. Similarly, after a man was killed and quartered
in 2004, hundreds of people again went to protest (CNRR 2009).5 Rural
campesino municipios have also used “constituent assemblies” to coordinate
citizens on participating in local politics and supporting mayors in blocking
armed actors from involvement in municipal affairs (Sandoval 2004). As I
discuss later and in the case studies, individual campesino junta councils have
also taken on some functions of the more formal organizations dedicated to
“peace” and “autonomy,” including methods of dispute resolution.
Throughout these movements, women have played integral roles alongside
men in advocating for the autonomy and protection of their communities.
Many women have suffered violence during the armed conflict, been widowed,
and also served as combatants. However, their work for autonomy is also
widespread. Many of my interview respondents were women of all ages, and
they offered unique reflections on peace and conflict. But, more than that, they
also sought to navigate the conflict as participants in junta councils, as leaders
of these councils, and as mayors of municipalities in conflict zones. Within the
ATCC, women have served as conciliators alongside men to defuse threats of
violence. Women are the backbone of autonomy-seeking organizations, such as
the Afro-Colombian COCOMACIA, and they carry the symbolic baston staffs
as part of the Nasa’s nonviolent Indigenous Guard. In quieter ways, women
have strived to keep their families intact and inculcated norms of peace and

5
ATCC#3, La India, 7/2008.
76 The History of Conflict and Local Autonomy in Colombia

A community planning exercise near La India, Santander, Colombia, 2008.

Photograph by Oliver Kaplan.

nonviolence to steer their children away from the conflict. In many commu-
nities, women also form back-channel information networks that share news
about the war and efforts to organize against it. Although the gendered aspects
of seeking autonomy are not the focus of this study, the specific contributions of
women to peace in Colombia cannot be overstated.
As is evident, formal organizations for autonomy are impressively diverse.
They have many names: peace communities, peace laboratories, zones of peace,
no-conflict zones, humanitarian zones and spaces, sanctuaries, territories of
nonviolence, constituent assemblies, and peace experiences (León 2004, Sandoval
2004, Páez Segura 2005, Rettberg 2006, Bouvier 2009). The unique origins and
nomenclature of these organizations is part of what makes them interesting to
study and compare. They also present difficult classification issues for determin-
ing what constitutes a peace community since they are also diverse in terms of
their mechanisms, composition, political alignments, legal status, rhetoric, and
geographic extent. For formal organizations that often have formal declarations
of autonomy-oriented policies, there are still many ambiguities about their poli-
cies and how well they are implemented. This creates a complication for social
scientists because if these organizations cannot be compared or even categorized
and classified, how can one identify which mechanisms are operating or what are
plausible counterfactual cases? Other political actors in Colombia and elsewhere
have had similar difficulties interpreting the nature of these organizations.
Peace Communities and Formal Autonomy Organizations 77

The Carare River, La India, Santander, Colombia.

Photograph by Oliver Kaplan.

The Politics of Autonomy and Neutrality

Autonomy movements have arisen as responses to pressure from armed groups
in conditions of state absence. Armed actors have responded in diverse ways,
with movements being received and respected differently at different times by
different armed actors.
The Colombian state’s position toward civilian autonomy movements has
been two-sided. Subnational accords with illegal armed actors are seen as an
affront to state sovereignty and are technically “illegal” in Colombia.6 Parts of
the government including the Colombian military have had little patience for
“peace communities.” As an example of this attitude, former Defense Minister
Jorge Alberto Uribe is quoted as saying, “There cannot be Peace Communities
without the presence of the Public Forces” (Navarrete-Frías and Thoumi 2005,

6
Military#1, 2, Bogotá, 8/2009. The term “neutrality” has been politicized in Colombia, so one
cannot necessarily draw conclusions about the actual stance of a community from the use of the
word. Former President Álvaro Uribe, as governor of Antioquia and later as president, supported
what was called “active neutrality,” which in reality meant that communities could be neutral, as
long as they sided with the government. This may partly explain why communities with practices
of noninvolvement in the conflict have come to favor the more acceptable term of “autonomy.”
78 The History of Conflict and Local Autonomy in Colombia

El Tiempo 2005). Based on my conversations with certain military officials

(colonels and generals), I found some suspicion of civilian peace and autonomy
movements on the part of the military, in part because the military expects and
prefers full allegiance and cooperation from the population.7
Still, autonomy organizations have been tolerated by the government in a
number of regions, and the Indigenous group movements are tolerated even
more widely (because of their constitutional legitimacy). Other parts of the
government have more explicitly aided local communities’ initiatives for peace,
including the Social Action Agency’s Peace Program, the office of the Human
Rights Defender (Defensoría), and the Colombian Vice-presidency’s Human
Rights Observatory’s reporting on community peace initiatives (Vicepresiden-
cia 2001). More recently, there has been evidence of an evolution in views
toward local peace communities, neutrality, and autonomy. In a groundbreak-
ing statement in 2013, President Juan Manuel Santos asked for forgiveness
from the San José de Apartadó Peace Community in Urabá for the previous
stigmatization of its residents by government and military officials, saying, “We
do not agree with phrases or attitudes of stigmatization [of] those who search
for peace and reject violence. . . we consider that every defender of peace and
human rights should be praised and protected. For this, we ask for forgiveness.
I ask for forgiveness” (Presidencia 2013).8 Santos not only acknowledged the
suffering of the community and the “mistake” of “unjust accusations,” but
even went so far as to praise the community as a positive example for peace,
noting that such local models are indispensible for national-level peace.
Although his remarks may have only garnered momentary attention, they
essentially validated the entire Colombian nonviolent peace movement in one
fell swoop.
The state’s perspective is of course different than that of the communities,
who originally selected their organizational strategies precisely because of state
absence and anarchy. The state’s differing perspective is a result of differing
interests and its sporadic rural presence (and thus limited information). But the
interests of the state have some overlap with those of communities. Many
communities want the full and permanent presence of the public forces and
other state institutions, and have at various times called for the army, the police,
and the Human Rights Defender (Defensoría) to protect them. This distin-
guishes these movements from the broader question of political autonomy since

7
As this chapter and Chapter 2 show, communities’ policies are diverse, and it is not clear if all
autonomy strategies are viewed as threatening or futile or if only certain aspects are disliked. It is
also unclear how well different military officials understand and have analyzed the details of these
communities’ policies and whether their judgments are based more on general preconceptions.
8
Remarks translated from the Spanish by the author. Santos’s statement is noteworthy as coming
from someone who previously served as the Minister of Defense in the 2000s, when atrocities
were committed by all parties to the conflict.
The Junta Councils and Their Relevance for Violence 79

most communities proclaim their apparent willingness to relinquish their

autonomy stances (at least in the realm of security) if the state could sustain
its institutional presence and deliver good governance. They seek a beneficial
and secure social contract. At least in Colombia, such communities have
primarily advocated for protection and have not called for enduring political
autonomy or separatism, as is sought by insurgents in some countries. How-
ever, because of the state forces’ sporadic presence in the daily lives of rural
communities, they are somewhat less relevant than other armed actors for the
study of civilian autonomy in Colombia.
Paramilitary blocs and guerrilla fronts have at different times blatantly
transgressed against formal autonomy organizations, often due to increasing
contestation or resource interests. Even though armed groups of some regions
are known for being repressive, the influence of civilian organizations and
strategies is understudied, with no controlled tests to verify whether these
repressed regions are unique in their conflict conditions or types of civilian
organizations. The Colombian guerrilla groups tend to be centralized, have
greater conformity across fronts, and have peasant origins, so they may exhibit
less variation in their treatment of civilians (Gutiérrez Sanín 2008). Still, they
can differ in their commanders, local dynamics, resource bases, and recruits.
The greater variation among the blocs of the more decentralized paramilitary
groups and their general suspicion of civil society makes them more variably
influenced by social organization. These stylized facts are examined as testable
hypotheses in Chapter 5.

junta councils and their relevance for violence

As the principal local organization across Colombian communities, junta coun-
cils have a surprisingly long yet surprisingly un-storied history. I argue that the
politics behind the juntas makes them appropriate for the study of local
responses to armed conflict. In the discussion that follows, I review their
origins, growth, and politics and show why they have (re-)gained a discreet
though important role in the present conflict in Colombia.
The idea and initiative to encourage the formation of local councils emerged
in 1958 at the end of the brutal La Violencia conflict (Law 19 of 1958). The
destruction was a warning to the incoming National Front unity government of
the pernicious impact of interference in communities by bands and outsiders.
As an early publication on juntas characterized this history, “The violence that
tormented the nation during many years clearly demonstrated the need to
change the situation that affected the local communities” (Triana y Atorveza
1966, 19).
For the government, local councils were “a response to the climate of
internal conflict.” They were viewed as nonmilitary pacification and a way
of encouraging reconciliation and economic development from the bottom
up in violent as well as peaceful areas (Ministerio de Gobierno 1993, 13;
80 The History of Conflict and Local Autonomy in Colombia

Edel 1969).9 As such, the juntas were a distinct break from prior community
organizational forms of the period before La Violencia, when development
policy was more top-down and did not consider local-level community inter-
ests. The past paternal-yet-distant relationship between the national govern-
ment and local communities not only failed as a model of community
development, it also failed to integrate the country given that the Colombian
state had limited reach into its vast mountainous, rural hinterlands. The juntas
embodied not just a strategy for economic development, but also an alternative
form of state-building. The Colombian state could not quickly and easily
increase its capacity in the wake of La Violencia’s devastation, but it could
quickly encourage capacity from the bottom up and then establish links of
communication and coordination, and incentivize development through match-
ing project funds.
The juntas as an organizational form were not imposed by the state. Rather,
the state created and encouraged a legally recognized vehicle that communities
could freely adopt – it attempted to institutionalize local councils. Virtually any
community could be organized into a junta and recognized by the government.
Although juntas are sometimes thought of as a local level of government, they
clearly fall in the realm of local civil society. According to law, they are defined
simply as “Civic nonprofit corporations composed of the neighbors of a place,
who unite efforts and resources to meet the most important needs of the
community” (Art. 1, Decreto 1930, 1979). It was intended that juntas would
help a community solve its own problems rather than look to the government
and keep communities informed about government programs (Fals Borda
1960). Juntas select their own development projects and rely on voluntary
contributions of labor and cash from their own residents (which may then be
matched by state funds). Juntas are involved in providing public goods for
economic development such as building roads, schools, and housing, as well as
cultural and social events such as festivals.
With the start of the communal action program, the number of juntas grew
rapidly. The government did promote juntas in some areas that were affected
by violence, but in most cases, communities decided to start a junta themselves.
According to the 1993 government census of juntas, 78 percent (23,690) of the
juntas were formed by initiative of the community. In contrast, only 12 percent
were formed by “promoters” (3,525; e.g., government staff or Peace Corps), 2
percent (568) were formed by the initiative of national officials/authorities, and
another 2 percent (587) were formed by the initiative of local/municipal
politicians (Ministerio de Gobierno 1993).10 As Cubides (2006) notes, since

9
The government also implemented a separate, more targeted Civic Military Action program (at
the suggestion of U.S. advisers) to provide short-term aid to some of the most war-torn commu-
nities (Rempe 1999).
10
Unfortunately, it appears the micro data from this remarkable study has been lost to eternity.
The Junta Councils and Their Relevance for Violence 81

table 3.1 Formation of junta

councils in Colombia, 1960–1993

Year Juntas
1961 1,000+
1966 9,000*
1970 16,108
1974 18,000*
1980 30,007
1987 34,842
1993 42,582 (inventory;
30,362 in census)
+ Estimates in Edel (1969)
* Estimates in Bagley (1989)
Source: DIGIDEC 1993.

many formal kinds of rural organizations met their demise during La Violencia,
the slate was clean for new forms to take root, “By more or less spontaneous
fashion, new independent organizations began to arise” (emphasis added).
Indeed, as shown in Table 3.1, the number of juntas has grown steadily since
the program began fifty years ago, with two-thirds of existing juntas created by
1980. The 1993 inventory found that nearly 2 million Colombians were active
members of juntas in their communities (of a population of roughly 40 million;
the mean size of a junta is forty-five people).11 According to Bagley, although
juntas became increasingly common, their numbers belied variation in their
functioning, “By the late 1960s, half of rural villages had juntas, although
many remained inactive for years” (Bagley 1989, emphasis added).
The history of juntas suggests there are four main reasons why juntas can
contribute to limiting violence in the Colombian conflict. First, juntas have
become reasonably widespread, including in historically conflictive areas, yet
are not everywhere. This distribution is helpful for making analytical compari-
sons since it has meant that some areas with recent conflict have juntas while
others do not (or have weak ones). Second, the national politics of the alternat-
ing National Front governments – both their goals and tensions – contributed
to the political positioning of juntas as independent (apolitical) and generally
centrist. This has meant that today they may have a degree of freedom in
decision-making (and even autonomy) when dealing with armed groups.
Third, and more specifically, by being democratically elected, juntas reflect
broad-based and unifying local collective action. This allows juntas to clearly

11
There can be a maximum of one junta per village or neighborhood. According to Article 36,
juntas must at minimum have eighty affiliates/members in Bogotá, sixty in other urban areas
(forty in urban areas of commissaries), and twenty-five in rural areas (twenty in rural areas of
commissaries, which were nonincorporated departments).
82 The History of Conflict and Local Autonomy in Colombia

represent their communities in relations with outside actors and limit the
number of spoilers who can disrupt their processes. Fourth, juntas have proven
they are adaptable through their ability to take on new functions and endure
over time. More specifically, when faced with armed conflict they have been
able to consider new strategies such as those previously discussed that surpass
their original mandates.12
Anecdotal evidence suggests that juntas have indeed been players in the
armed conflict today and that there are a number of ways in which they can
impact conflict dynamics. First, some of the juntas’ own activities are important
for reducing the social divides that might invite intrusions by armed actors. For
instance, in addition to the role of juntas in development projects, another
important function is to provide local order by adjudicating local disputes or,
according to the legal articles, “Seek harmony in interpersonal relations of the
community to achieve an environment that facilitates its natural development”
(Sec 300, 1987, Art. 11). This task appears to have been widely pursued by
juntas since, according to the survey of juntas in 1993, 87 percent of juntas
(26,474) had a conciliation committee.13 Second, juntas have become the
organizational building blocks for larger, more formal, and more cohesive
organizations that more openly advocate for civilian protection and autonomy
with specific reference to the armed conflict (e.g., “peace communities,”
“humanitarian spaces,” etc.). For instance, the farmers’ association of
ASOPROA in eastern Antioquia was formed through the coordination of
various junta leaders in the midst of conflict in 2002. Though not denominated
as a “peace community,” with united junta leaders, the association adopted
various community-strengthening measures to “protect human rights” and
avoid the dangers of multiple armed groups.14

12
Based on these characteristics of juntas, I conclude that juntas are better positioned than other
historic organizations, such as the at-times radical and even militant ANUC land reform
councils, to be relevant for reducing violence.
13
Although the primary function of conciliation committees is to adjudicate ambiguities in the
implementation of junta rules, in many cases they have been adapted to deal with other forms of
intracommunal conflict as well, or what in Colombia is called “conciliation in equity” (inter-
view, AC#1, leader in the junta movement, Bogotá 8/2008).
14
According to a document on this association, “The violence reached its highest peak and
generated an unprecedented humanitarian crisis in the region that demolished everything that
had been built. The leaders noticed there was a very fragile system of organization, only of
community juntas, very atomized (de grupo). It was necessary to build a system of organization
with a higher profile and greater coverage of regional character . . . With a very diminished group
of promoters that were left and a few juntas comunales in the middle of 2002 an association
began to form, holding meetings, crystalizing the idea and bringing in other communities. It
wasn’t easy, the region was very militarized and the meeting sites were chosen with much
caution; even though nothing illegal had been done, there was much fear that the army would
find the leaders meeting . . . The communities have achieved a certain level of . . . independence
regarding the management and denunciation of problems or situations . . . Interlocution with
authorities of every level, [and] . . . directly denouncing violations before national and
Summary 83

Third, there are various examples of juntas themselves playing important

roles in the crossfire for protecting residents and asserting autonomy since the
1980s (Cubides 2006).15 Cubides and others view juntas as occupying a special
place in the landscape of the conflict, “The juntas are the only [organizations]
that seem to be above all suspicion for the contending [armed] groups . . . Each
armed group has wanted to adopt the JACs, coopt them, ensnare them in their
mobilization strategy, but none have been able to completely succeed” (Cubides
2006).16 For instance, as the conflict escalated in the early 2000s, the govern-
ment sought to co-opt the juntas as part of the “Democratic Security” plan.
And, although some juntas were variably influenced by the government, armed
actors, or political bosses, national junta leaders decried this policy and its
inherent risks to and encroachments upon them.17 They advocated for the
juntas to remain independent and not take part in the conflict.18
In sum, the implications of the history of juntas for their present-day abilities
to blunt the effects of conflict are tantalizing for students of human rights and
peace studies and deserve careful assessment. Chapter 5 tests empirically
whether preexisting organizational structures such as juntas help civilians
effectively coordinate when conflict comes to town.

summary
Social movements for peace in Colombia are accounted for historically as a
response to the spread and intensification of conflict. They are also the results of
particular social landscapes and trends in rural organizing that were shaped by
a diverse set of prior causes and motivations, including politics. The national
and sometimes local-level tensions that determined the evolutionary course of
different social organizations, such as the juntas, provide crucial context for

international human rights organizations . . . has permitted the wide improvement in security
conditions of the communities that were directly affected by threats from one or other of the
contesting [armed] groups, in this way recovering the confidence for team/ group work”
(ASOPROA 2006; translated).
15
A#1, Anolaima, 3/2009.
16
As Cubides elaborates, “In zones controlled by the guerrilla, in disputed zones, or even in zones
controlled by paramilitaries, [the juntas] in contrast are the only form of civil society power.
They fill a gap. They are indispensable as forms of authority.”
17
El Tiempo. 2004. “Comunales Preocupados Por Politización Y Amenazas” (Juntas Worried
About Politicization and Threats). El Tiempo, December 7, 2004.
18
As one leader asserted during this period, “The juntas will be neither army informants nor
instruments of armed groups like the FARC . . . .This message that they don’t involve us in the
war is also directed at the paramilitaries and ELN.” As the article continued paraphrasing, “The
junta leaders arriving from remote villages lamented why they are neglected in development
plans but included in plans to involve them in the armed conflict. . . . Popular power should stem
from the autonomous will of the people and not from the pressures of armed actors. They oppose
being used for war” (El Tiempo. 2002. “No Les Serviremos Ni A Uribe Ni A Las Farc” [“We
Won’t Serve Uribe nor the FARC”]. El Tiempo, July 29, 2002).
84 The History of Conflict and Local Autonomy in Colombia

testing hypotheses and relationships over long periods of time. The political
battles among national-level elites along with the varying social landscapes
from town to town explain why the juntas’ ties to the central government
weakened over time and why some juntas remained more vibrant than
others. These patterns positioned these organizations for autonomy within
the armed conflict.
With many existing works on Colombian history, my argument that Colom-
bian social history witnessed competing political visions is not an innovation.
But connecting it to the conflict-relevant social landscape, the behavior of civil
society organizations, and the prospects for local autonomy today is. As
Romero (2008) observes, urban-rural divides persist even to the present day.
My field observations and the case studies of subsequent chapters indicate that
urban zones are experiencing greater security but rural areas are still neglected
and face resurgent criminal “bands.” It is under these conditions that civilians
seek to manage and adapt their organizations for autonomy.
The history reviewed in this chapter is helpful for understanding the empir-
ical chapters that follow. The history is first informative for interpreting what
juntas may do to help civilians deal with violence. It also suggests there may be
a relationship between organizations and the reporting of information on
indicators such as violence. Lastly, a deep understanding of long-term trends
and historical events can help disentangle circular relationships (e.g., between
social organization and violence) by pointing to how factors can be measured
before the circular relationships became entrenched. In the next chapter,
I elaborate on the methods I use to study questions of civilian autonomy and
how Colombian history guides the application of these methods.
4

Living to Tell About It

Research in Conflict Settings

atcc leader: Oye Gringo, do you hear that? Listen . . . They’re coming for you.
me: Who?
atcc leader: La Guerrilla! . . . (laughter)
– Heading upriver from La India, Santander, Colombia, 8/2009

One of the aims of this study is to bring methodological structure and rigor
to the question of civilian agency. In this chapter, I outline my multimethod
empirical strategy for studying the question of civilian autonomy. The chapter
serves as a guide for the next four empirical chapters. I start by describing the
broader research design of the study and how the different methods and pieces
of evidence fit together. I also discuss the different choices and trade-offs of
particular methods and their benefits and limitations. The design deals with the
issues of reverse causality between the impact of civilians and armed groups as
well as possible bias in case selection. I then describe the research process and
preview the data sources I collected during eleven months of fieldwork in
Colombia spread over four years.
The research design seeks to understand the central counterfactual question
of whether armed groups would have used more violence if not for civilian
autonomy strategies and the organizations that enable them. The empirical
work is useful for theory building, but its main purpose is to test clearly stated
and falsifiable implications of theory. The use of multiple methods in a subna-
tional study in a single (post-)conflict country is beneficial for providing deep
and comparative understanding. While no single test or method alone provides
conclusive answers for the outcomes being investigated, each method plays an
important role in inquiry. The combination of quantitative analysis, purposive
case selection, and fieldwork aims to push the methodological boundaries of
civil war studies. Taken together, the approaches tend to point toward the same
conclusions and paint a coherent picture of civilian behavior.

85
86 Living to Tell About It: Research in Conflict Settings

This research embodies an interplay between inductive theory building and

deductive theory testing. When I began this project, I was principally engaged in
building theories. I had little initial knowledge about the potential security con-
cerns that could have made this work infeasible or the details or effectiveness
of civilian organizational processes. I began my fieldwork interviewing a broad
cross-section of subjects from many communities and with diverse experiences
to gain background on civilian choices and formulate my research questions.
I then carried out a plausibility probe of the apparently successful case of the
ATCC. Within this community case there is theory building based on initial obser-
vations and initial validation – that some organizations effectively protect civilians
at least some of the time. I also found opportunities for theory testing based on
additional data collection and new units of analysis such as individuals, villages,
and events. From this initial research, I gained a sense of the variation in civilian
social cohesion, organizations, and strategies across formal autonomy organiza-
tions, such as “peace communities,” and learned about the role of junta councils.
With a theory in hand about organizations, juntas, autonomy, and armed
groups, I sought to generalize it to a broader universe of cases by collecting
necessary additional datasets (e.g., on organizations like juntas) and visiting
new cases in additional field trips. The additional case studies from Cundina-
marca are “out-of-sample” cases, which is to say they are unknown, unstudied
cases and not cases from which hypotheses were derived (though they are
“nested” among the large-n universe of cases). After tests with these new cases,
I again transitioned back to the inductive mode of research. Some inaccurate
predictions in these cases – for instance due to the effects of clientelism in
weakening organizations – suggest refinements to my theory. Such refinements
of theory can then be better tested in future research.

why the case of colombia?

Why study civilian autonomy in a single-country study and why study Colombia?
To begin, violence and social organizing are micro-behavioral phenomena that
occur at that level. The Colombian civil war also provides a good opportu-
nity to gain a systematic understanding of the choices of civilians since there
are widely varying experiences across local communities. A substantial number
of communities have overtly implemented autonomy strategies to resist vio-
lence and displacement, while others have either implemented more covert
processes or been forcibly displaced. In Colombia, like in other countries, there
can be drastic differences in social cooperation even from one town to the next,
with certain towns benefiting from idiosyncratic and socially beneficial histor-
ical events.1 Variable conflict conditions, levels of violence, and types of armed

1
Some examples of pockets of highly cooperative communities include Las Gaviotas in the
eastern plains (Romero 2009), the highly cooperative Afro-Colombian community of Sanquianga
(Cárdenas 2008), the Palenque of San Basilio (De Friedemann 1979), and the religiously founded
town of Jericó (Jericho; Otis 2010), to name but a few.
Why the Case of Colombia? 87

groups are also found across these towns. The Colombian case also has many
high-quality subnational datasets. As I later illustrate, this has payoffs as it
allows for sophisticated comparative methods to select cases for qualitative
study. Lastly, I chose to do research in Colombia because its increasing physical
security allowed broader access to case sites and because of my Spanish ability.
A potential concern with a single-country study is whether the context of the
Colombian conflict is unique, causing findings not to generalize to other set-
tings. For instance, a problem for inference could be that the phenomenon of
civilian autonomy in Colombia is broad enough to be studied there but not in
many other locations. Colombia is unique in some regards. For a middle-
income country, it has a relatively strong legal system. Its war has also been
relatively long-lasting, with at times relatively stable and slow-moving conflict
dynamics (though punctuated by periods of rapid change and intense fighting).
Yet, similar to other conflict-ridden parts of the world, Colombia is a
relatively capable state with weak reach into the periphery. The law is unevenly
enforced, if at all, in large parts of the country and illegal armed groups
dominate (O’Donnell 1999; Palacios 2006). Some of these areas have high
levels of poverty and receive few state services.2 Colombia is a nominal democ-
racy, however there is high inequality, clientelism, and corruption, especially in
the countryside. Colombia’s war is certainly not the only one that has econom-
ically motivated armed actors and illicit crop cultivation. Colombia may have
been an outlier in levels of foreign aid, but it is not clear if it has had relatively
great NGO involvement or international attention beyond the drug issue.
Comparative data helps put Colombia and its conflict in context. Sudarsky’s
(2007) analysis of the 1998 World Values Survey shows that 50.5 percent of
Colombian respondents were active members in voluntary organizations (when
religious organizations are excluded), putting Colombia in the lower half
(middle quintile) of a set of twenty-seven developed and developing countries
(and according to Inglehart et al. 1998, a Catholic tradition is correlated with
low interpersonal trust).3 This description is consistent with an uneven social
landscape with low overall “civicness” but also pockets of social cooperation.
Similarly, though comparing conflict intensity across countries can be messy,
Colombia’s conflict falls in the upper-middle range for countries in civil war. It
is ranked 86 out of 114 conflicts for total battle deaths and 68 out of 114
conflicts for total battle deaths per capita. However, after accounting for the
conflict’s long duration, it is lower for average annual battle deaths per capita,
ranked at 30 out of 114 conflicts (for conflicts from 1945–2002, based on Lacina
2006 data). This makes Colombia neither the least nor most intense of conflicts.
These data give reason to believe that Colombia is not such an outlier on
important variables that some of the findings here cannot generalize. Colombia

2
The impoverished Pacific department of Chocó is frequently categorized as one of the poorest
places in the Americas.
3
However, this comparison set may undersample developing countries where conflicts are prone to
occur.
88 Living to Tell About It: Research in Conflict Settings

naturally shares greater similarities with some country-conflicts than others, but
its generalizability is supported by the additional countries classified as having
autonomy actions in Chapter 1 and those studied in greater depth in Chapter 9.

the framework for subnational analysis

The analytical framework in this book allows for the subnational comparison
of hypotheses for outcomes of violence and social cohesion against alternative
explanations with both quantitative and qualitative methods. I combine a large-n
statistical overview with analysis within and across cases. The qualitative and
quantitative analyses both draw on a broad array of original and existing data,
and I triangulate data from multiple sources and perspectives on the same
concepts, such as violence. I use this data to test implications of theory across
time and space using several units of analysis, ranging from cross-national
examples down to individual events of threats and violence. Topics I consider
include: How do individuals view the conflict? How do villages behave? Which
municipios are least likely to suffer violence? How do different armed groups
and fronts behave?
The different methods have different advantages, with some able to speak
better to particular outcomes than others. The large-n methods alone can suffer
from the problems of missing data, insufficient control variables, omitted
variables, and measurement error. The quantitative methods are most useful
for establishing correlations between relatively easily measurable concepts such
as social organization and violence, testing for conditional effects, and sampling
on a broad distribution of cases. However, with limited availability of this
historical data, the methods are not very suitable for measuring or testing
civilian strategies since they are often hidden, change rapidly, or are difficult
to categorize. Quantitative methods are challenged to distinguish cooperative
autonomy mechanisms from alliance or fleeing strategies.
Qualitative methods alone can be subject to problems of selection bias, inter-
pretation, forgetting, and replicability. The case studies are useful for the process
tracing of decision points, mechanisms, accurate measurement, and identifying
idiosyncratic details not easily observable from a distance or across many cases.
The case studies can give faithful accounting of the details of particular commu-
nities, but can only be more broadly representative if they are carefully selected.
Analysis of only few cases may also not be helpful for assessing causality if there
are many potential causal variables. In sum, any one of these methods alone
would provide an incomplete picture of behavior. Next I explain how the
methods are joined to deal with selection issues and trace decision-making.

The Choice to Use Historical Data

This study is based on historical sources of data and qualitative fieldwork. A
historical approach is taken since social landscapes are the product of long-term
The Framework for Subnational Analysis 89

historical trends and events. Further, with the concerns surrounding selection
bias and reverse causality, a long view of history helps incorporate social
landscapes prior to the occurrence of violence. This book encompasses over
sixty years of Colombian history, focusing on 1990–2005 but going back to the
1940s. As with all historical studies, there can be challenges in collecting data
on certain events long after they have occurred. With the passage of time,
memories can lapse and relevant people and sources can disperse.4
Studies undertaken in the midst of conflict can avoid some of these problems
but collecting reliable data during conflict entails other challenges and trade-
offs. Such studies are usually best able to describe conditions and generate
theory. They are challenged at collecting candid data from subjects, testing
theory with methodically selected cases, and porting hypotheses to new samples
or units of analysis.5 Such insights are valuable but can be subject to the
conundrum of testing the generalizability of findings, where it is difficult to
gauge their reliability and how far they reach. To balance these concerns, I
opted to conduct research in areas that had recently emerged from conflict.
An alternative method I do not use is surveys because I felt it was important
to measure variables before the start (expansion) of conflict, rather than con-
temporaneously. Surveys can also be too blunt for understanding the details of
complex strategic interactions. Even in post-conflict, it seemed participants
would not have enough confidence to reveal their true preferences, decisions,
histories, or reasoning. Furthermore, the sensitivity of some questions and the
unpredictable security situation across the country could have impeded random
sampling or sampling enough units.
To test the effects of juntas and other organizational variables, I analyze
observational data on violence by armed groups from 1990 through 2005.
Beyond reasons of data availability, these years are also the most relevant
time period for study of the emergence and impact of autonomy organiza-
tions since many were formed during or immediately prior to this period, and
there are also examples of junta activity, renewal, and increasing self-govern-
ance. This era in Colombia also saw an apex of violence and combat that
affected much of the country. It encompasses the height of paramilitary
violence with the AUC expansion and the implementation of the Plan Patri-
ota counterinsurgency campaign. The later years then saw a de-escalation
beginning around 2003 with the gradual repulsion of guerrillas and the
beginning of the demobilization of paramilitary blocs. Some of the paramili-
taries then quietly remobilized and new “emergent” bands began to appear
(CNRR 2007). These years are a tough test but are also when we might
detect an effect of juntas on violence.

4
However, researchers may be able to access clusters of relevant subjects in refugee camps or
demobilization centers.
5
Field experiments, which are growing in use and popularity, can mitigate some of these issues but
also have ethical considerations.
90 Living to Tell About It: Research in Conflict Settings

Joining Methods to Deal with Selection Bias Issues

Understanding how cases are selected for study is important for gauging exter-
nal validity and making inferences about other contexts. Bias from selecting on
the dependent variable (violence) could arise from looking at well-known,
organized communities because the independent variable (organization) could
be correlated with violence in unknown ways. Social organizations could, for
instance, exist in more violent areas (or not), which would bias against observ-
ing an effect on violence. Alternatively, organized communities could have
initially faced less violence than other communities, possibly contributing to
their very survival and the later emergence of formal autonomy organizations
(with prior violence being correlated with future violence). This encapsulates
the related problem of reverse causality, where the effects of conflict could
weaken or destroy social organization, causing social organizations to survive
and persist only in relatively peaceful places.
The options for selecting cases can be limited for several reasons. Especially
in conflict settings, our attention can be drawn to noteworthy cases, while wide-
ranging fieldwork is costly. In developing countries, where conflicts tend to
occur, there is also usually poor historical and quantitative data to more
broadly sample additional cases (though this is improving). A second issue
is selection bias from limits on researcher access to sites. At the time of the
fieldwork for this study, Colombia was not a post-conflict country, but could
rather have been classified as a “conflict” country. A main limitation in field
research is the risks of entering conflict and guerrilla-controlled zones.
I use a number of strategies to deal with selection and reverse causality
issues. The power of analyzing many cases (large-n) is the overview it provides
to understand where particular cases fit within the broader universe of cases.
I use standardized units of analysis by looking at violence and the coverage
of juntas and other organizations at the municipio level, with over 1,000
municipios in Colombia. My complete sample permits the inclusion of the full
range of variation on the dependent variables. It also helps construct valid
counterfactuals to compare cases of organized, “resisting” community cases
with “nontreated” (unorganized, no autonomy strategy) cases – the “dogs that
did not bark.” I do this by controlling for potential confounding factors (e.g.,
conflict dynamics) that might cause any observed relationships between social
organization, civilian strategies, and violence to be spurious.6
Data for these variables come from numerous micro-level governmental and
private sector sources that provide multiple measures of social cohesion and
organization and violence. Though it can be hard to come by historical data on
social indicators in developing countries going back much before 1990, I was

6
I attempt to control for a variety of preexisting characteristics of municipios. The inclusion of
time-invariant historical measures of different organizations precludes the use of fixed-effects to
deal with unobserved heterogeneity.
The Framework for Subnational Analysis 91

able to assemble data on independent variables from prior to the spike in

conflict around that time (indeed, I found several studies and censuses had
lamentably been lost or destroyed!). For example, data on the junta councils
comes from a rare dataset published by the Colombian census bureau along
with qualitative reports from interviews I conducted with members of the juntas
movement. Linking data on historical conflict from La Violencia to the variables
on the subsequent growth of organizations helps address reverse causality con-
cerns by at least accounting for the distribution of junta councils across towns.
My case selection methods join the quantitative and qualitative approaches
of the study. The narratives of the selected cases are a key element of the book.
Like some other studies, they use structured, focused comparisons (George
1979) to trace decisions and outcomes, meaning all of the cases are studied
with similar (causal) structures and data collection procedures. But, in contrast
to other narratives on civil war experiences, this study’s narratives try to do
something different. The narratives, based on difficult and potentially danger-
ous fieldwork, are embedded in a larger quantitative overview.
In Chapter 6, I use the quantitative data to identify neighbor towns that are
matched on relevant characteristics (covariates) that might contribute to violence,
including conflict dynamics, but have differing levels of organization. Within these
quasi-experimental comparisons, any differences in outcomes should be attribut-
able to the variation in the difference in the “treatment” of social organization. In
other words, at least within a zone, key differences between cases should not be
attributable to broader regional historical factors or conflict conditions because
they are largely the same. In this way, the matching is an additional solution for
avoiding spurious inferences due to reverse causality. There are clear expectations
about how these different communities should have behaved during the conflict,
and the qualitative variables I collect on those cases can be sensibly compared with
the indicators from the large-n datasets. The close link between theory and
empirics allows for the identification and systematization of new, unknown
strategies and social processes from previously overlooked cases.
I selected case sites in recent post-conflict areas both for safety and because in
the midst of conflict, interview respondents would be harder to recruit and
perhaps less candid. Gaining access to combatant perspectives would also be more
difficult. The zones in Santander and Cundinamarca had guerrilla presence until
fairly recently, but were safely controlled through government actions to regain
territory under Plan Patriota by the time I arrived.7 While visiting areas of past

7
There is a risk of bias from not being able to access some long-term guerrilla strongholds that had
not yet been secured by the military. However, several of the zones I study had guerrilla presence
for decades and my quasi-experimental design helps minimize this concern (as does the analysis of
FARClandia in Chapter 9). Additional case studies of other communities as they become safer in
the future can further assuage concerns of selection bias based on safety criteria. The fact that the
sites I visited were safe enough to visit at the time may set these cases apart from other zones that
continued to have conflict, but there is little else I could do about this as a researcher.
92 Living to Tell About It: Research in Conflict Settings

rather than ongoing conflict may introduce possible selection bias due to limits on
where a researcher can safely venture, the zones I study were not quite beacons of
tranquility and some persisted with low-level conflict and violence. I interviewed
many subjects who were currently threatened or had been previously, as well as
previous kidnap victims. In some of the zones I visited, people were killed before,
during, and after my visits (by either the military or criminal bands). There were
also written threats in the form of pamphlets from armed bands. Nevertheless,
I strove to minimize this potential form of bias by talking with civilians from zones
with ongoing conflict in safer sites as well as interviewing combatants.8
This research design and case selection process is helpful for structuring
information for later analysis, but it made the process of data collection
relatively difficult because information is easier to come by in some cases than
others. Prior scholarship that has examined accessible cases has suffered from
case-selection bias: well-known peace organizations are well known because
they are good at managing and disseminating information and data. This is the
case with the ATCC, where the rich sources of historical information available
attract scholarly attention. However, in the ATCC neighbor areas and the other
case-study sites there were no local formal civilian or human rights organiza-
tions that consistently monitored the consequences of the conflict, and the
civilian organizational responses to deal with those consequences were also
subtler. The result was much less available data. Studying the full spectrum of
cases thus called for modified data collection procedures for the “unorganized”
portion of my sample. I return to this issue later in the discussion of interview
procedures and in Chapter 8. Compared to the data-rich conflict environment
and formal process of the ATCC, studying the other towns also meant reduced
expectations for the detail of analysis that could be undertaken.
The case matching research design also made the task of identifying potential
differences between towns through fieldwork even harder. If I employed my
matching techniques well, and by virtue of the selected towns being neighbors,
they should in many ways not be that different (putting my measure of junta
councils to a true test!). Furthermore, because of how the cases were chosen, as
a researcher I had little background knowledge of these towns and had virtually
no acquaintances or research contacts before I selected them and arrived in the
zones. While not knowing what to expect made it all a surprise and very
exciting (and at times nerve-wracking), it also meant networking to get infor-
mation would be slower and created an even greater need for creative and
precise data collection tools. For instance, I had to devise some new, compar-
able qualitative measures of cohesion and violence, such as counts of machete
fights, which are described in Chapter 8.

8
It is conceivable that respondents that left these zones could have distinct views (compared to an
average remaining resident) related to the reasons why they decided to leave. There is not much a
researcher can do to mitigate this concern other than to try to include respondents with a variety
of reasons for their departures (personal security, economic prospects, family reasons, etc.).
The Case Studies and Fieldwork: “To the Villages!” 93

the case studies and fieldwork: “to the villages!”

I visited and interviewed subjects from many locations, but the majority of
my qualitative evidence focuses on five core case “towns”: the ATCC, its
neighbor villages, and the Cundinamarca towns of Bituima, Quipile, and
Vianí.9 I began my fieldwork for the project in 2007 and made multiple short
trips to Colombia through 2009 (with follow up visits in 2011 and 2013, and
work in the Philippines in 2012). The daily lives and social relations of some
of these communities are astonishingly similar to the caricatures of Gabriel
García Márquez’s novels – small and insular, and with many interwoven fami-
lial relationships (though as García Márquez’s novels also attest, this does not
necessarily make them cooperative or amiable). People meet at the village store,
or sit around in plastic chairs and chitchat and drink (many!) beers, perhaps
while playing games of tejo.10 Farmers help neighbors harvest coffee or sugar-
cane. Women wash clothes in the stream or nurse their babies together. Coffee
growers come to the town center for meetings, and some residents come for
Sunday mass. Many residents are related to each other as some kind of cousin
or in-law. With the slower pace of life, information and gossip about others and
about newcomers abounds.
My research involved work in cities but mostly involved extensive work in
the campo, or countryside – in town centers, in corregimientos (rural sectors),
in small villages, and at lone houses. My contacts and guides were enthusias-
tic and adamant that I go “a la vereda” or “to the village” to get the true
perspective of the campesinos, and I found that getting the views from different
parts of these municipios was invaluable. For many people, taking me to the
village was a point of pride: a place they own, where few outsiders go, with
guarded secrets, and a sacred, serene, tranquil lifestyle compared to the bustle
of the city. These visits served both to find examples of collective action for
autonomy from the “affirmative” (autonomous) towns and, perhaps more
crucially, to verify that there were few or no collective actions from the
“negative” towns, such as Quipile, which required more extensive visits.
Reaching these sites involved many and only occasionally comfortable modes
of transportation: planes, taxis, buses, chivas (country bus), jeeps (actually,
Willys, as Colombians refer to them), motorcycles, bestias (“beast,” or mule),
lanchas (motor canoe), and unbelievably steep mountainous hikes in panta-
neras (rubber mud boots) on what some might call trails.
An ongoing concern was how a “gringo” like myself could safely access
these regions. In my fieldwork preparations, I would analyze the security situa-
tion with existing data and news reports and identify at least one local contact.
I would then ask the questions of whether I could get there safely (i.e., will I get

9
I say “towns” in quotation marks since these rural areas include villages that surround the town
centers.
10
Tejo is a popular game in Colombia in which players score points by tossing iron discs at packets
of gunpowder that are set in clay basins and explode when hit.
94 Living to Tell About It: Research in Conflict Settings

Campesinos playing a game of tejo in a village in the ATCC region, 2008. The game
involves tossing metal discs at small packets of gunpowder.
Photograph by Oliver Kaplan.

kidnapped right away?) and whether people are likely to talk to me. I was
fortunate to be accompanied in most of my travels by church leaders and
priests, NGO staff members, or local residents. For security reasons, I did not
stay in zones for long periods at a time (more than three weeks or so) because
word travels fast and, before long, everyone came to know there was an
American in town. In my research process I would visit, then leave to reflect
and analyze, and then repeat. This reality may have limited access to infor-
mation and understanding of especially deep social relationships. However, on
follow-up visits I would return with more focused research questions and
broader comparative lenses through which to view the communities.
Even with my precautions, I still encountered some complicated situations
that required spur-of-the-moment judgments. In one instance, I was leaving La
India at the end of a field visit and my acquaintances helped arrange a ride for
the stretch of dirt road to Cimitarra in a friend’s car, from where I would catch
the night bus back to Bogotá. It sounded faster and better than the last bumpy
school bus of the day, so I waited for the car to arrive. When it came at dusk, a
gregarious man got out and said hello and shook everyone’s hands. Yet I
opened the passenger door only to see a revolver lying on the seat. A series of
panicked thoughts raced through my mind: Who is this guy!? And why does he
The Case Studies and Fieldwork: “To the Villages!” 95

A motorcanoe (lancha) heading upriver from La India, Santander, Colombia on the

Carare River, 2007.
Photograph by Oliver Kaplan.

have a gun!? Should I get in, or stay another night!? I looked to the others, who
did not seem concerned, and not wanting to offend, I anxiously got in and the
driver tossed the gun to the back seat. Once moving, I casually asked him about
it, and he said he had it because he was a city councilman and, as I had been
warned, robberies on the road were common after dark. His answers were
hardly reassuring, but we thankfully made the journey without incident.
Although planning is key, one can only prepare so much for these dilemmas.

Interview Techniques
As part of my fieldwork I interviewed more than 200 people. Most were
civilians from the core case-study towns as well as residents or former residents
from many more communities in both rural and urban settings.11 I ended up
interviewing between fifteen and forty-five residents in each town. The civili-
ans comprised a broad cross-section of individuals, including campesinos,

11
Some but not all of these interviews are incorporated into this text. Some were conducted for
background purposes to understand the variety of Colombians’ general experiences with social
organization and conflict.
96 Living to Tell About It: Research in Conflict Settings

A bus ride in eastern Antioquia, Colombia.

Photograph by Oliver Kaplan.

Indigenous group members, Afro-Colombians, women and men, and youths

(over 18 years old). These people often had various roles: farmers, business
owners in the town centers or corregimientos, leaders of formal organizations,
local junta leaders, national officers of the acción comunal (junta) movement,
and even former coca growers. There were also victims of various kinds of
violence: torture, kidnapping, forced displacement (some of the displaced per-
sons had lost everything and were living day-to-day), widows and people who
lost family members in the conflict, and even people who were under death
threats at the time. I also interviewed members and regular participants in
community organizations and juntas as well as nonparticipants in communities
with formal organizations or juntas.
Beyond regular community residents, I also interviewed “elite” subjects as
diverse as current and former mayors, city councilmen, municipal officials
(Human Rights Ombudsmen and Defenders – Personeros and Defensores),
members of churches, NGO workers, government officials, and current and
former police and military officers, as well as former guerrillas and paramilitar-
ies. Community leaders and elites were invaluable for gaining historical over-
views of the communities. It was also helpful to contrast elite responses with
nonelite responses to see how daily life and experiences from the villages at
times differed from “official,” more central accounts.
The Case Studies and Fieldwork: “To the Villages!” 97

A chiva bus in Pensilvania, Caldas, Colombia, 2009.

Photograph by Oliver Kaplan.

Interview subjects were recruited in several different ways in hopes of

obtaining broad and representative viewpoints. First, I would “snowball,”
asking one person to recommend other people to talk with who might be
knowledgeable about a certain topic, hold a different opinion on a given
subject, or have been involved in community affairs at a different period in
time. The right introduction was essential for entering a community, arranging
initial interviews, and gaining trust.12 Prior planning and networking was
crucial, as these introductions were only possible with the help of many people,
including representatives of organizations such as the Catholic Church, NGOs,
and fortunate acquaintances I made in Bogotá. At times, gaining trust was
additionally aided by participating in community events, joining in soccer
games (though playing poorly), or drinking unfortunate amounts of liquor
and beer to become “part of the tribe.”
Second, in some instances, quasi-random interviews were possible. For
instance, on market day I was able to speak with villagers who by chance came
into town and ate at a restaurant where I was loitering. At various large,

12
I had been told before I began fieldwork that, being an American, people might think I was a CIA
agent. In at least one instance, this concern was dispelled with irony when I saw a teen walking
down a village lane with a “CIA” ball cap on one of my early field visits.
98 Living to Tell About It: Research in Conflict Settings

community-wide meetings I was able to speak with people who happened to

have had time to stay afterward. Sometimes I was able to meet the guys who
happened to be having drinks at the corner store.
All interviews were conducted in Spanish by me and were anonymous,
which is one reason why respondent descriptions are vague. As a researcher, I
sought to present myself as objective and impartial. I was there to understand
histories and protection processes, but not to make value judgments regarding
other end goals, particular sides in disputes or conflicts, or people’s past
behaviors. Most interviews were conducted in the private settings of a person’s
home or an NGO office, though if a person consented and felt comfortable, the
interview sometimes took place at a more public place like a café or a corner
store. Interviews varied in length, with some lasting only thirty minutes and
others involving multiple sessions lasting hours. While most interviews were
with lone individuals, some interviews were with groups of subjects who
consented to collective discussions. All interview notes were written by hand
and no recording devices were used. In addition to helping protect the identities
of subjects, this technique also helped make subjects feel more comfortable.
This meant writing in shorthand, though I would write out important quotes
word for word and ask subjects to repeat if necessary.
Participants in interviews were almost always excited to share their stories
with me. Whether they were tales of desperation or inspiration, most people
wanted to be heard and have a voice. Some subjects were fearful or had been
previously traumatized and spoke in hushed tones, and conversations required
extra empathy and calm for comfort. Some potential subjects declined to
participate. Other individuals were ebullient and, even if it could have been
risky to share certain details with me, were adamant that I report every detail as
a testament to their community’s strength, what they suffered or endured, or
the truth about atrocities and victimizers. Some were also adamant that I
attribute the stories to them by name (which I of course could not and did
not do). Despite the possibility of censoring out of fear or danger, I found that
people were surprisingly open, perhaps because I was a foreigner, or perhaps
because of appropriate introductions by trusted mutual acquaintances.
The interviews were semi-structured. I began my research and interviews
with a loose plan of standard interview questions and topics of interest, seeking
information that would be helpful for qualitatively understanding key variables
and trends. At the same time, I let conversations evolve depending on the
insights provided by the subjects.13 To minimize danger or potential trauma
to subjects, I asked general questions about communities and trends rather than

13
In gaining consent and explaining the purpose of my project, I described my interests in general
terms to avoid guiding subjects’ answers from the start. I told them I was interested in social
relationships in the midst of conflict settings and how different towns experienced the years of
conflict.
The Case Studies and Fieldwork: “To the Villages!” 99

questions about specific or personal events. Subjects who felt comfortable

would at times volunteer more personal information or anecdotes. In some
cases, I introduced hypothetical scenarios and asked subjects to respond about
what would have happened or what people would have done in X or Y
situation. Some questions were tailored for each case based on the local history
and context. As my thinking and knowledge as a researcher evolved and grew,
I adapted my interview approach to ask about relevant and comparable
phenomena from town to town to assemble qualitative measures such as
the prevalence of macheteras (machete fights), bazaars (fairs), and responses
to cattle theft, which I did not know to ask about when my fieldwork began.
In discussions of social organizations I invited subjects to comment on both
positive and negative aspects.
I faced two principal challenges in grounding interviews for useful compari-
sons across space and over time. First, there was the risk that interview subjects
might fixate on more current and accessible ex post conceptualizations of social
conditions and explanations for violence. While I was interested in information
on outcomes, what I really needed were historical assessments of pre-conflict
conditions and independent variables. It was also hard for some civilian sub-
jects to clearly remember details of brief events that occurred in times of stress
five years prior, ten years prior, or even earlier.
To deal with these issues I used two different interview techniques. First,
I asked subjects to think about different periods of time by prompting them not
only with years, but also by grounding questions with reference to specific
national events, local events, and personal events. Second, instead of interview-
ing only current long-time residents, I also interviewed people who migrated to
towns at specific times in the past, people who left towns at specific times (and
were living elsewhere), or returnees who moved away from regions at specific
times and maybe only recently moved back, after the conflict had subsided.
These kinds of respondents were helpful for pinpointing information at specific
moments in time and describing pre-conflict characteristics since time in the
town effectively stopped for them when they left or began when they arrived.
Finally, interviews with certain elderly residents provided descriptions of La
Violencia based on their own recollections or from stories told to them by
their parents.
A second challenge of cross-town comparison stems from the inability to
visit many different research sites given the rough terrain, poor roads, and
potential safety concerns in parts of rural Colombia. I was able to overcome
these limitations with some craftiness in interview strategies. I first maximized
geographical coverage of interview subjects by leveraging interviews with
people at meetings they would attend in town centers or Bogotá. For instance,
I interviewed victims of forced displacement in cities that had come from a wide
variety of rural towns across all parts of the country. In cities and towns I was
also able to interview people who came from dangerous settings that I might
not have been able to reach easily.
100 Living to Tell About It: Research in Conflict Settings

In addition to the challenge of gaining geographical breadth, I was also con-

cerned about the comparability (“anchoring”) of observations and interview
responses across towns and villages. For instance, even if people from many
different villages or towns were to get together in the same room, they might
not have a basis for comparing or ordering their communities on dimensions of
interest to a social scientist because the community each person knows best is
his or her own. By visiting different towns I was able to make some of my own
comparative judgments, but it can also be challenging for an outsider to make
accurate comparisons, especially with limited mobility and time and arriving in
the post-conflict period after causal processes of interest have occurred.
As a solution, I sought out peripatetic individuals who had broad geographic
familiarity from traveling across or living in different villages or towns. These
key respondents included people who for business or other reasons had traveled
extensively within the broader region surrounding their community, such as
merchants, bill collectors, agronomists, drivers, priests, etc. As I elaborate on
later, ex-combatants were also often more mobile than the civilian population
and could provide further insight into cross-town differences. I also encoun-
tered people who had recently moved to a case-study town from one of the
neighboring case towns and so had familiarity with both. Lastly, some archival
documents contain dialogues with residents of multiple villages.
The interviews were varied and yielded mixed though generally insightful
accounts. Some people were minimally informed about local history or had
little experience with conflict and social organization and could only provide
basic background information. Other respondents were keen students of social
organizing, local politics, and civil war dynamics. Some were enthusiastic about
community processes and organizations, while others dissented and were embit-
tered, perhaps because they felt excluded from a process or program, disagreed
with decisions, or had personal issues with leaders, among other reasons.
As a consequence of the larger research design, interview recruitment and
information-gathering processes differed slightly between communities with
formal organizations like La India with the ATCC and other towns with no
formal organizations or NGOs. These latter towns did not have the same kind
of centralized collective memory or central point of contact from which to
network. Interviews and data collection were therefore more scattershot.14
Further, since violence suffered by civilians in areas with no organization (or
few juntas) can be prone to inaccurate reporting and a relative underreporting

14
The effects of socialization and experience were also evident in formally organized La India.
ATCC respondents were more self-conscious of social and conflict processes than those from
Cundinamarca, having pondered many explanations for their lot. In the Cundinamarca towns it
was less common that any single person would directly attribute broad and abstract factors such
as the unity or division of populations as an explanation for different outcomes. Perhaps this was
because they tended to only hear about individual events of violence and were less exposed to
collective discussions, news, ideas, and narratives.
The Case Studies and Fieldwork: “To the Villages!” 101

A game of mini-soccer (micro-fútbol) in La India, Santander, Colombia.

Photograph by Oliver Kaplan.

bias, enhanced qualitative measurement techniques were required for compar-

ability.15 The potential existence of subtle protection mechanisms in the non-
formally organized cases also created the need for adapted techniques to discern
their effects. Inferences from areas without formal organizations are still pos-
sible, but less complete reports make it harder to construct relevant “counter-
factual” cases to understand how social variation relates to threats and killings.

Interviews with Ex-Combatants

Discussions with ex-combatants contribute a valuable additional perspec-
tive. Interviews with ex-combatants involved slightly different procedures
and techniques (and bias concerns) and merit a separate discussion. I fielded
twenty-eight total interviews with ex-combatants, with three-quarters from
guerrilla groups and one-quarter from paramilitary groups, and including three

15
I rely upon standard indices of violence in the large-n analysis with the hope that, if there are
similar reporting biases across municipios, a large enough sample will average them out and that
control variables can account for them. However, in any particular handful of cases, there are
greater risks of mismeasurement, and police statistics and press reports could be inaccurate if
residents do not report incidents out of fear.
102 Living to Tell About It: Research in Conflict Settings

Loading coffee in Pensilvania, Caldas, Colombia, 2009.

Photograph by Oliver Kaplan.

female participants.16 These were the later interviews I conducted, in August

2009, having already selected and visited case zones and heard the perspectives
of civilian residents.
I was aided in the recruitment of subjects by the Colombian government’s
High Advisory for Reintegration (ACR; now the Colombian Agency for
Reintegration), which helped me identify and schedule interviews with geo-
graphically relevant ex-combatants – individuals that operated in or around
my case-study areas – that dispersed after they left the conflict and were living
in the capital of Bogotá. The majority of these interviews were conducted in
ACR field service centers and were done so voluntarily, anonymously, and
in private.17 While the setting could have affected the subjects’ openness or
candor, the responses seemed open and truthful. For instance, the location
of the interviews did not keep subjects from giving detailed accounts of
interactions with civilians, recalling gruesome atrocities, or criticizing the
demobilization program.

16
There were two additional interviews that were incoherent and not used. Approximately eight
other subjects were invited to participate but did not show up for interviews.
17
I was certain to emphasize to subjects that their participation was voluntary and was in no way
related to their receipt of government benefits.
The Case Studies and Fieldwork: “To the Villages!” 103

Obtaining ex-combatant perspectives was prioritized over military and

police perspectives because these latter accounts tended to be less relevant or
informative. The case-study zones were conflict zones in a weak state and, even
with the state forces’ periodic patrols, there tended to be little permanent state
presence and inconsistent state contact with the campesinos of the far-flung
villages of the countryside. It is also difficult to identify military and police
personnel that served in a particular zone at a particular time in the past
because troops rotate in and out of zones as part of their tours of duty. Despite
these considerations I was able to obtain several accounts from current and
retired military officers.
The transience of the ex-combatant population combined with the ACR’s
limited information in its individual profiles meant I could only geographically
target subjects with moderate precision. The armed group fronts, blocs, and
zones represented in the sample are summarized in Table 4.1. I was fortunately
able to track down ex-combatants that had operated in my case-study areas
even though it had sometimes been from five to ten years since they had
demobilized. Two-thirds of the participants came from fronts or blocs that
were involved in my case-study regions. The rest operated in other parts of
Colombia but still provided valuable comparative insight.
The group of subjects comprises those people who were enrolled in the
ACR’s reintegration program and showed up at the service centers to partici-
pate in the study. This obviously excludes combatants who had not yet
demobilized, been killed, or ex-combatant no-shows.18 Nevertheless, the
sample still includes a range of subjects in terms of their experiences in the
conflict and attitudes toward their former patron armed groups. Subjects
include individuals who never wanted to be in the conflict as well as individuals
for whom it was their life (i.e., defectors as well as captures).
The quasi-random recruitment of ex-combatant subjects yielded a variety of
interviewees according to their roles and ranks within the armed groups and
their experiences with civilian populations. The participants ran the gamut
from informants, to “rasos” or foot soldiers, to middle commanders. Some
were merely collaborators who did not wear uniforms and kept their places as
civilians living in their communities. Others were soldiers whose principal work
was to serve as interlocutors with civilians and organize meetings with them.
Still others spent almost all their time camped in the mountains or jungle
seeking out enemy forces and had almost no contact with civilian populations.
Some subjects had experiences in several different fronts or blocs as they
rotated across different regions of the country. This was especially true among

18
Ex-guerrillas and paramilitaries arrived in the demobilization program through different pro-
cesses. Guerrillas were frequently captured or fled and turned themselves in to state authorities.
Most ex-paramilitaries participated in the formal “complete” demobilization of paramilitary
blocs beginning in 2003 (for information on this process and the number of demobilized from
each bloc see Oficina del Alto Comisionado para la Paz 2006).
104 Living to Tell About It: Research in Conflict Settings

table 4.1 Characteristics of ex-combatant interview subjects

Guerrilla Front (F) or No. of Departamentos/Municipios From case-

AUC Bloc (B) subjects where operated in study zone?
F42 8 Cundinamarca: Viotá, Pulí, Yes
Quipile, Vianí, Bituima
F22 2 Cundinamarca: Quipile, Vianí, Yes
Bituima
B Omar Isaza 1 Caldas: Pensilvania, Samaná Yes
F47 4 Caldas: Samaná Yes
Antioquia: Nariño
F9 2 Caldas: Samaná Yes
B AUC Puerto Boyacá 2 Santander: San Vicente de Yes
Chucurí
Boyacá: Puerto Boyacá
F27/ F Antonio 1 Meta No
Navarro
F Urbano 1 Bogotá/ various fronts No
AUC/ ACB/ ACCU 2 Casanare No
BCB 2 Bolívar No
Nariño: Tumaco
F22 (urban) 1 Bogotá No
F24 1 Santander: Barrancabermeja No
ELN 1 Nariño No

former FARC guerrillas. In the Cundinamarca case towns, the bulk of infor-
mation from ex-combatant interviews is from former guerrillas. Unfortunately,
little is known about the paramilitary groups that arrived there later because
they were more clandestine and informal, short-lived, and did not formally
demobilize. Instead, they disintegrated and disappeared.
The ex-combatant interviews followed a more structured progression. I
began most of these interviews by asking general questions to make the subjects
feel comfortable, gain their confidence, and get to know their background.19
What are they doing now? How do they feel about the demobilization pro-
gram? Where are they from? How long ago did they enter the reintegration
program?20 I then asked them general questions about their experiences:

19
More than in the civilian interviews, the ex-combatants preferred not to talk much about their
personal experiences, either because they preferred not to look back on their former lives or
because they feared possible stigma for their actions. For these reasons and for my interest in
general information about particular groups and blocs I had no reason or interest to ask
questions about personal behavior.
20
Analysis of these comments was expanded upon and published in two related studies (Kaplan
and Nussio 2015, 2016).
The Case Studies and Fieldwork: “To the Villages!” 105

A banner from the Colombian Agency for Reintegration (ACR) promoting the
reintegration of ex-combatants (“We build peace from the countryside”), 2016.
Photograph by Oliver Kaplan.
106 Living to Tell About It: Research in Conflict Settings

Which front or bloc were they in? For how long? Where did they mainly
operate? What was their role in the group?
Some subjects were more open to discussing these themes than others. Of
course, I had expectations about how combatants in different areas would view
different towns based on my fieldwork, but to avoid tipping them off about
my interests or hunches, I would begin with general queries and then probe
progressively closer toward my topic of interest. Depending on their cogency
and willingness to discuss prior questions, I asked more substantive questions,
phrased somewhat open-endedly to encourage free association: Did they notice
differences between villages or towns? Was there a lot of combat or pressure
from other groups? What was the group’s attitude toward using violence?
I would raise the topic of civilian resistance toward the end of interviews if
it did not come up previously: Did civilians ever resist their group’s control?
Did they act collectively? What did they tell the armed group? How did the
group respond? Why?21 I would sometimes conclude with some hypothetical
questions such as: What civilian strategies did they think were best from the
civilians’ point of view?
The inclusion of ex-combatants enriches the study by highlighting their
groups’ reasons for using violence and transgressing against communities,
and other themes. They add textured descriptions of the intensity of conflict
to the rough quantitative indicators of armed group activities. Through their
mobility the ex-combatants are also able to provide comparative perspectives
across towns on civilian characteristics that can be hard to obtain from the
mostly stationary civilians. Some ex-combatants were also formerly junta
leaders prior to being soldiers and so were able to see issues from both points
of view. For some, being in an armed group seems to have reinforced the
importance of juntas for protection. Interviews of ex-combatants from areas
beyond my case-study zones reveal additional examples of apparent move-
ments for autonomy from armed groups at the village level. Lastly, it was
“interesting” to converse with these subjects (especially former guerrillas) about
how they viewed gringos.
As valuable as these ex-combatant interviews are, extracting useful infor-
mation is not without pitfalls. First, the accuracy of ex-combatants’ memories
was variable since the subjects I interviewed had demobilized between two and
twelve years prior to the interviews. I was asking them to recall events from five
to fifteen years ago when they were living different lives. Second, it was compli-
cated to temporally and spatially piece together different accounts. Since fronts
can cover several municipios, even if subjects were contemporaries from the
same front they may not have operated in the same place or even known each

21
I also asked clarification questions during the interviews to confirm whether the subjects and I
had mutual understandings of what they were reporting regarding civilian autonomy and
advocacy. This was a relative advantage of the interview technique compared with surveys.
The Case Studies and Fieldwork: “To the Villages!” 107

other. Further, some units were mobile and moved across wide swathes of
terrain, causing trouble in pinpointing where individuals saw or remembered
certain events or trends. This was a larger concern for ex-guerrillas, since many
were transferred between different fronts during their careers. Third, interview-
ees had varying capacities to recall their group’s interactions with civilians,
either because they were less exposed to, less sensitive to, or less observant of
civilian issues. Fourth, in a similar vein, the subjects’ variable positions in the
groups meant some ex-combatants had greater exposure to group leaders
and decision-making processes than others. Fifth, ex-combatant reports could
suffer from confirmation bias. They might either be inclined to confirm ex post
that they respected civilians or that civilians had agency to whitewash the
darker aspects of their group. Subjects could also dissemble and tell half-
truths to confirm what they expected I, the researcher, wanted to hear (social
desirability bias).
Ex-combatants’ ex post justifications for their actions can be unreliable. Like
any such reflections, they are not complete, definitive accounts of decision-
making and require further confirmation. Still, there are also reasons to believe
in the validity of these accounts. Many subjects were able to recall certain
autonomy events in great detail. Further, they did not solely recall civilians
advocating for their rights but instead also distinguished many regions where
civilians were subjugated. Some accounts are also corroborated by civilians’
stories, archival documents, and secondary sources with interviews conducted
by other researchers studying other questions. In general, when I encounter
ambiguity or contrasting accounts – about levels of control, attacks, violence,
and explanations of behavior and decision-making – I try to show the different
versions and indicate levels of uncertainty and doubt. This transparency
includes examples from their statements that do not support my theory.

Participant Observation
Participant observation during visits to communities was also an important
part of understanding social differences across towns. I spent many months
in small towns and gained insight into the daily lives of residents. I was also
able to attend specific events and meetings. I observed village, junta, and
cooperative meetings and saw the struggles and successes of various decision-
making processes. I saw trainings of junta leaders. I saw how villagers interact
with NGOs and international organizations. I observed protest marches and
funerals, and attended church services. It was often informative just to see who
would and would not show up to these meetings and events – who and how
many people would seek assistance from village leaders, organization leaders,
or mayors? Who would go out to work in the fields? I could also observe
features such as how the local economy functioned, how much common
knowledge and communication there was between residents, and how close
together the houses were.
108 Living to Tell About It: Research in Conflict Settings

Participant observation in the post-conflict settings I studied from 2007–

2009 pointed to obvious social and attitudinal differences between towns. It
also provided intangible validity to the statements made by interview subjects.
However, a problem with post-conflict observations is that, though it is safer
than conflict zone research, they are ex post to the causal relationships of
interest. The state of the world one can observe is “contaminated” by the very
experiences of conflict itself. Any such observations are therefore only moder-
ately helpful for measuring differences that existed between communities before
armed groups arrived.

In the Archives
A last set of data is from various archives. I accessed documents from govern-
ment archives, community archives, and personal collections of individuals
involved in the acción comunal movement. The ATCC’s musty, dusty, rat-filled
community archive was a particular treasure trove of information (and, yes,
rats!). It contained personal journal entries; verbatim meeting minutes from
civilian organization meetings; verbatim transcripts of meetings between civil-
ians of organized and unorganized communities and various armed groups;
original documents and acts; and correspondence between the community and
armed groups, government officials, and international organizations. These
documents hold special validity because they were not recorded for the pur-
poses of posterity or academic research, but rather to hold potentially deceitful
actors to their words. They provide a snapshot in time and let us reconstruct
particular moments in history. They also provide a glimpse inside the heads of
armed group leaders, highlighting their interests, approaches, decisions, and
rhetoric toward communities (at least inasmuch as they would communicate to
civilian communities in private settings). The archival data show consistencies
with both civilian and armed group testimonies and thus serve as a useful
validity check for potential reporting biases in some interviews (and some
interviewers!).

Secondary Sources
The case-study analyses are supplemented with existing secondary sources and
news reports. A goal of using these sources is to obtain adequate coverage of
events in both organized and unorganized case towns. Violence data based on
press reports, fieldwork, secondhand sources, and existing datasets was com-
piled, coded, and geo-referenced and then matched to case-study towns and
villages.

summary
This chapter has sought to transparently present the methodological choices,
challenges, and contributions of this book. Civilian autonomy is a question
Summary 109

that is ripe for careful research design to critically scrutinize the claims of
“peacebuilding.” Reflection on methods is also important given the reality of
studying a conflict setting and because the topic involves sensitive issues, subtle
strategies, and “hidden transcripts” (Scott 1992). Since understanding civilian
agency requires understanding civilian organizational problems and institu-
tions, I measure and compare social characteristics, organizations, institutions,
strategies, and conflict processes. I map problems of violence that civilian
mechanisms aim to solve and then empirically link them to outcomes.
The larger design holds several advantages. It controls for hypotheses of
violence and tests these hypotheses across standardized units of analysis. To
reduce selection bias, the design samples broadly across these many units,
including organized, unorganized, violent, and peaceful communities. The
historical context of the cases is used to assess complications from reverse
causality. The case selection procedure joins the different methods and helps
structure information. And, since field experiments are often not practical in
conflict settings, it is helpful that the cases that are identified for study simulate
experimental conditions. Overall, the design embodies a triangulation of many
methods, data sources, and viewpoints to provide both breadth and depth
of knowledge.
The different methods have different roles. I lay bare some of the limitations
of these methods and the sources of potential biases and mismeasurement.
Quantitative methods are good for an overview and verifying broad correl-
ations but are less able to test hidden strategies and are subject to measurement
error. Qualitative methods allow for the careful process tracing of history to
provide both the top-down view of the armed groups’ perspectives on civilians
as well as the civilians’ bottom-up opinions and observations of armed groups.
However, qualitative cases can be susceptible to selection bias and problems of
making broader inferences.
This study’s integration of methods means the whole is greater than the sum
of its parts. Yet, even with the technical and social scientific aspects of this
study, I have sought to breathe life into it and not lose the texture of the setting
and culture in which I worked. The social science is needed, but it must not be
forgotten that my subjects are real people and many have suffered torture, loss
of loved ones, trauma, and displacement, and have still persevered.
The empirical chapters ahead apply these methods to test implications of
theory. In the next chapter, I provide a statistical overview of the relationship
between autonomy-enabling social organizations and violence across Colom-
bian municipios.
5

How Civilian Organizations Affect Civil War Violence

“In conflict zones, there always has to be someone there in the community to
advocate for the community.”
–Ex-FARC combatant (Exc#9), Bogotá, 8/2009.

Even as the La Violencia conflict still raged in some parts of Colombia, a small
team of sociologists went to study the municipality of Chocontá, Cundinamarca
in 1959. They focused on the small village of Saucío as a microcosm of a new
community development program that held the promise of social repair (Fals
Borda 1960).1 The now-famous early account of the “communal action” pro-
gram starts with the parable about the construction of a new school for the
community. Desperate for a better educational facility for their children, residents
of the village first joined together to hold a bazaar (fair) to collect donations, but
they got burned when local authorities squandered the funds. With little to show
for their efforts, decaying cooperative traditions, and a mistrust of authorities, the
village formed a junta (board) in a second attempt to complete the project.
A promoter (technical advisor) arrived to help break the community’s iner-
tia, and reach consensus on the need for the school. As a catalyst (Fals Borda
1960: 51), the promoter helped form the junta, broker with authorities, and
train community leaders in organization, bookkeeping, and project manage-
ment. But the promoter’s stay was only temporary, since he was too costly to
keep around for long (52). The decisions came from the people, with the junta
taking the lead and dedicating the school when it was completed the next year.
The Saucío junta next set its sights on electricity and road projects (33). An
agricultural cooperative also sprang from the junta, including a store that
provided credit and maintained the shared tractor and sewing machine (43).

1
Translated and paraphrased from Fals Borda (1960) by the author.

110
How Civilian Organizations Affect Civil War Violence 111

The communal action in Saucío was credited with producing “a resurgence

of collective energies that otherwise would have been left dormant” (Preface).
In his description of an awakening, Fals Borda notes the junta helped form a
collective identity and ethos – “a liberation of the traditional campesino.” With
newfound pride and independence (60), the campesinos were no longer “sub-
missive and unsophisticated.” Instead, when the farm boss would pass, “they
would no longer take off their hats and salute with reverence and fear” and
would “act with dignity, and demand and command.” As one resident said,
“We no longer need mayors who are only concerned about collecting our
taxes” (60).
Communal action also altered the social order within the community as well
as the community’s relations with other actors and institutions. It was credited
with “reducing the brawls in stores and rowdy weekend drunks to only those
marginal individuals outside of communal action.” The beer bottle began
“losing its social prestige” and the junta provided “healthier avenues for the
expression of prestige” (61). The junta also embodied a challenge to insti-
tutions, such as the Catholic Church, that resisted social change or that had a
“cultural monopoly.” The scholars emphasized the resulting “autonomy,”
which “implies recognizing the talents and efforts of the people, united, that
as a general rule have been ignored by the dominant classes.”
The precursors for confronting the armed conflict are evident even in this
early study. The authors show how the junta of Saucío contributed to commu-
nity order and managed the community’s autonomy against outside actors.
Little did they know the juntas would continue to serve in this role for decades
to come, and against different and more lethal armed actors. This chapter
empirically tests observable implications of the theory from Chapter 2 about
the protective roles of civilian organizations such as juntas – like Saucío’s – in
conflict settings with statistical analysis across the universe of Colombian
towns. The chapter aims to be accessible to readers without backgrounds in
statistics. The statistical models and other technical information are included in
an appendix (see Tables 5.1 to 5.7) for readers interested in those details.
This book argues that civilians themselves, through their varying cohesion
and local organizations, can impact real-world outcomes and explain variation
in violence. Local civilian organization is essential for reducing violence because
individual civilians, similar to individual fighters, have little recourse and are
ineffective when navigating dangerous, complex, and changing environments
such as those of a civil war. The key empirical challenge to assessing this
argument is overcoming the identification or “measurement” problem of sep-
arating the impact of civilians on civil conflict from the reverse – whether
conflict destroys or stimulates social organization, and whether social organiza-
tion dampens or accelerates violence. Past studies of civilian movements have
had limited generalizability because of the problems of few cases (small-n), case
selection bias, nonstandardized units of analysis, and the poor accounting of
reverse causality.
112 How Civilian Organizations Affect Civil War Violence

For the analysis of how local civilian organization affects civil war violence, I
exploit data across counties (municipios) on various community organizations
in Colombia including the village or neighborhood juntas councils through
which residents coordinate to solve local problems and provide public goods.
This is because their existence is broad enough and their history is long enough
to attempt to distinguish causal relationships. Juntas are theorized to affect
violence because they proxy the high levels of coordination and social capital
necessary for communities to implement more complex “autonomy” (or other)
strategies to preserve themselves in the face of conflict. Because the presence and
functioning of juntas are more simply measurable and comparable than larger
or more formal autonomy organizations (such as “peace communities” or
farmer associations) and their presence and effectiveness vary across many
localities, they offer a convenient way of getting leverage on the challenging
research design issues. Data on these councils comes from a rare dataset
published by the Colombian census bureau as well as interviews I conducted
with members of the juntas movement.
As an implication of my theory, the main tests I conduct on the effects of
juntas focus on forms of selective violence. After controlling for combat and
contention among armed groups, I find that a variable representing juntas has a
negative effect on violence. In other words, organized civilian communities on
average tend to suffer fewer effects of violence than unorganized communities.
The next chapter adds confidence to these results by showing that the formation
of civilian organizations was unlikely entirely a result of earlier patterns of
conflict. The findings provide insights about general civilian behavior in civil
war and the recent wave of civilian resistance to the armed conflict in Colombia
that accelerated in the early 1990s.
In the sections that follow, I first conduct a basic analysis of conflict-related
drivers of violence. Second, I analyze the effect that the juntas of campesino
(mestizo) communities have on violence when accounting for the conflict-
related drivers. I also examine the violence-reducing effects of highly cohesive
Indigenous and Afro-Colombian populations and other organizations. Third, I
conduct some additional tests to better understand the scope of the juntas’
effects on violence. Fourth, I review the history of the outlier case of the town of
Belén de Los Andaquies to contextualize the findings. I conclude with a sum-
mary of findings and possible avenues for further research.

the junta councils’ impact on violence

Universe of Cases and the Unit of Analysis

To test the effects of juntas, I analyze data on violence by armed groups from
1990 through 2005. Beyond reasons of data availability, these years are also
the most relevant time period for study of the emergence and impact of
autonomy organizations since many were formed during or immediately prior
The Junta Councils’ Impact on Violence 113

to this period and there are various examples of junta activity, renewal, and
increasing self-governance as discussed in Chapter 3. As such, these years are a
tough test but are also some of the more likely years when we would detect an
effect of juntas on violence.
I use the municipio, or county, as the relevant unit of analysis (municipio-
year). This is in some ways not the ideal unit, as there are lower levels of
geographical division in Colombia that might more aptly represent “commu-
nities.” It may also gloss over some important differences between urban and
rural areas or between the county seat and peripheral villages for certain
municipios. However, the municipio is the lowest unit for which broad and
reliable data is available on the dependent and independent variables of interest
(in other chapters I test implications of my theory at lower levels of analysis).
Municipios also still allow a much finer-grained analysis than Colombia’s
thirty-two departments. The sample of municipios is heterogeneous. Some are
highly populated, some are nearly empty. Some are vast, some are small. The
mean area of the municipios is 1,174 sq. km and the mean population for 2005
is 38,530 people.

The Dependent Variable

To measure violence against civilians, I use the annual civilian homicide rate for
municipios as the dependent variable, measured as homicides per 100,000
residents. I use two sources for this data. First, I use data from the Colombian
NGO CINEP, which is based on independent press reporting from twenty-six
national newspapers and includes only armed conflict violence – “political”
homicides – and not those due to common crime. This source is advantageous
because it has broad coverage across the country and is arguably unbiased.
Although only events that are reported in the press are recorded and many
events are surely absent, this source should still permit relatively accurate
comparisons across municipios. According to this data, 7 percent of political
homicides are attributed to the ELN, 11 percent to the military, 15 percent to
the FARC, and 67 percent to paramilitaries (Figure 5.1).
I also use data collected by the Colombian National Police. This data has
coverage in all municipios with police presence and reporting should not
depend on the (inconsistent) presence of press reporters. However, it also
depends on people reporting crimes to the police (and that the police report
this data) and it is a “catch-all” statistic that covers more than just conflict-
related killings. In addition to including deaths from common crime such as
murders resulting from brawls and robberies, the police data also includes
deaths of noncombatants at the hands of all armed actors, including the
guerrilla groups, paramilitaries, the police, and the armed forces (excluding
combatant deaths). While the data certainly includes episodes of selective
violence, some of the violence counted as homicides is more indiscriminate in
nature, including civilian deaths from combat, massacres, acts of terrorism, and
114 How Civilian Organizations Affect Civil War Violence

figure 5.1 Political homicides by presumed perpetrator, 1990–2005.

Source: CINEP.

land mines. The map in Figure 5.2 shows that while violence is fairly wide-
spread, there are also areas of calm.

A Conflict-Based Model of Violence

I first analyze when and where armed actors will have strategic incentives to
contest territory and commit violence. I do this because the ability of civilians
themselves to affect violence is most likely observable after conflict variables are
controlled for. Further, any relationship found between civilian organization
and violence could be spurious if highly organized communities suffered less
violence than other communities because they faced less pressure from armed
groups. In the next section, I add other civilian-related variables to the analysis.
One of the main conflict-related explanations tested with this data from Colombia
is Kalyvas’s (2006) theoretical arguments and empirical conclusions (based on
other countries) about how the balance of territorial control affects violence.
In brief, Kalyvas’s research suggests that much of the violence against
individual civilians in civil wars is a result of battles for territorial control
among armed actors (the control-collaboration model). Violence is strategically
and selectively used against enemy collaborators to coerce support among the
civilian population. Violence is both used to gain control of territory and is also
the result of the dynamics of territorial control. According to his logic, violence
is produced “jointly” by the main mechanism of denunciations of enemy
The Junta Councils’ Impact on Violence 115

figure 5.2 Map of mean annual political homicide rates by municipio, 1990–2005.
Source: CINEP; Author’s calculations.
116 How Civilian Organizations Affect Civil War Violence

collaborators to the armed actors by civilian informants. These informants only

risk implicating a neighbor for one group if they are confident their patron will
be able to protect them from retaliation from the enemy group. Denunciations
and therefore selective violence are thought to be most commonly carried out
by the stronger armed actor in zones of dominant but incomplete control. In
contrast, violence is relatively low both at the front lines where control is evenly
contested, and in areas where one army or the other enjoys complete control.
This theory predicts an upside-down “U” relationship between violence and the
military balance of control.
Examining this relationship with statistics at the micro level can be labor-
intensive if not altogether impossible. Kalyvas (2006) uses a survey methodology
to classify variations in the military balance of control based on the responses from
civilians about the presence of armed groups near their homes. He traveled vast
areas of the Greek countryside to code civilian perceptions of control during that
country’s civil war and finds evidence to support his theory of selective violence.
Kalyvas and Kocher (2009) use existing U.S. government surveys of hamlets for a
test on the Vietnam War. This data collection technique is prohibitively difficult in
the case of the ongoing conflict in Colombia, but existing data can be used to
approximate the balance of military contestation. I use data graciously provided
to me by the Colombian Vice-presidency’s Human Rights Observatory, econo-
mist Fabio Sánchez of the Universidad de Los Andes (see Fabio Sánchez 2007),
and press-based data from the Jesuit research center CINEP.2 My initial analysis
of conflict takes into account both the magnitude and balance of armed actor
activity in predicting violence. A variable for Total offensive actions (attacks)
captures the magnitude of armed activity in a municipio in a given year by all
armed actors occurring between two or more actors. This variable is constructed
as the counted sum of a variety of belligerent activities.3 This variable and the
results of the regression analysis are described in the appendix.
Variables for economic resources are also included as explanations for
violence. Resources may represent strategic interests for armed groups and
therefore the intensity of contestation and incentives to commit violence against
civilians. Resources may also predict abusive, undisciplined armed actor organ-
izations (Weinstein 2006). As Colombia is known to have drug cartels and
cocaine production, I include a variable for the area of coca cultivated in each
municipio-year in hectares based on United Nations aerial survey data (United

2
Sánchez’s data comes from information from both the Colombian Government and the nonprofit
organization Fundación Social (Sarmiento Anzola 1998).
3
To measure the military balance of control, I include a squared term of the Total offensive actions
by the various armed actors in the specification since Kalyvas’s theory would predict an upside-
down “U”-shaped curve explaining selective violence. I also devise a second way to measure the
military balance of control by creating a variable based on the proportion of attacks by each
armed group “side” in a given municipio-year. This variable is used in the police data regressions
because the higher counts of attacks make it a more feasible proxy.
The Junta Councils’ Impact on Violence 117

Nations 1998–2006). In some models I also control for oil infrastructure, since
armed groups have been known to siphon and sell oil on the black market at
contraband gas stations, and a count of the number of mineral mines in each
municipio (East View Cartographic 2002).
I also include a number of additional control variables. To control for
variation in demographics across counties, I include variables for population,
population density, and the percent of population from minority groups from
the 1993 national census. The percent of households with dirt floors and the
adult literacy rate (1993), also from the census, are used as interchangeable
measures of socioeconomic status (SES) and poverty.
To account for geographic variation, I include a measure for rough terrain
since Fearon and Laitin (2003) find cross-nationally that mountains are a proxy
for rebel group activity and areas where rebels can hide. I measure rough
terrain as the elevation above sea level of the county seat of each municipio
and in some specifications as the standard deviation of elevation in meters. To
account for isolation and state strength, I include a measure of the distance of
each municipio from its departmental capital and measures for a municipio’s
lengths of rivers and paved road access (DANE 2000; and created from GIS
VMAP data). Region indicators (dummies) are included to account for region-
specific variation and department effects are also tested. Lastly, a one-year lag
of the homicide rate dependent variable is included to account for serial auto-
correlation of homicides in municipios and year dummies are included to
account for national trends in violence over time.
The results lend some support to Kalyvas’s balance of control theory from
the Colombian case (see Table 5.2 with CINEP data and Table 5.3 with Police
data). They show that the total offense and balance of control variables both
affect violence at statistically significant levels (p < .01), with marginal improve-
ment in the amount of variance explained (r-squared = .34). Total offensive
activity is significant and positive in all models and, on average, every additional
action is associated with an increase of about 2.5 in the homicide rate. Nearly
half of all municipio-years experienced at least some offensive activity. This
suggests that magnitude of combat itself could be capturing some of the incen-
tives for violence against civilians reflected in the balance of control theory.
The area of coca cultivation has a statistically significant and positive effect
on violence in most police models, even after controlling for combat. On average
there is an increase of four deaths per 100,000 residents for every additional
1,000 hectares of coca cultivated (the municipio with the most coca has 15,000
hectares). This is consistent with the narrative in Colombia about the harms
associated with coca. Many civilian cocaleros (coca growers) are viewed as tacit
participants in the conflict and have regular interactions with armed group and
drug cartel buyers, increasing the risk of homicide. Other variables that predict
important strategic interests and contestation are discussed in the appendix.
Overall, these models suggest that the conflict variables reasonably approxi-
mate the strategic incentives of armed groups to commit violence.
118 How Civilian Organizations Affect Civil War Violence

Testing the Explanation of Social Organization and Social Capital

Civilian autonomy theory suggests communities in civil wars might gain
protection through several mechanisms if they are organized versus if they are
not. The relevant counterfactual to assess whether juntas reduce violence is
that, given two similar municipios with similar conflict histories, a municipio
with more juntas should suffer less violence. A community achieves a degree of
“de facto” autonomy if it suffers fewer than expected civilian killings given the
local configuration of conflict. This leads to a first hypothesis:
H1: Municipios with higher densities of local junta councils should have lower levels
of homicides, all other things equal.
To measure the variation in the presence of juntas across municipios, I use data
on the number of juntas in a municipio in 1985 as published by the Colombian
census bureau (DANE 1985 and DANE 1987).4 As a measure from a single
point in time, this variable will best explain variation in violence across muni-
cipios as opposed to across time. Of the 990 municipios that existed in 1985
there is data on juntas for 967 of them. The 1985 data shows there were 18,458
total juntas. Information from the 1993 junta census shows that 27 percent
were in urban sectors and 72 percent were in rural sectors (implying that if
juntas are found to have an effect, it is likely occurring in rural areas).
To approximate the extent to which local communities (with specifically
rural villages and small towns in mind) are covered by juntas across counties,
I calculate an indicator of juntas per capita by dividing (normalizing) the
number of juntas in each municipio by its census population in 1985. For ease
of interpretation of this indicator, the fraction is multiplied by the mean size of a
junta of forty-five members to estimate the percent of “communities” that have
juntas out of the (imagined) number of those that could (hence, “Juntas Per
Communities”).5 The Juntas Per Communities measure ranges from a low of
zero percent to high of 48 percent with a mean value of 6 percent. The distribu-
tion of this measure of juntas is mapped across municipios in Figure 5.3.

4
While published in 1985, the initial collection and compilation of this information on junta
councils may have occurred a few years earlier. Although I could not find earlier data, there are
two references to a juntas “registry” or “directory” produced in 1978 and 1981, respectively.
Since I could not access these sources, for the purpose of this analysis, I refer to 1985 as the time of
the final count. However, to the extent that the measurements were taken earlier, it would mean
they are even less contaminated by the growing violence of the 1980s.
Ministerio de Gobierno and DANE. 1978. Registro Juntas de Acción Comunal. Bogotá:
DANE, as cited in: DANE. 1982. Indicadores Socioeconomicos de Desarrollo Rural en Colom-
bia, Bernal E., Alejandro (coordinator). Bogotá: DANE. There is also a reference to a “Directorio
Nacional de Juntas de Acción Comunal-1981” in: DANE. 1993. Las Estadísticas Sociales en
Colombia. Bogotá.
5
The use of an estimate of villages from 1970 (DANE 1971) to create a Juntas per Villages
indicator yields similar results, but it has more missing data.
The Junta Councils’ Impact on Violence 119

figure 5.3 Map of juntas per communities by municipio.

Source: DANE 1987; Author’s calculations.
120 How Civilian Organizations Affect Civil War Violence

The strongest evidence for my hypothesis on the effect of civilian organiza-

tion would be a direct, negative correlation between juntas and homicides after
controlling for conflict variables. But the theory of armed actors’ sensitivity to
civilians suggests it is more likely that the effect of juntas will be attenuated at
high levels of conflict. Juntas should affect violence in some regions and
moments in which violence might plausibly be in the interests of armed actors
for coercion or other ends. They may do this through autonomy mechanisms
that manage the costs and benefits to armed actors such as vouching, normative
beliefs for avoiding the war, and resolving community conflicts. However,
juntas are less likely to affect violence in areas where armed groups are fighting
especially hard for objectives or have strong interests in targeting the popula-
tion. This implies an additional hypothesis to be tested:
H2: The effect of juntas on homicides is conditional and will interact with and be
attenuated by the level of armed actor combat and contestation, as represented by the
total number of attacks and the balance of attacks.

Direct Effects of Juntas

There are strong results for a direct effect of juntas on violence. Analysis using
CINEP violence data shows a significant, negative bivariate relationship between
juntas and the political homicide rate in Model 1 of Table 5.2. Results from the
regression models in Table 5.3 using the police homicide rate show that the Juntas
per communities coefficient is also negative once the intensity of combat and
balance of control are controlled for (significant at the .05 level). Figure 5.4
(based on Model 4 of Table 5.2) shows the significant substantive effect the junta
councils have on violence: based on the mean values for the CINEP homicide
rates. A change in the level of juntas from 0 to 0.11 (to the 75th percentile) can
lead to an average estimated decrease of 25 percent in the homicide rate, reflecting
a modest but significant degree of protection for most communities. As expected,
the conflict-related variables account for a relatively large share of the variation in
violence in the models, but the Juntas variable still has a small amount of
explanatory power above and beyond the base model variables and conflict
indicators (moving the r-squared about 0.02, or about one-third the change
compared to the change from the introduction of the conflict variables).
Additional tests in the appendix contain more refined tests of the effects of
juntas. For instance, the juntas are also effective in areas where multiple groups
are operating and potentially endangering the population. In these tests, the
sample is restricted to municipios with at least one or two attacks in a given
year and municipios where both guerrilla and government/paramilitary actions
were registered (not shown). These tests lend support to the theory that juntas
represent a vehicle for protection through autonomy strategies since these are
the most likely conditions where juntas are representing and implementing the
autonomy protection strategies theorized in Chapter 2, and not just strategies
of aligning with dominant actors.
The Junta Councils’ Impact on Violence 121

figure 5.4 Estimated mean municipal political homicide rates by levels of junta
councils, 1990–2005. Low Junta councils is the 0 percentile; High Junta councils is the
75th percentile (0.107); Error bars indicate 95% confidence intervals; Estimated
based on Model 4 of Table 5.2 (CINEP data).

A key issue that could affect the reliability of these results is reporting bias, a
risk not just in this study of civilian organizations, but for all studies of civil
conflict. It is possible that the presence or absence of social organizations like
juntas can lead to a reporting bias. If juntas are stronger in rural, isolated
communities and CINEP news and police coverages are weaker in those places,
it could produce the negative relationship that is observed.6 A similar finding
could arise if people rely on juntas instead of the police in places where juntas
are strong (as a result of state absence or fear of entanglements) and these
juntas do not report to the authorities. On the other hand, if communities with
juntas are more organized than those without juntas, they may be more likely to
report homicides to the police. If this were true it would actually bias against
results confirming my theory that juntas suffer less violence.
I gauge the severity of this problem and address it in several ways. First,
evidence from Cubides (2006) corroborates that juntas tend to be key reporting
sources. Cubides mentions the following example:
In 1997, in a zone that had been recently dominated by the guerrilla, a unit of the army’s
elite forces was annihilated. The only civil power acting in the zone to register the dead
and the survivors was the local junta. The media sought the official story and came back

6
The homicide rates from the CINEP and police datasets are correlated at r = 0.3; see Table 5.7.
122 How Civilian Organizations Affect Civil War Violence

to the junta’s president, the only one who could give public faith to what happened. The
junta was the only organization to manage to survive in the middle of the war zone and
with the tacit recognition of both sides.
As this episode shows, if juntas help communities retain autonomy and increase
information flows about acts of violence relative to communities without
juntas, counts of violence could appear to be highest in areas with juntas. This
would bias against finding any violence-reducing juntas effect.
Second, comparisons of these large-n data sources with qualitative data I
collected from the ATCC case and the Cundinamarca towns (see Figure 8.2)
also suggest organized communities are more likely to report violence. Third,
underreporting bias that could lead to spuriousness should be partially
accounted for in the statistical tests that control for and subset cases based on
“ruralness” and state presence with the inclusion of the variables of distance
from department capital, paved roads, and population.7
The junta measure’s inaccuracy may also mask a larger junta effect on
violence. This could be for two reasons. First, some juntas that were likely
measured in the dataset appear to be shells today, existing mainly in name (as
found in some of the Cundinamarca towns in Chapter 8). Juntas may be
inactive because they successfully met communities’ public goods needs or
because of clientelist depredations. This would weaken any observed “treat-
ment effect” of juntas (i.e., they are measured as having a violence-reducing
effect even though in reality they are inactive or clientelist). Second, if juntas are
underreported in some areas where they actually exist, it means some cases that
actually got the treatment of juntas are classified by mistake as part of the
control group that does not have juntas (meaning if the junta effect holds, these
towns should suffer less violence). This would cause the mean violence of the
“control” (low junta) group to drop closer to the “treatment” group’s because
the juntas are actually reducing violence but are not observed as doing so.

The Juntas: Organizational Effects or Favorable Conditions?

I consider three additional factors that might undermine the suppressive effect
of juntas on violence related to community homogeneity and coexistence: the
La Violencia conflict, inequality, and polarization. To test an additional impli-
cation that juntas resulted from earlier violence, I also add a dummy (zero or
one) variable reflecting the severity of La Violencia to the estimates of present-
day violence to account for an area’s unobserved, or omitted, “historical
propensity for violence.”8 I coded dummies for whether or not a municipio
experienced “high levels” of violence based on historical information in

7
Additional tests exclude department capitals and municipios that are within 20 kilometers of the
capitals from the sample. Other control variables such as government officials per capita and the
presence of police stations and inspections (administrators) should also help to account for
reporting biases to official sources.
8
This is similar to pulling this latent “propensity” out of the regression’s error term.
The Junta Councils’ Impact on Violence 123

Guzmán et al. (1963) and Roldán (2002) on the La Violencia period from 1948
through 1963. There are dummies for the first wave of violence and the second
wave of violence (1954–1958), as well as remnant violence after the peace
accord was signed and the National Front government was installed (1958–
1963).9 I also coded additional dummies from Guzmán reflecting whether or
not a municipio was the headquarters of any local bandit bosses during the
entire episode (1948–1963) as an additional indicator of conflict and disorder.
Consistent with historical accounts, municipios in the central departments of
Tolima, Cundinamarca, Huila, Boyacá, Santander, Caldas, Antioquia, and
Casanare tended to suffer high levels of violence while the departments of
Nariño and Cauca in the southwest and Magdalena (today spanning Cesar,
Bolívar, Magdalena, and Guajira) in the northeast suffered very little violence
(Figure 5.5).
Experiencing La Violencia appears to be significantly and positively associ-
ated with greater violence today. This points to a serial correlation in violence
and suggests La Violencia is working to account for a municipio’s latent
propensity for conflict. But, the observed effect of the junta councils is not
washed out; it remains significant and negative. This suggests that even after
correcting for a possible selection bias in the distribution of juntas due to a
circular relationship between violence and social organization, juntas continue
to work to suppress violence in the 1990–2005 period. In an additional test of
only La Violencia municipios, the significant and negative effect of juntas
persists, arguing against juntas solely reflecting some kind of post-Violencia
“learning” effect about how to get along (not shown). It also hints at a
“reversal of fortune” of sorts (e.g., Acemoglu et al. 2002), where areas that
were previously violent during La Violencia were revitalized upon an increase
in coverage of juntas. I return to the effects that La Violencia may have had on
junta formation in the next chapter.
Inequality is one of the first factors to which analysts attribute Colombia’s
social and armed conflict woes. By distribution of land and income, Colombia
has historically been one of the most unequal countries in Latin America and
the world (Lorente 1985). Albertus and Kaplan (2013) show this inequality has
persisted in part because of poorly implemented agrarian reforms. Could juntas
mainly be reflecting minifundios (small farms) and equal social relations? Could
degrees of inequality produce qualitatively different armed groups and
incentives to use violence? Could inequality impact conflict beyond mere levels
of poverty? I examine these questions by using municipio-level variables for
GINIs of land area and land value estimated based on the cadastre (land
registry) by Offstein (2005) for the late 1980s.10

9
A designation of “highly violent” does not necessarily mean that all parts of a municipio
experienced violence. Results are similar using data on La Violencia coded by Oquist (1980).
10
The GINI is a commonly used estimator of what proportion of land is held by a given proportion
of the population. These measures are missing data for several departments, including the
department with the most municipios, Antioquia.
124 How Civilian Organizations Affect Civil War Violence

figure 5.5 The geography of La Violencia

Source: Compiled by author based on Guzmán et al. 1963, Roldan 2002.
The Junta Councils’ Impact on Violence 125

The land value GINI is not a significant predictor of violence regardless of

whether the juntas councils are included in the estimation, suggesting it adds
little explanatory power above and beyond the other conflict and societal
variables in the models (Table 5.4). However, the GINI land values do washout
the significance of the junta councils (land area GINI has no effect). This
suggests that juntas are somewhat correlated with inequality. Still, while juntas
may be more likely to be formed in areas with minifundios, the junta councils
are a better predictor of violence than inequality in these models and the
entirety of their violence-reducing effect cannot be reduced to background
levels of equality. This issue is taken up again in the following chapter on the
origins of juntas and in the case studies.
I account for the amount of political unity or discord across municipios
immediately after La Violencia when the first juntas were being formed. I
constructed an indicator of the political polarization (Polarization) in the vote
for president in the National Front government in the 1962 election.11 This
variable is not a significant predictor of subsequent violence in the 1990s and
2000s (even though violence has been directed at leftist social movements) and
the juntas effect persists. This is perhaps because the conflict has been of a less
partisan nature and the juntas or the indicators of economic stratification may
capture more important aspects of social unity.

Conditional Effects of Juntas

Testing Interactions with Conflict Intensity. To check the limits of the effects of
juntas I test for an interaction effect between Juntas per capita and Total
offense. The interaction is statistically significant using the police data, as both
the constitutive terms and the interaction term are significant (p<.05; Table 5.3
Model 6) and in the predicted direction. The estimate of Juntas is negative,
suggesting that it is working to decrease homicides, while Total Offense and
interaction terms are positive. Together, these effects suggest the violence-
reducing effect of juntas weakens as conflict intensity increases (Figure 5.8).
The significant relationship between juntas and homicides is not conditional on

11
Polarization in a municipio is calculated based on the following formula: 1 - (%Conservative
Vote - %Liberal Vote)2. Using this data to measure political polarization in 1962 can be
problematic because of the rule under the National Front pact that the presidency alternate
parties from term to term. Still, there is some variation in voting that can be exploited. Although
1962’s presidential election was supposed to automatically go to a Conservative and was in fact
won by Conservative León Valencia, Liberal candidate López Michelsen made a rogue run for
office as part of the Liberal Revolutionary Movement (MRL) party. Votes for Michelsen were
annulled but were still tallied and summed to 24 percent of the ballots. This amount certainly
does not reflect a balanced vote but should still be able to provide a relative estimate of
polarization in many parts of the country. Unfortunately, this municipio-level indicator cannot
say much about political preferences at the village level.
126 How Civilian Organizations Affect Civil War Violence

the intensity of combat when using CINEP data (though is nearly conditional
for paramilitary violence; Table 5.5 Model 3).
Testing Interactions with Political Ideologies. To examine how the junta
councils perform in conditions of ideological stigmatization (and against the
kinds of hybrid armed actors that may target enemy collaborators based on
such stigmatization), I constructed a variable to indicate the fifty-three munici-
pios where Communist Party or Patriotic Union Party (leftist) mayors were
elected between 1988–2000. Partly a result of the FARC’s “all forms of
struggle” approach that mixed political organizing and the use of force, the
politicians and organizers of these splinter movements were perceived as leftist
rebel supporters and were exposed and therefore targeted by right-wing armed
groups (a “politicide”; Gómez-Suárez 2007). They also contributed to the
stigmatization of the communities that voted for them. Many civilians of these
communities became perceived as leftists and were also left open to politically
motivated retaliation. However, in line with civilian autonomy motives, I test
whether the juntas might have a mediating effect on violence in these historically
“leftist” areas by interacting the juntas variable with this partisanship variable.
While a Patriotic Union legacy itself correlates with higher violence (an ideo-
logical effect of the anti-UP stigma), and some junta leaders have been targeted,
the juntas dampen this effect (a countering-stigma effect; Table 5.5 Model 5).
This effect is consistent with the juntas’ general efforts to keep communities
impartial in the conflict and dampen violence due to political stigmatization (or
other forms of collaborator stigmatization). This finding further speaks to the
theoretical scope conditions for civilian autonomy, suggesting that particular
types of armed actors such as the “hybrid” groups can be influenced by local
social organizations to keep from using violence (e.g., groups with particular
political motives for violence that lack ideological restraints against using
violence). By staving off politicide, or at least limiting its effect on the general
population, the power of the juntas may also be relevant for countering mass
killings and genocide.
Effects of Juntas on Additional Conflict Outcomes. I tested several add-
itional dependent variables of other manifestations of violence to get a clearer
picture of the scope conditions for when juntas and the strategies they enable
may defuse violence. I use data on forced displacement from the NGO
CODHES (Consultancy for Human Rights and Displacement). I also test
indicators of massacres (events where four or more individuals are killed at
once) from both Colombian government and CINEP data. Regression results
(not shown) indicate some possible massacre-reducing effects of juntas but little
relationship with displacement (though the act of displacement also involves
response behavior on the part of civilians).12 This suggests that, consistent with

12
The relationship between juntas and massacres may be sensitive to model specification, as
juntas is negative and significant using Zero-Inflated Negative Binomial or Logit models, and
so may forestall massacres. Juntas is positively associated with displacement data from
The Junta Councils’ Impact on Violence 127

both theory and the interaction effect encountered earlier, juntas are decreas-
ingly effective as conflict conditions intensify or armed actors have strong
motives for killing.
The Effects of Juntas Over Time. The conditional nature of the effect of the
junta councils can also be explored by looking at how the juntas’ effectiveness
varies over time. A key implication is that junta effectiveness should vary
sensibly with the facts of Colombia’s historical conflict patterns. To see the
juntas’ trend, I regressed the Juntas and conflict variables against violence in a
series of annual cross-sectional models. The Juntas coefficients are plotted by
year in Figure 5.9 with 95 percent confidence intervals.
This analysis illustrates the varying capacity and limits of civilian
cooperation. The nationwide effect of juntas is found to be negligible during
the early 1990s, which can plausibly be explained by the relative calm and low
levels of conflict. At this time, the guerrillas were expanding and paramilitary
groups were still being conformed, so conflict conditions may have been either
too calm to spark civilian responses or not yet sufficiently widespread to
observe a juntas effect. A strong, significant, and negative juntas effect is
observed from 1995–2000. This period saw escalation in armed group presence
and combat. Juntas may have worked to protect residents through organiza-
tional learning and possibly the diffusion of best practices over time. There was
then a period of severe brutality until the paramilitaries demobilized in 2003–
2006, during which time the juntas’ effectiveness again appears to have dimin-
ished. There are several possible reasons: armed groups became more oppor-
tunistic and cartel-like, less political, and fought harder; juntas and other
organizations became increasingly targeted beginning around 2001, likely
diminishing their organizational capacity (El Tiempo 2004); and the state
increased its reach through Plan Patriota, making social organizational strat-
egies less necessary in secured areas.
This interpretation of the events is supported by the trend in violence
directed against the juntas themselves. I coded the killings of junta leaders
based on the press reporting data compiled by CINEP. For the eleven-year
period from 1996–2006 there were 423 events with 546 victims, indicating the
juntas themselves were not immune to violence.13 Killings spiked around 2000–
2001, suggesting a change in the nature of the conflict that corresponds with the
reduced effect of juntas on violence. If juntas are directly targeted, it becomes

the Presidency’s Social Action agency. This data is based on registration of displacements in
receptor municipios for government social benefits and is viewed as an underestimate, raising
questions of data reliability. Reporting bias could be present if registration is correlated with
community organization.
13
This is surely an undercount, though perhaps less so than counts of violence against the general
population because there is likely better reporting from organizations and about semipublic
figures.
128 How Civilian Organizations Affect Civil War Violence

harder for them to protect their own populations.14 This timing also corres-
ponds with an increase in the intensity of conflict and the FARC modifying its
political strategy, which likely further endangered junta leaders with both
armed groups.
Junta Effects by Armed Groups. Another way to explore the conditional
effects of juntas is to disaggregate the conflict homicide rate for armed groups in
the conflict as they may have different reactions to the juntas. I created variables
for killings of civilians by the guerrillas and paramilitaries based on which
actors are “presumed responsible” for the acts in CINEP’s press reporting data
(for many cases, no perpetrator is identified). The junta effects are strongest for
predicting homicides by paramilitaries, with no significant effects associated
with violence by guerrillas (Table 5.5; paramilitary effects are significant at
p < .05 even when land value GINIs are included in the model).15 I also tested
whether the violence-mediating effects of junta councils were stronger after the
AUC paramilitary umbrella group formed in 1997 (the interaction between the
Juntas variable and a Post-1997 dummy variable is negative though not signifi-
cant). Consistent with these results and with the temporal trends in the reduc-
tions in violence, the juntas indeed helped mediate violence during the height of
the paramilitary surge.
A plausible interpretation of this result is that the juntas are more effective
in protecting against and credibly signaling to out-groups that enter com-
munities seeking to purge enemy collaborators (paramilitaries surged in the
1990s and dislodged guerrillas in many areas; the army surged post-2002). The
paramilitaries’ default may be to target juntas, historical Patriotic Union party
strongholds, and other social organizations, but they may be more susceptible
to influence in areas where juntas push back against them or where they feel
they can safely delegate some order-maintenance responsibilities to juntas. The
largest effect of juntas on violence would then appear to come from paramili-
taries improving their treatment of civilians. This could be consistent with the
guerrillas being more disciplined overall or having greater general affinity for
the peasants or better baseline information to identify enemy collaborators in
rural communities. In such circumstances, guerrilla behavior should not be
more greatly influenced or “reformed” by strong local councils.
Further support for these possible differences in armed groups’ attitudes
toward the juntas and civilian communities is found in additional data and

14
For instance, see: “Por ser de izquierda, asesinaron a líder comunitario en San Vicente del
Caguán” (“For Being of the Left, a Community Leader Is Killed in San Vicente del Caguán”).
2008. El Tiempo, December 15, 2008.
15
These differences should be considered with caution since there is underreporting in the press
data, with many acts with unidentified perpetrators and likely more inconsistent coverage of
guerrilla killings relative to paramilitary killings since paramilitaries are usually stationed closer
to towns. However, as noted, various control variables should help account for such reporting
biases. Further, a social capital interpretation is consistent with interview accounts of paramili-
taries attempting to build juntas in some areas where none exist to consolidate territory.
The Junta Councils’ Impact on Violence 129

historical evidence. When the data I compiled on killings of juntas leaders is

broken down by armed groups, one sees there are many more victims at the
hands of the paramilitaries than the guerrillas. This finding is consistent with
the different approach that the FARC took toward the juntas. A news report in
El Tiempo (2002) with the headline “The FARC Stalk the Juntas Comunales”
describes a document recovered in an army raid that outlines a new FARC
“popular power” political strategy to weaken state institutions from the
bottom up. They planned to attack the political system by threatening govern-
ors, mayors, and city councilmen. They would then win over the population
and control politics by having the junta movement replace the elected officials
they forced to flee. This turn of events greatly worried the national juntas
umbrella organization, the National Juntas Confederation (CNAC). CNAC’s
human rights advocate responded, “We don’t want to fall into the trap where if
we don’t obey we’ll be displaced or killed . . . the FARC wants to intimidate us
and we want to strengthen participatory democracy and local development, but
based in respect and tolerance.”
This suggests three conclusions. First, the FARC is generally more amenable
(than the paramilitaries) to respecting civil society and civilian organizations.16
This may mean that the juntas were mainly effective in gaining autonomy in the
conflict by protecting communities from incoming paramilitary forces. Second,
although the FARC (and likely paramilitaries) became more and more oppor-
tunistic it still valued local political support. Third, the increased “political”
competition over the juntas is consistent with its leaders being more directly and
violently targeted and with the waning effect of juntas over time. This rising
FARC pressure is confirmed in case studies and ex-combatant interviews in
later chapters.
Overall, juntas appear to decrease violence on average in conditions of
moderate conflict intensity and, in some cases, may also be able to reduce
violence at more severe levels of conflict. That juntas decrease violence after
controlling for resources and coca cultivation, the severity of armed combat,
and the balance of control provides some reassurance they are effective in
at least some regions where armed groups might have incentives to kill.
These results fit with the theoretical expectation of the third part of my theory
about the conditions for “de facto” autonomy. They suggest that social organ-
izations can be effective under many conflict conditions but have the best
chances for reducing violence among hybrid armed groups and moderate
conflict conditions.

16
According to a FARC bulletin, the guerrillas seek to work with the juntas, saying “[The
Communal Action movement] strikes out on an independent path, outside of the politicking
and corruption of chiefs and bosses. Modernizing the Communal Action and making it a
democratic participative alternative will contribute all its long history to the design of the New
Colombia” (Boletín informativo del Comité Temático de las FARC-EP N 11).
130 How Civilian Organizations Affect Civil War Violence

additional social organizations

Percent of minority population. I include the percent of a municipio’s popula-
tion that is of Indigenous or Afro-Colombian descent both as a control variable
for the effect of the Juntas variable and as an indicator of organization in
its own right. First, minority group population is an important control vari-
able since Indigenous and Afro-Colombians form juntas less frequently than
(mestizo) campesinos and their unique local councils and organizations are not
included in the tally of juntas in a municipio. Without this measure, the effect of
Juntas per capita would be increasingly underestimated as the minority popu-
lation increases. Second, the dense social ties among minority group popula-
tions imply they may be an additional relevant indicator of civilian organization
in conflict settings. This points to a third hypothesis:
H3: The greater the proportion of minority populations in a municipio, the lower the
homicide rate, all other things equal.

The minority population variable has a significant and strongly negative

effect, suggesting that a large presence of Afro-Colombian and/or Indigenous
minority groups in a municipio reduces the homicide rate, all other things equal
(Table 5.3). This could be due to such groups existing in fringe areas of the
country where armed groups may have little presence. However, these groups
actually tend to reside in more conflictive zones and the result stands after
controlling for combat activity, geography, and population density. A possible
interpretation consistent with the juntas finding is that these minority groups,
through their ethnic identities and intragroup ties, have strong community
cohesion and organizational structures that help them deal with conflict. For
instance, the Afro-Colombian consejo comunitario and Indigenous cabildo
organizational structures were essential for winning rights to territory as well
as legal autonomy of local decision-making, beginning in the late 1980s and
early 1990s and continuing through today. The same organizations have also
been known to play important roles in preserving order in the community
(“Indigenous justice”) as well as managing interactions with outsiders, be they
regular visitors, the army, or illegal armed groups (see Guerra Curvelo 2004,
Wirpsa et al. 2009).17 In a different study, I find that Indigenous communities
with stronger shamans suffer relatively less violence than Indigenous commu-
nities with weaker tribal authority structures (Kaplan 2013).
Evidence on participation rates across ethnic groups is consistent with the
interpretation that the organizations of minority groups help them limit vio-
lence. As shown in Figure 5.6, based on my calculations from 2005 census
data, the Indigenous population in the rural sector participates in communal
organizations at three times the rate of either mestizos or Afro-Colombians.

17
Though unlikely, there could also be reporting bias for homicides if these groups or municipios
underreport homicides to the police because they have their own justice procedures.
Additional Social Organizations 131

25%

20%
Urban
Rural
households

15%
Pct. of

10%

5%

0%
Indigenous Afro-Colombian Mestizo / white
Ethnicity
figure 5.6 Participation in community organizations by ethnicity, 2005
Source: Author’s calculations based on 2005 census.

Indeed, when the percent indigenous population variable is separated out from
other minority groups, the estimated reduction in violence is even greater, all
other things equal.
I also examine several additional organizational forms that are important in
rural Colombia for their possible effects on violence. Since these additional
organizations are theorized to be less useful for seeking civilian autonomy per
the organizational characteristics developed in Chapter 2, they resemble a
“placebo” test of the junta councils. First, I include a measure of the number
of economic cooperatives existing up through the year 1976 from the Direc-
torio Nacional de Entidades Cooperativas (DANE 1978). This count includes
cooperatives for agriculture, transportation, credit, and mutual aid. In con-
trast to the junta councils, cooperatives are not found to significantly reduce
violence across rural municipios and their effect is even positive in some cases
(Table 5.4; this measure does not disaggregate between types of cooperatives).
This could mean that cooperatives reflect robust social capital (“cooperation”)
but still fail to protect civilians. It could also mean that cooperatives could be
targeted for being viewed as excessively political or Communist, be insuffi-
ciently inclusive of community members to implement broadly effective protec-
tion policies, or be found in target-rich areas that are not captured by other
measures. To the extent that cooperatives do reflect cooperation, the findings
could indicate there is something unique about the junta councils as organiza-
tions that make them more effective.
Lastly, I consider the potential role of ANUC (National Peasant Association)
land reform councils as indicated by the prevalence of their land invasions
132 How Civilian Organizations Affect Civil War Violence

during the 1970s (from Zamosc 1982). As mentioned earlier, although the
ANUC councils were not as widespread as the junta councils, for a time they
were a central organization for economic development in certain parts of
Colombia. I find the ANUC councils indicator does not appear to reduce or
significantly affect recent violence. This is plausibly explained by their gradual
disappearance or their being more political and more easily stigmatized (as
leftist) than the other kinds of social organizations tested.
In sum, as “placebo” tests, the cooperatives and ANUC councils proved to
be like sugar pills – they look like juntas but do not have the same effects
because they have different and less useful intrinsic characteristics. This is
consistent with Cohen and Arato’s (1994) perspective on the boundaries that
constitute civil society organizations, as entities such as economic cooperatives
and land reform councils are born from civil society but are not a part of it.
The results on these organizations provide added support for the uniqueness
of the juntas’ (and ethnic minority communities’) ability to unify residents to
seek autonomy.

learning from an outlier: the case of belén

de los andaquies
The data indicate that the municipio of Belén de los Andaquies in the depart-
ment of Caquetá is the biggest outlier in junta councils per capita (2.47 juntas
per 45 residents). The extreme juntas value for this case does not exert great
leverage in the analysis because it is well predicted (it suffered an expectedly low
amount of violence given its many juntas). But the case presents a good
opportunity to learn about the validity of the juntas indicator from qualitative
confirmation since there should be a strong expectation of vibrant juntas and
collective actions for protection.
In my fieldwork I was fortunately able to interview a resident from the rural
area of this town.18 She confirmed that the town historically had strong juntas,
which were still active in the 1990s. The juntas there were responsible for many
public works before the guerrillas came. In one instance, a junta organized the
town to build a new water tank after the first one was swept away in floods.
Many residents pitched in and they held a bazaar (fair) and fiesta to raise funds,
with people donating money, a roasted pig, sancocho (traditional Colombian
stew), etc. (Molano et al. 1994, 186).
There are several stories behind the origins of the town’s strong juntas. First,
the town had an active priest in the 1960s named Juan Salateo who would visit
the villages in the countryside and helped organize them to cooperate and build
public works for the community, such as chapels. Second, according to one
interpretation, the juntas were adopted early and easily as a carry-over from the

18
Bel#1, Bogotá, 2/2009.
Conclusions 133

culture of caciques, or chiefs, that organized society among the previous Anda-
quies Indigenous group (the group died out in late 1800s, though some resi-
dents have indigenous features from intermixing, or mestizaje). Third, colonist
settlers escaping La Violencia from Huila were reportedly very cooperative in
founding the town in the shadow of state neglect (Tulio Rodríguez 1982, aptly
titled Forgotten Municipios). As my subject summarized, the town was histor-
ically “peaceful and united.”
During the more recent years of armed conflict, the junta leaders played an
active role in advocating for their communities and dialoguing with armed
groups. For instance, they would attempt to recover youths who had been
forcibly recruited and would sometimes try to conciliate and reverse death
threats (mainly issued by the FARC guerrillas) against members of the commu-
nity. In the villages, the junta leaders also played the role of conciliator to
resolve disputes between neighbors over issues like property lines (linderos)
or disagreements among cheating couples, etc. There was also a notable
collective act of resistance when the guerrillas tried to attack the police station.
The townspeople came out dressed in white and protested by forming a
human chain to hold off the impending attack (interview; Sandoval 2004 citing
El Tiempo).
The complete picture is more varied though and not quite so rosy. Even with
the examples of advocacy and autonomy, some junta leaders had to displace
from the municipio under threat (although many junta leaders were able to
remain). The guerrillas also exerted pressure and manipulated the juntas in
some ways. In a report in Molano et al. (1994, 187) that contrasts with what I
found earlier, the juntas did what guerrillas told them to and “the guerrillas
organized everything in that town,” from forcing residents to hold a bazaar and
tejo tournament to repair roads to holding a beauty pageant to fund a bridge. In
sum, while there is strong anecdotal evidence that the level of social organiza-
tion in Belén de los Andaquies corresponds with the junta measure and helped
civilians assert their autonomy in the face of armed conflict, neither is social
organization invincible to high levels of coercion. Some of the insights from the
case of Belén de los Andaquies, including the importance of the church, are
explored more broadly in the next chapter on the origins of juntas and corrob-
orated in Chapter 8.

conclusions
The juntas in Colombia not only helped to repair social relations after the
devastation suffered during the 1950s, but have also persisted to become
important civilian organizations in the conflict today. Beyond their immediate
role in public goods provision, juntas are also shown to play a role in
dampening violence in the armed conflict in Colombia. They signal that vari-
ations in characteristics across communities such as the strength of civil society
matter for conflict outcomes, even after controlling for variation in armed
134 How Civilian Organizations Affect Civil War Violence

conflict. Tests of juntas as well as other social organizations in Colombia help

refine our understanding about the degrees and characteristics of civilian
cooperation that are most helpful for limiting violence. There is also some
evidence that the effects of juntas are conditional on and mediated by the
intensity of conflict conditions and different types of violence.
This research shows that civilian-based explanations of violence are comple-
ments to the macro-actor explanations from the existing literature. The analysis
also provides an out of sample test for Kalyvas’s preeminent model of violence
against civilians. Despite issues of measurement error and incomplete data,
these tests show promise for extending the balance of control explanation of
violence and measuring it in new ways. However, this explanation for violence
in the Colombian context is found to be relatively weak compared to measures
of raw levels of armed contention (offensive actions) in an area. More broadly,
this chapter contributes a methodological framework for evaluating the impact
of civil society initiatives for peace building.
These conclusions also deserve a note of caution. The results are not meant
to be an explanation of the various strategies that civilian communities might
select to attain autonomy or the impact of specific strategies. These findings can
only hint that juntas effectively allow for civilian autonomy of decision-making
as a strategy (independence, opting out). Since conciliation councils appear to
be widespread in juntas, a plausible interpretation of these results is that juntas
keep local conflicts between neighbors from spiraling to reach the macro-armed
actors, thus reducing denunciations even in zones where armed groups have
incentives for coercive killing. It is also possible that juntas provide a platform
for dialogue and negotiations with armed groups. It is plausible, however, that
in some areas juntas may be the vehicle through which civilians ally with parti-
cular macro-armed actors to preserve their communities.19 Precise junta-based
mechanisms are studied more carefully in the case study chapters.
Overall, juntas have shown to be “sticky,” persisting over time, implying
that such organizations may represent effective interventions for solidifying
local stability and order in the longer term. These implications for policy raise
an inevitable question – if juntas are so effective, why does not every commu-
nity already have a strong junta? This question of where juntas arise is closely
intertwined with whether their effects are derivative of conflict conditions. The
next chapter analyzes the factors that drove the formation and growth of the
junta councils. It deals head-on with the possibility that juntas are only passive
and largely shaped by the conflict and casts doubt on this as an explanation for
the correlations found here.

19
For instance, according to Ramírez (2001), the juntas of coca-growing civilians in some muni-
cipios of Putumayo and Cauca seem to have tacitly allied with guerrilla groups and other coca
purchasers in the mid-1990s to protest government fumigation of crops, civilian harassment,
and violence. Ramírez and Mitchell (2009) also show that these communities later took more
autonomous positions.
Appendix

This appendix describes the measurement of variables and statistical tests in

this chapter in greater detail.

Additional Information on the Units of Analysis

This study accounts for long historical processes in violence and social capital,
meaning some measures of the independent variables are from many years prior
to the violence. This creates a concern for specifying the unit of analysis because
Colombian political geography changed over the timeframe of study – since
1985 when juntas are measured to the present. In that window, about 10
percent of existing municipios in the sample were split to create new jurisdic-
tions. In 1985, there were 990 municipios. Today, there are nearly 1,100
(DANE 2007). This could be problematic since it could cause indicators meas-
ured in 1985 to not correspond with indicators measured later on (e.g., if a
measure of juntas is taken in 1985 and the municipio subsequently splits,
measurements of homicides, attacks, and even population will be underesti-
mated because they will be attributed in part to the offspring municipio).
I minimize the extent to which changing geography and splits in municipios
affect the results by remerging the variable values for “mother” and “daughter”
municipios where possible (DANE 2000; slight discrepancies exist between the
count of municipios in the census and the administrative data). This is simplest
for daughter municipios that were created from only one other municipio since
when new municipios are created from multiple municipios, it is nearly impos-
sible to know which daughter characteristics should be attributed to which
mother municipio. To address this issue, in some cases I created quasi-mega-
municipios that merged values of the connected mothers and daughter(s). I say
“quasi” because some of these “mothers” were always separate and never
actually existed as conglomerate entities. For these few cases I had to merge

135
136 How Civilian Organizations Affect Civil War Violence

municipios that existed in 1985 because they had multiple overlapping splits.
The municipios of Bagadó, Tadó, Condoto, Nóvita, Istmina, and Lloró in
Chocó and Santa Ana, Plato, Chivolo, Ariguaní, Pivijay, and Fundación in
Magdalena and their daughters are each taken as single municipios. As an
additional check, I also ran specifications with only those municipios whose
boundaries remained intact in the subsequent twenty years.

Additional Information on the Juntas Per Capita Variable

For the data from 1985 to be relevant today, one must assume some amount of
path-dependence – that the distribution of juntas in the past is correlated with
juntas today. Although using data on juntas that is measured much prior to the
dependent variable may entail a trade-off in accuracy of measurement of
(correlation with) juntas today, it has advantages for dealing with endogeneity.
Data from 1985 is a reasonable compromise for balancing these two goals.
Although the absolute count of juntas may differ from other published counts,
the measure still likely has relative validity across towns.
Census population counts for certain municipios appeared problematic com-
pared to population censuses from other years and population estimates from
the National Planning Department. Because these figures produced extreme
outliers for the juntas per capita variable, they were reestimated based on the
additional sources. For 1993, population data for Cabrera in Santander and
Recetor and Chameza in Casanare were abnormally low and were recoded.
For 1985, extreme population data was recoded for Chita, Cienaga, Combita,
Chiquiza in Boyacá, Toribío in Cauca, Uribia in Guajira, Fuente de Oro in
Meta, Mosquera in Nariño, Charalá in Santander, and Lérida in Tolima. The
statistical results remain robust whether these cases are included or excluded
from the sample. Because this measure is sensitive to counts of population, I also
test additional specifications based on population data from the 1993 census
and averages of the 1985 and 1993 censuses. Although the 1985 data may be
more accurate because it is contemporaneous with the measurement of juntas,
the 1993 census is generally considered more reliable for most other statistics.
Figure 5.7 graphically displays the distribution of juntas by this measure,
which is bimodal.

Additional Information on Control Variable Measures

Sanchez’s dataset includes the following measures as reflecting offensive initia-
tive by an armed actor: confrontations initiated by the group, ambushes,
attacks on installations, terrorist bombings, other terrorist acts, incursions on
populations, assaults on private property, roadblocks, harassment, robberies,
and illegal roadblocks. The activities plausibly reflect initiative and control
through the ability to act freely, although some of the activities are perhaps
aimed more at civilians than enemy armed groups (however, incursions,
Appendix 137

figure 5.7 Distribution of juntas in Colombian municipios.

robberies, and roadblocks are relatively few compared to the other categories).
Government actions only include attacks.
The measure of total offensive actions excludes more blatant activities
against civilians such as homicides, political homicides, displacement,
massacres, and kidnappings (indices that might reflect the dependent variable
of homicides). This measure of balance based on total observed events may be
partially endogenous to control to the extent that armed groups use actions to
consolidate power. Nevertheless, this does not preclude the measure from being
a relative guide for levels of control, for instance, if these attacks can be roughly
mapped to Kalyvas’s zones of control – that most areas with none to few
attacks are completely controlled by a given group (e.g., either the state or
the guerrillas), areas with low to moderate attacks exhibit dominant but
contested control, and areas with many attacks are the front lines of battle
(where indiscriminate violence against civilians may occur but there is little
selective violence). These data are validated in an analysis of the relationship
between land reform and insurgency by Albertus and Kaplan (2013).
Data on elevation was taken from the 1985 Divipola. Missing data was filled
in by the author from geographical information at www.fallingrain.com. Other
tests (not shown) use standard deviations of elevation calculated for each
municipio to account for variation in rough terrain.
138 How Civilian Organizations Affect Civil War Violence

The Carribean Coast dummy includes the departments of Guajira, Cesar,

Atlántico, Magdalena, Bolívar, Sucre, and Córdoba; the Pacific Coast dummy
includes Chocó, Valle del Cauca, Cauca, and Nariño; the Central Andean
dummy includes Antioquia, Caldas, Risaralda, Quindío, Tolima, and Huila;
the Eastern Andean dummy includes Norte de Santander, Santander, Boyacá,
and Cundinamarca; the Eastern dummy includes Arauca, Casanare, Meta,
Caquetá, and Putumayo; the Amazon dummy includes Guainía, Guaviare,
Amazonas, Vaupés, and Vichada.
The summary statistics for the dependent and independent variables are
displayed in Table 5.1. A comparison among municipios above and below
the median level of the Juntas variable helps to highlight some basic relation-
ships among variables. First, consistent with expectations, the high-juntas
group suffers a lower mean level of violence than the low-juntas group. Second,
some other variables (population, population density, and cooperatives) appear
to substantially differ across the groups of municipios, suggesting these are
important factors to include in the analysis since they may be associated with
both the development of the juntas as well as violence (Table 5.6). Lastly,
missing data is not a large concern. For most of the models, 929 of 990 possible
cases enter the regression, with only about sixty to seventy cases missing data,
or about 7 percent of the sample.

Statistical Results
To test the effect of junta councils on violence I specify a panel regression model
with municipio-year data, with some variables varying across time and space
and others only varying cross-sectionally (time-invariant). I test GLS models
with random effects:
Homicide Rateit = a + BHomicide Ratei,(t-1) BTotalOffenseit + BTotalOffenseDummyit +
BBalance of Controlit*BTotalOffenseDummyit + BBalance of Control2it*BTotalOffense-
Dummyit + Bz + e

The estimates of the effect of juntas are generally consistent across specifica-
tions, including when using robust standard errors and errors clustered by
municipio. The effect of juntas is significant after controlling for roads, rivers,
mines, oil, socioeconomic status, and population variables. The effect is also
robust to removing outliers with high juntas such as Belén de los Andaquies,
outliers with high homicides or massacres, only municipios with positive counts
of armed group attacks, the department of Antioquia (which may have under-
reported junta counts), and for small municipios with populations of less than
30,000. The effect is somewhat sensitive to population estimates used for
assessing junta coverage, however. The estimates using 1993 population are
less statistically significant although still negative. The effect of juntas is also
seen in a propensity-score matching model, where the homicide rate for cases
above the median level of juntas is 24 percent lower (Table 5.6).
Appendix 139

In additional robustness models, I tested the Juntas variable interacted with

year dummy variables as well as muncipio fixed-effects (dummy variables
for each municipio). These models are helpful for reducing the influence of
possible omitted variables. These tests face limitations, however, given the time-
invariant structure of the cross-sectional variables in the models. In some of
these tests, the results for the Juntas variable are consistent and exhibit a
significant and negative correlation with violence (however, in some specifica-
tions the Juntas variable is not significant). Overall, these models provide some
additional confirmation of the results, but are not conclusive.
The effect of juntas on violence may be understated. First, the use of a lagged
dependent variable of homicides to control for time trends can suppress the
coefficient of the Juntas variable if juntas have the effect of reducing homi-
cides in the past as well as in the present. The long-term effect (i.e., the effect
of juntas that does not act through the lagged homicide rate) using CINEP
data in Table 5.2, Model 4 is represented by a B of −28.17, which is actually
38 percent larger than the short-term effect.20 In the long run, then, the juntas
reduce homicides by up to 20 percent on average.
It is possible that the correlation could be susceptible to alternative interpret-
ations. However unlikely, I examine two possibilities here: mismeasurement
and only effects on criminal homicides. First, rather than reflecting the import-
ance of communal organizational processes, it is possible that the juntas vari-
able reflects levels of urbanization, development, state strength or reach, or
proximity to cities, which all could reduce violence. This is not likely, however,
because much of the variation in state strength should be captured by the
geographic and demographic variables of elevation, rivers, population, popula-
tion density, SES (dirt floors and literacy rate), coca cultivation, and distance to
departmental capital.21 In alternative specifications (not shown), junta results
are robust to the inclusion of indicators for state strength for 1995 of the
number of police stations in the county seat and locations of military bases,
as well as an indicator of government officials per capita.22
Second, it is also possible that the observed effect of juntas could just
be limiting criminal homicides among neighbors, and not by armed groups
(which would not be so terrible). This is difficult to parse with the police
homicide data. Even if this were the case, it still could be considered an
important effect of juntas since some scholars argue that all civil war violence
is, at its root, local. Still, this interpretation is unlikely, since homicides track
closely with levels of combat and juntas interact with levels of offensive activity.

20
The long-term effect is calculated using the formula: Bx1/ (1-BlagDV). If the lagged variable has no
effect (B=0) then short-run effect equals the long-run effect. As B of the lag goes to 1, the long-run
effect goes to infinity. In Table 5.2, Model 4, (-20.423 / (1-.275)) = -28.17.
21
See Fearon and Laitin (2003) for cross-national arguments about state strength and civil war. See
Collier and Hoeffler (2004) for arguments about poverty and related grievances.
22
Only the government officials per capita is statistically significant (and negative) after the other
controls are included.
140

table 5.1 Summary statistics

Variable Observations Mean Std. Dev. Min Max

Political homicide rate (CINEP) 14,745 9.75 28.18 0 502.74
Total offense (CINEP) 14,995 0.66 1.87 0 49
Total armed actions (1998–2005) 8,040 2.43 6.21 0 134
Police homicide rate (1998–2005) 7,864 67.77 86.14 0 2,034.26
Coca area (ha.; 1998–2005) 8,040 110.10 843.63 0 22,732.4
Number of junta councils 1,005 18.48 21.43 0 187
Juntas per communities 941 0.07 0.06 0 0.48
Pct. minorities 998 0.06 0.19 0 0.99
Cooperatives 1,005 1.99 11.13 0 262
ANUC raids (land invasions) 1,005 0.88 2.87 0 38
Population (1993) 983 33,556.90 182,199.00 584 4,900,000
Log 1993 population 983 9.47 1.06 6.37 15.41
Elevation (m) 993 1,222.86 912.01 0 3,657
Std. dev. elevation 995 340.56 246.37 1.07 1,560.98
Distance to dept. capital 993 121.69 97.85 0 850
Pct. Dirt floors 974 0.23 0.18 0 0.93
Literacy rate 998 0.67 0.15 0 0.87
Population density 998 115.04 480.93 0 10,609.10
La Violencia 983 0.40 0.49 0 1
 Churches 953 4.64 22.94 1 622
Paved road access 961 0.39 0.68 0 1
Polarization 740 0.50 0.36 0 1.00
Land area GINI 857 0.68 0.14 0 0.98
Land value GINI 857 0.66 0.13 0 0.91
Total guerrilla attacks (1985) 994 0.36 1.76 0 39.00
Year municipio founded 1,005 1858 109.28 1525 2002
Villages 725 25.91 19.52 2 207
Patriotic Union 1,005 0.05 0.22 0 1.00
Caribbean region 1,005 0.12 0.32 0 1
Pacific region 1,005 0.15 0.35 0 1
Andean region 1,005 0.25 0.43 0 1
East Andean region 1,005 0.35 0.48 0 1
Eastern region 1,005 0.07 0.26 0 1
Amazon region 1,005 0.03 0.17 0 1
141
142 How Civilian Organizations Affect Civil War Violence

table 5.2 Models of CINEP political homicide rates, 1990–2005, by municipios

(1) (2) (3) (4)

Political homicide rate 0.360** 0.332** 0.324** 0.275**
(1-yr lag CINEP) (10.82) (9.56) (7.23) (8.64)
Juntas per communities −20.248** −12.514* −16.472* −20.423**
(−3.89) (−2.11) (−2.15) (−3.47)
Total offense (CINEP) 4.743** 4.486**
(9.74) (11.91)
Total offense2 (CINEP) −0.103** −0.079**
(−5.79) (−4.67)
Log 1993 population −0.062 −2.214** -2.320**
(−0.16) (−4.54) (−5.66)
Elevation (m) −0.002** −0.002** −0.002**
(−3.50) (−2.77) (−3.10)
Distance to dept. capital 0.004 −0.010** −0.003
(1.05) (−2.58) (−1.07)
Pct. Dirt floors 3.099^ −1.208 −0.010
(1.66) (−0.55) (−0.01)
Population density −0.001** −0.001^ −0.000
(−2.77) (−1.70) (−1.29)
Pct. Minorities −4.718* −2.029
(−2.26) (−1.23)
La Violencia −0.770 −0.134
(−0.94) (−0.23)
Coca area (ha.) 0.002
(0.61)
Constant 7.383** 8.155^ 26.195** 29.173**
(13.86) (1.74) (4.42) (6.40)
Region effects Yes Yes Yes Yes
Year effects Yes Yes Yes Yes
Observations 14,115 13,950 6,503 13,935
R-squared 0.14 0.17 0.26 0.21
Municipios 941 930 929 929
Robust t-statistics in parentheses; Errors clustered by municipio
** p < 0.01, * p < 0.05, ^ p < 0.1

Population density, SES, and regional effects should also be decent controls for
crime. More strongly ruling out this interpretation, I find consistent results in
additional tests when excising rates of common crime by using CINEP’s polit-
ical violence data.

Discussion of Control Variable Results

A puzzling result is that, after controlling for combat and population density,
poverty or income as measured by the percent of residents with dirt floors and
Appendix 143

literacy rates is associated with decreased homicides in the police data. This
subnational finding potentially contradicts Collier and Hoeffler (2004) on their
claims about poverty and civil war onset. It is possible that poverty is proxying
areas that are less strategically vital to the armed groups (or poorer commu-
nities report less to the police). Alternatively, from a collective action point of
view, poor populations might rely more on each other and have greater com-
munity cohesion or place greater importance on communally provided public
goods, even after accounting for the presence of juntas.
An indicator for rough terrain is included because it may reduce the mobility
of armed groups and thus combat activity. The elevation variable is negative
and significant, suggesting that living in mountainous terrain affords civilians
some degree of protection (the standard deviation of elevation is not significant
in the CINEP specification). Interestingly, the effect of this variable is robust
after controlling for combat variables. This suggests that mountainous terrain
could be a proxy for some other explanation that reduces violence. For
instance, in Colombia, mountains may be a proxy for state strength since the
high-up cooler climates were settled first, with many low-lying zones colonized
only recently. It is also possible that mountainous terrain could reflect some
amount of social capital among residents, since communities in mountainous
areas may be more isolated than their lowland counterparts and so have to rely
more on each other (for similar arguments about paved roads in Sierra Leone,
see Humphreys and Weinstein 2006b). Alternatively, mountains may alter the
way combat unfolds – armed groups may have much more sporadic presence or
civilians may be able to hide more easily when combat breaks out.
The presence of oil pipelines in a municipio does not have a statistically
significant relationship above and beyond the conflict model, even though oil is
known to be a lootable resource in Colombia. The absence of this relationship
could be because oil infrastructure has been increasingly secured by the state
and multinational corporations (MNCs). The presence of mines (gold and
emeralds, for example) has a significant and negative effect on violence. This
could be because these areas are also more secured, have labor unions, or
involve MNCs.

Estimating the Conditional Effects of Junta Councils

The Juntas variable in interaction with Total Offense has a moderate impact on
homicide rates. Since interaction effects can be difficult to interpret, I assess the
average marginal effect of juntas on violence conditional on the level of offen-
sive activity graphically in Figure 5.8 (based on Model 6 of Table 5.3). A main
question is whether the negative effect of juntas is significantly different from
zero for relevant values of attacks from the sample data. On average, juntas
have a dampening effect on violence conditional on there being about nine or
fewer offensive events in a municipio in a given year, although this negative
effect can only be distinguished from zero with 95 percent confidence when
144 table 5.3 Models of police homicide rates, 1999–2005, by municipios

(1) (2) (3) (4) (5) (6)

Homicide rate (1–yr lag) 0.367** 0.299** 0.368** 0.297** 0.293** 0.296**
(33.26) (27.37) (32.73) (26.46) (25.97) (26.43)
Juntas per communities 14.743 −37.562* −42.982* −52.568**
(0.76) (−1.99) (−2.24) (−2.66)
Log 1993 population 0.148 −9.516** 0.106 −10.374** −10.680** −10.316**
(0.14) (−8.66) (0.09) (−8.83) (−8.43) (−8.78)
Population density −0.001 0.005* −0.001 0.005* 0.005** 0.005**
(−0.57) (2.25) (−0.46) (2.55) (2.62) (2.61)
Elevation (m) −0.008** −0.008** −0.008** −0.006** −0.006** −0.006**
(−5.76) (−5.54) (−5.43) (−4.54) (−4.52) (−4.62)
Distance to dept. capital 0.009 −0.003 0.010 −0.006 −0.000 −0.007
(0.77) (−0.27) (0.86) (−0.57) (−0.04) (−0.60)
Pct. Minorities −44.279** −58.818** −44.669** −58.220** −53.228** −58.443**
(−6.31) (−8.71) (−6.29) (−8.41) (−7.51) (−8.45)
Pct. Dirt floors −46.642** −62.899** −46.746** −59.827** −58.344** −59.648**
(−7.61) (−10.63) (−7.59) (−10.04) (−9.19) (−10.01)
Total offense 2.520** 2.439** 2.473** 1.917**
(14.77) (13.91) (13.97) (7.01)
Offense dummy 23.094** 21.806** 22.006** 22.534**
(6.10) (5.67) (5.68) (5.84)
Balance of control 52.154** 50.599** 50.981** 53.605**
(3.17) (3.03) (3.04) (3.20)
Balance of control2 −61.486** −58.813** −59.176** −62.190**
(−4.22) (−3.99) (−3.99) (−4.20)
 La Violencia 4.864* 5.038*
(2.35) (2.46)
Coca area (ha.) 0.000 0.004** 0.006** 0.004**
(0.27) (2.62) (3.68) (2.65)
Mines −3.563*
(−1.96)
Oil pipeline 4.675
(1.30)
River length −5.604**
(−3.14)
Paved road access 2.292
(0.92)
Juntas * Total offense 5.715*
(2.49)
Year effects Yes Yes Yes Yes Yes Yes
Region effects Yes Yes Yes Yes Yes Yes
R-sq between/ overall .77/.28 .77/.34 .78/.29 .77/.34 .77/.34 .77/.34
Observations 6,748 6,748 6,510 6,503 6,447 6,503
Municipios 964 964 930 929 921 929
z-statistics in parentheses
** p < 0.01, * p < 0.05, ^ p < 0.1
Model 1: Base model
Model 2: Base model plus conflict variables
Model 3: Base model plus juntas
Model 4: Base model, conflict variables, and juntas
Model 5: Base model, conflict variables, La Violencia, juntas, mines, oil, rivers, and roads
Model 6: Base model, conflict variables, La Violencia, juntas, and juntas interacted with armed actions
145
146 table 5.4 Models of additional organizations, 1990–2005, by municipios

(1) (2) (3) (4) (5)

CINEP CINEP CINEP CINEP CINEP
homicide rate homicide rate homicide rate homicide rate homicide rate
Political homicide rate 0.274** 0.268** 0.232** 0.212** 0.263**
(1-yr lag CINEP) (8.65) (8.10) (5.98) (5.65) (6.84)
Juntas per communities −19.930** −21.320** 1.992 −24.166**
(−3.37) (−3.56) (0.36) (−3.52)
Pct. Minorities −2.057 −1.418 −2.136 −2.276 0.163
(−1.25) (−0.81) (-1.27) (−1.33) (0.08)
Cooperatives −0.042** 0.312^
(−3.12) (1.82)
ANUC land invasions −0.011 −0.030
(−0.12) (−0.25)
Total offense (CINEP) 4.487** 4.634** 3.728** 3.772** 4.055**
(11.88) (9.81) (10.48) (10.92) (11.36)
Total offense2 (CINEP) −0.077** −0.073* −0.065* −0.062* −0.075**
(−4.60) (−2.33) (−2.19) (−2.09) (−4.62)
Log 1993 population −2.158** −2.268** −1.227** −1.483** −2.090**
(−4.93) (−3.80) (−2.93) (−3.66) (−5.35)
Elevation (m) −0.002** −0.001** −0.002** −0.002** −0.001*
(−3.10) (−2.76) (−3.82) (−4.01) (−2.40)
Distance to dept. capital −0.003 −0.005^ −0.004 −0.005 −0.002
(−1.13) (−1.66) (−1.33) (−1.46) (−0.54)
Pct. Dirt floors 0.016 −0.083 −0.993 −1.304 0.941
(0.01) (−0.05) (−0.56) (−0.75) (0.53)
 Population density −0.000 −0.007* −0.001* −0.001^ −0.000
(-0.47) (−2.15) (−2.09) (−1.79) (−1.34)
La Violencia −0.192 −0.056 0.070 0.175 0.035
(−0.33) (−0.09) (0.13) (0.30) (0.06)
Land value GINI 1.901 1.019
(0.60) (0.31)
Land area GINI 0.220 1.339
(0.07) (0.40)
Polarization 0.247
(0.24)
Constant 27.637** 29.121** 19.331** 22.251** 26.042**
(5.77) (4.99) (4.14) (4.87) (6.12)
Year effects Yes Yes Yes Yes Yes
Region effects Yes Yes Yes Yes Yes
Observations 13,935 12,840 12,195 12,600 10,890
Municipios 929 856 813 840 726
R-squared 0.214 0.219 0.167 0.152 0.216
t-statistics in parentheses; Errors clustered by municipio
** p < 0.01, * p < 0.05, ^ p < 0.1
Model 1,3,4,5: All municipios
Model 2: Rural municipios (>20 km from Department Capital)
147
148 How Civilian Organizations Affect Civil War Violence

figure 5.8 The conditional effect of juntas on the homicide rate by levels of conflict

there are two or fewer attacks. Conditional on there being nine or fewer
attacks, the average marginal effect of a percent increase in junta coverage on
violence ranges from reductions of −0.53 (at zero) to −0.02 (at nine). Even at
two attacks (B = −0.41) this is a nontrivial percent change relative to the mean
homicide rate, as discussed earlier.
Since the effect of juntas is conditional, it is possible that juntas may have
different effects at different levels of armed conflict. For levels of attacks greater
than 2.8 but less than about 41, the conditional effect of juntas is not statistic-
ally significant because zero is within both of the upper and lower bounds of the
confidence intervals (for this range of attacks it cannot be distinguished from
zero). When there are 41 or more attacks, the mean conditional effect of juntas
flips positive suggesting that juntas are no longer effective at providing protec-
tion at such a high level of attacks and actually increase violence. But what is
the substantive importance of this result in relation to the actual distribution of
attacks across cases? How prevalent are these different effects? The long tail of
the distribution of attacks indicates that the positive effect of juntas on violence
only holds for less than 1 percent of the sample with a high severity of armed
conflict. It is not theoretically obvious why juntas might be measured as
increasing violence at extremely high levels of contestation.
The distribution of attacks speaks to the substantive importance of the
conditional effect of juntas. It has a long tail, with most municipios suffering
few (registered) attacks per year. Approximately 55 percent of the municipio-
year cases had zero attacks, about 15 percent had one attack, 8 percent had two
 figure 5.9 The effect of juntas on violence over time
149
150 table 5.5 Models of killings according to armed group perpetrators

(1) (2) (3) (4) (5)

Paramilitary FARC Paramilitary Homicide Homicide
homicide rate homicide rate homicide rate rate rate
(CINEP) (CINEP) (CINEP) (CINEP) (CINEP)
Political homicide rate 0.217** 0.157** 0.218** 0.279** 0.269**
(1-yr lagged DV) (7.55) (4.59) (7.61) (8.86) (8.24)
Juntas per communities −21.737** −0.814 −17.785** −16.308* −15.956**
(−6.03) (−0.59) (−5.41) (−2.48) (−2.86)
Pct. Minorities 0.052 0.911* 0.332 −2.044 −2.436
(0.04) (2.57) (0.24) (−1.25) (−1.54)
Cooperatives −0.041** −0.043**
(−3.05) (−2.76)
ANUC land invasions −0.010 −0.016
(−0.12) (−0.18)
Total offense (CINEP) 1.966** 0.461** 1.675** 4.477** 4.388**
(7.36) (5.03) (4.08) (11.74) (11.83)
Total offense2 (CINEP) −0.038** −0.003 −0.079** −0.076**
(−4.48) (−0.52) (−4.62) (−4.82)
Log 1993 population −1.376** −0.265** −1.230** −2.147** −2.246**
(−5.77) (−3.57) (−5.34) (−4.93) (−5.13)
Elevation (m) −0.001* −0.000 −0.001* −0.002** −0.001**
(−2.35) (−1.42) (−2.22) (−3.10) (−2.66)
Distance to dept. capital 0.000 −0.000 0.000 −0.003 −0.004
(0.21) (−0.14) (0.23) (−1.13) (−1.45)
 Pct. Dirt floors 0.575 1.177** 0.605 0.011 −0.562
(0.53) (3.38) (0.55) (0.01) (−0.35)
Population density −0.000^ −0.000 −0.000^ −0.000 −0.000
(−1.68) (−1.38) (−1.74) (−0.48) (−0.53)
La Violencia −0.289 0.153 −0.205 −0.192 −0.157
(−0.75) (1.20) (−0.53) (−0.33) (−0.27)
Juntas * Total offense −251.676
(−1.45)
Post-1997 2.252**
(3.33)
Juntas * Post-1997 −6.543
(−0.90)
Patriotic Union 10.907**
(3.18)
Juntas * Patriotic Union −94.363**
(−2.72)
Year effects Yes Yes Yes No Yes
Region effects Yes Yes Yes Yes Yes
Observations 13,935 13,935 13,920 13,935 13,935
Municipios 929 929 928 929 929
R-squared 0.12 0.08 0.12 0.20 0.22
t-statistics in parentheses; Errors clustered by municipio
** p < 0.001, ** p < 0.01, * p < 0.05, ^ p < 0.1
151
152 How Civilian Organizations Affect Civil War Violence

table 5.6 A matching model of the effect of juntas

(1)
Probit
Juntas per communities
Political homicide rate −0.003**
(1-yr lag CINEP) (−6.64)
Pct. Minorities −0.479**
(−4.95)
Cooperatives 0.027**
(2.66)
ANUC land invasions 0.023**
(4.14)
Churches −0.088**
(−11.99)
Total offense (CINEP) 0.034**
(2.64)
Total offense2 (CINEP) −0.001
(−0.79)
Log 1993 population −0.205**
(−9.15)
Elevation (m) −0.000^
(−1.75)
Distance to dept. capital −0.001**
(−7.43)
Pct. Dirt floors 0.327**
(3.56)
Population density −0.005**
(−18.97)
La Violencia 0.113**
(3.84)
Region effects Yes
Year effects No
Pseudo R-squared 0.31
Municipios 899
Observations 13,485
z-statistics in parentheses; Juntas is split at the median of 0.062
** p < 0.01, * p < 0.05, ^ p < 0.1

panel b Average treatment effect of junta councils on CINEP homicide rates

Variable sample Treated Controls Diff T-stat

Unmatched 7.52 11.72 −4.20** −8.84
CINEP Homicide ATT 7.52 9.83 −2.31** −3.25
 table 5.7 Pair-wise correlations of conflict variables and juntas

Juntas per CINEP CINEP CINEP CINEP CINEP Police Total Guerrilla Para- Gov’t
communities Hom Rt. total Guerrilla Para- Gov’t Hom Rt. attacks military
attacks military
Juntas per communities 1.00
CINEP Hom Rt. −0.06 1.00
CINEP total attacks −0.03 0.29 1.00
CINEP Guerrilla −0.04 0.19 0.73 1.00
CINEP Paramilitary −0.06 0.23 0.47 0.39 1.00
CINEP Government −0.02 0.26 0.94 0.49 0.24 1.00
Police Hom Rt. 0.01 0.43 0.28 0.16 0.16 0.27 1.00
Total attacks 0.00 0.24 0.69 0.45 0.26 0.69 0.34 1.00
Guerrilla 0.02 0.22 0.57 0.41 0.20 0.55 0.28 0.91 1.00
Paramilitary −0.04 0.12 0.22 0.20 0.29 0.15 0.14 0.41 0.23 1.00
Government −0.01 0.20 0.69 0.38 0.20 0.73 0.32 0.86 0.61 0.23 1.00
153
154 How Civilian Organizations Affect Civil War Violence

attacks, 6 percent had three attacks, etc. Only about 7 percent of municipio-
years suffered ten or more attacks. So, in about 80 percent of the sample
(51 percent of the sample suffering at least some armed activity), the juntas
decrease violence to some degree (with 95 confidence). Figure 7.1 of attacks
(CINEP) for the ATCC profiled in Chapter 7 and the Cundinamarca towns
in Chapter 8 (Figure 6.4) show they suffered this level of low to moderate
intensity for most of the 1990s and 2000s, with a few years registering
extremely intense fighting. Other parts of the country, such as zones in the
departments of Arauca and Casanare and the Urabá region, were heavily
contested by armed groups and suffered many more attacks.
6

Why Some Communities Are More Organized

than Others

“Community organization in Nariño (Antioquia) is pretty weak . . . Altruistic

solidarity is scarce for collective needs like the construction of roads, schools,
and health centers . . . the juntas are greatly decayed.”
– Henao Delgado and Arcila Estrada (1993, translated)
“Jericó (Antioquia) . . . sprung up in the 1850s and its founding fathers were
picky. They didn’t allow miners, cowboys and roughnecks to take up resi-
dence and allowed in only those considered decent, God-fearing people. Mission-
aries built convents and schools. The town is now home to four museums . . . a
library, a cultural center, a botanical garden.”
– Otis (2010)

The previous chapter showed that different levels and forms of social organi-
zation can reduce levels of civil war violence suffered by civilians. Intertwined
with this relationship are the questions of where social organizations come
from and where they are likely to arise. This chapter addresses these questions
and does so for three main reasons. First, knowing where social organizations
arise in a developing country is itself an interesting question, with implications
for social and economic development. I find that geographic, demographic, and
social factors all play roles in the formation and spread of local junta councils
(qualitatively, I also consider promotion efforts by government personnel and
private actors).
Second, understanding the causes of juntas helps deal with the threat to valid
inference of reverse causality – that junta councils might only exist and survive
in historically peaceful places. Introducing the indicator of the violence suffered
during the La Violencia conflict of the 1950s shows that juntas were actually
more likely to be formed later in the 1960s and 1970s in areas that suffered
relatively more violence in the prior conflict, easing concerns of spurious
correlation. These statistical tests along with a close reading of Colombian

155
156 Why Some Communities Are More Organized than Others

history between the period of La Violencia and when juntas are measured –
both up to 1960 and from the 1960–1985 interim period – show that conflict
conditions were low and relatively stable through most of the country. The
tendency for juntas to be formed in historically conflicted areas was therefore
not likely disrupted during this period.
Third, with a better understanding of junta origins I am also able to unite the
statistical analysis with the case studies to better rule out confounding explan-
ations. An ideal research design to study this question would be an experiment
that randomly assigns juntas to communities. While we cannot go back in time
and rerun the history of junta formation, the statistical techniques in this
chapter can be used as a tool to simulate these experimental conditions by
more precisely matching sets of towns with similar likelihoods of having juntas
(treatment) but that in reality developed quite different patterns of social
organization. I elaborate on this method of case selection and argue that it
can be an appealing option for researchers in the field of political science when
facing constraints such as time, safety, or resources that preclude the use of
other methods such as random case selection.
After outlining this method I apply it to select a set of towns for qualitative
study of social organizations, civilian autonomy strategies, and violence. To
preview this method, I statistically match towns on observable indicators and
then select neighbor towns with similar conflict histories to help minimize issues
with unobserved variables. I arrive at sets of towns in case study sites in the
departments of Cundinamarca and Caldas (discussed in Chapter 8). Given the
dangers of fieldwork in a country like Colombia that is still experiencing
conflict, the case selection techniques balance representativeness with safety
and help assure that the research findings are not vulnerable to unmeasured
global factors that might affect all these towns similarly.

explaining the formation of junta councils

If juntas as an organizational form seem to correlate with reduced violence, a
natural question to ask next is why some communities decide to organize juntas
and others do not? This section tests various social and structural explanations
for variation in juntas per capita across municipios. As noted previously, while
juntas are sometimes encouraged by actors external to the community, they are
most frequently the creations of the communities themselves. Juntas are often
the product of networking in the absence of state provision of services. They
should therefore be found to arise in areas with strong social relations and
social capital (of course, effective juntas are also an indicator of cohesion).
Variation in political geography and demographics can be crucial determi-
nants of these intracommunity relations, as can shortages of and demand for
public goods.
A municipio’s population is a strong determinant of its coverage of juntas.
Less populated municipios are much more likely to have more juntas per capita
Explaining the Formation of Junta Councils 157

than more populated municipios (population density has minimal impact once
population is controlled for; see chapter Appendix). The number of villages in a
municipio from the 1960 agricultural census (DANE 1962) also strongly and
significantly predicts a larger number of juntas. This could indicate the possible
number of communities that could form juntas.
Geographically, there is a fairly strong relationship (p < 0.1) that municipios
that are closer to department capitals are slightly more likely to have more
juntas per capita. Juntas tend to be more prevalent in municipios in the Eastern
region followed by the East Andean region. An indicator of rough terrain
(standard deviation of elevation) is similarly significant, reflecting zones where
settlements are more broken-up. Surprisingly, whether or not there is paved
road access to a municipio is not a significant determinant of juntas once these
other factors are controlled, even though isolation might create greater
demands for community self-reliance.
The results for indicators for poverty and socioeconomic status (percent of
households with dirt floors; literacy rate) suggest that poorer communities (in
1993) tend to have a greater density of juntas (although juntas may have
affected economic growth). This speaks for a needs-based, demand-driven
explanation of junta formation. Wealthy communities likely have more private
goods or get public goods more easily from the state.
A variety of indicators of civil society strength are also strong predictors
of juntas. The number of churches in a municipio’s county seat (Catholic or
other denominations, Sarmiento Anzola 1998) is associated with increased
coverage of juntas. Consistent with the characterization of Jericó, churches
may provide an important form of technical assistance for community organiz-
ing, a curious contradiction with the World Values Survey finding that Catholic
countries have relatively lower levels of social capital. A measure of coopera-
tives is not significantly correlated with juntas. The percentage of the popu-
lation of minority groups is also not significantly associated with juntas
even though the presence of minority populations may suppress the number
of juntas. This could be because, rather than organizing through juntas,
Indigenous and Afro-Colombian communities have their own organizational
structures, such as cabildos or consejos comunitarios, which are not counted
as juntas.
I also test the impact of land (in)equality on junta formation. Inequality of
land values (in 1985) significantly reduces the likelihood of having juntas – they
are more prevalent in more equal municipios.1 This equality finding could
reflect greater cooperation because of shared socioeconomic preferences;
closer proximity of farms, which aids networking; or that clusters of small
minifundios have greater needs for public goods. As I find in tests of violence in

1
The results on polarization, villages, and inequality should be taken with caution since more cases
are missing data on these variables.
158 Why Some Communities Are More Organized than Others

Chapter 5, the effects of juntas on violence are dampened by inequality but

are also more robust than inequality and more strongly associated with levels
of violence.
As a check of whether the juntas are more prevalent in areas with preexisting
cooperation, I attempt to account for the amount of political unity or discord
across municipios immediately after La Violencia when the first juntas were
being formed. I again use the indicator of the political polarization (Polariza-
tion) in the vote for the 1962 presidential election in the National Front
government. I also test the percentage of Liberal votes as a measure of polar-
ization, possibly reflecting holdouts who did not support the National Front
(not shown). I find that polarization is significant and negatively associated
with juntas, indicating that less polarized (more cooperative?) places went on to
form more juntas (although polarization may be overpowered by inequality).
Again, as noted in Chapter 5, junta effects remain significant when controlling
for prior political polarization (a possible form of social cohesion) as well.
In sum, both social and structural factors predict the formation of local
councils in Colombia. In the next section, I elaborate on the relationship
between La Violencia and juntas.

the role of la violencia and reverse causality

Instead of juntas explaining violence, it could be that juntas are endogenous
to – shaped by – conflict patterns and are merely an indicator of histories of
low violence. To attempt to minimize this possibility, I use a measure of juntas
that was taken in 1985, five to ten years prior to the episodes of violence being
studied. This should effectively rule out the possibility of a contemporaneous
reverse causal relationship: the past distribution of juntas should be highly
correlated with the current distribution of juntas, but it should be impossible
that violence after 1990 affected the distribution of juntas in 1985 (and violence
spread to new areas during this time). However, even with the large time lag
between the measures of juntas and violence, the path-dependent nature of both
these variables (suggested by both the quantitative and comparative historical
analysis) could still potentially bias these results.
The argument goes like this: violence is correlated over time, so violent areas
tend to stay violent and peaceful areas stay peaceful; high historical violence (or
combat, which act like omitted variables) destroys social relations over time so
that juntas only survive in historically peaceful areas; these same areas with
juntas are then observed in the future to be relatively more peaceful than their
counterparts without juntas for some unobservable reason having nothing to
do with the autonomy mechanisms of juntas.2 In short, historical violence

2
It is also conceivable that juntas could survive in combative areas, but not play an important
combat-related role. For example, they may just provide public goods, in which case, even if many
juntas exist in combative areas, there would be no observable effect on violence.
The Role of La Violencia and Reverse Causality 159

might affect both the distribution of juntas observed in the 1985 data as well as
subsequent violence, causing the true effect of juntas on violence to be spurious
and overestimated. Juntas would then reflect peaceful conditions, rather than
bring such conditions about.
The severity of this issue depends on how one thinks violence might affect
the formation and survival of juntas. If one believes that a history of violence
impedes the formation or survival of social capital and therefore of juntas, the
observed beneficial effect of juntas might be a byproduct of peace (epiphenom-
enal). This would suggest a reciprocal relationship where low social capital
leads to violence, which leads to low social capital, etc. Alternatively, if a
history of violence stimulates social cooperation in some places – for instance,
through necessity or government intervention – and violence is then tamped
down, it would lend support to the explanation that juntas are indeed inde-
pendently benevolent. So, for juntas to be at least partially exogenous to the
conflict, they must have formed and persisted as much if not more in historic-
ally conflictive areas as in peaceful areas prior to the more recent era of conflict
(for which they are theorized to causally impact).
To get a true understanding of the impact of juntas, prior violence must be
taken into account. I review two episodes of twentieth-century Colombian
history to better understand how violence may have affected the process of
junta formation and civilian social capital: La Violencia of the 1950s and the
subsequent years through 1985, when the juntas are measured. A careful
reading suggests that a reverse causal relationship in a (negative) direction that
would bias against the junta results is unlikely. First, there is evidence that La
Violencia, in addition to reflecting a propensity for violence, increased commu-
nity cohesion and stimulated the creation of juntas. Second, I find there was
relatively little civil war activity that would affect juntas from 1960–1985.
Third, I find that conditions that promoted social networking in the absence
of the state were important stimuli for the creation of juntas.

La Violencia and the Distribution of Juntas

There is no question that La Violencia was one of the most devastating conflicts
in Latin America in the twentieth century. What is less clear is its impact on
social relations in Colombia. It is true that many communities were torn apart
and there were migrations from the countryside to cities, implying a negative
relationship between violence and community cohesion. However, there are
other reasons to believe the relationship is positive and, in fact, the historical
record favors this account. Information on La Violencia can account for part of
the distribution of juntas because the government encouraged juntas to pacify
ravaged areas and because some aspects of the violence itself engendered
community cohesion.
To statistically measure an area’s historical “propensity” for violence
and test whether this affects junta formation, I again use the dummy variables
160 Why Some Communities Are More Organized than Others

figure 6.1 Comparison of La Violencia and junta coverage

coded based on Guzmán et al. (1963) and Roldán (2002) that are discussed
in Chapter 5. A simple comparison of towns with “high” and “low” junta
coverages in Figure 6.1 shows that juntas are slightly more common in La
Violencia areas. Similarly, the La Violencia variable is significant and positive in
the OLS juntas selection models that predict the distribution of juntas across
municipios. This suggests that places that suffered violence in the 1950s were
actually more likely to have formed juntas that persisted through 1985, not less
(p < .05). Experiencing La Violencia on average translates into about 1 percent
more juntas per “communities.” Only when inequality is included in the
specification does La Violencia no longer significantly predict juntas (though
it retains a positive relationship). This suggests that inequality likely contrib-
uted to La Violencia.
An examination of department-level data on the distribution of juntas shows
similar trends. In the early 1960s, juntas were no less prevalent in La Violencia
areas (r = 0.01) and this distribution of juntas is associated with the juntas
measured in 1985 (r = 0.69; calculated based on Ministry of Government
data in Edel 1969). I also find positive correlations between where the
early government promoters were assigned, whether a town experienced La
Violencia, and levels of juntas in 1985. These findings fit with the historical
evidence that juntas were encouraged in areas that had seen significant vio-
lence (Bagley and Edel 1980). This finding also argues against the possible
The Role of La Violencia and Reverse Causality 161

spuriousness of the earlier results that juntas reduce violence and is consistent
with considering juntas a policy “treatment.”
The perspective that La Violencia increased community organization actu-
ally appears fairly consistently within the sparse existing literature. In perhaps
the only prior econometric study of juntas, Edel (1971) sampled ninety-six
communities with juntas to understand variation in the implementation of
community development projects. Contrary to his expectation, Edel found that
residents of communities in the sample afflicted by La Violencia actually made
greater investments in community goods with community-contributed funds
(which he views as an indicator of community “effort”) during his 1964–1965
time frame.3 Elsewhere, Edel suggests that not only did communal action occur
in areas with legacies of violence, but that it was also beneficial. He claims the
reunification of feuding populations around community projects and the isol-
ation of unpopular bandit leaders were frequent results of community develop-
ment (Edel 1969, 42).
Other sociological reports of the era corroborate several possible reasons
why La Violencia may not have been as detrimental for social cooperation as
might be expected. For instance, Torres (1963) argues that communal
responses to banditry engendered community solidarity where neighbors
banded together to defend themselves.4 Similarly, Guzmán et al. (1963 Vol.
2, 423) observe that community action actually helped to end violence in many
areas. Torres also suggests that the violence and guerrilla movements of this
time led to a greater desire for economic development and increased people’s
expectations for upward mobility when peace was restored, which could have
increased the appeal of juntas as a solution. These explanations speak to the
robustness of the violence-reducing effect of juntas and are also consistent with
civilian autonomy theory as an effective response to conflict and violence – just
as much yesterday as today.

Assessing Armed Conflict from 1960–1985

The period between the beginning of the juntas program in 1960 and the
measurement of juntas taken in 1985 was a relatively tranquil period in
Colombia’s recent history, with a lull in violence (see the homicide rate time
series graphed in Figure 3.1). It is therefore unlikely that conflict in this
time period affected the subsequent distribution of juntas across Colombian
municipios in a prejudicial way. According to Zamosc, the resolution of
La Violencia of the 1950s was fairly abrupt: “After 1953 the disturbances

3
However, Edel does not find that the government invested more or was likely to have more
promoter staffers in La Violencia areas (in contrast with the correlation I find), though Edel’s
cases are not a random sample.
4
Camilo Torres Restrepo was a priest, sociologist, and founding member of the ELN. He became
an early martyr of that group.
162 Why Some Communities Are More Organized than Others

receded to marginal mountainous areas. This created conditions for a rapid

normalization” (Zamosc 2001, 107–108).5
The ELN and FARC rebel groups formed in the 1960s, but their activities
remained sporadic, limited, and contained by the armed forces. As Zamosc
continues, these insurgents faced difficult beginnings, “The guerrillas . . . in the
hills were unable to recruit for what the peasants seemed to regard as a useless
continuation of violence” (Zamosc 2001, 112). The army also routed the
communist holdouts known as the Independent Republics in 1964–1965 (in
Tolima and Huila departments; Ruhl 1980). According to Ruhl’s characteriza-
tion, the 1970s were relatively calm in terms of guerrilla activity, “By the early
1970s, the army had reduced the guerrillas to more of a nuisance than a serious
concern . . . Colombian newspapers were praising the armed forces ‘final’
eradication of the guerrillas” (Ruhl 1980).6
The historical record indeed confirms the sporadic presence of guerrilla
groups through the 1970s and into the 1980s (see the growth of guerilla fronts
in Figure 3.2). The ELN only had three fronts through 1980 (Echandía 1999).
The FARC was formed in 1966 but only registered activity in ten municipios
(fronts) until a broader expansion as a “people’s army” in the early 1980s
(Echandía 1999). Furthermore, early paramilitary groups such as MAS (Death
to Kidnappers) only formed in 1983 in Puerto Boyacá and had only sporadic
influence until the late 1980s. Drug cartels, such as the Medellín Cartel led by
Pablo Escobar, also did not greatly expand until the cocaine boom of the mid-
1980s. By this time, statistics show that 30,007 juntas had already been created
(Ministerio de Gobierno 1993).7
Finally, the historical record also shows that the conflict greatly expanded
over time from the mid-1980s and onward. Figure 3.2 shows that both the
number of guerrilla fronts and the national homicide rate have increased since
the 1960s, but these trends really first accelerated in the 1980s. Furthermore, in
1985, the presence of guerrillas was still quite sparse compared to the subse-
quent growth in activity that would occur over the following two decades.
According to data in Echandía (1999), in 1985 only 173 municipios registered
guerrilla presence and only 123 had “active” guerrilla presence from 1987–
1989, affecting around 15 percent of the country. In contrast, data from 1995
shows 622 municipios had guerrilla presence and 190 had active presence,
affecting 20–60 percent of the country (from 1993–1995). This suggests that
there were many places that were equally peaceful when the juntas measure was

5
For instance, while there was still some guerilla violence in the early 1960s, Ortiz shows that there
were 16,000–18,000 deaths between 1958 and 1965, but that by 1963, violence greatly receded.
6
This is not to deny that certain parts of Colombia did indeed continue to experience armed
conflict, just that it became far less common. For data on events of army “repression” during this
period, see Torres and Barrera 1982.
7
Regression analysis also shows that 1985 data on the distribution of insurgent activity by
municipio does not correlate with juntas.
Selecting Cases Under Constraints in Multimethod Projects 163

taken and that, where juntas were prevalent, they went on to reduce violence in
some of the areas that were later touched by the expansion of the conflict.
This discussion casts doubt on the possibility that the legacy of La Violencia
and any post-La Violencia violence significantly affected the distribution of
juntas in a way that would raise the specter of biased results due to reverse
causality. Colombian history is actually fairly consistent with my theory. Even
today, anecdotal evidence suggests that, in certain areas, violence has been the
impetus behind many of the social movements to protect human rights.

selecting cases under constraints in

multimethod projects
The statistical analysis of the drivers of juntas was used to choose cases for
qualitative study. I outline a new set of procedures to select cases for qualitative
analysis in multimethod research projects when flexibility or quantity in choos-
ing cases is constrained. The constraints could include researcher time or
resources, safety, missing information, or other factors that commonly crop
up in the field of comparative politics and preclude large qualitative samples,
randomized selection, or gaining more detailed information on certain cases.
The method should be especially useful for the increasing trend of multimethod
subnational research designs that involve fieldwork and many potential cases
from which to choose (it could similarly be helpful for systematically incom-
plete archives). This method proves useful here for selecting municipio-cases in
Colombia for qualitative study to test the relationships between social organ-
izations and levels of violence in a civil war/post-civil war context where
selection is constrained by safety and resource considerations. In the end, I
used the model to select six additional towns from two case study areas, which
are profiled later in this chapter.8
Regardless of the various goals of case-study research, as Geddes (2003)
discusses in her chapter entitled “How the cases you choose affect the answers
you get,” a clear rationale for selecting cases is crucial for minimizing selection
bias. Case selection that is based on a statistical (large-n) model of relationships
and unit characteristics helps resolve this problem by providing insight about
where particular cases lie among the entire universe of cases on observable
traits. Fearon and Laitin (2008) call for randomized case selection to minimize
researcher bias from selecting cases with which they are already familiar and
increase representativeness. This helps limit omitted variable bias but requires
the ability to potentially study any case in the sample and cull a large number of

8
I originally selected two additional towns to study in the western department of Caldas, Pensil-
vania and Samaná, as well as the neighboring town of Nariño in Antioquia. Insights from
fieldwork I conducted in Pensilvania are consistent with civilian autonomy theory, but I was
not able to conduct similar research in the other towns. These comparisons are therefore not
included in this study.
164 Why Some Communities Are More Organized than Others

The town of Pensilvania, Caldas, Colombia, 2009. During field visits it was observed
that villages in the surrounding countryside exhibited similar patterns of juntas
de acción comunal advocating for civilian autonomy to towns in Cundinamarca
and other parts of the country. Photograph by Oliver Kaplan.

qualitative cases to achieve representativeness. Lieberman (2005) calls for

“nesting” cases within large-n models by selecting cases on the regression line
that have wide variation in the independent variable of interest. This strategy
allows the researcher greater flexibility and takes into account observed differ-
ences between cases but does not mitigate unobserved heterogeneity not meas-
ured in the large-n analysis.
The method I outline here aims to balance latitude in selecting nested cases,
as may be required by circumstances, with representativeness. Further, it aims
to provide guidance in making smart choices given the sometimes overwhelm-
ing number of possible cases for study when researchers may not have clear a
priori impressions of where they lie within distributions of potentially con-
founding or unmeasured variables. Multimethod projects involving field
research components in conflict settings can especially benefit from this
method. There are often obvious limitations of safety, costs, or difficulties to
traveling to certain regions in the midst of conflicts. Even post-conflict situ-
ations can be dangerous, as security situations can change rapidly and be
difficult to assess.
Selecting Cases Under Constraints in Multimethod Projects 165

It is almost impossible to avoid the critique that any safe place for the
researcher risks selection problems and is therefore not likely representative
of all cases. But with this method, at least within the realm of “safe” sites, the
researcher can have increased confidence that conflict intensity is controlled for
and that cases are similar. The method’s transparency helps readers make their
own assessments of the severity of this issue. This method could therefore be
called the method of “living to tell about it,” since it allows for safety but also
helps assure that the “it” – the story and information collected – will be
worthwhile, with representativeness and inferential value.
The main challenge, then, is how to minimize both observed and unobserved
heterogeneity biases. Using propensity-score matching techniques with a bivari-
ate dependent variable or picking cases on the regression line can help deal with
matching on observable characteristics (Seawright and Gerring 2008). To set
up a quasi-experiment, the next step is to choose cases among those with
similar propensities for independent variable “treatment” so as to maximize
variation on the independent variable(s) of interest (in an Ordinary Least
Squares regression framework, this entails minimizing the residuals while
maximizing the dispersion on the independent variable; Lieberman 2005).
The statistical techniques are really shortcuts that help deal with the complexity
of a multidimensional (multivariate) problem of assignment to the “treated”
group. It is therefore also helpful to compare the actual variable values across
the pool of matched cases to see how closely they each correspond and ensure
that great discrepancies in certain observable factors are not overwhelmingly
driving differences in propensities (since these too could ultimately be correl-
ated with the dependent variable of interest).9
Still, there can remain the problem of unobserved bias or global (nonregion
specific) explanations for an observed correlation between the independent and
dependent variable.10 Traditionally, omitted variable bias is addressed in infer-
ential sciences in three principal ways: (1) measuring variables (so they are no
longer omitted), (2) randomization (so that omitted variables are orthogonal to
treatment), or (3) assumptions about how cases can be matched on unobserved
variables. When researchers reach the case selection stage of a project they have
usually already collected all the large-n measures that are available or thought
to be relevant. If they then face a (security) constraint on the number or types of
cases they might access, randomized selection may become impractical. At that

9
This procedure has limitations for matching since the calculated propensities are additive
combinations of variables that allow for greater differences on some variables to be offset by
similarities on others. Exact matching on certain variables is one solution. This matching
technique may better stand on its own for large-n statistical use than picking particular cases
where the precise values of a particular variable can matter a great deal. Balance tests can be
useful for identifying whether, overall, treatment and control were assigned as-if randomly to
their respective groups by sharing similar values on covariates.
10
Since there is matching within the region, these are global factors that might cause the selection
of the matched region itself to not be representative of the larger universe of cases.
166 Why Some Communities Are More Organized than Others

point, assumptions about unobserved similarities – how cases fit within a given

strata of a variable of interest – can be employed to further reduce bias.
A useful assumption is Tobler’s first law of spatial analysis, which states
that, “Everything is related to everything else, but near things are more related
than distant things” (Tobler 1970).11 Applying this law would call for the
researcher to select proximate or neighbor cases. Of course, a researcher might
equally apply a different “law” that could provide guidance on how character-
istics of units are generally related or stratified.12 Selecting matching neighbors
on the geographic dimension should also have the added advantage for the field
researcher of making it easier to transit from one site to another if there is access
between them (roads, paths, waterways, etc.).13 For increased representative-
ness, clusters of pairs of matched cases should vary on the geographic dimen-
sion as well, drawing from diverse regions. The researcher may then apply
other considerations based on constraints as necessary, such as safety, distance
to travel, available case histories, etc., or to stratify or match on additional post-
“treatment” (or secondary treatment) variables that do not belong in a selection
equation.
The development and application of this method for case selection sets this
study apart from prior micro studies on civil war. Few if any studies have
selected cases rigorously and transparently to deal with problems of reverse
causality or omitted variable bias. Lacking transparency and theoretically
grounded procedures, it can be difficult to have a solid a priori understanding
of the representativeness of the cases chosen by an analyst. Wood’s (2003)
insightful study of civil war in El Salvador does not use quantitative methods
and selects what she deems “typical cases” that also meet the criteria of safety,
military contestation, political tractability, and variation in agro-economies.
Without a large-n component, however, it is hard for readers to evaluate
whether the villages in her comparison set are useful contrasts or are truly
representative of all villages and therefore assess tests of her theory. The case
selection therefore renders the analysis more useful as a theory-building exer-
cise (she originally began with a different research question but then shifted to
studying civilian participation upon visiting her sites). Kalyvas’s (2006) book
discusses what he deems to be “representative” towns from the Argolid region
of Greece. Through the wide-ranging interviews upon which some of his

11
The neighbor criterion could also be incorporated into the regression framework, either through
contiguity indicators or latitude and longitude for closeness. Technically, this geographic
assumption requires the additional assumption that the likelihood of encountering different
neighbors on the independent variable is orthogonal to the dependent variable. In this applica-
tion, it means that towns with differences in juntas do not happen to cluster together for some
reason that is also correlated with violence.
12
E.g., perhaps a temporal rule, such as towns that were founded in a given period.
13
With contiguous units, there is the chance of contamination or spillover effects from one case to
the other. Hopefully, selecting cases based on variation in the key independent variable limits this
problem.
Selecting Cases Under Constraints in Multimethod Projects 167

quantitative measurements are based, he assembles a broad coverage of cases,

but does not directly confront selection issues or match villages.

Selecting Cases to Study Junta Councils

The main goal of studying cases in depth is to trace the mechanisms that might
account for the effect of juntas on violence encountered in the statistical
analysis and probe how resilient these mechanisms are to armed group pressure
in different settings. The new cases should help answer a number of questions
that are difficult to study with large-n data. For instance, do junta mechanisms
account for trends in violence, or are these trends accounted for by other
explanations? Is the juntas variable a good measure of social capital? Were
other organizations besides juntas more important for explaining violence? Did
juntas represent community interests, or were they subjugated to clientelist
relations? Did juntas align with armed groups or instead implement tactics to
maintain their autonomy from influence and violence? Are there confounding
factors that keep theorized junta mechanisms from transferring to affect vio-
lence? New cases can help uncover additional empirical support for the mech-
anisms and highlight new, unknown processes (theory building).
In selecting cases, I faced constraints of safety and resources when I opted to
conduct research in “post”-conflict Colombia. It is a country with wide vari-
ation on civilian movements for autonomy from conflict but has also been
known for kidnapping of foreigners. As I would later find out, even the towns
I visited that were recently “post-conflict” had security concerns. These were
towns that most urban Colombians had never heard of (epitomizing the coun-
try’s urban–rural disconnect). I was warned of the possibility that informants,
militants, or gangs were possibly lying in wait, and found that information
travels fast in small towns, so that everyone and anyone would know within
days that a gringo had arrived.14 I also learned that not long prior to my visits
to some towns and even possibly during them, there were FARC spies on the
very bus lines I sometimes used to get in to and out of the case-study regions
(fortunately for my peace of mind, I only found this out on the way back from
my site visits). Youths were killed in nearby villages shortly before, after, and
even during some site visits. Militants were still turning themselves in to the
town Ombudsmen for demobilization (in one instance, this happened the night
before I interviewed one). Propaganda “pamphlets” intended to intimidate
residents were also distributed.
Instead of aimlessly setting out into the Colombian countryside, and with
time constraints, my approach led me to fewer and more purposefully selected
cases. The selection modeling is useful as a way of thinking about which towns

14
Fortunately, many of my contacts were looking out for me and were also plugged in to the local
situation (coyuntura) or had sufficient prestige to keep undesirable events from transpiring.
168 Why Some Communities Are More Organized than Others

were selected into the “treatment” group of having juntas and approximating
quasi-experimental conditions. So, in addition to the plausibility probe case
studies on the ATCC (Carare Peasant Workers Association), I conduct three
new comparisons of matched municipios as theory-testing cases. Since there are
990 municipios in Colombia and many variables to control, I first use the
statistical propensity-matching procedures to determine a town’s likelihood of
having juntas.
To test the first proposition from Chapter 5 that juntas decrease violence,
I matched cases on all other characteristics using the propensities and looked
for pairs that diverged in levels of juntas. I looked for cases that had relatively
equal probabilities of receiving the “treatment” of having juntas from the
middle range of propensity scores (support) that were neighbors or in the same
department.15 Selecting cases from this middle range of support should help
assure that the cases are fairly typical municipios (expected to have average
levels of juntas but actually have more or less) and reduce the likelihood that
omitted variables or “noise” are causing observed differences in juntas between
pairs.16 Choosing neighbor towns helps further reduce unobserved differences
among towns and also eases passage between towns once in the field.
I was also interested in the histories of armed conflict since I hypothesize
that civilian organizations will only be likely to reduce civil war violence in areas
that actually suffer from civil war. I further cull from the pool the group of towns
that suffered at least some attacks according to the data and manually match
cases on conflict dynamics in the 1990–2006 period of the study by both intensity
and mix of active armed groups. This measure of conflict intensity is not included
in the propensity model because it is “post-treatment” (i.e., it is measured subse-
quent to the juntas and so does not explain them; it could be considered a second
treatment whose propensities could be modeled separately). As an additional
assurance, I also restricted the possible cases to those that are coded as having
experienced high levels of violence during La Violencia of the 1950s to decrease
the likelihood of including places that have been historically peaceful.
To test the second proposition about the possible attenuating effect of the
intensity of armed group activity on junta council capability, I matched a
second set of cases (in Caldas) with high levels of juntas but varying levels of
armed group activity in the form of attacks (recall from Chapter 5 that this
variable is measured apart from incidences of violence against civilians).

15
I dichotomize the dependent variable of juntas per capita and use a probit model to identify cases
that cluster around probabilities of 0.5. I also consider an OLS model since the actual juntas
variable is interval and look at cases that cluster around the mean of predicted values. Compari-
son sets could be chosen from different propensity strata for generalizability if there is confidence
that the model is complete and does not suffer from omitted variables.
16
For example, pairs at an extreme end of the propensity distribution could have one case that is
extremely mispredicted and was expected to receive (not receive) the treatment but did not for
some reason possibly correlated with the dependent variable. One can of course still learn from
such cases, but they pose a greater challenge to solely examining the comparative statics of a few
key variables.
Selecting Cases from Cundinamarca 169

Armed group Low to medium High

actions

Juntas councils
Prediction: No juntas effect Displacement/violence
Quipile (Cundinamarca)
Low

ATCC neighbors
Prediction: Juntas decrease violence Weak or no junta effect
Bituima, Vianí
High (Cundinamarca)

ATCC 1990s ATCC pre-1987; post-2000

figure 6.2 Comparisons of selected cases

Although I could have constructed a second propensity model for attacks, I

instead “eyeballed” cases with junta propensities close to 0.5, high levels of
actual juntas, and that are in the same region but vary in average numbers of
attacks over time. The application of these case selection methods ended up
being fairly “blind” for the researcher – I knew little about the cases beforehand
and the actual levels of violence do not on the surface strongly support or
undermine my hypothesis.17 The cases I arrived at and their predictions for
juntas and violence are shown in Figure 6.2 (and map of Figure 6.3).
As noted in Chapter 4, I spent between two and four weeks in each of these
sites interviewing (randomly where possible) past and present junta leaders and
members, and members of other relevant organizations. I also interviewed
populations of ex-combatants, displaced residents, and church and NGO
leaders in the zones and nearby cities. Although these different clusters of case
towns may appear close to each other on the map of Colombia, the case zones
are different in their subcultures and conflict dynamics, with different armed
group blocs, fronts, and patterns of contestation, and are at least a day’s travel
apart. They should therefore be helpful for assessing the generalizability of
theory in different subregions of Colombia.

selecting cases from cundinamarca

There are not many regions in Colombia where the number of juntas varies
among neighboring municipalities and yet these municipalities also have similar

17
Whether this constitutes “peeking” and should be done is a question worthy of debate. I argue
that having an understanding of the dependent variable values from the large-n analysis does not
tarnish the case selection as a test so long as one has few a priori qualitative understandings
about the case that would drive selection.
170 Why Some Communities Are More Organized than Others

figure 6.3 Map of case study sites

likelihoods of having high levels of organization and experienced at least some

armed conflict. However, in the western region of Cundinamarca, there are a
handful of counties that match these requirements. The two small neighboring
municipios of Bituima and Vianí both have many juntas for their levels of
population in 1985 (90th–99th percentiles). Quipile, their immediate neighbor
to the south (as well as the nearby towns of Chaguaní and Anolaima), is
measured as having a below-average number of juntas per capita (approxi-
mately 25th percentile).18 Table 6.4 and Figures 6.6 and 6.7 display how the

18
This case would ideally have zero juntas, but I decided it had sufficiently few juntas to be a viable
test and thought it more important to geographically match neighbor towns than squeeze slightly
more variation out of the Juntas variable. This choice may also help avoid falling prey to possible
measurement error in the Juntas indicator if zeros are relatively more likely to be mistakenly
measured. The irony was not lost, however, that I may only be able to access and learn about
“low” junta cases and their levels of organization by interviewing junta leaders.
Selecting Cases from Cundinamarca 171

cases compare on actual versus predicted values for juntas (Figure 6.8 displays
their predicted homicide rates based on Model 4 of Table 5.2). This mountain-
ous region with colonial origins is about three hours west of Bogotá and it has
historically suffered from state neglect (poor roads, water quality, health ser-
vices, etc.). Other than slight differences in elevation, the towns look similar
from a distance, with small populations of between seven and ten thousand
people and agriculture-based economies (coffee, cane sugar).
These towns are contiguous and share similar geographic settings, all lying
between road corridors that radiate northwest and southwest out of Bogotá.
However, there are some apparent differences. Vianí and Bituima are two
different towns but for analytical purposes I decided to consider them as a
group, since together they have similar values on the Juntas indicator, closely
match Quipile on important observables such as population and poverty, and
were originally a single municipio until Vianí split off. While the analysis of two
town centers and municipal administrations in two treatment towns could
cause problems for inference (if somehow correlated with both social networks
and later violence), Quipile also has several corregimientos (noncounty-seat
“urban” areas) that could approach the size of the county seat of a small
municipio. By way of comparison, I decided the inclusion of the low-junta
town of Anolaima did not make sense. Even though it is in the same region, its
level of urbanization means state presence might be greater, leading to different
conflict dynamics and a smaller potential role for social organizations.
An advantage of this pairing is its proximity to Bogotá and the feasibility of
traveling to the region. Despite this location, it apparently historically suffered
moderate to high levels of armed group influence. The towns in the subregion
seemed to have experienced similar degrees of influence by the same armed
groups, likely due in part to their compact sizes (Figure 6.4). They had been
largely dominated by the FARC, and even the same guerrilla fronts, during
most of the time period of study. For some periods, paramilitaries briefly
extended their reach, and the army mounted counterinsurgency operations over
about a four-year period from 2000 through 2004. A key question to be
investigated in Chapter 8 is whether the conflict dynamics fit my scope condi-
tions for junta-based autonomy of experiencing the presence of multiple armed
groups and ongoing uncertainty about levels of contestation.
These counties should provide a tough test for junta autonomy mechanisms.
First, their dynamics of armed group control – possibly switching abruptly from
FARC control to army control – may not have created conditions where civil
society independence would be helpful. Second, there are no known, formal
meta-organizations such as peace communities in these areas. Third, even with
the measured historical disparities in junta coverage from one town to the next,
as close neighbors, these towns may still share social organization characteris-
tics (because of proximity, diffusion, and possible measurement error), making
it unlikely to observe effects based on differences in juntas. Fourth, most of
these towns are recorded as having suffered high levels of violence during La
172 10

8
Total armed group actions

6
Bituima & Vianí
Chaguani
5
Quipile
Anolaima
4

0
1990 19911992 199319941995 19961997 1998 1999 2000200120022003 2004 2005 2006
Year
figure 6.4 Armed group actions in Cundinamarca case towns.
Source: Government of Colombia.
Conclusions 173

Violencia, so any observed negative relationship between juntas and violence

should not stem from these areas being historically peaceful. Lastly, a peek
at the dependent variable of violence suggests that the high-junta counties
actually suffered high homicide rates for some years. This prompted the goal
for the qualitative research of verifying the accuracy of the large-n violence
indicators, whether all these cases really did suffer similar levels of contestation,
and, counterfactually, whether they would have suffered more violence had
they fewer juntas. Overall, these towns provide a fruitful contrast with the
culturally distinct, highly organized ATCC case of the low-lying Magdalena
Medio region.

the atcc cases in santander

The ATCC case (Carare Peasant Workers Association) of Chapter 7 was
chosen prior to my statistical modeling as a plausibility probe. While this
organization spans several municipios, I still try to approximate a match on
other community characteristics by comparing the ATCC with nearby neighbor
communities. I also exploit within-case variation inside the ATCC to test
theories at smaller units of analysis (regions, villages, and incidents) and over
time. These “cases” include:
1. ATCC area over time:
i. Pre-organization (1975–1987)
ii. Post-organization (1987–2000)
iii. Post-resurgence of violence (2000–2007)
2. ATCC area vs. grouped neighboring areas (the large villages of San
Tropel, Santa Rosa, Miralindo, La Sabana, and other smaller villages)
3. Intra-ATCC comparison of villages and threat incidents.

conclusions
This chapter provided a better understanding of the origins and prevalence of
the junta councils. In doing so, it helped to confront some of the potential
reverse causality and selection bias issues presented in Chapter 5. Beyond the
relevance of this analysis for the study of violence, it is also one of the first
systematic studies of the historical social and organizational landscape of
Colombia. This involved identifying the role that the monumental event of La
Violencia played in reshaping Colombian society as well as the importance of
inequality as a determinant of social organization.
This analysis explains why, if juntas are so effective, every community does
not already have a strong junta as of today. Both social and structural factors
influence the formation and persistence of junta councils: they are found to be
more prevalent in areas that are less populated, are poorer, have more
churches, have greater land equality, are less politically polarized, and received
174 Why Some Communities Are More Organized than Others

support from promoters. Juntas are also less prevalent (though not signifi-
cantly) where there are sizable ethnic minority populations, who tend to form
their own social organizations. Still, the question is open to further study, as
there are several possible additional reasons why juntas are not (active) every-
where. One major reason could be that isolation and limited information and
communication has slowed policy diffusion – that only relatively recently did a
consensus begin to form among different Colombian campesinos (or churches
and NGOs that support them) that organizational structures can make a
difference. Juntas were also not originally envisioned as being active or useful
in conflict settings (but were intended to prevent conflict recurrence). Instead,
the distribution of juntas may have been partly determined by communities’
differing needs for public goods provision, or other organizations, such as labor
unions, may have filled a collective action gap.
The chapter deals directly with the possibility that juntas are derivative of the
conflict through path-dependent processes and plausibly rules this out as an
explanation for the observed association of more juntas with less future vio-
lence. I find that juntas formed more prevalently in La Violencia zones or at the
least were no less prevalent in such zones. The finding conforms to the historical
consensus on the effects of La Violencia and of junta development and promo-
tion. For Colombia, at least, this is further evidence that the formation of
civilian organizations such as juntas and their effects on violence are not
completely determined by the preferences of armed actors, the dynamics of
conflict, or latent propensities for conflict.
Lastly, the analysis of selection bias issues aided the development of a case
selection methodology that is particularly useful for coping with researchers’
constraints, including limited resources or working in conflict conditions. After
deploying these techniques in this chapter, the next two chapters discuss the
small-n research findings from the selected matched cases. As a product and
benefit of this methodology, these cases are not “just so” stories that might be
convenient examples of theory (Elster 2000; Bates et al. 2000). They are
embedded in a larger explanatory framework and are structured, focused
comparisons (George 1979) that take careful measures of key variables. The
cases have known variation in variables, but are “out-of-sample” with regard
to the cases that were used to theorize about specific protection mechanisms.
With their mechanisms unknown beforehand, they help set up civilian
autonomy theory for possible falsification.
Appendix

This appendix describes the measurement of variables and statistical tests in

this chapter in greater detail.
I model the variables that explain variation in juntas as a simple OLS
regression with region effects since panel data on these variables are not
available for the time period prior to 1985. The summary statistics for the
dependent and independent variables are displayed in Table 5.1 in Chapter 5.
The regression results are displayed in Table 6.1, Table 6.2, and Table 6.3, and
show that demographic, social, and geographic variables are all significant
predictors of junta coverage (to ease interpretation, Juntas per communities is
multiplied by 100 and re-scaled to reflect percentages). Table 6.1 is a base
model and Table 6.2 and Table 6.3 add additional covariates (though they
drop cases). For comparison, the dependent variable for Table 6.2 Model 4 is
the raw number of juntas instead of the normalized Juntas per communities
variable.
Table 6.3 Model 1 uses a binary dependent variable for the propensity-score
matching case selection procedure. This makes sense since, according to Figure
5.7, the distribution of juntas per capita is bimodal despite a long tail. In OLS
models of junta formation I consider closeness of cases to the regression line
(minimizing residuals). Below, I model a cross-sectional selection equation to
explain the likelihood of having juntas.19 Figure 6.5 displays the distribution of
treatment and control cases by propensity quartile and shows that this method
is effective at separating “experimental” groups.

19
This model used to select cases is based on the data I had collected up to this point in my research
process. The other junta selection models presented earlier were developed later and are more
complete. However, the predicted case matches do no vary much among the various models
because they share many similar characteristics.

175
176 Why Some Communities Are More Organized than Others

table 6.1 Models of juntas per “communities” in 1985, by municipios

(1) (2) (3)

Log 1993 population −1.863** −1.896** −1.770**
(−10.17) (−9.98) (−7.81)
Population density −0.000 −0.000 −0.002^
(−0.10) (−0.20) (−1.83)
La Violencia 0.850* 0.855* 0.791*
(2.38) (2.39) (2.16)
Elevation (m) −0.000 −0.000 −0.000
(−0.29) (−0.24) (−0.31)
Distance to dept. capital −0.003^ −0.003^ −0.004^
(−1.79) (−1.78) (−1.77)
Pct. Minorities −0.497 −0.487 −1.464
(−0.41) (−0.40) (−1.16)
Pct. Dirt floors 2.078* 2.004^ 1.345
(2.01) (1.93) (1.21)
Paved road access −0.556
(−1.59)
Total guerrilla attacks (1985) 0.068 −0.224
(0.66) (−1.47)
Churches 0.083*
(2.48)
Caribbean region 0.259 0.260 0.579
(0.27) (0.27) (0.59)
Pacific region 5.146** 5.067** 5.338**
(4.98) (4.87) (4.98)
Andean region 3.186** 3.123** 3.322**
(3.21) (3.13) (3.23)
East Andean region 5.621** 5.560** 5.862**
(5.61) (5.52) (5.68)
Eastern region 8.592** 8.537** 8.828**
(7.63) (7.56) (7.66)
Constant region 20.068** 20.411** 19.324**
(9.22) (9.12) (7.96)
No. Municipios/Observations 929 929 896
R-squared 0.34 0.34 0.34
t-statistics in parentheses
** p < 0.01, * p < 0.05, ^ p < 0.1
Appendix 177

table 6.2 Models of juntas per “communities” in 1985, by municipios

(1) (2) (3) (4)

Juntas per Juntas per Juntas per Number of
communities communities communities juntas
Log 1993 population −2.899** −2.825** −3.255** 7.760**
(−11.04) (−9.98) (−11.32) (8.09)
Pct. Dirt floors 2.917* 3.927** 3.571** 5.510
(2.34) (2.99) (2.80) (1.30)
Population density 0.001 0.001 0.001 −0.007*
(1.22) (1.45) (1.15) (−2.13)
Distance to dept. capital −0.001 −0.001 −0.001 0.010
(−0.44) (−0.46) (−0.40) (1.03)
Std. dev. elevation 0.003** 0.003** 0.003** 0.007*
(3.77) (3.36) (2.93) (2.45)
Pct. Minorities 2.605 2.611 1.759 4.571
(1.24) (1.15) (0.84) (0.65)
La Violencia 0.121 0.280 0.008 0.215
(0.30) (0.63) (0.02) (0.14)
Land value GINI −8.297** −19.801**
(−3.66) (−2.62)
Polarization −1.277* −0.892 −2.551
(−2.17) (−1.52) (−1.30)
Cooperatives −0.015 −0.021 −0.040 −0.181
(−0.44) (−0.61) (−0.84) (−1.14)
Villages 0.052** 0.052** 0.064** 0.337**
(4.67) (4.59) (5.71) (9.01)
Churches 0.035 0.037 0.093^ 0.509**
(0.90) (0.96) (1.90) (3.13)
Year municipio founded −0.000 0.001 −0.000 −0.010^
(−0.15) (0.46) (−0.07) (−1.83)
Caribbean region 0.151 −0.052 −0.138 5.929
(0.12) (−0.03) (−0.10) (1.30)
Pacific region 5.210** 5.634** 5.899** 24.550**
(3.94) (3.66) (4.07) (5.07)
Andean region 4.169** 4.302** 6.582** 24.561**
(3.24) (2.83) (4.62) (5.16)
East Andean region 6.329** 6.388** 5.613** 22.040**
(4.91) (4.19) (3.96) (4.66)
Eastern region 10.314** 10.126** 9.094** 15.871*
(5.00) (4.43) (4.32) (2.26)
Constant 27.898** 25.889** 37.112** −51.191**
(6.10) (5.26) (7.40) (−3.06)
Observations 682 585 507 507
R-squared 0.42 0.41 0.51 0.59
t-statistics in parentheses
** p < 0.01, * p < 0.05, ^ p < 0.1
178 Why Some Communities Are More Organized than Others

table 6.3 Junta case selection models

(1) (2)
Probit OLS
Juntas per communities Juntas per communities
Log 1993 population −0.457** −1.781**
(−6.04) (−7.86)
Elevation (m) 0.000** −0.000
(5.14) (−0.11)
Distance to dept. capital −0.001* −0.004*
(−2.16) (−2.05)
Pct. Minorities −0.337 −3.026*
(−0.81) (−2.01)
Pct. Dirt floors −0.153 0.340
(−0.48) (0.29)
Literacy rate −0.537 −4.542*
(−0.81) (−2.00)
Churches 0.002 0.049*
(0.07) (2.01)
Paved road access −0.281* −0.481
(−2.44) (−1.37)
Population density −0.005** −0.001
(−4.43) (−1.23)
La Violencia 0.360** 0.796*
(3.45) (2.17)
Caribbean 0.477
(0.49)
Pacific 5.216**
(4.92)
Andean 3.136**
(3.07)
East Andean 5.847**
(5.68)
Eastern 8.659**
(7.52)
Constant 4.399** 22.946**
(4.93) (7.75)
Observations 896 896
Pseudo R-squared/R-squared 0.21 0.34
z-statistics / t-statistics in parentheses
** p < 0.01, * p < 0.05, ^ p < 0.1
Appendix 179

300

251
250
Number of Municipio-cases

200
168 167
Number Control
150 Number Treated
114
104
100

51 56
50
30

0
0–25th% 25–50% 50–75% 75–100%
Propensity Quartiles
*Treatment group if Juntas Per Community variable >.08
figure 6.5 Distribution of junta-treated and control cases by propensity-score quartile

figure 6.6 Propensities for junta “treatment” of Cundinamarca towns

180

table 6.4 Characteristics of Cundinamarca towns

Selected Towns Nonselected Comparison Towns

Municipio Quipile Bituima Vianí Anolaima Chaguaní Guayabal de S.
Juntas per communities (1985) 4.4% 20.8% 16.5% 0.0% 2.7% 9.4%
Predicted prob. of high juntas “treatment” 0.52 0.61 0.71 0.31 0.7 0.61
(if Juntas Per Comm.>.08)
Number of juntas (1985) 9 14 13 0 3 8
Number of villages (1965) 36 14 17 49 25 11
1993 population 10,033 2,932 4,107 12,959 5,080 3,835
Number of city blocks 55 33 36 181 26 37
Elevation (m) 1,443 1,412 1,498 1,657 1,200 1,630
Std. dev. elevation 305.9 209.9 221.3 426.0 472.1 237.3
Distance to dept. capital (km) 82 111 104 71 113 62
Pct. Minority population (1985) 0.1% 0.1% 0.1% 0.2% 0.2% 0.3%
Pct. Dirt floors 32.6% 24.0% 14.2% 13.1% 25.4% 16.4%
Literacy rate 72.0 71.0 75.2 75.1 70.5 77.7
Total churches (town, 1995) 1 1 2 3 2 1
Paved road access (1985) 0 0 0 0 0 0
Population density (1985) 78.4 48.1 60.4 106.2 30.4 65.0
1950s La Violencia 1 0 1 0 1 0
 Year founded 1900 1772 1853 1882 1700 1845
Officials per 1,000 residents 1.1 5.8 2.1 1.2 2.5 0.5
Number of police stations (town) 1 1 1 1 1 1
Number of police inspections (town) 0 1 0 1 1 0
Coca area (ha) 0 0 0 0 0 0
Total AUC actions 0 0 0 0 2 2
Total FARC actions 11 5 4 4 6 13
Total ELN actions 0 1 2 0 1 0
Total army actions 3 2 2 3 11 10
Total armed group actions, 1990–2006 (gov’t) 14 8 8 7 20 25
Total armed group actions, 1991–1999 (CINEP) 3 1 2 2 1 1
Total armed group actions, 2000–2005 (CINEP) 4 2 0 3 4 3
Avg. police homicide rate (1990–2007) 42.8 42.1 44.4 54.5 31.2 84.3
Avg. displacement rate (1998–2005) 868 550 831 205 1,029 1,076
181
182 Why Some Communities Are More Organized than Others

figure 6.7 Predicted junta values of Cundinamarca towns

figure 6.8 Predicted vs. actual homicide rates in Cundinamarca towns

7

The Institution of the ATCC

Protection through Conciliation

Thank God for the Association. If it weren’t for the Association’s mediation on
my behalf, I’d be dead now. They’re good people. Many accused owe their lives to
the Association.
– Interviewee (ATCC#2), La India, Colombia, November 2007

Late one February night in 1987, a small group of leaders from various
village councils along the Carare River, in the heart of Colombia, met in secret.1
They were respected family men who had been in the region for some time and
knew each other well. In the back room of a house, they discussed the pressing
topic of how to respond to a threat of violence against the community, an
ultimatum from multiple armed groups giving residents a choice to displace,
join one of them in the conflict, or be killed. The community had already seen
years of atrocities, and even their small gathering risked great danger – should
anyone have seen them or passed word of their discussion on to any of the
armed groups, they would have been killed.
They were caught between armies, but what were they to do? If they threw
their lot in with the army or paramilitaries, the guerrillas would surely find out
and kill them. And yet, if they joined the guerrillas in hopes of protection, the
paramilitaries would have no mercy. “Well,” one man proposed, “we could
find some weapons – take up arms and defend ourselves.” Others demurred,
arguing that they were not soldiers, had no weapons, and would easily be
crushed by the standing armies. “What’s worse,” another said, “we would be

1
This vignette is based on secondary sources and several interviews, some of whom were present in
these discussions (ATCC#3,6,7,8, La India, 10/2007). The contents of this chapter were previ-
ously published in Kaplan 2013a.

183
184 The Institution of the ATCC: Protection through Conciliation

no better than the armed groups, and then they would have every right to
target us.”
To manage this problem of stigmatization, the discussion came around to a
fifth option. From then on, they would manage their own affairs and would not
take any part in the conflict among the armed groups. Unsure how the armed
actors would respond, they sought them out in motor-canoes to declare they
would neither leave nor take any sides. Surprisingly, after many months of
discussions, the various armed groups acceded to the civilians’ policies. The
result was the formation and survival of an organization called the Peasant
Workers Association of the Carare River, or ATCC.2 Over the following years
and decades, the organization ostensibly developed the agency to create phys-
ical and political space and deal with the uncertainty and continuing risks of
civil war. But was the organization itself really effective over time and across
space at providing certainty and security to civilians? How did it function? In
what ways is this experience unique? This chapter analyzes the case of the
ATCC by bringing methodological structure to Colombia’s culturally rich rural
communities.
The case of the ATCC illustrates some of the mechanisms through which
local organizations, which were found in the statistical analysis to reduce
violence, actually function to protect civilians. The ATCC is a good case for
studying the phenomenon of civilian autonomy and dispute resolution strat-
egies because it exhibits variation over time (within the case and within its area)
on the prevalence of violence as well as the presence and functioning of its
institutional procedures. The ATCC’s municipios also register above-average
historical levels of local junta councils and therefore have experienced less
violence than would have otherwise been expected.3
Violence against civilians arising from armed groups’ exploitation
of neighborly disputes has been shown to be pervasive in civil wars. Sometimes
the victims of killings have participated in the conflict by collaborating with a
rival army. In other instances, suspects are killed when residents make false
accusations against their neighbors and denounce them to armed groups to
achieve personal ends. This chapter argues that local justice procedures can
substitute for state justice to solve these problems in areas where there is a “fog
of war” and state presence is weak.
The ATCC experience is remarkable for the apparent effect it has had
on violence. Although approximately 10 percent of the population in its
region was killed over a twelve-year period leading up to its formation,
it was able to negotiate accords and procedural rules with the various
local armed groups. The subsequent period from 1991 to 2000 saw an
absence of violence during which time there were reportedly no civilian

2
As previewed in Chapter 1, La Asociación de Trabajadores Campesinos del Carare in Spanish.
3
The importance of juntas to the ATCC process is described below.
The Institution of the ATCC: Protection through Conciliation 185

victims.4 The evidence suggests it is hard to deny some independent impact

of civilian institutions on levels of violence. Neither the acceptance of the
accords nor the cessation of violence in the case of the ATCC can be
completely explained as resulting from permissive preferences of armed
actors.
I make three claims to argue how and why these institutions can manage the
“identification” problem common in civil wars, when there are incentives to
implicate even upstanding civilians as collaborating with armed actors. First, a
civilian institution that is larger than any one person and persists over time can
act as an investigatory body to evaluate denunciations by armed actors and
send a signal that “separates” (exonerates) pacifist civilians from belligerents.
Since individuals may have incentives to misrepresent their private preferences
and actions in the “fog of war” (e.g., Fearon 1994), these signals help resolve
uncertainty and should make killing more difficult or even costly and reduce
mistakes in targeting and costs of governance. Second, civilian norms of non-
violence and nonparticipation in the conflict are necessary to minimize the
participation of individual civilians in the conflict in the first place. Third, there
must be some minimal degree of joint interests between armed actors in not
harming civilians so that they will abide by the civilian institutional
arrangement.
Evidence for the theoretical argument is based on qualitative and quantita-
tive data collected through field research in 2007 and 2008. I conducted
approximately forty-five interviews of ATCC residents in both Bogotá and
the ATCC zone, including founders and presidents, as well as residents who
were currently or formerly involved in the coca economy, recently migrated to
the area from neighboring zones (or traveled the region widely), and had
personally been threatened by armed groups. I was also able to access the
community archives at the ATCC offices, which contained verbatim minutes
of meetings between the ATCC and guerrilla and paramilitary leaders and army
officials, and discussions among the ATCC’s governing council; diary entries;
local census data; and copies of the ATCC’s institutional rules. Lastly, to serve
as a check for the interview data and expand data coverage, I also accessed
existing quantitative data from secondary sources for the ATCC and neighbor-
ing regions.
On a methodological note, this case study chapter is careful to be explicit
about the various within-case cases it uses to test theories. First, variation in
both independent and dependent variables is exploited over time across case
periods. Second, I make comparisons between the ATCC area and its adjacent
neighboring areas that are not part of the organization (but otherwise experi-
enced similar conflict dynamics). Third, the analysis draws on counterfactual

4
Tragically, as the ATCC’s procedures were being consolidated, three of the ATCC’s founding
leaders, Josué Vargas Mateus, Sául Castañeda, and Miguel Ángel Barajas Collazos, as well as
journalist Sylvia Duzán, were assassinated by paramilitaries in the cabecera of Cimitarra in 1990.
186 The Institution of the ATCC: Protection through Conciliation

cases and reasoning. Lastly, I use a within-case dataset at the individual and
village levels of analysis to further test explanations for violence.
The chapter proceeds as follows. First, the ATCC case is put in context with
a description of its setting and recent history. Second, alternative explanations
for patterns in violence are analyzed with qualitative and quantitative data for
the region and found to be incomplete. Third, the ATCC conciliation process is
analyzed as an explanation for violence using an original dataset of threat
conciliations. Fourth, insights about the stability of the ATCC’s procedures
are put to a further test with data from the most recent period of conflict in the
region.

the atcc in context and trends in violence

The ATCC’s “area of influence,” where its members reside and it exercises
decision-making and protective authority, extends across 100,000 hectares of
territory (1,200 square kilometers, or about 400 square miles) and encompasses
thirty-two villages (with thirty-six local councils). The area is nestled along the
banks of the Carare River in a steamy, forested valley crisscrossed by river
tributaries. The area lies about fifteen miles from the county seat of Cimitarra,
but it cuts across the neglected rural peripheries of six municipios. Today,
the ATCC area is home to about 5,000 people, with about 2,000 of them
(300 families) residing in the village center of La India.
The Carare region is an ethnic microcosm of Colombia, with a population
that includes a small group with Indigenous origins, Afro-Colombians who
migrated from the Pacific department of Chocó, and mestizos who migrated
down from the mountain towns of Boyacá and Santander. Yet because of its
diverse ethnic composition and with few roads, the region had few natural bases
for cooperation. The integration of residents came about, in part, through the
complementary productive activities of these different groups. The mestizos
were farmers, while the Afro-Colombian residents, who had previously lived
along the Atrato River, were boatsmen who helped transport goods. The river
network was also an important lifeline of communication and exchange: milk,
produce, and wood would flow down the river, while cases of beer and other
supplies were shipped upriver.
Like many other frontier regions in Colombia, the Carare has historically
had minimal state presence. Since armed actors first entered the region more
than thirty years ago, its history can be separated into approximately three
distinct eras. Across these eras, and even within them, there is variation in levels
of violence, the presence and functioning of the institution, and other significant
independent variables.
The first period began in 1975, when the FARC and ELN guerrilla groups
first moved into the region, and continued through the rise of paramilitaries and
the ATCC’s founding in 1987. In this crossfire, providing aid to one side often
The ATCC in Context and Trends in Violence 187

brought retaliation from the other. This bloodiest period saw an estimated
530 to 585 civilians killed through 1987 (more than 10 percent of the popula-
tion), with 60 percent of those killed at the hands of paramilitary groups and
40 percent by the guerrillas (Restrepo 2005, 72; CNRR 2009).5 Residents said
that it was common to see the bodies of the dead floating down the Carare
River. To counter the insurgency, the army implemented carnetización, an ID-
card monitoring program that required residents to report to a local base every
one or two weeks.6
When the army presented the residents of La India with the ultimatum at
a meeting on February 20, 1987, the ATCC civilians demonstrated strategic
thinking in their response. I found that there were real debates among
leaders about how to respond to the armed groups.7 Agreeing that without
their land they had nothing, yet not wanting to participate in the conflict,
they decided for the “fifth” option – not arming and staying in the zone as
unarmed and neutral. Despite initial hesitation, the guerrilla and paramili-
tary groups each conditionally accepted the civilians’ proposal as long as
their counterparts did as well.
A second period of consolidation began as the ATCC’s institutions and
norms were put in place and armed actors became accustomed to dealing
with the civilians. There was virtually no conflict-related violence from
1987 until the next millennium. The perception of the effectiveness of the
ATCC is reflected in a 1989 journal entry from a former association presi-
dent where he declares, “Today we have passed two years of living better.
There’s no war, no thirst, no hunger. Long live the Association” (ATCC
Archives). Nevertheless, from around 2000 through the present, violence
returned and a third era began for the region. I later analyze this third era
as new data for a test of my mediation hypothesis as an explanation for
violence.
As a skeptical researcher, the purported vanishing violence seemed suspi-
cious. In interviews, I pressed subjects with questions about violence in this
period. Interestingly, in some interviews, people would begin to tell me there
were deaths and massacres in that time. But, when I would press them
harder, asking, “Really, like in 1995?” they would think for a minute and
respond, “No, wait, that was earlier (or later).”8 Overall, my findings
confirmed that there were death threats in the 1990s, but few if any
conflict-related deaths (though some died in machete fights among drunks
or fights over women).9 Impressive as it seems, only a structured analysis can
resolve to what extent this can be attributed to civilian processes as opposed
to causes that lie elsewhere.

5 6
ATCC#9, La India, 10/2007. ATCC#3, 6, La India, 10/2007.
7 8
ATCC#1, La India, 10/2007. ATCC#10, La India, 7/2008.
9
ATCC#4, La India, 8/2008.
188

table 7.1 Key variables in the ATCC region, 1975–2007

Period Civilians ATCC ATCC norms Territorial control Illegal group

killed rules resources & discipline
1975–1987 530 total victims; No rules Fear; threats; weak FARC dominant, then Emeralds, cattle, oil
Average of norms; civilian contested by army and
44 victims informants paramilitary
annually
1987–2000 3 estimated conflict- ATCC; Strong, widespread Contested; then Emeralds, cattle, oil;
related deaths; no informal norms; violent past paramilitary dominant FARC front received
victims from accords for and new process but with infighting, coca funds from
1991–2000; investigation inspire continued but central committee;
selective violence of accused participation sporadic FARC FARC executed some
defused/ presence through abusive commanders;
diminished 1995; FARC presence paramilitaries merged
diminishes around into AUC organization
1995 and paramilitary in 1997; coca from
gain complete control 1998
2000–2007 Selective killings ATCC, formal Migrants to the area, New contestation; then Emeralds, cattle, oil,
resume; tribunal, youths, and coca increasing coca, coca taxes
approximately 35 village growers who did paramilitary control;
conflict-related delegates not experience then some
victims violence or ATCC paramilitary
formation violated demobilizations and
neutrality norms new bands emerge
and participated in
conflict
Evaluating Explanations for Violence: The Balance of Control 189

evaluating explanations for violence:

the balance of control
This section assesses how various explanations account for the observed trends
in violence over time with special attention to the surprising era of the absence
of violence during the 1990s. These explanations include: general (national-
level) trends in violence and peace negotiations; shifts in territorial control;
changing rebel organization, resource bases, and discipline; and increased
international human rights advocacy. I assess these hypotheses with various
“cases” within the ATCC meta-case. I derive eleven real and counterfactual
cases across the ATCC and its neighboring regions (see Appendix B). While
I find these alternative explanations do not completely account for violence,
I only discuss the territorial control hypothesis in the body of the chapter. The
alternative explanations are summarized in Table 7.1 and discussed in greater
depth in Appendix B.
Changes in Military Contestation and Control. Kalyvas (2006) argues that
much of the violence against individual civilians in civil wars stems from battles
for territorial control among armed actors. Violence is selectively used against
suspected enemy collaborators to coerce support among the civilian popula-
tion. Violence is used to gain control of territory and is also the result of
contestation for territorial control. Violence is “jointly” produced as civilian
informants denounce enemy collaborators to armed actors when they feel
protected from retaliation. Such denunciations may arise from war-related
motivations, but may also be false and involve local disputes among neighbors.
Denunciations, and therefore selective violence, are thought to be most
common by the stronger armed actor in zones of dominant but incomplete
control. By implication, neutrality strategies – fence-sitting and double-dealing –
are permitted by armed actors for individuals and local committees in com-
pletely contested areas, but nowhere else. Neutral civilian organizations might
therefore exist only because they reflect a stalemated balance of power, and
thus have little independent impact. Institutions to clarify the “fog of war” are
thought to get eliminated or be useless where they might be most effective.
Can the ATCC’s apparent mitigation of violence be attributed merely to
existing in zones of either complete control or evenly contested control (epi-
phenomenality), rather than in a zone of dominant control (where it might
independently affect violence)? To answer this question, I compare the military
balance in the ATCC region with that of other regions. Kalyvas operationalizes
complete control as when the enemy army has no presence day or night
(according to civilian perceptions), dominant control as when the enemy army
only has presence at night, and contested control as when the government does
not have control at night (Table 7.2). Although there are challenges in applying
this standard to assess control, I adapted several interviewing procedures for
this task. I asked interviewees from among various villages across the ATCC
region the following two questions related to military control: first, “Around
190 The Institution of the ATCC: Protection through Conciliation

table 7.2 Predictions of the balance of control theory of violence

Control description Control Violence

zone type prediction
Full control by incumbents; adequate 1 No violence
security both day and night
Mainly but not fully controlled by incumbents; sporadic 2 Incumbent
covert activity by insurgents violence
Balance of forces: incumbent security adequate by day, 3 No violence
only marginal at night; sporadic to regular covert
insurgent activity at night
Mainly but not fully controlled by insurgents. Incumbent 4 Insurgent
presence marginal at day; regular covert and overt violence
insurgent activity
Insurgents are the primary authority day and night 5 No violence

year x, which actor had control of the area during the day? At night? Was the
control strong or weak?” and second, “Did you have to pay a tax to an armed
actor? Which one(s)?”
The ATCC’s antecedents should be a textbook case for Kalyvas’s theory. The
FARC maintained initial dominance in the region through the early 1980s and
little violence occurred (the guerrillas had an estimated 500 fighters in 1978;
García 1996). Guerrilla control began to erode and they retreated for a time as
the army and paramilitaries moved into the department of Santander.10 As the
guerrillas ceded control, they strengthened their offensive against the population
as a last-ditch attempt at coercion (García 1996). With the rise of the paramili-
taries and army presence through the mid-1980s, all of the elements predicted in a
zone of dominant control were present – denunciations, varying degrees of
contested control, and selective violence. As one resident put it, “There were
people that were seriously implicated by the sapos (frogs, or informants), people
that worked for one group or for the other. There was an information campaign
and many people were ‘marked’ by a certain group . . . The dark waters of this
river are a silent witness to the numerous dead they dumped in there” (Hernández
Delgado 2004; translated by author).
My interviewing and secondary sources indicate that in the next period,
from 1987 through 2000, the army (and paramilitaries) had increasingly dom-
inant yet still incomplete control (hence the ultimatum).11 This should have

10
ATCC#3, 11, La India, 10/2007.
11
E.g., ATCC#3, 12, Bogotá, 8/2008. Figure 7.7 below shows the balance of threats across villages
between guerrillas versus the army and paramilitaries and indicates that a number of villages had
continual threats, as well as relatively more threats by a dominant actor.
Evaluating Explanations for Violence: The Balance of Control 191

predicted intense, continued selective violence by the army and paramilitaries,

yet in reality violence dropped off dramatically. Although the guerrillas largely
withdrew their forces, they remained active in the area (partially confirmed by
García 1996). I heard of further examples of guerrillas threatening civilians
during the 1990s from camps in the nearby mountains. In almost every inter-
view, when asking whether the guerrillas had left the zone, people consistently
responded, “Always, there has always been guerrilla presence here.”12
Beyond the interview evidence I collected, there are additional signs of
guerrilla presence in the mid- and late 1990s. For instance, in a meeting of
ATCC leaders early in 1996, one leader noted that, “Three months ago the
FARC called the directors of the ATCC for a meeting” (ATCC Archives 1996).
The individual went on to make an additional observation about the conditions
during this time period in the Carare region, “The armed groups that operate in
the region have informants right here and are committing violence for gossip – a
problem that the ATCC has to fix, and not the armed groups.” They also noted
dialogues with the paramilitaries during this period.
The National Commission on Reparation and Reconciliation (CNRR 2011)
study provides similar evidence of guerrillas contesting the ATCC region during
the 1990s. In a section on “The interference of armed actors in the area of
influence of the ATCC,” the CNRR study notes that, “From the mid-1990s,
armed groups began to gain terrain in the Carare, including the ATCC zone of
influence and the corregimiento of La India. This influence meant a permanent
tension between the armed actors and the ATCC, which persisted in convincing
the campesinos to not plant coca and not interact with armed groups” (CNRR
2011, 179). The section in the CNRR study on “The reign of armed actors, the
persistence of violence, and the expansion of coca, 1994–2010” makes particu-
lar reference to the guerrillas and notes, “[T]he persistence of the guerrilla, with
the 23rd Front locating its base in the mountains of Landázuri and Peñon,
projecting toward other places in the Carare” (CNRR 2011, 167). The strategic
importance of the Carare region to the guerrillas is also noted in subsequent
pages: “The guerrillas continued to have presence in the Carare region . . .
considering the zone a passing place or a strategic corridor” (CNRR
2011, 171).
Quantitative data from the Jesuit think tank CINEP tends to corroborate
these characterizations. In most years of the 1990s, guerrilla groups registered at
least some armed activities (Figure 7.1) and human rights violations (Figure 7.2)
in or near the ATCC zone (data is missing for 1995).
My interview instrument allowed me to further cast doubt on armed actor
control as a determinant of ATCC autonomy and impact. By phrasing the
question on control about an “actor” rather than an “armed actor,” I did not
force respondents to answer with the name of an armed group. Indeed, some

12
ATCC#3, 13, La India, 10/2007.
192

figure 7.1 Armed actions in ATCC municipios, 1990–2006

Source: CINEP, author’s calculations.
Evaluating Explanations for Violence: The Balance of Control 193

5
Number of violations

0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Year
Paramilitaries Guerrillas State forces

figure 7.2 Human rights violations in ATCC municipios, 1990–2006

Source: CINEP, author’s calculations.

respondents naturally responded that the “actor” who had control was in fact
the civilian ATCC.13 Furthermore, many respondents noted that they did not pay
protection taxes to anyone since the ATCC was able to negotiate an end to this
practice in the region.14 More than once during the 1990s the ATCC civilian
leaders were able to brush off the dominant paramilitaries’ efforts to install a
base in the town of La India by arguing that it would only cause them more
problems with the guerrillas.15
Perhaps the most important evidence against the explanation of territorial
control for patterns in ATCC violence is that its causal mechanism persisted,
but did not lead to violence. If the balance of control theory were right – that
the ATCC could only have thrived under contested or complete control –
denunciations, threats, and violence all should have ceased. It is incongruous
with the theory that denunciations and threats continued to occur, but did not
lead to the killing of civilians (see interviews, García 1996, Hernández Delgado
2004). The ATCC therefore either persisted for some amount of time in a
dangerous zone of dominant control or in other conditions that were neverthe-
less quite dangerous. Without prompting, one respondent characterized the
situation and threats of the 1990s as “selective violence,”16 while another

13 14
ATCC#14, La India, 10/2007. ATCC#3, 6, La India, 10/2007.
15 16
ATCC#10, La India, 7/2008. ATCC#3, La India, 10/2007.
194 The Institution of the ATCC: Protection through Conciliation

reported that armed actors’ strategies were “psychological” and designed to

control through “threats and fear.”17
In sum, the theory of territorial control appears to explain some but not all
periods of violence and its absence in the ATCC region. The patterns of control
would have predicted higher selective violence than actually occurred. I next
explore how the ATCC dealt with the denunciation mechanism so that it would
not lead to violence.

civilian institutions as an explanation: the process

of the atcc
Levels of control or violence may indeed contribute to the establishment and
persistence of institutions such as the ATCC, but they provide an incomplete
picture of the decrease in violence observed after the founding of the ATCC.
This leaves open the possibility that it is explained by something inherent to the
ATCC. The ATCC has fulfilled many roles and functions for the citizens of the
region, including economic development planning, operating a community
store, lobbying government officials, and giving civilians early warning ahead
of impending battles. Many of these functions can potentially affect the dynam-
ics of conflict and violence in direct and indirect ways. However, the function
that most directly affects violence is the institutional procedures the civilians
developed to deal with threats from armed actors and resolve disputes over
civilian allegiances. This provision of “order without law” (Ellickson 1991)
goes to the heart of the mechanisms of other explanations of violence and short-
circuits them.
The ATCC’s institutional procedures are activated when a civilian resident of
the ATCC region has been accused of aiding one armed group or another,
becoming “comprobado,” or implicated, and is threatened with execution.
Rather than being killed outright as he or she normally would be absent the
institutional rules, the procedures call for turning the accused over to the
ATCC’s governing council, the Junta Directiva (below, I analyze the conditions
where armed actors would agree to this).18 In a region where government
authority is distant, the Junta acts like a court and conducts investigations of
the implicated person, using its advantage over armed groups in local
information. ATCC leaders meet about the case and interview acquaintances
of the accused, including family, friends, and neighbors. They further draw on
local village committees that monitor compliance with agreements and inform
on violators of neutrality.19 If the person is an ATCC “socio” and has signed a

17
ATCC#4, La India, 10/2007.
18
This is a stylization of the ATCC procedures, which sometimes unfold differently from one case
to another. The ATCC’s Junta Directiva is a different structure than the village-level juntas de
acción communal previously discussed.
19
ATCC#3, La India, 10/2007.
Analysis of Threats and Conciliations, 1987–2007 195

membership contract letter, they will present this letter to the accusing group as
a form of character witness.20 They may also leverage their bilateral relation-
ships with each of the armed groups to confirm the accusation with the rival of
the accusing group. The information is then compiled and discussed with the
accuser. If the implicated person is found to be a noncollaborator, by the
agreement he or she is absolved of wrongdoing by the accuser.
If the implicated is found to be a collaborator, he or she has two options
depending on the response of the accuser. Conditional on good behavior and
“correcting,” he or she can stay in the area (if he or she is found “guilty” again,
the armed actor will have the “right” to mete out punishment). Alternatively he
or she might be given funds from the ATCC (and sometimes even the armed
actor) to leave the region and find land elsewhere. If he or she still decides to
stay, the ATCC acknowledges it can no longer provide protection.21 This
procedure effectively sorts noncollaborator civilians from collaborator ones
who participated in the conflict. It reduces both the potential for false
accusations and the incentive for residents to participate in the conflict since
it becomes more costly to do so (and gain whatever selective benefits they may)
in secrecy.
There is substantial evidence that this procedure has been effective. The
impression among the civilians of the ATCC (and academic analysts) is that it
has saved many lives. According to one resident:
Of note is the right to life that has been achieved through dialogue. There have not been
deaths but people have been threatened by armed groups. So, we asked for meetings
with them and asked them to respect life. They told us that we would have to remove
[the accused] for this reason or that reason . . . We have had to remove some people from
the zone or turn them over to the competent authorities so they would pay their sentence
there [or to other armed groups]. But that has been a big achievement in the region, and
has been able to rescue the lives of many people. (Hernández Delgado 2004, 356)

If accurate, this characterization does much to explain the lull in violence

observed in the region during 1990s. The ATCC clearly fostered an environ-
ment where pacifist civilians had fewer entanglements with armed actors,
noncollaborator residents could be liberated from threats by investigations,
and exposed collaborators could reform or flee.

analysis of threats and conciliations, 1987–2007

To better understand the severity of the threats and the ATCC’s ability to
protect residents, I collected and compiled a new dataset on threats as well as
deaths from the region. The main purposes of this effort are twofold: first, to

20
ATCC#11, La India, 11/2007. Early on, membership was conceived loosely as support, partici-
pation, and adherence to pacifist norms but has been increasingly formalized over time.
21
ATCC#4, La India, 10/2007.
196 The Institution of the ATCC: Protection through Conciliation

assess if differences in violence outcomes depend on whether the civilians’

investigations confirmed the reasons for threats, and second, to describe the
information flows of the region and how civilians generate credible
information.
Since the conciliation of threats has been an oral, day-to-day process with
almost no written records available, the dataset – this oral history – was
recovered based on in-depth interviews, in some cases with multiple follow-
ups, with over ten “conciliators” with specific knowledge of and participation
in conciliations.22 I also spoke with a number of people who were actual
victims of threats and expressed their gratitude to the ATCC for saving their
lives. These interviews yielded detailed information on sixty-seven named
threat episodes involving ninety-eight people between 1988 and 2007, with
the majority of recorded cases occurring from 1998 onwards.23 Based on
conciliator estimates, these figures likely represent only about a third to a half
of the total episodes of conciliation in the ATCC’s history. Although many
episodes are certainly missing, the cases that were recalled for the dataset are
likely the most representative and important cases (i.e., extreme cases of either
saving lives or of killings).24

22
ATCC#1,3,4,6,10,17,18,19, 8/2008; ATCC#20, Cimitarra, 8/2008; ATCC#15, 11/2007,
ATCC#3,12, Bogotá, 8/2008. Some written confirmation of approximately fifteen conciliation
cases could be found in the meeting minutes of dialogues although the data collection process
depended on the memory of conciliators. A number of conciliators (including women) were
present at each dialogue, and the dates, results of investigations, outcomes, and other concili-
ation characteristics were compiled to the best of their knowledge. To improve subjects’ recall,
they consented to discussions with other conciliators in small groups to help jog their memories.
They then helped revise the compiled lists of cases for accuracy. Unfortunately, it was more
difficult to collect information on threats from the distant past – from the early 1990s – because
some conciliators had either died (of natural causes), moved away from the region, or simply did
not remember cases clearly. During this early period, conciliation processes were also less formal,
with smaller groups of conciliators or conciliations carried out only by ATCC presidents,
meaning there were fewer people to recount the history.
23
Data on the following variables was collected for each threat: year; gender and age of accused;
threatening group; reason for threat (informant, material aid to enemy, coca, etc.); how the
ATCC learned of the threat; how the conciliation took place and why it worked (if it did);
whether investigation showed the charge of the threat to be true or false; and result for the
accused. It should be noted that, according to the ATCC, regardless of whether the threat victim
is found to be guilty or innocent, death threats are never considered a legitimate way of resolving
disputes. This information was collected not to cast blame, but rather to give a fair accounting of
why different events unfolded as they did. Seven episodes did not involve armed groups or have
validated accusations and a small number of individuals were accused in multiple episodes. This
does not include an additional thirty-five cases of killings associated with the coca economy in
the post-2000 period that were not conciliated.
24
One leader from the early 1990s reported investigating somewhere between 70 and 100 such cases.
These implications would usually first begin with light threats to deter certain behaviors and, if those
were not effective, would then escalate to death threats (although only a handful of cases went this
far; ATCC#15, Bogotá, 11/2007). Other conciliators who served toward the late 1990s also recalled
dealing with at least seventy such cases (ATCC#1, ATCC#4, La India, 11/2007).
Analysis of Threats and Conciliations, 1987–2007 197

figure 7.3 Outcomes of threats according to whether the victim was believed to have
collaborated
Source: Interviews.

Conciliators also provided their assessments of the validity of the accusations,

yielding examples of both “neutral” people being “vouched for” and saved and
collaborators being identified and turned over. Indeed, the analysis of the thirty-
six threat cases where the reason for the threat was determined to be “false” (in
some cases ex-post), fifty-five people were saved while seven were killed and
three were forced to displace. With only about 15 percent of the threat victims
killed, the difference between these cases and those where the reasons for threats
were confirmed (“true,” i.e., participating in the conflict) is noteworthy. Of these
twenty-four “true” cases involving thirty-seven people, nine people reformed
and stayed, nine people were killed, and four were forced to displace (the
outcomes for some of the people remain unclear). Figure 7.3 show the difference
in homicide rates between these two groups, which is found to be statistically
significant with a chi-squared test.25
Narrations of actual cases illustrate more clearly the danger the victims face
and how the cases are resolved. In one example from 1995, a man who was
marked for death was protected by intervention from the ATCC:
Don Diego, a middle-aged wood-cutter, was accused by the guerrillas of providing aid to
and being an informant for the paramilitaries. When Don Diego got word they were
going to kill him, he went to the ATCC for help. The ATCC (along with the guerrillas)

25
It is conceivable, although unlikely, that judgments about whether a threat victim was falsely or
validly accused were influenced by hindsight bias based on the outcome (e.g., the kind of
thinking that “if they were killed or continually threatened, then they must have been guilty of
the charge”).
198 The Institution of the ATCC: Protection through Conciliation

investigated his case by talking with neighbors and monitoring his actions going forward
and found the accusation to be false. They determined that another wood-cutter had
informed the guerrillas on him and had been spreading lies for his economic benefit – to
kill Don Diego so he could take his wood and push him out of business. Upon word of
this, the guerrillas relented and Don Diego remained safely in the zone. Instead, they
punished his accuser.26

This case is emblematic of the many instances in which the ATCC leverages its
local information network to protect “innocent” civilians (as residents refer to
them) from the dangers of the armed conflict. While in this episode, guerrillas
made a death threat based on a denunciation, underscoring their ability to
project into the zone, it was also common in the 1990s for paramilitaries to
bring similar threats.
In another instance, a man who was identified as aiding the paramilitaries
was exposed and sentenced. He did not ultimately “correct himself” or comply
with the ATCC’s finding against him and suggestion that he leave the region,
essentially renouncing his protection from the ATCC. He was eventually killed
by a guerrilla assassin:
Señor George was implicated by the guerrillas for supporting the AUC around 1999.
The FARC was going to kill him, but they notified the Directiva that he had given help
and food to the paramilitaries. Testimony of George’s neighbors gathered by the ATCC
confirmed the FARC’s belief that he provided the aid and had a revolver (an armed
civilian). The ATCC also asked the AUC if they had received help from him. George was
given 200,000 COP (about US$90) to leave the region but he instead spent the money.
The ATCC could not force him to leave since he was a student and son of La India, but
said they were not responsible for what might happen. Before long, George was killed by
a guerrilla assassin.27

A case like this is tragic, but outcomes of this sort where an actual ATCC
member insists on breaking covenant appear to be relatively rare (although
more frequent among nonmember residents). Studying episodes of mediation
with various outcomes illustrates how the process of vouching works and when
it fails.
In a last example, an ATCC leader was falsely implicated in the early 2000s,
and even though the evidence cleared him, he opted to displace from the region
for a time because he feared his life remained in danger:
An (false) accusation was made against Don Franklin by paramilitaries for supposedly
giving aid to the guerrillas. He was implicated because guerrillas earlier attacked a
paramilitary motor canoe on the river and then fled past his house in their escape. The
ATCC interviewed the man’s neighbors to collect evidence of his participation and the

26
ATCC#1, 4, La India, 10/2007.
27
ATCC#4, La India, 10/2007. This did not constitute a death for the ATCC’s 1990s count
because he was killed in 2000, beginning the new era of violence (or effectively became a
combatant).
Analysis of Threats and Conciliations, 1987–2007 199

figure 7.4 Reasons for threats (and killings) by armed groups

Source: Interviews.

neighbors cleared him. Although he left the zone for several months to Bogotá as a
security precaution, he soon returned and resumed his daily activities.28
The ATCC was once again effective in saving a threatened leader. Nevertheless,
this case also illustrates the complexities of removing threatened civilians from
dangerous situations. Although the man in this case was able to return to his
normal life, the process is not a utopia and not all, but many, cases of non-
collaborators investigated by the ATCC are resolved so satisfactorily.
The dataset of threat conciliations also provides descriptive insights about
how the armed conflict and concomitant threats unfolded and were dealt with.
First, there were threats from many actors, and the conciliations demonstrate
how the ATCC was able to adapt to dynamic conditions over time and across
space. Second, hinting at the blend of armed group motives, there were diverse
reasons for threats, including providing aid and information to the enemy,
coca cultivation, eliminating demobilized fighters, countering delinquency, and
coercing leaders (Figure 7.4). Third, the ATCC obtained advance information
about threats – sometimes from residents, sometimes from armed actors, but
always with the help of a dense interpersonal network – which allowed the
organization to take action in a number of different ways (Figure 7.5). Fourth,
the ATCC relied on a variety of appeals and investigation techniques to
enhance the credibility of the information presented to the armed groups about
threat victims (Figure 7.6). Fifth, the ATCC was able to intervene and save
people even in various cases where armed groups were intent to kill (even with
lists of targets) and did not first approach the ATCC and inform them of the

28
ATCC#9, La India, 10/2007.
200 The Institution of the ATCC: Protection through Conciliation

figure 7.5 Information channels: how the ATCC learned of threats

Source: Interviews.

figure 7.6 Information and appeals used by the ATCC to investigate threats
Source: Interviews.

threat. Lastly, in some of the cases where conciliations were not successful, the
ATCC was able to rescue victims and get them to safety.
The data on threats and the presence of informants also provide insight on
the geographic variation in the balance of armed group control by villages in
the region. For instance, as shown in the graph in Figure 7.7 and the map in
Figure 7.8, the villages with the most threats by the guerrillas were La Ceiba,
Mate de Guadua, and El Pescado, on the east side of the Carare River, closer to
the alleged guerrilla camps in the mountains.29 Paramilitary threats were most

29
In some cases, individuals were threatened by multiple armed groups.
 figure 7.7 Distribution of threats and killings among armed groups by ATCC villages, 1991–2007
201

Source: Interviews.
202 The Institution of the ATCC: Protection through Conciliation

figure 7.8 The balance of control among armed groups across ATCC villages,
1987–2007
Source: Interviews.

prevalent in villages such as Campo Banda and La Zarca, closer to the para-
military bases in San Tropel and Puerto Araujo. The data again reinforce the
characterization that there existed conditions of largely dominant (though
incomplete) control by the paramilitaries where one would expect violence.
Analysis of Threats and Conciliations, 1987–2007 203

Lastly, the threat conciliation data also enable comparisons between violence
in the ATCC “area of influence” and the immediate surrounding areas. This
comparison suggests the ATCC experienced similar levels of armed conflict
danger as its neighbors but suffered fewer actual killings. I pooled data on events
from CINEP, Equipo Nizkor (based largely on press reporting and some police
reports), Zamora (1983), and Vargas (1992) at the village level to classify
killings by whether they were committed within the ATCC area of influence or
in (rural) neighboring areas in the six municipios in which the ATCC is located.
This data was then matched with the interview data I collected on (successful)
threat conciliations within the ATCC region to gauge the number of killings that
might have occurred absent the ATCC (conciliations plus the actual killings).
I was able to classify data back in time over the thirty-two-year period from
1975, twelve years prior to the ATCC’s founding, through 2007. These series
are displayed in Figure 7.9 and mapped in Figure 7.10.
From about 1980 leading up to the ATCC’s founding in 1987, the ATCC zone
suffered much higher conflict homicide rates (per 100,000 residents) than neigh-
boring villages. Because the greater repression might have made social organiza-
tion even less likely, the trends argue against the ATCC forming solely because of
mild conflict conditions. By contrast, in the post-1987 period after the ATCC was
founded, the homicide rate in the ATCC region over time trends slightly lower
than in the neighboring areas.30 However, the dashed line, which represents the
number of killings and threats that were resolved in the ATCC region – the
counterfactual scenario – rises up to approximate the actual killings that occurred
in the neighboring areas. Consistent with the ATCC becoming more effective by
strengthening its conciliation institutions with a delegation of conciliators after
increasing threats in the late 1990s, the number of successful conciliations is also
shown to increase over time (the gap between the dashed total ATCC victims line
and the solid line representing number of ATCC victims killed). These trends
corroborate that the ATCC suffered far less violence than its neighbors despite
experiencing similar or greater levels of danger from armed groups.31 They also

30
The interview data has some gaps in coverage due to memory lapses, especially in the early years.
Press data is also missing for the year 1995 due to a gap between the two press datasets. There
are slight discrepancies in the timing of events in the two datasets, again likely a memory issue for
interview subjects. Despite these issues, although the graph displays the count of people killed in
the ATCC area from the interview data for purposes of consistency with the threat counts, the
CINEP/ Nizkor data of people killed yields a similar total number of victims over the same time
period (sixteen versus twenty-two).
31
By raw numbers of fatalities, the ATCC suffers relatively fewer killings over the time period. Note
that while there is little reason to believe there were many threat victims that were saved or spared
in the neighboring zones, this is not counted nor verified through these data because of the limited
types of events in press coverage and challenges in covering the broad neighboring zone territory
during fieldwork. This may produce an underestimate of rescues in neighboring zones. But by the
same token, the ATCC’s superior coverage based on its level of organization, press reporting, and
interview data biases toward killings in the ATCC zone being overreported.
204

Rate of Victims Killed, Threated (per 100,000 residents)

500

450

400

350

300

250

200

150

100

50

0
1975 1980 1985 1987 1990 1995 2000 2005 2007

Neighbor zone victims ATCC killed threat victims ATCC killed + saved threat victims
*CINEP, Nizkor, Vargas, Zamora *Interview data *Interview data
*CINEP, Nizkor, Vargas, Zamora *CINEP, Nizkor, Vargas, Zamora
figure 7.9 Threat and killing rates in the ATCC zone vs. neighboring areas
Source: Interviews and other sources.
 (a) (b)

figure 7.10 Maps of actual and counterfactual violence in ATCC and neighboring regions, 1987–2007
Source: Interviews and other sources.
205
206 The Institution of the ATCC: Protection through Conciliation

demonstrate how well-organized communities such as the ATCC are capable of

suppressing violence even better than the average junta council.

conditions for the maintenance of local

order in wartime
The ATCC functioned in a context of continued threat, danger, and denunci-
ations, yet little to no violence occurred against its members. The success of the
ATCC rests on three interlocking conditions: an institutional process to deal
with denunciations, a shift in civilian preferences to abide by neutrality and not
aid armed actors, and “favorable” preferences of armed actors (and favorable
policies to “sweeten” the deal).
Condition 1: Institutional Investigatory Capacity. A civilian organization can
mediate the flow of information and provide credible, balanced signals to armed
actors about the participation of residents in the conflict.32 As an institution, it
has set procedures and is larger than any single person. This allows the process to
be applied to any particular case without being ad hoc – it endures over time for
more than a “single shot” game.
The institution need only provide information, as it can rely on “out-group
policing” by the armed actors to enforce punishments when necessary (a
variant of Fearon and Laitin’s (1996) “in-group policing”). Similar to the
Law Merchant institution that helped resolve disputes among traders in medi-
eval Europe (Milgrom et al. 1990), the onus is on the individual in the commu-
nity to stay out of trouble with the armed actors. Since the institution only
provides information, it cannot restrict residents (whether few or many) from
participating in the conflict and thereby running the risk of being denounced
and killed if they are “outed.” In other words, it stays an execution, but only
until a verdict is reached. The institution alone cannot broadly eliminate
violence against civilians, only against “virtuous” ones.
I found that investigations by the ATCC, or clarifications (“aclaraciones”), as
the residents refer to them, actually unfold in a variety of ways. While it is most
common for armed actors to bring suspects before the ATCC for inquiry, they do
not always defer to this procedure and may be intent on eliminating targets on
their lists – the ATCC does not always learn about threats or gather information
strictly “by the book” as I first expected. In these circumstances, ATCC civilian
informants that know of threats may leak that information to the Junta. Alterna-
tively, some combatants in the armed groups with affinities for the civilians
(referred to as “angels” or “friends of peace”) may oppose a particular “lim-
pieza,” or “cleansing,” and surreptitiously alert the ATCC.33 Usually once the

32
Participation can include active fighting but also acts short of this, including giving information,
food, shelter, etc.
33
ATCC#11, La India, 11/2007.
Conditions for the Maintenance of Local Order in Wartime 207

ATCC is made aware of threats it is able to intervene – even when the armed
groups were originally quite determined to eliminate suspected collaborators.
Condition 2: Civilian Pacifist Preferences. To eliminate violence against
civilians, the civilians must also confront the “preference problem,” or the
problem that some civilians prefer to participate in the war more than others
(or alternatively, are less “deterrable” from collaborating with an armed
group). Preferences for participating in the conflict can be influenced by various
sources. As discussed in other literatures, there may be substantial selective
benefits to aiding or joining an armed group (Lichbach 1994). Examples from
the ATCC region include payments for aid; threats for not providing aid;
payments for cultivation of coca, from which cocaine is made; or gaining an
advantage against one’s neighbor. These rationalist/materialist benefits may be
counterbalanced by other factors that work to keep civilians out of conflicts.
For one, economic development and economic opportunities can reduce the
desperation that may drive poor residents to grow coca or seek other selective
payments from armed groups.
Norms of nonviolence and pacifism or the philosophical and ideational
belief that peace and nonviolent advocacy are morally superior to war and
killing, are another way to keep civilians from getting entangled with an armed
group. The pervasiveness of religion and spirituality can shape these moral
antiviolence beliefs, as some ATCC residents asserted the Seventh-Day
Adventist theology did in their region.34 For instance, civilians that obey strong
norms of nonviolence may be less inclined to participate in coca cultivation,
even if they are economically disadvantaged. These nonviolence norms and the
“culture of peace” are often the focus of peace-building strategies.
Various events and circumstances surrounding the founding of the ATCC
likely contributed to building a “culture of peace” and norms for neutrality,
nonviolence, and nonparticipation in the conflict. First, the ultimatum to dis-
place led to a sorting process where less resolute residents left the region or
joined their armed actor patrons (Hernández Delgado 2004, 329). Second,
once the ATCC was founded, solidarity was strengthened by the persisting
threat environment (García 1996, 251). Third, there was some social indoctrin-
ation and awareness raising (“conscientización”) by the ATCC in its early
years. Large meetings were held in the plazas and the recordings of meetings
with armed actors would be played on loudspeakers as a form of ideational
coordination. There were also educational programs in the villages to explain
the purpose and functioning of the ATCC and educate residents about human
rights. The community also later founded a peace radio station that would
politely decline requests to play narcocorridos or other violence-themed

34
ATCC#1, 7, La India, 10/2007. Many of the leaders of the Adventist church were also leaders in
the ATCC, though the organization also featured prominent participation of other denomin-
ations, including Pentacostals, Evangelicals, and Catholics.
208 The Institution of the ATCC: Protection through Conciliation

songs.35 Fourth, the institutional investigation process gave further motivation

to civilians who were considering aiding armed groups to straighten up or leave
the organization or the region for fear of being targeted (as did the protection of
upstanding residents). Fifth, a “Group of Conciliators” was established to
mediate interpersonal disputes so they would not be resolved through outside
actors. Sixth, the accords also won commitments from the armed actors to
refrain from imposing on civilians by asking for support, making nonparticipa-
tion in the conflict more permissible.
Norms can be a powerful force for nonviolence but alone are not sufficient
to protect civilians from violence since the politics of civil war show that many
upstanding people can be falsely implicated as enemy supporters (Kalyvas
2006). This is confirmed by evidence on the variation in the strength and
pervasiveness of ATCC norms over time. For instance, although many residents
were very spiritual and pacifist before 1987, violence was rampant and many
exemplary residents were killed by false accusations.36 There was no way to
“vouch” for pacifist citizens who had been implicated, so pacifist and oppor-
tunistic citizens alike were killed. While norms were strengthened after 1987, it
is unlikely that the degree of faith and spiritual purity of residents alone
brought about the extreme reduction in violence.
Condition 3: Armed Actor Incentives for Compliance. Civilian information
systems and pacifist norms can be helpful but will not eliminate violence if
armed actors have no incentives to abide by (agree to) the civilians’ institutional
procedures in the first place. Why should armed actors not simply kill a civilian
implicated with helping their enemy? Some amount of joint interest among the
armed actors to preserve or not directly target the civilian community must
exist (i.e., this argument is less likely to apply to genocidal situations or
“draining the sea”). However, this does not mean that an armed actor alone
has no interests in committing violence against civilians (this might depend on
the choices of its rivals). For instance, all armed actors may seek to coerce
support of the population through violence, though they all may also prefer
accuracy to avoid angry backlashes.37
The ATCC civilians pursued a cooperative strategy through negotiations
and dialogues to get armed actors to buy into their process. They negotiated
symmetrical, transparent agreements so that armed groups could be confident
they were not losing civilian support to other groups (and they tape-recorded
meetings for added transparency). The investigation process reduces the burden
on armed actors to carefully select their victims to deter enemy collaboration

35
To such requests, the disc jockey would say, “So sorry, but we can’t help with that.” (“Que
pena, pero no le podemos colaborar con eso”).
36
ATCC#3, La India, 10/2007.
37
As one respondent suggested, “Armed actors want to feel like they’re needed by the civilian
population. They want to solve all civilian problems, even intrafamily disputes.” ATCC#3, La
India, 10/2007.
Conditions for the Maintenance of Local Order in Wartime 209

since the ATCC bears these costs. The ATCC also discarded strategies and
policies that might run against armed actor preferences and upset the insti-
tutional equilibrium. For instance, since the armed groups wanted to be seen as
the legitimate law of the land and were sensitive to bad publicity, the ATCC
more conciliatorily opted not to publicly denounce suspected perpetrators of
acts of violence by name.38 The ATCC also chose not to prohibit the armed
groups from passing through their territory since it belongs to “all Colom-
bians” (but insisted they did not bring arms into communities), which allowed
for the independent verification of the fair implementation of its institutional
procedures.39
Civilians may also even provide some benefits to armed actors by acting as
neutral arbiters and serving as a channel of communication between enemy
groups. First, the ATCC helped the armed actors negotiate various prisoner-of-
war exchanges.40 Second, the ATCC has facilitated negotiations for armed
actor demobilizations by guaranteeing the security of combatants as they
reintegrate into civilian society. In the mid-1990s, some members of the local
FARC fronts (including a commander) took advantage of the ATCC’s arbitra-
tion and laid down their arms.41 Third, ATCC peace overtures also reportedly
facilitated cease-fires – a mutually beneficial “descanso,” or rest. As one resi-
dent eloquently described the armed groups’ desire to avoid unnecessarily
antagonizing their enemies, “When passing a beehive, don’t throw stones.”42
This jibes with accounts from World War I of troops from opposite trenches
tacitly colluding to not fight (Axelrod 1984). Although not directly related to
the armed actors’ accession to the ATCC’s investigation procedures, these
benefits help “grease the deal” and can allow for bargaining leverage across
issues, including civilian security.
In the ATCC’s experience, armed actors have generally fought through the
civilian population, being loath to engage in direct confrontations with the
enemy.43 Limiting civilian defections by limiting casualties appears to have
become a second-best option to paying the costs of winning the civilians’ full
allegiance. The perception of fairness of this civilian arrangement turns out to
be central to its stability and compliance. By implication, the process depends

38
ATCC#3, 6, La India, 10/2007.
39
This indeed was a concern of the guerrillas and later the paramilitaries. In response to the initial
proposal, a guerrilla commander said, “Compañeros, these conditions that you are demanding
are not fair, you would have to also impose them on the army and the paramilitaries, who are
your greatest enemies” (García 1996: 196).
40 41
ATCC#3, La India, 10/2007. ATCC#6, La India, 11/2007.
42
ATCC#4, La India, 10/2007. “Si yo voy a pasar por el lado de las abejas, mejor no tirar la
piedra.”
43
ATCC#15, Bogotá, 11/2007. One respondent told me, “It’s not convenient for the armed groups
to fight” (“no le conviene”). Instead they prefer the safer option of fighting a dirty war for
civilian support. Another respondent said, “The entire armed actor effort of the 1990s was to
regain control of the people.” ATCC#3, La India, 10/2007.
210 The Institution of the ATCC: Protection through Conciliation

on the armed actors’ access to high-quality, transparent information to inde-

pendently verify the workings of the ATCC. Ironically, sapo armed actor
informants embedded within civilian society are a central source of confidence
in the institution.44
Accounts from the combatants (from archives, interviews, and secondary
reports) support this view of their preferences and strategies toward civilians.
The AUC paramilitaries had relatively greater control and capability than the
guerrillas, and therefore greater incentives and opportunities to commit vio-
lence. However, there is evidence that both groups could be persuaded by
community processes. Consistent with the paramilitary bloc apparently com-
mitting more abuses outside the ATCC zone than within it (as illustrated by the
San Tropel killings and the data in Figure 7.9), the narratives indicate that the
armed groups of the region were not simply especially respectful or did not
generally prefer to avoid targeting civilians.
The accounts instead suggest that the use of violence is conditional on
community organization and collective action. This insight is borne out in a
verbatim transcript of a meeting between the ATCC leader, the representatives
of the relatively poorly organized villages of San Tropel and Santa Rosa on the
border of the ATCC, and a paramilitary subcommander held in 2001 in Santa
Rosa, Cimitarra (from the ATCC archive; see full dialogue in Appendix B).
When Santa Rosa residents voiced concerns about being stigmatized by the
guerrillas, they probed whether the paramilitaries would leave and allow them
the kind of autonomy enjoyed by the ATCC. The paramilitary subcommander
present tellingly responded, “The entire community would have to decide . . .
But if only two or three people don’t want our presence, then we’ll continue to
be here.” Freedom from armed group incursions would depend on the level of
community cohesion, indicating the guarantees of the ATCC process itself were
pivotal in affecting the group’s calculus and diminishing violence, rather than
some inherent characteristics of or changes within the armed group.
Another ex-paramilitary subcommander from the bloc provides additional
confirmation.45 He noted that violence was more frequently employed prior to
2000 because they had not yet learned how to interact with civilians. Worried
about losing support and seeking less costly strategies, they became increasingly
willing to delegate the maintenance of order, but mainly to well-functioning
village councils.
The AUC paramilitaries would be expected to have relatively fewer
incentives to commit violence in the zones neighboring the ATCC where their
control was even more dominant. However, as noted by a village representative
from neighboring San Tropel, there was both greater control and greater
repression, “For us it hasn’t gone very well, since we’re 100 percent dominated
by the Autodefensas” (2001 meeting in Santa Rosa, ATCC Archives). A similar

44 45
ATCC#1, La India, 8/2008. Exc#7, Bogotá, 8/2009.
Conditions for the Maintenance of Local Order in Wartime 211

pattern can be seen with the guerrillas. One resident who moved to the ATCC
region from another part of Sucre municipality under heavier guerrilla control
said that the guerrillas would not investigate gossip there, but would simply kill
“at once.”46
Greater background on the motivations behind the paramilitaries’ stances
toward communities and their ambivalence is found in additional archival min-
utes from a meeting with a different AUC subcommander in an ATCC village
from September 2003.47 In the verbatim exchange, Comandante Montoya, true
to his group’s counterinsurgent, “self-defense” mission, initially proclaims his
solidarity with the campesinos, “We truly believe our work should go hand in
hand with the community . . . to free this zone from the guerrilla. We’re here
because there are campesinos . . . It is for this reason in some circumstances we
accept your opinions, but in others we disregard them because this war is
difficult.” He later derides the guerrillas, professes his own group’s humaneness,
and also acknowledges the weight of the accords signed with the ATCC:
It’s the guerrillas that attack you. In our ideological principles we respect life and come
from the communities – we aren’t ordered to kill campesinos. Commander Botalón
talked to me about the accords and we believe we are complying with them. We accept
your claim to the right of neutrality. But . . . since the conflict is intensifying, the
population should choose a side.

Montoya allows the right of neutrality and claims to have respected it, but also
maintains the possibility of an exit from this clause. His blunt words express his
first-best preference for civilian allegiance. However, he later vacillates between
allowing civilians to live independently and public displays of strength through
tacit threats for civilian support.
Montoya goes on to more clearly express his main concern: that of civilians’
defection to the guerrillas. Acknowledging that the AUC does not depend much
on the population for material support, he wants to assure the accords are being
reciprocally upheld:
We don’t need things [drugs, food, and arms] from you the way the guerrillas do.
I recommend you don’t compromise yourselves [with the guerrillas]. That’s not a
threat, it’s a suggestion . . . The campesino compromises himself when he conceals
information about where the guerrillas are and about their activities. This is indeed a
problem. Such houses [and traitors] shouldn’t be allowed to do this because they’ll be
killed. The campesino that dedicates himself just to his family doesn’t have any
problem; if he acts to the contrary, he will see (emphasis added).
Montoya acknowledges that they try to avoid pressuring civilian involve-
ment, but again, there is a tension in his rhetoric between “suggestions” to
gain civilian support and using “threats” to deter defections. The contradic-
tions suggest preferences that, although conflicted, are settled on the second-

46 47
ATCC#21, La India, 10/2007. Pseudonym; edited for clarity.
212 The Institution of the ATCC: Protection through Conciliation

best option. He would like civilians to inform on guerrillas, but his main
priority is that they at least do not aid them. In the ATCC’s solution to this
dilemma of armed groups, the threats are mainly against enemy
collaborators, while noncollaborators are left alone.
The Jekyll-and-Hyde balancing act of these paramilitaries is confirmed by
Colombian scholars Gutiérrez Sanín and Barón (2005). They concur that the
paramilitaries have been willing to allow civilians space for autonomy within
certain constraints, “[Commander] Botalón tolerates trade unionism and col-
lective action not controlled by the paramilitary, as long as it clearly distances
itself from the guerrilla” (20). The authors observe the group’s interest in
maintaining order on the cheap, though also suggest it is tenuous, as the
possibility of the breakdown of that order is never far away, “Botalón . . . has
learnt to calculate keeping in mind long-term horizons, which involves higher
levels of self-control [and] replacement of pure repression by less expensive
mechanisms . . . Naturally, this does not prevent occasional outbursts of mur-
derous violence” (22). With the theory developed here, these outbursts are now
better accounted for.

explaining a resurgence of violence, 2000–2007

Starting in 2000, violence returned to the ATCC region after nearly a decade
with few if any civilian victims. Why did this occur? What changed? The
fluctuation in this phase provides a good opportunity to test the theory of the
effect of civilian institutions on new data. Although the new violence could
reflect some amount of institutional breakdown of the ATCC, I find that the
critical change was the degradation of norms of neutrality and nonviolence
among the population due to exogenous factors.
The ATCC’s residents were again put at greater risk as the guerrillas
increased their presence in the region. Levels of contestation increased between
the guerrillas and the then-dominant paramilitaries. With the contestation,
violence increased as well. In 2001, paramilitaries began killing campesinos
who resisted selling their land to coqueros, coca-growing campesinos involved
with the AUC.48 Guerrillas began collecting a “vacuna,” or “vaccination”
payment from civilians for protection, and when residents did not pay, they
started killing too.49 Conciliator estimates suggest about thirty-five civilians
were killed in all.
In this period, the ATCC institution was actually strengthened, not
weakened. For example, a formal Tribunal consisting of thirty elected
members was established to improve the process of investigating
denunciations. Information gathering was improved with the formation of
village committees and delegates. An economic development report was

48 49
ATCC#6, La India, 8/2008. ATCC#3, La India, 8/2008.
Explaining a Resurgence of Violence, 2000–2007 213

conducted and the ATCC also formalized its bylaws and began a membership
drive to reassociate its “socios” (ATCC 2006; a membership contract docu-
ment is displayed in Appendix B). Mediation efforts were also increasingly
formalized with teams of conciliators and do not appear to have decreased in
effectiveness.50 Variation in the ATCC’s institutions therefore does not
appear to be the source of increased selective violence.
Instead, the change in the distribution of people in the population who were
normatively committed to neutrality and noninvolvement in the armed conflict
is associated with the increase in individuals killed.51 The distribution shifted
for two main exogenous reasons, both related to the fading of important norms
that were formed from earlier experiences of suffering great violence and
overcoming it. In the post-2000 period, there were two new populations in
the ATCC region that were not committed to the norms inculcated at the
ATCC’s founding. First, beginning around 1999, the region saw the arrival of
new migrants who had not experienced the ATCC’s history, did not have the
same average level of commitment to avoid the conflict, and did not have a
good understanding of how the organization worked. The population in the
region swelled by as many as 2,000 people, many of them coming from the
department of Bolívar to plant coca. With the influx, La India was full of
people. Prices rose, brothels opened, and a resident complained about the
“drinking and loud music all day long.”52
These migrants and newcomers were drawn by the region’s tranquillity.
Some initially joined the ATCC but did not uphold their commitments and later
violated the ATCC’s rules. These newcomers tended to try to have their cake and
eat it too: they wanted the protection (and perhaps camaraderie) of the ATCC’s
“public umbrella,” yet also wanted the easy gains from illicit activities. Coca
divided the people, as the ATCC made an agreement with the armed actors not
to get involved with the crop because “it was their thing.”53 As an ATCC leader
stated in a meeting with coqueros, “He who plants coca is not to get involved in
the organization” (September 20, 2002, in the village of Ahuyamera). Indeed,
the ATCC’s members report not knowing many of the victims from this time,
although they suggest that many of the thirty-five murders occurred against
residents of villages some distance to the southwest of La India, such as La
Corcovada and La Zarca – villages with coca farms and coqueros. This account
corresponds with data on coca cultivation across villages I calculated based on
UN aerial surveys, as displayed in the map in Figure 7.11.
The immigrants were, either by nature or experience, more opportunistic
than the ATCC population, or were simply less morally committed to neutrality
and so were enticed by armed actors into the coca economy. Given the rela-
tively low incomes of these coquero civilians, coca profits were a tempting way

50 51
ATCC#12, Bogotá, 8/2008. ATCC#3, 11, La India, 8/2008.
52 53
ATCC#17, La India, 7/2008. ATCC#11, La India, 8/2008.
214 The Institution of the ATCC: Protection through Conciliation

figure 7.11 Estimates of coca production in ATCC villages, 2003

Source: UN Aerial Survey using 2003 data; calculations and map by author.
Explaining a Resurgence of Violence, 2000–2007 215

to increase their earnings.54 But it also entailed risks: once a civilian begins
growing coca and selling it to one of the armed actors, he or she is seen as
participating in the conflict, sometimes triggering a response from the enemy
armed group. In these cases, despite the ATCC’s commitment to protecting the
lives of all civilians, there was little the ATCC could do. At times, these
incidents of violence had spillover effects, intensifying competition and
conflict among armed actors, causing some pacifist residents to be seen as
“guilty” merely by association. These concerns are confirmed in meeting min-
utes from discussions in 2002 between the ATCC and the coca growers in the
region.55 However, unlike in the cases of collaborating coqueros, the ATCC
was still generally able to mediate successfully on the behalf of
noncollaborators.56
In a sense, the ATCC was a victim of its own success. Anecdotal evidence
suggests that the “peace” and growth the ATCC created during its nonvio-
lent era of the 1990s created the moral hazard of attracting these new
migrants to the region to share in its prosperity.57 For instance, a current
ATCC member said he moved to the sanctuary of the ATCC region from
elsewhere because of the danger of gossip and slander he faced there.58 Yet,
similar to how the gentrification of a run-down neighborhood can push out
longtime residents, success can sow the seeds of its own demise. As the
American baseball player Yogi Berra once said, “Nobody goes there any-
more, it’s too crowded.” This reflects a partially “endogenous” source of
change in the equilibrium of violence, as the institution itself, while benefi-
cial, can also produce instability and be “self-undermining” (Greif and
Laitin 2004). Residents said the ATCC was unable to counteract this trend:
as an informal organization, it does not have the governing powers to set
boundaries and keep migrants out.59 These examples demonstrate the perni-
cious influence that coca and the prospects of easy money can have on a
local civilian institution.
Second, a growing number of youth who were born around the time of the
ATCC’s founding came of age around 2000. These youths were not old
enough to remember the formation of the community and so were less likely
to be instilled with the community’s neutrality norms (recall the plight of
Señor George). They were born into the ATCC system and had not agreed to

54
As one coquero declared in a meeting with the ATCC, “I am the owner of this farm here and
I also plant coca, but I am not ashamed since we are in a precarious economic situation . . . I’d
like it if those that have coca crops and are landowners were to plant agricultural crops and buy
cattle, that will help us end with coca” (Archive, 2002).
55
As one ATCC leader stated, “The policy of the ATCC is to not get involved in questions of coca.
If some campesino has some kind of problem with the state, he’ll have to face the consequences.
The problem is that this brings disputes between the [armed] groups over the territory and forced
displacement” (Archive 2002).
56 57
ATCC#4, La India, 8/2008. ATCC#15, Bogotá, 10/2007.
58 59
ATCC#22, La India, 8/2008. ATCC#3, La India, 8/2008.
216 The Institution of the ATCC: Protection through Conciliation

live in the area of their own volition (rather by that of their parents).60 These
youths “created disorder” and were often paid by armed actors to be
informants (often using the money to buy small prizes, such as soft drinks
or new shoes).61 Again, once a youth is involved in the conflict, there is little
the ATCC can do. If a family’s adolescent is found to be continually causing
trouble, the Junta tells the family to leave. The parents cannot go against the
community’s request without losing the ATCC’s protection and facing poten-
tially lethal retribution from an armed group.
That the ATCC could do little for these victims is not to say that members of
the Association were not greatly pained by the killings. This became clear to me
one night when an ATCC founder and I got to chatting, sitting in plastic chairs
on his cement-strip porch and drinking sugary soft-drinks as the heat of the day
finally began to fade.62 I gently asked about the resumption of violence, a
period that few had discussed with me in detail. He looked up and, gazing off
into the darkness, listed the names of the victims, one after another, in cadence,
but pausing between each one for emphasis, perhaps, or respect or remorse.
When he finished, he was fighting back a tear. I was amazed that, even after a
few years, he could recall almost every single one. Even though the conflict-
related activities were “their thing” – a choice of most of the victims and
separate from the ATCC – I sensed regret and sorrow that the ATCC was not
able to do more.
As the ATCC and other communities have realized, maintaining their pro-
cess requires continuing collective action and active management. An implica-
tion from this discussion is that, in addition to mediation procedures,
community processes to maintain norms are central to staving off the return
of violence in the long term. Stability requires policies to counteract the self-
undermining processes triggered by their successes. Alternative development
programs can directly affect violence against civilians insofar as they can
prevent civilians from participating in the conflict out of opportunism or
desperation. Although coca cultivation in the region has declined in recent
years, the ATCC has continued working with residents to keep them from
growing coca.63 Some residents have also participated in the government’s
“guardabosques” (forest ranger) subsidy program to eradicate coca on their
farms. Programs to provide opportunities for the community’s youth are
equally important for stopping violence.

conclusions
This chapter profiled some of the unarmed, nonviolent strategies that are used
against heavily armed combatants in civil war settings to protect human rights.

60 61
ATCC#3, La India, 8/2008. ATCC#3, La India, 7/2008.
62 63
ATCC#11, La India, 8/2008. ATCC#11, La India, 8/2008.
Conclusions 217

The ATCC experience as a single but important community suggests that

civilians are not powerless and can effectively organize against repression to
make life in lawless wartime settings a little more predictable and ordered. The
ATCC civilians did not wait for intervention or focus on national-level peace
negotiations or military strategies. Instead, they took matters into their own
hands. I explored how and whether their efforts – the mediation, production of
credible information, and behavioral norms – functioned as an explanation for
reduced violence.
I developed an empirical framework and methods for measurement to study
this form of “peace building” in the midst of conflict. The dataset of threat
conciliations helped trace the ATCC process and even provided quantification
of how the armed conflict and concomitant threats unfolded and were dealt
with. Under the ATCC investigations mechanism, individuals found to have
been wrongly suspected of collaborating were less likely to be killed than those
found to have collaborated. The ATCC’s cooperation and institutions set it
apart in the eyes of armed groups from its violence-suffering neighbors. These
findings suggest that existing theories of violence such as the balance of control
have limitations since they do not completely explain violence: denunciations
against individuals were short-circuited and the production of violence did not
lie solely with armed groups. The findings have broad implications for civilian
agency, community autonomy and later resilience, and peace building in
civil wars.
Perhaps one sobering implication is that peace is not simply or easily
“created” or “built.” The absence of violence emerged through a subtle inter-
action between mediation, nonviolent civilian norms, and armed actor
preferences. There are also limitations on where civilian organizational processes
succeed – they are again not a panacea. Along with successes, communities face
challenges and failures. The ATCC suffered continued pressure from armed
actors and, at times, outbursts of violence. As the ATCC and other communities
have realized, stability requires continuing cooperation to counteract the self-
undermining processes triggered by their successes. These may include
strengthened mediation procedures, community processes to maintain norms,
or even alternative development programs that limit civilians’ participation in
the conflict out of opportunism or desperation. The ATCC has continued
working to keep residents from growing coca and provide opportunities for
youth.64
Organizations such as the ATCC exemplify what could be, what is possible.
These kinds of communities brave great risks and costs in resisting pressures
from armed groups and yet are crucial inspirational models. To the extent that
local peace institutions take hold across many communities, grassroots move-
ments may have broader effects on belligerents’ behavior at the macro level,

64
ATCC#11, La India, 8/2008.
218 The Institution of the ATCC: Protection through Conciliation

including resolving uncertainty, the reduction of violence, and supporting

national peace negotiations to bring conflicts to a close. The replicability of
experiences such as that of the ATCC should certainly be studied further, but
there is reason to believe they can generalize to other communities.65
The ATCC’s context of a frontier area with shifting conflict dynamics and
little state presence or rule of law resembles other conflict-ridden parts of the
world. Some of the ATCC’s features may be distinct, but within Colombia
similar investigation procedures have been implemented by certain village
councils and Indigenous groups in the crossfire, including the Nasa Indians in
the Cauca department. Many communities may be lying in wait to mimic these
processes, though less organized or highly endangered communities may con-
sider less institutionalized protective strategies. By unpacking the details of how
local, nonviolent protection institutions work beyond the mantra of “resist-
ance,” new communities and NGOs will hopefully be better able to understand
and apply these models when and where they are needed. The next chapter
continues this task by comparing towns in the department of Cundinamarca.

65
Similar mediation programs have been implemented in some American cities to end cycles of
gang violence (e.g., see Project CeaseFire, today known as Cure Violence: www
.cureviolence.org).
8

Discovering Civilian Autonomy in Cundinamarca

I’m going to tell you a story/ about why my town cried/ It happened in the early
morning/ disturbing a deep quiet/ The uniformed troops/ knocking as they could/
awakened many people/ according to them guerrillas/ They broke into homes/ of
whom they never should have/ asking for papers/ along with the prosecutor/
hurting feelings/ and opening many wounds . . .
We struggle to keep ourselves/ united in love/ and forget that the State/ scarred our
heart/ May the experience we lived/ help us not falter/ We ask the God of all/ give
us your grace and strength/ so Quipile won’t cry/ and that way is reborn.1
– “The Day Quipile Cried”
Berenice Cabra Jímenez
Para adelante cuando unidos; solos jodidos.
United we move ahead; alone, we’re screwed.
– Resident of Vianí (V#1, Vianí, 3/2009)

If you head west from the Colombian capital of Bogotá, just after dropping
off the central plateau you will find a number of small, isolated, mountainous
coffee-growing towns. In the 1990s, FARC guerrillas came to these towns,

1
Translated to the English by the author. “El Día Que Quipile Lloró”: Voy a contarles la historia/
Por la que lloró mi pueblo/Ocurrió muy de mañana/Perturbando un gran silencio/Las tropas de
uniformados/A golpes como pudieron/Despertaron mucha gente/Según ellos guerrilleros/Pene-
traron en las casas/de quienes nada debían/y pidiendo documentos/Junto con la fiscalía/Maltra-
taron sentimientos/y abrieron muchas heridas . . .
Luchemos por mantenernos/Unidos en el amor/Y olvidemos que el estado/Hirió nuestro
corazón/Que la experiencia vivida/Nos sirva pa` no caer/Pidamos al Dios de todos/Nos de su
gracia y poder/Pa` que Quipíle no llore/Y así vuelva a renacer.

219
220 Discovering Civilian Autonomy in Cundinamarca

massing their forces as part of their strategy to eventually cordon off and lay
siege to the capital. These towns were subjected to pressures and violence that
many had not experienced since the bipartisan violence of the 1950s. No
known formal civil society peace organizations emerged to respond to the
conflict. Yet even in this region, could the variation in the social and organiza-
tional landscape have impacted how this new period of armed conflict would
affect the civilian population?
This chapter explores additional town cases in the department of
Cundinamarca that were selected with the aid of universal data and statistical
models (as discussed in Chapter 6). The cases are similar or “matched” on
many of their characteristics except for differences in their historical densities of
junta councils. The goal of this exercise is to further test theory as well as assess
the accuracy of the statistical analysis.
I compare here the neighboring rural towns between Bogotá and
the Magdalena River of Quipile (key-PEE-lay), which historically had
a low number of junta councils, with Vianí (vee-ah-NEE) and Bituima
(bee-TWEE-mah), which together are similar in size and population to Quipile
(and were historically the same county) but had many more juntas in 1985 rela-
tive to their populations.2 I qualitatively investigate several central questions
about these cases to assess their fit with theory: What was the status of junta
councils on the ground and what other forms of social unity or divisions have
existed? What role if any did they play in affecting the nature of the armed
conflict and violence? An additional task throughout this chapter is to assess the
quality of the case matches using qualitative data and confirm whether the
similarities and differences encountered in the statistical analysis actually exist.
I argue by way of the Millian method of difference that the towns have many
conditions in common – including conditions that might predict violence – but
exhibit social differences. Despite being only several hours away from Bogotá
by car today, they have been historically and similarly isolated from state
presence. They are wedged between the touristic and commercial towns of
Anolaima, the self-proclaimed “fruit capital” of Colombia, and Villeta, the
self-proclaimed “panela” (sugarcane) capital of Colombia. They are also
wedged between roads running northwest and southwest away from Bogotá
(and therefore roughly equidistant from the city) but are isolated, and only
recently accessible by paved or semipaved roads.3 The populations are purely
“campesino,” or mestizo, with almost no Indigenous or Afro-Colombian resi-
dents. The towns are impoverished and have poor provision of public services.
Critically, and consistent with being small neighbor municipios, these towns

2
It is appropriate to consider Vianí and Bituima together because they are historically almost a
single area. One resident described them as “brother” towns. B#1, Bogotá, 3/2009.
3
These towns are not only isolated, but also forgotten. When I presented my preliminary research
at a think tank in Bogotá, the Colombian audience did not even realize that these counties existed.
They did not know where they were and thought they were perhaps villages.
Discovering Civilian Autonomy in Cundinamarca 221

faced similar projections of force by armed groups. The army only really came
to these regions during the implementation of Plan Patriota to disperse the
guerrillas and keep them from approaching the capital. For these reasons,
Quipile and Bituima-Vianí are nearly ideal selections for representativeness of
many Colombian towns. Further, their cultural conservativeness and relatively
long institutional legacies make them tougher tests for collective action.4
Contiguity implies similar topographical, geographical, political, and cultural
environments, but the physical separation between these towns makes contamin-
ation effects unlikely. Although the counties are situated in the mountainous
terrain of majestic, verdant Andean peaks, they are divided by a small set of
higher peaks. Historically, some communication passed through the corregi-
miento of La Sierra, but because of the peaks, the towns developed separately.
Indeed, they are located in separate geographic subregions (provinces) of
Cundinamarca, with Quipile known as the “ceiling” of Tequendama, while Vianí
and Bituima are situated at the southern end of Central Magdalena (Figure 8.1).
Based on the differences in the junta councils data, I expected to see differ-
ences on the ground in historical levels of organization and horizontal social
relations. Given civilian autonomy theory’s predictions for how variation in
these organizations affects violence, I therefore also expected to see differences
in how civilians were able to respond to the dynamics of armed conflict,
conditional on their levels of organization. However, I did not expect to find
full-blown peace organizations such as “peace communities.” If these organiza-
tions had existed, they surely would have already been reported in the press.
Rather, I expected I might find subtler and less well-known forms of social
cohesion and responses to the conflict. In correspondence with their measured
values of juntas then, I expected to find levels of organization increasing from
low levels in Quipile (which had few juntas), to a high degree of collective
action in Vianí, and even more in Bituima.
The analysis of the social life and armed conflict in the towns in this chapter
is based on firsthand field research carried out in 2009. I conducted forty
interviews with a variety of people with historical knowledge of social processes
and the armed conflict in these zones as well as additional interviews with
ex-combatants that operated in the region.5 When possible and as much as

4
In other words, their eras of colonist cooperation are in the distant past. Bituima and Vianí were
founded in 1772 and 1853, respectively, and were on the old road from the Magdalena River to
Bogotá. Quipile was founded in 1900.
5
As noted in Chapter 4, to increase the precision of historical knowledge and periodization, the
body of interviews benefits from people who arrived in the towns at different times or became
involved with the juntas at different times. An interesting outcome of the interviews was the
revelation of the different cognitive frames the residents of the different towns would subtly
express as they would respond to questions and think about social concepts through common
catchphrases. These sayings that came out during conversation are telling of cross-town differ-
ences (it is equally telling what is not said). I include them in their original form in direct
quotations as much as possible.
222 Discovering Civilian Autonomy in Cundinamarca

figure 8.1 Villages of the Cundinamarca municipios

Source: Cundinamarca Secretaría de Planeación.

possible, I visited the villages in the countryside and spoke with villagers when
they traveled to the town centers or Bogotá. Despite the absence of formal
peace organizations, I was surprised at how openly many subjects were willing
to discuss not only the history of the conflict and acts of violence, but also
depravities in the local political system. The information from these interviews
is supplemented by available secondary sources.
Discovering Civilian Autonomy in Cundinamarca 223

Despite my own initial skepticism at detecting differences on the ground,

I did in fact encounter the expected variation in precursor social conditions and
autonomy examples in my field research. Quipile historically had divided social
relations and few if any responses to armed groups.6 Vianí was characterized by
more unified and pacific social relations, and some “weapons of the weak”
tactics. Bituima, with the highest density of juntas, had both unity of social
relations and instances of collective civilian resistance to armed group hostility.
However, there were also displacements and collaboration with armed actors in
many villages across the towns. I assess the effect of civilians’ actions on
violence in three ways: civilians’ accounts, ex-combatant accounts, and add-
itional indicators of patterns of violence. In line with theoretical predictions, the
fieldwork highlights even greater differences in levels of violence across the
towns than is seen in the quantitative data.
The residents of Quipile were disadvantaged in being able to cope with
armed groups relative to those of Vianí and Bituima due to several factors.
Quipile is relatively impoverished in experience with the voluntary provision of
public goods and projects. It historically had lower levels of education (or a
smaller educated class to help organize), greater (perceived) inequality of social
relations, was more geographically and politically fractured, and there were few
outsiders to nurture communal cooperation. In contrast, while Vianí and
Bituima were not greatly organized compared to some towns in other parts of
the country, they were in some ways predisposed for social organization. Not
only were there stronger juntas, more unity, fewer social problems, and other
local organizations and associations, but there were also fortuitous (exogenous)
twists of fate (“critical junctures”) that helped them build these foundations.
The story of organization does not end there. On a deeper level, I also find
that conflict hurt social organizations such as juntas. So, even though there
were some collective actions against violence, they were not always coordinated
directly through the medium of the juntas. As I discuss at the end of the chapter,
I find a possible new, intervening explanation in clientelism as another reason
why juntas were weakened. Comparisons with the ATCC point to reasons why
clientelism can be the death knell of local organizations in some contexts but
not others.
This chapter differs in its content and purpose from the analysis of the
ATCC on several grounds. First, it serves as a qualitative test of observable
implications of theory and is more tightly linked with the statistical analysis.
These cases are separate from those used to generate theory. Second, rather
than directly testing new mechanisms by linking them to micro-level outcomes,
the emphasis is on showing that new civilian mechanisms are at least plausible.
This is because in the absence of formal human rights organizations, there is

6
That Quipile historically had fewer juntas does not necessarily mean that the juntas there never
played important roles, just that they were generally weaker.
 224 Discovering Civilian Autonomy in Cundinamarca

less reporting of information about violence and how civilians interacted

with armed groups compared to what is available for the ATCC investigations
mechanism. Third, and as a result of this different purpose and these infor-
mational challenges, this chapter is structured differently than the previous one.
The sections that follow are organized by causal and historical progression
up through the recent period of armed conflict. Within each section, I first
explore Quipile – the “negative” case – and then contrast it with Vianí
and Bituima. This chapter proceeds by first recounting experiences with La
Violencia, then preexisting social capital and juntas, then the incidence of
conflict, then responses to conflict and civilian interactions with armed groups,
then clientelism, and lastly outcomes of (the dependent variable of) violence.
I conclude with a summary of findings and methodological contributions and
indicate how the insights from this chapter compare with the findings from the
other chapters and may generalize more broadly.
This, then, is the story of how these communities made it through the war
years . . .

manifestations and impacts of la violencia

Understanding the local manifestations of La Violencia of the 1950s is
important because historical tranquility could mean any later relationship
found between civilian social organization and violence could be spurious.
There are also additional lessons about how peace was restored and whether
historical conflict triggered local collective action and cooperation in the form
of junta councils. Information on the events of this era was gathered by
speaking with elderly residents of these communities who were youths when
La Violencia occurred as well as asking middle-aged adults what they remem-
bered being told by their parents.
Their stories confirm that La Violencia affected all three towns: residents
spoke of fear, threats, violence, and displacement, often with partisan motiv-
ations, and of the presence of guerrilla bands, or “chusma.”7 Impressively,
residents could often recall specific events and episodes. One town had only
bipartisan conflict, while others had guerrilla bands in addition to bipartisan
conflict and killings. In support of the larger claims of this book, some of the
residents also recall civilian responses to deal with the violence even in these
early times. Most people attributed the decline in political tensions and the
ultimate end of violence to the National Front pact at the national level, but

7
The term “chusma,” translating to “rabble” or “mob” in English, was used in the vernacular as a
catchall phrase for an armed band. In some parts of Colombia, where there was conflict between
multiple bands, there was fighting between the Chusma and the Contra-Chusma. Interestingly, my
findings do not completely concur with the categorizations of La Violencia by Guzmán et al.
(1963). I found violence in all three towns, whereas Bituima is not listed in Guzmán et al.
Manifestations and Impacts of La Violencia 225

some suggested that juntas at the local level helped repair social relations and
build unity (and possibly political homogeneity as well).8
Residents of Quipile report that it was primarily a politically Conservative
town in the 1950s and experienced Liberal–Conservative partisan conflict and
killing, with Liberals tending to suffer greater persecution. According to what
one woman’s mother told her, there was also sporadic presence of “La
Chusma,”9 and a man remembered that the armed Chusma passed through
the sector of La Palestina (on the southwest edge of the municipio).10 These
bandoleros (gunslingers) were Conservatives and purportedly killed many
people. Some residents also mentioned the Chusma leader Sangre Negra, or
“Black Blood” (though it is not clear whether his band operated much in
Quipile).11 In La Sierra, on Quipile’s northern border with Bituima,
Conservative bandits would take Liberals down from the buses and kill them
(La Virgen, at the southern boundary, was a Liberal bastion and suffered
less).12 Illustrating how widespread the violence was, a man said that his
Conservative father, who had been living in the predominantly Liberal neigh-
bor town of Anolaima, had to come to Quipile to escape the violence against
Conservatives there.13 At one point, the residents united to form a local guard
to keep the bandoleros out.14
Vianí was a historically Liberal town and residents recall some levels of
conflict and partisan tension through 1965. One person attested that many
leaders in the town were killed in the 1950s.15 Another cited the occurrence
of threats, rape, and the burning of farms. One woman from a Liberal family
recalled a specific episode when Conservatives came to burn her in-laws’
farm.16 By contrast, a different woman’s family received threatening letters
because her father was Conservative.17 She remembers hearing of the
Chusma, or “Chulavitas,” and recalls that her father had to hide because
Liberals were going to kill him. Another man said that while there was
violence in the 1950s, the old city councilmen tended to get along.18 Resi-
dents also mentioned the bandit Sangre Negra and said that, while he did not
directly attack Vianí, his band was still operating in neighboring areas to the
northwest (toward Villeta and Chaguaní) in the early 1960s.19 Similar to
Quipile, Vianí also saw organization for self-defense during this time with
help from residents who were former soldiers.20 The priest of the era also
reportedly tried to defuse threats before they were acted upon.21 Some

8 9 10
E.g. B#2, Bituima, 3/2009. Q#1, Quipile, 3/2009. Q#2, Quipile, 3/2009.
11
Q#3, Quipile, 3/2009. Sangre Negra was finally shot dead by the army in Tolima department in
1964 (Time Magazine 1964).
12 13 14
Q#4, Quipile, 3/2009. Q#5, Quipile, 3/2009. Q#3, Quipile, 3/2009.
15 16 17
V#3, Vianí, 3/2009. V#2, Bogotá, 3/2009. V#3, Vianí, 3/2009.
18 19 20
V#4, Vianí, 3/2009. V#5, Bogotá, 3/2009. V#4, Vianí, 3/2009.
21
V#4, Vianí, 3/2009. The priest, who arrived in the early 1960s (see later in chapter), used Sunday
Mass to discuss social problems and shame thieves, and would promote dialogue to resolve
conflicts between neighbors. The man was also sufficiently brazen that he would at times
226 Discovering Civilian Autonomy in Cundinamarca

believed the arrival of the juntas around 1962 did help unite people by at
least highlighting community-wide interests through the completion of public
goods projects, though the juntas originally only tended to manage commu-
nity funds.22
In contrast to neighboring Vianí, and like Quipile, the population of Bituima
was characterized as largely Conservative going into the 1950s.23 According to
some residents, their grandparents said small groups of bandoleros would
attack political enemies.24 Some Liberal families were run out of town or killed.
A single large and powerful Liberal family persecuted many Conservative
families during this time. A man I met in Quipile said his family was run out
of Bituima and fled to Quipile for refuge.25
The record is clear that these towns of Cundinamarca experienced violence
during the 1950s. But did this harm collective action everywhere? Why or why
not? If all towns suffered from violence and this violence can increase social
homogenization or organizational responses, including the development of juntas,
why did Quipile have less widespread and sustained junta presence than Vianí and
Bituima? This history suggests that the social differences I find between the towns
in the next section are not solely due to being historically peaceful (or for that
matter solely suffering, since all towns are similar in experiencing past violence).

the juntas de acción comunal and preexisting

social capital
How do the municipios vary in terms of the breadth and depth of junta councils,
other social organizations, and horizontal social relations (social capital) from
the 1960s through the beginning of armed group presence and guerrilla incur-
sions of the early 1990s? Answering this question helps verify whether the cross-
municipio juntas variable from the statistical analysis is a faithful descriptor of
different municipios’ social lives – understanding whether, according to people’s
memories, the juntas were created, existed, and were active when the data say
they were. It also provides a picture of what real juntas actually do.
What emerges from the interview responses are coarse but identifiable differ-
ences in the social landscape and images of social relations. I first review the
challenging social landscape of Quipile and then recount the greater levels of
organization, activity, and “convivencia,” or coexistence, found in Vianí and
Bituima. In the next section, I proceed to examine whether the variation in
conditions mapped here predisposed these latter towns to be able to respond to
the conflict. For ease of comparison, the synthesis of indicators from this qualita-
tive analysis is displayed in Table 8.1.

purportedly approach suspects of crimes and grab them by the collar to rebuke them, practically
scaring them into reforming.
22 23 24
V#1, Vianí, 3/2009. B#1, Bogotá, 3/2009. B#2, Bituima, 3/2009.
25
Q#6, Quipile, 3/2009.
 table 8.1 Qualitative indicators of social cohesion and organization

Juntas and participation Machete Cattle/ coffee theft Other organizations Equality
fights and responses
Quipile Some juntas, but low Frequent Thefts of cattle, Failed efforts to form “Three climates” and
participation and activity tools, coffee cooperatives (only related social
some success recently) separation
Vianí Juntas everywhere and examples Uncommon Some cattle theft; Crop warehouse, etc. Equal plots carved
of successful cooperation response of local from latifundios;
(within and among juntas) watch committees village elders
conciliate
Bituima Juntas everywhere and examples Uncommon Some cattle theft; A local NGO; some Church divided land
of successful cooperation Response of local villages with many equally; village
watch committees stores (places to elders conciliate
socialize)
227
228 Discovering Civilian Autonomy in Cundinamarca

Quipile
Quipile certainly has some examples of social capital and junta activity, but the
more common tendency has been difficult and distant social relations and a
predominantly weak and narrow penetration of junta councils in the munici-
pio. In exploring the modern origins of cooperation to provide public goods in
Quipile, residents did recall an auspicious episode from the late 1950s. At that
time, Quipile benefitted from a proactive pastor who helped organize the town
to pave roads.26 With his help, they contracted for the use of two bulldozers
(once of which they aptly named the “Conqueror”) from the departmental
government.
For a variety of projects, both in the past and more recent, however, the
townspeople themselves did not have to contribute effort or labor. A resident
recalled that in the 1950s and perhaps into the 1960s, prison “chain gang”
laborers were brought in from outside the municipio to build roads.27 The
residents were not brought together to maintain associative traditions for the
common good. This challenge of energizing voluntary contributions for public
goods carried through to later years. According to a former resident with a
long-running history of involvement in a junta, it was a struggle, for example,
to build an aqueduct system for the town.28 He requisitioned funds to buy pipe
and tried to get people to participate and contribute work, but little collabor-
ation was forthcoming.29 Indeed, other residents lamented that Quipile has
never planned to develop and take advantage of its abundant water resources.30
Similarly, the aforementioned roads constructed during the mid-twentieth cen-
tury were allowed to fall into disrepair as residents and mayors could not
cooperate to repave some sections that connect Quipile to neighbor municipios
until a few years ago.31
The distribution and activities of juntas in Quipile are consistent with the
town’s general trend of collective action problems as residents recall historic-
ally few and ineffective juntas. This jibes with the municipio’s statistics on
juntas: the 1985 dataset shows that there were only eight juntas in the entire

26 27 28
Q#7, Quipile, 3/2009. Q#2, Quipile, 3/2009. Q#5, Quipile, 3/2009.
29
Q#2, Quipile, 3/2009. This man felt “he accomplished in one year as a leader what many
presidents couldn’t do in twenty,” but finished so exhausted and embittered by the experience
that he washed his hands of communal action. Others noted that a later effort to build a small
hotel complex with a pool to bring tourist revenues to the town was a similar boondoggle. The
pool was built but the complex was left incomplete and now sits in disuse.
30
Ironically, during one of my field visits the aqueduct to the town center broke due to a few days
of heavy, sustained rain that washed out the pipe connections up the mountain, leaving the town
without water for three days. I was informed this was a frequent occurrence but that not much is
done about it.
31
Q#8, Quipile, 3/2009. For instance, a junta leader from the early 1990s reported she tried to
repair a bulldozer and repaired roads to a few villages, but there was insufficient commitment
and the people “got tired of working.” Q#1, Quipile, 3/2009.
Juntas de Acción Comunal and Preexisting Social Capital 229

municipio (which encompassed thirty-two villages, four urban centers, and a

population upwards of 8,000 people). One man recalled that, when he arrived
in Quipile in 1980, the urban centers of the county seat, La Sierra, La Botica,
Santa Marta, and La Virgen did have juntas, but many villages did not have
juntas.32 Residents described a push to form some juntas in the 1960s and
1970s, but the councils did not become ubiquitous. Even when they were
founded in some places they suffered from low levels of participation and
meeting attendance, were not sustained and were therefore short-lived.33
Furthermore, it is telling what people did not mention in interviews – residents
recalled few if any major projects implemented by the juntas. Even today, there
is no designated communal junta meeting room in the municipio (or “salon
comunal”).
The variation of junta activity within municipios is of course more complex.
Although the Quipile juntas have generally been weak, there have been some
geographical pockets of organization. For instance, several people concurred
that La Sierra (close to Bituima) has had more solidarity and been better
organized.34 A visit to La Virgen to the south also revealed greater cooperation
and junta activity. A current priest observed some of the better-organized
villages in the municipio were Guadalupe Bajo and La Unión but that only
about three of the eleven villages in his ecclesiastical district had active juntas.35
A villager felt that some villages such as Los Guayabos were united and
successfully worked in groups, but that others such as Sinaí were not very
united.36 Another person believed distant villages like La Joya did not have
juntas at all.37 Multiple people further observed that there was historically little
contact between the juntas in the municipio and that the juntas were themselves
not very united.38
When and where the juntas have functioned in Quipile, they would, among
other things, communicate urban planning needs to the mayor.39 Juntas could
also requisition resources for projects from the Committee of Coffee Growers
(Comité de Cafeteros; although coffee production declined in the 1980s).40 One

32
A different person disagreed and thought that juntas were in fact widespread in the municipio by
the 1980s, but inconsistencies with this statement during the interview provide reasons to doubt
this assertion.
33
Q#11, Quipile, 3/2009. Another long-time resident believed many juntas did not “legalize”
(registering with the government to get recognition for resources – a necessity for organizational
survival) because there was not enough “push,” or participation. Q#3, Quipile, 3/2009.
According to someone else, some disorganized juntas only legalized about fifteen years ago.
Q#9, Quipile, 3/2009. A third person conjectured that many villages perhaps did not legalize
juntas out of “laziness” (“pereza”) because it may have required making a trip to Bogotá to
register. Q#10, Quipile, 3/2009.
34
Q#8, Bogotá, 3/2009. “They’re fighters, really moving ahead, awesome” (“son luchadores . . .
muy echada pa’ adelante, bacana”).
35 36 37
Q#12, Quipile, 3/2009. Q#13, Quipile, 3/2009. Q#10, Quipile, 3/2009.
38 39 40
Q#1, 2, Quipile, 3/2009. Q#10, Quipile, 3/2009. Q#9, Quipile, 3/2009.
230 Discovering Civilian Autonomy in Cundinamarca

resident noted that up until about 1985, junta meetings were “important,” but
others voiced more skepticism. One person said many people would not go to
junta meetings and another believed the juntas were basically only a way to get
resources or for candidates to get elected to the next office.41 The juntas
encountered problems with “anti-progressive” mayors who would skim money
from some projects.42 Juntas were strong in some areas, but were later manipu-
lated and weakened by clientelist politics.43 These problems with juntas are
indicative of other organizational problems in the villages of Quipile. For
example, even though the farmers live on thin margins they have not been able
to cooperate to more cheaply transport crops to market.44
Widely cited explanations for the patchy functioning of juntas include the
problems of lack of leadership, passivity, complacency, and dependence on
local authorities to drive collective action. A man observed that juntas “only
survive if they have good leaders, but there was not a strong enough culture [in
Quipile] for junta leaders to keep being produced [everywhere].”45 The “cul-
ture” issue was seconded by another woman, who used the adjective “individu-
alist.”46 Another man, in noting the “lack of commitment” (“falta de
compromiso”), similarly pointed out that not all villages have good leaders.47
The former mayoral liaison to the juntas observed from his work with the
councils that “the residents were not united and that the people needed a
guide.” Another former leader of the juntas of the municipio in the early
1990s noted a big problem was that the juntas never received national govern-
ment support.48 This statement is telling about both the juntas’ ineffectiveness
as well as the weary, passive attitude of people involved in the councils –
waiting for outside help instead of undertaking new ventures on their own.49
In this vein, a resident from one village said they felt they were the most
forgotten village (using the diminutive, “somos olvidaditos”), again reinforcing

41 42 43
Q#7, 13, Quipile, 3/2009. Q#5, Quipile, 3/2009. Q#3, Quipile, 3/2009.
44
Q#7, Quipile, 3/2009. Some aqueduct associations and crop cooperatives have succeeded
(though only most recently) and persist today, but several past efforts to form cooperatives for
plantains and sugar failed. Q#2, Quipile, 3/2009. The Comité de Cafeteros used to be more
active in the past.
45 46
Q#5, Quipile, 3/2009. Q#8, Quipile, 3/2009.
47 48
Q#9, Quipile, 3/2009, Q#13, Quipile, 3/2009. Q#7, Quipile, 3/2009.
49
Q#12, Quipile, 3/2009. The impression of a current priest is that, equally in the countryside as in
the town, the people are “a little difficult to get organized” (“durita para organizarla”) and that
they are lazy about coming to meetings or workshops (“perezosa para formarse”). They are
more “individualist” and tend to “keep to themselves” (“cada uno por su lado”; perhaps in part
from fear from conflict). He has also observed low attendance at his village masses, saying that
the people believe in God but do not congregate, “They don’t integrate/gel” (“no se integra”).
This suggests that even outsiders who have come to Quipile recently have been surprised at the
lack of organization in the municipio. Most organization today (in the cabecera) appears to be
pushed by people from outside. Even today, differences between Quipile and the other munici-
pios are notable to the casual observer: life in Vianí is, at least on the surface, more vibrant than
in Quipile, with people associating out in the street.
Juntas de Acción Comunal and Preexisting Social Capital 231

The main street of Quipile, Cundinamarca, Colombia, 2009. Photograph by Oliver

Kaplan.

their reliance on outside groups and a certain “learned” helplessness and lack
of initiative.50 As one person aptly summarized, there is “ignorance” in Quipile
and it does not have an associative, “do-it-yourself” colonist culture.51
Whatever junta activity did exist through the 1980s, the juntas of today are
much weaker. These already feeble organizations collapsed further due in part
to effects of the armed conflict, such as fear and pressure from armed groups.
One person stated that, from about 1994 on, the juntas “only became a vehicle
to fill out paperwork.”52 When a village junta leader came to Quipile around
1999, he was surprised to find that communal action practically did not exist,
that the juntas were “ungrounded” (“desarraigada”), and that each village
existed as a “loose wheel” (“rueda suelta”), disunited.53 Juntas have really
only been strengthened since about 2005, as part of the nationwide juntas

50 51 52
Q#13, Quipile, 3/2009. Q#2, Quipile, 3/2009. Q#10, Quipile, 3/2009.
53
Q#14, Quipile, 3/2009. From that time up until a year ago there was no full assembly of juntas in
the municipio. He perceived that the junta movement in Quipile “had a lack of will or precedent
for working together” and that “people never proposed many ideas to move ahead – a lack of
leaders.” Once again, Quipile depended upon outsiders to push organizational processes for-
ward.
232 Discovering Civilian Autonomy in Cundinamarca

An Acción Comunal trash bin in Quipile, Cundinamarca, Colombia, 2009. Rather than
symbolizing successful cooperation, the bins came to be lampooned as representing the
listlessness of the cabecera’s junta council. Photograph by Oliver Kaplan.
Juntas de Acción Comunal and Preexisting Social Capital 233

“renewal” movement. They now exist in almost all villages, but in some places
only in name (and the impression of many is they would fail without mayoral
support). For example, around 2008, 300 people signed up to participate in the
junta in Quipile’s cabecera, or county seat, but only 50 showed up to the first
meeting and there was not enough interest for continued attendance.54 Perhaps
most emblematic of the problems with juntas in Quipile, about the only thing
the junta of the county seat was recently able to accomplish, and almost as a
final afterthought before ending its term, was to put up “acción comunal”
garbage bins around the town (what’s worse, they have been graffitied, resi-
dents do not use them much, and some people actually removed them from in
front of their houses because of the smell!).55
In Quipile and the other municipios I also asked about important social
trends and indicators of social harmony, order, and how conflicts are resolved.
First, I asked about the prevalence of machete fights, or “macheteras,” in hopes
of understanding the severity of social divisions and cleavages that armed
actors might later exploit. In many parts of rural Colombia, as well as many
parts of Latin America and the developing world, farmers carry small garden
swords called machetes. The machete is a useful tool for cutting plants and
brush and clearing trails and, in some cases, is the campesino’s only means of
self-defense. As a sociological phenomenon in Colombia, the machete is also
the implement of choice for committing violence or brawling.56
Residents of Quipile were fairly unanimous in describing macheteras as a
frequent occurrence from the 1970s up until the early 1990s.57 During this
period, with more difficult access to villages, many people would come to town
for Sunday markets, where people would drink large quantities of alcohol and
machete fights would occur over such banalities as love or loans. As one person
reported, there were many machete fights “everywhere” (and what is worse, the
town had no ambulance!).58 Sometimes, four or five people would fight for up
to thirty minutes. The police often did not get involved in these incidents, either
arriving (conveniently) late or letting the brawlers just fight it out (though the
police might later fine the fighters). A resident of a village specifically remem-
bered that certain villages like La Candelaria were especially prone to machete
fights and other incidents with drunks.59 He recalled many fights – practically
every week – in Boquerón de Hilo and Botica. Another reported they were
moderately common in La Virgen.60
Cattle theft, an indicator of the level of disorder and local solutions for
justice, is another prevalent social problem in Latin America and other parts

54 55
Q#11, Quipile, 3/2009. Q#2, 11, Quipile, 3/2009.
56
A resident of Vianí shared a revealing saying about parties, “If there’s not a machetera it’s not
really a fiesta!” V#3, Vianí, 3/2009. Machetes have also been used for macabre ends. In Rwanda,
machetes were the main implement used in the killings of that country’s genocide.
57
Another sign of belligerence there is the frequent cockfights, or galleras.
58 59 60
Q#9, Quipile, 3/2009. Q#13, Quipile, 3/2009. Q#8, Quipile, 3/2009.
234 Discovering Civilian Autonomy in Cundinamarca

of the developing world where state presence and policing is scarce (livestock is
a key asset of rural residents for preserving wealth).61 The extent of theft of
cattle and other kinds of robberies as well as collective responses to these
incidents can be another useful standard for cross-town comparisons. Though
Quipile is not a predominantly cattle-ranching municipio, some residents
reported the town historically suffered many incidences of abigeato (cattle
theft) as well as robberies of trapiches (cane presses), farm tools, and even
coffee and panela (sugar cane) crops.62 Nobody I spoke with ever recalled any
local committees or organized efforts to deal with this issue.63

Vianí
The municipio of Vianí has exhibited subtle but meaningful differences from
Quipile in social cooperation and junta councils. With a high number of juntas
and juntas per capita – fifteen of sixteen villages had juntas in 1985 – the
expectation is to find stronger signs of cooperation in Vianí. The qualitative
evidence shows that Vianí had several advantages for collective action and
junta persistence, which resulted in greater junta activity, additional social
organizations, and harmonious social relations. Residents recalled that Vianí
was a UNESCO “model town” in the 1950s and an elderly woman remembers
there was much social life. There was also a group of educated literati –
intellectuals, lawyers, dentists, business people – in Vianí in the 1960s.64
Compared to Quipile, various indicators confirm Vianí’s historically better
civil relations and procedures for social control.
Even with better social conditions, somewhat random events were critical for
catalyzing social cooperation and organization. Most prominently, people
spoke of the arrival of a Catholic priest in 1962, after La Violencia, who helped
transform social life in the municipio.65 In this year, the story goes, the roof of
the town church collapsed, and the priest formed the first junta around the goal
of bringing the residents together to raise a new roof and repair the church. The
roof collapse and this priest appear to have (exogenously) helped set Vianí on
an organizational path. Witnessing this early success, the priest was spurred on
to continue forming juntas. According to an old coffee grower I spoke with, the
priest went from village to village organizing juntas.66
When I was able to seek out this now-elderly priest, retired and living in
Bogotá, he recalled that, at least while he was there in the 1960s, the people of

61
See Gitlitz and Rojas (1983) about the origins of the Rondas Campesinas in Peru. In Latin
America there are several words that refer specifically to the theft of livestock and cattle-rustling,
including not only “robo de ganado” but also “abigeato” and “cuatrero,” perhaps reflecting its
importance.
62
Q#2, 8, Quipile, 3/2009.
63
Although at times a junta fiscal (officer) might report problems to police inspectors.
64 65 66
V#3, Vianí, 3/2009. V#3, 5, Vianí, 3/2009. V#6, Vianí, 3/2009.
Juntas de Acción Comunal and Preexisting Social Capital 235

The church in Vianí, Cundinamarca, whose roof collapsed in 1961 (2009). A priest
helped organize the community to rebuild it, kick-starting the junta councils in the
municipio and the neighboring town of Bituima. Photograph by Oliver Kaplan.

Vianí were very collaborative (“muy colaboradora”) and that there were
frequent junta meetings.67 Vianí also received (technical) support for the juntas
in the form of a government junta promoter because he was requested by the
priest as well as a U.S. Peace Corps volunteer (around 1967–1968), both of
whom helped coordinate and develop the juntas. The same coffee grower
believes that if this priest had not come to the municipio, the juntas would
not have formed as they did (i.e., they would have been more dependent on the
government and less unified). He said the priest’s legacy was that the “people
never let the juntas die.”68
The priest’s work was likely made easier by organizationally favorable
preconditions in the municipio. A series of land reforms in rural areas had
the effect of creating a class of small farmers and fostered social cohesion.

67
V#8, Bogotá, 3/2009. It was also his impression that Vianí had more junta activity than the next
town he was assigned to by the Diocese (Viotá).
68
As I find later, this had not turned out to be completely true, at least not for all parts of Vianí, but
that in general the juntas have historically been strong.
236 Discovering Civilian Autonomy in Cundinamarca

The passage of two national laws in the 1930s, the agrarian reform Law
200 of 1936 and the 1936 Labor Law, impacted certain large property
holders in Vianí and required that latifundios be split apart. As a resident
told me, some of these “latifundistas” feared that rough and violent types of
people might come from outside the area to claim pieces of land.69 To
prevent this from happening, a latifundio owner in what is today the village
of Manillas decided to slowly sell off pieces of his property to his own
sharecroppers, who he at least knew – who were trusted and loyal and with
whom he had good relations. As the laborers worked off the price of the
plots, he gave them deeds. As generations passed, land kept being divided
among the descendants of these original sharecroppers, who live there
harmoniously to this day.
The preexisting cosmopolitan population, the fortuitous efforts of the priest,
and the legacy of amicable land reform in Vianí helped produce a history of
early, strong, widespread, and lasting juntas. Residents recalled the existence of
juntas from the 1960s and 1970s in most parts of the municipio.70 Even in the
large, isolated village of Manillas, a resident told how a group got the perso-
nería jurídica (legal charter) on August 11, 1970, and that the junta remained
active (up until the armed conflict and guerrillas came).71 Similarly, the old
coffee grower recalls that residents also obtained the charter for his village’s
(Vianicito) junta early on, around 1970 (which required going to Bogotá to
legalize).72 He also recalled there was “always much” (“siempre, mucho”)
contact among junta leaders in the municipio.73
There is also proof of participation in juntas and various successful projects
and activities. The coffee grower recalls many public works led by the junta in
Vianicito and that the juntas in the municipio would hold fairs (“bazaars”) to
raise funds for projects.74 A resident pointed to the example of the roads in the
municipio, which were built through communal action, and noted that today
all villages are accessible by roads.75 Even more recently, in the early 1990s, a
large volunteer work group of about forty-five community members united to
build aqueducts with help from the Comité de Cafeteros, the juntas, and the
mayor.76 Also in this same period, all (seventeen) junta leaders from Vianí met
to discuss education and school infrastructure in the municipio.
Despite the juntas’ propitious beginnings, the juntas in Vianí today, like
those in Quipile, are weak and poorly organized, existing in some cases in

69 70 71
V#2, Bogotá, 3/2009. V#4, Vianí, 3/2009. V#2, 9, Bogotá, 3/2009.
72
V#6, Vianí, 3/2009. Organizations are important, he said, because being able to demonstrate
you can commit to contribute labor for projects makes it easier to get money and resources (from
the departmental government or private organizations like the Comité de Cafeteros). As he
commented, “Without the charter, people [in government/outside the community] wouldn’t pay
attention to you.”
73
He observed it was an advantage to have a junta in each village, as otherwise it can be difficult to
reach agreements among various villages for joint projects.
74 75 76
V#6, Vianí, 3/2009. V#4, Vianí, 3/2009. V#3, Vianí, 3/2009.
Juntas de Acción Comunal and Preexisting Social Capital 237

name only. One man said that village leaders were powerful up through the
1970s and 1980s but have since lost power.77 A priest observed that when he
arrived in 2003, the juntas were weak, and have really only been revived since
2007.78 Even with this decline, a person involved with the juntas believed some
good junta leaders remain. One of the strongest juntas at present is in Manillas,
where residents were able to organize to build their own village chapel.79
Today, the main activities are town cleanups, since there have been few funds
for projects after the congressional grants known as “auxilios parlamentarios”
were eliminated in 1993.80 Problems are also attributed to historical paternal-
ism, broken promises by mayors, exclusion of female leaders, and varying
support among department governors for local juntas.81
Consistent with the strength of the juntas in Vianí, there are various other
examples of unity and cooperation from the municipio. A person cited the
village of Calambata as historically being very cohesive and said that, through
their solidarity, they formed a collective sales center (“centro de acopio”)
to avoid having to pay high quotas to intermediaries when selling crops
(something residents of Quipile failed to collectively achieve).82 In Manillas, a
high-elevation coffee-producing village, the Comité de Cafeteros supported
many public works. The villages of Hatillo and Cuchimira were also viewed
as well-organized villages.83 These specific examples coincide with the resi-
dents’ general views of the town’s social climate. As one man described it, “In
Vianí, when someone needs something or is suffering, nobody closes their
doors. They pitch in to help.” Another man reiterated the early priest’s senti-
ments about the people being “very collaborative.”84
By the same token, the unity in Vianí has meant there is little crime and few
social divisions or quarrels (especially in the 1970s and 1980s).85 There are few
property line disagreements and various people testified that the residents are
not “bellicose” or “prone to fighting” (“la gente no es peleadora”). Various
residents said that while in the past there were perhaps a few feuds between
families or villages and some macheteras, they were not very common. For
instance, many people would come to town for the county festival, but there
would not be many fights. One longtime resident said they could hardly
remember any machete fights, except perhaps a few between migrant coffee
pickers. In Manillas, a resident recalled that there were “some” but not many in
the 1970s and 1980s (and also attributed the “convivencia” in part to the
population being 90 percent Liberal).86 As a city councilman explained, “The
people of the countryside are wholesome” (at least in Vianí).87

77
V#6, Vianí, 3/2009. A different resident who started with the junta in 1996 said in 2002 there
was no junta participation. V#12, Vianí, 3/2009.
78 79 80
V#5, Vianí, 3/2009. V#10, Vianí, 3/2009. V#11, Vianí, 3/2009.
81 82 83
V#12, Vianí, 3/2009. V#2, Bogotá, 3/2009. V#1, Vianí, 3/2009.
84 85 86
V#11, Vianí, 3/2009. V#6, 9, 11, Bogotá, 3/2009. V#9, Vianí, 3/2009.
87
“Gente de campo es gente ‘sana.’” This was noted by about half the interviewees.
238 Discovering Civilian Autonomy in Cundinamarca

Vianí has some traditions of conciliation for when conflicts did arise. An old
man from Manillas recalled how he and two other men were informal mediators
for neighbors’ conflicts because they were respected and viewed as impartial
(instead of the alternative of going to the distant police inspector).88 They would
deal with problems like disputes over property boundaries or cows trampling
fences and eating crops (where, for example, they would be called upon to
impartially value the damage) by hearing the positions of the interested parties
and then proposing a solution. After reaching an agreement, the deal would
generally be solidified by the ever-important ritual of drinking beer together.89
During the 1970s and 1980s, there were also reports of the church helping to
mediate conflicts.90 Several people also noted that junta presidents became
important authorities in the countryside for conflict resolution. One person
recalled how a junta dealt with the “commons” problem of water overuse by
taxing the resource so residents of the village would share and conserve it.91
A more recent junta leader said she was called on to resolve a variety of domestic
and neighborly conflicts, as well as deal with some cases of rape.92
Even with the general unity and traditions of conciliation in Vianí, there were
still some instances of insecurity in the form of cattle theft (abigeato), with
varying reports as to its prevalence. In certain villages, prior to the arrival of
the guerrillas, friends joined up to maintain order, policing certain families
suspected of robbing and fighting off thieves from other towns.93 One man
recalled “quiet” strategies to maintain order from the 1970s and 1980s in the
village of Hatillo, “I tried to unite neighbors for security against (cattle) theft and
also against rape and muggings (atraco). We had secret, quiet discussions to
organize collective vigilance. Neighbors would get together at night with arms,
with revolvers to prevent theft – like a civil defense (defensa civil).”94 The
purpose of these committees was more to frighten and deter would-be miscre-
ants, though a few perpetrators were actually purportedly killed. As the man
explained, “When there’s no army present, one must take justice into his own
hands.” These committees were said to have lasted about five years and, even
though neither they nor the problem of cattle theft were long-lasting, the experi-
ence did instill trust among neighbors (the guerrillas also suppressed cattle theft
when they entered the municipio).95 The result was that residents saw it was

88 89
V#9, Bogotá, 3/2009. This type of mediation was suspended when the guerrillas arrived.
90 91 92
V#1, Vianí, 3/2009. V#6, Vianí, 3/2009. V#12, Vianí, 3/2009.
93
V#2, Vianí, 3/2009. Even before La Violencia, justice was often meted out at the local level due
to the long distances it was necessary to travel to reach official institutions. For instance,
latifundios would have their own small holding cells for accused criminals where they would
have to serve a period of incarceration. There was also a local “commissary” who would task
criminals with communal labor.
94
V#1, Vianí, 3/2009.
95
Their arms were useless against armed groups and the committees ended around 1990, when the
guerrillas arrived in Vianí.
Juntas de Acción Comunal and Preexisting Social Capital 239

better to resolve conflicts within the town and rely on lines of authority embed-
ded between and within families.96
The existence of similar associations in Manillas to deal with a relatively
small number of incidences of cattle theft was also described by multiple
interviewees.97 The neighbors would form “commissions” (posses) to
investigate an allegation. They would call on the suspect(s) to turn themselves
in and, if they did not come willingly, the commission of four to eight people,
armed with revolvers or machetes, would go to capture (but not hurt) them.
Once captured, the suspects would then be turned over to the local authorities.
In sum, the social history of Vianí contains ample episodes of initiative to deal
with local problems rather than waiting for help from the outside.

Bituima
Bituima has historically had an even wider coverage of juntas than Vianí and
demonstrates similarly high levels of social cooperation. According to the 1985
data, Bituima had juntas early on and in all its villages and a high number of
juntas per capita. In Bituima, as in Vianí, there were some early and key
exogenous precursors to the formation of juntas that disrupted latifundio
land-holding patterns and helped set parts of the municipio on a path toward
local cooperation and organization. And, as in Vianí, the church also had a role
to play. According to one story, Concepción Romero de Bustos, a woman who
was a large landowner with property in the present-day villages of Caracol,
Aposentos, Volcán, and Montañas, died a widow and without a will or heirs
(inheritors). The land was received by the Catholic Church, which proceeded to
divide it into equal plots of small farms, or “minifundios.” These small plots
helped engender a proximity and density of interaction among rural residents,
as well as shared preferences, which were helpful antecedents for effective rural
organizations and cooperation.98
Bituima experienced the same kind of historical agrarian reform process that
occurred in parts of Vianí. As a result of the passage of agrarian reform laws,
some large latifundio properties were broken up. Owners sold plots to the
existing sharecropper renting families (arrendatarios) who had been there all
their lives. Since many of these families had peacefully coexisted and gotten

96
As the man elaborated on the issue, “When outside forces come and commit atrocities and
abuses, we don’t know who is ordering whom or why, so it’s hard to fix the problem. In contrast,
within communities, there are lines of authority and families can limit abuses of power. Families
have the moral power to deal with individual thieves/‘abusers’/delinquents. Four or five families
might unite and talk with the people or parents of youths who were stealing and would tell them,
for instance, to control their kids, or ‘there will be consequences,’ as in threats to kill (informal,
but it worked). They might also ask for a ‘contraprestación’/collateral/counterweight.”
97
V#2, 9, Bogotá, 3/2009.
98
B#1, Bogotá, 3/2009. Quipile is also primarily composed of minifundios, but this did not appear
to engender similarly close social relations.
240 Discovering Civilian Autonomy in Cundinamarca

along well, this fomented good social relations. The division of land also
produced well-defined property boundaries (linderos), which helped to reduce
conflict (even though residents did not always hold official deeds). Although
there is a general tendency of association in the municipio, consistent with these
accounts of land reform, the villages of Aposentos and Gualivá are perceived to
be the most educated and equal areas today.99 The same Quipile woman who
characterized her town as “individualist” goes frequently to Bituima and dis-
tinguished the people there as being “[socially] aware.”100
The organizational form of juntas was quickly adopted and became wide-
spread in Bituima when it was first established in the 1960s. One middle-aged
man recalled that, when he was a child (about fifty years ago), juntas existed in
every village.101 The residents also recalled the enthusiastic priest from Vianí.
Juntas were “always there” and helped with local planning and a variety of
public works projects, including building schools and a police post. The juntas
also generally received the collaboration of mayors. A former mayor reported
working well with the juntas and, even during the 1990s, mayors contracted
with juntas for projects.102 There is also evidence of unity and coordination
among the juntas, as representatives from the juntas would meet in the town
center. However, today, as in the other municipios, junta participation suffers
from greater apathy.
Bituima has also had other social organizations. From 1990 to 1993,
the community started its own nongovernmental organization (NGO)
called “Todos por Bituima” (All for Bituima).103 They received money from
outside sources to build their own self-sufficient and sustainable granaries
and seed banks. It was one resident’s perception that the organization
succeeded in improving communal work and cooperation, but that it
also faced challenges because people were self-interested (the NGO report-
edly dissolved once the armed groups arrived). Certain villages were also
highly cohesive because of their large number of stores, which serve as
important gathering places (those without stores were less integrated and
communicative).104
Bituima, like any community, experienced some social divisions. But like
Vianí, the recollections of residents point to a history of unity rather than social
disorder. In one episode from the 1970s, campesinos who wanted to build a
high school confronted the obstructionist (“anti-progress”) and politically
powerful cattle boss elites (gamonales).105 While the 1980s saw some partisan
conflicts and tensions because of ever-present poverty, they did not generally
escalate, and grave conflicts between neighbors were rare. The record of
machete fights tends to support this view. Although residents reported that
some macheteras occurred and increased in the 1970s and 1980s, they were not

99 100 101
B#1, Bogotá, 3/2009. Q#8, Bogotá, 3/2009. B#2, Bituima, 3/2009.
102 103 104
B#3, Bituima, 8/2009. B#1, Bogotá, 8/2009. B#3, Bituima, 8/2009.
105
B#4, Bituima, 3/2009.
Juntas de Acción Comunal and Preexisting Social Capital 241

all that common (a partially dissenting voice said, “Yes, macheteras were
common, but less so now”).106 In a joint interview with someone from Quipile
and someone from Bituima, both agreed that Quipile historically tended to
have more machete fights.107
Like in the other towns, there was also some cattle theft (abigeato) in Bituima,
but it was not as severe a problem. As one interviewee reported, residents unified
to deal with this threat and would advise neighbors when they saw strangers pass
by.108 Her father, a coffee-grower whose harvest was stolen several times at
night, led his neighbors in the collective responses of organizing a night watch
and later pooling resources to eventually build a police post. During the 1990s,
there were some reported increases in the amount of cattle theft and highway
holdups, but cell phones have helped with security today by facilitating
communication.
Bituima, like Vianí, has a prior tradition of local conflict resolution. Before
the creation of a police inspection, Bituima had conciliation centers (“centros de
conciliación”) in some villages. According to one woman, up until the 1960s,
village “elders” would mediate conflict in the countryside using their moral
authority to bring disputants and enemies together. They would use “registries
of trust” (“escrituras de confianza”) that documented statements of witnesses
to the dispute to reach agreements (similar to the ATCC’s conciliations). The
juntas were also mentioned as important arenas in this process. Junta leaders
fell into this class of “elders” and were viewed as impartial conciliators that
could promote dialogue. This tradition has dropped off more recently with the
greater role of the police inspector, but junta leaders say some of these
leadership aspects are conserved today.
In sum, the legacies of social cohesion in the various municipios demarcate
one town from another. Quipile has had some hints of collective projects and
unity, but was practically destined for difficulties. By contrast, both Vianí and
Bituima exhibited greater organization and unity among residents. The juntas
were traditionally more widespread across political subunits and more effect-
ive, despite signs of their waning in the 1990s.109 This qualitative analysis
uncovers that these differences were in part due to idiosyncratic, exogenous
factors, which were not easily measured or identified in statistical cross-
sectional analysis. The different origins of the juntas across municipios mean
that some legalized earlier, lasted longer, and left greater imprints on their
communities’ social relations. The resulting differences in junta activity correl-
ate with other forms of social cooperation and organization that are obvious
even today.

106 107
B#1, 2, Bituima, 3/2009. Q#8, Quipile, 3/2009. B#1, Bogotá, 3/2009.
108
B#1, Bituima, 3/2009.
109
Falling coffee prices also contributed to the economic and social decline of these coffee-growing
municipios.
242 Discovering Civilian Autonomy in Cundinamarca

the nature and severity of armed conflict

This section explores the dynamics of the armed conflict over time, including a
look at the severity of contestation between the armed groups and their inter-
ests, motivations, and strategies toward civilians. While assessments in the
statistical model based on counts of attacks, reported events, and measures of
state presence are helpful, they are less than ideal to understand the texture of
day-to-day life. The qualitative assessment here seeks to better verify whether
the towns were indeed in the same type of zone of control as the statistical
matching procedures suggest they were – a key determinant of violence
according to the balance of control theory. As in the statistical analysis, the
task is to measure control apart from violence.
The characterizations here draw from four main sources: press reports,
civilian and ex-combatant interviews, and work by Peña (1997), which con-
tains around 100 additional interviews with active guerrillas from the early
days of the fronts in the region during the mid-1990s. I find that these towns
had similar experiences and were subject to pressure from the same armed
group units, implying similar rules and modes of operation. However, while
I expected zones of dominant though incomplete control based on the data on
attacks, I encountered less state and paramilitary presence than expected and
more guerrilla dominance. The armed groups overlapped less in space and time
than I expected. According to my hypothesized scope conditions, guerrilla
dominance is actually not ideal for expecting civilian resistance to armed
groups, possibly creating a tougher test ex ante for civilian autonomy theory.
Nevertheless, I find there was still sufficient fear of the out-groups by the
civilians and guerrillas alike to create some incentives for autonomy.

The FARC’s 22nd and 42nd Fronts

With historically little state presence, the FARC guerrillas began incursions into
the towns of western Cundinamarca in the early 1990s. The counties began to
feel pressure from the FARC’s 22nd (Simón Bolívar) and (later) 42nd fronts
(Bogotá-Villeta) as they came from outside and slowly built up presence,
beginning with only a few cadres and resources.110 Their reach continued to
grow until the entire western Cundinamarca region became a conflict zone. The
guerrillas used the western mountainous flank of Quipile and Vianí as a key
north-south corridor (Peña 1997).
At first, the fronts claimed to want to “fight for the freedom of the people”
and indoctrinate the masses. Their main enemies were the “paramilitaries and
their patrons.” Early commanders Negro Alfonso and El Ciego had years of
experience and were seen as “clear in the military aspects,” “disciplined,”

110
After government forces attacked the FARC’s “Casa Verde” headquarters in Meta in 1983, the
FARC dispersed and shifted to Cundinamarca.
The Nature and Severity of Armed Conflict 243

“effective,” and bringing “a higher level of culture” to the front. Discipline

and behavior degraded over time, however, as new commanders such as
Negro Antonio were more abusive, extortionist, and rapacious.111 In general,
even though they targeted the rich, their application of violence was more
widespread.
The fronts financed themselves through a variety of sources. In the 22nd
Front’s beginnings, the FARC secretariat transferred resources from narco-
trafficking to help the front grow to forty men. They later came to finance
themselves through roadblocks, kidnapping and taxing wealthy residents based
on studies of economic productivity, and robbing banks (Peña 1997). They
were so successful that they came to be considered one of the FARC’s more
well-off fronts and were frequently able to transfer large sums of money to the
FARC’s central Secretariat for redistribution to other fronts.112

Conflict Dynamics by Towns

In Quipile, the guerrillas had their strongest presence from 1994 through 2002.
At first, the guerrillas tried to make friends with people, arriving with a vision
of helping the campesinos. But like a “courtship,” things turned bad and
Quipile was ultimately hit hard by the conflict.113 Under guerrilla dominance,
there were ominous signs on the roadsides with messages such as “Do Not
Proceed. FARC-EP.” Beginning in 1999, they imposed a 6 p.m. curfew and if
people went outside they would be tied to a pole. Even Colombian beer was
banned, driving people to make chicha (moonshine, usually from corn). There
were always police, but never enough, and few army patrols. The county seat
was “taken” or attacked by the guerrillas four times.114 They would attack the
police and rob the bank (and when unable to open the safe, they would take the
whole thing with them). In 2003, eight police officers were killed by the FARC
in an area above the town.
The guerrillas inserted themselves into residents’ daily lives and made
demands of them. They made themselves indispensable so people would
go to them to solve their problems. Gossip (“chisme”), invented stories, and
denunciations among neighbors were commonplace (“because there’s nothing
else going on, no production, so everyone’s in everyone else’s business”).115
The civilians had to help the guerrillas (and some times the army) by providing

111
Exc#1, Bogotá, 8/2009.
112
“In 1995, according to statistics, Simón Bolívar was the front that collected the most money
from roadblocks” (Peña 1997). It was estimated that the front spent 2 billion pesos per year and
sent 30 million pesos monthly to the central Secretariat.
113
Q#9, Quipile, 3/2009.
114
Q#2, Quipile, 3/2009. The “cabecera municipal,” or the municipio center around which rural
villages are oriented.
115
Q#2, Quipile, 3/2009.
244 Discovering Civilian Autonomy in Cundinamarca

food.116 The guerrillas also demanded protection taxes (vacunas) and would
ask people to bring them goods like prepaid cellular SIM cards. Several resi-
dents reported some forced recruitment of youth (the “sardines”), enticed by
motorbikes, cell phones, money, or liquor – “the good life.”
“Pico y plomo,” or “Obey or a bullet (lead),” was a frequent ultimatum and
one reason why, according to residents, fear was rampant.117 People variously
described the dynamic as “between two fires” (“entre dos fuegos”),118
“between a rock and a hard place” (“La espada y la pared,” or literally, “the
sword and the wall”),119 and “small town, but huge fire” (“pueblo pequeño y
fuego grande”).120 People talked little with each other and followed the “law of
silence.”
When mayors asked for security help from the central government, officials
responded that there were not enough troops to send.121 The army was finally
sent in by President Uribe in 2003, bringing security (by the time I arrived,
people told me the town was very “Uribista”). There were supposedly two
notable combat episodes between the army and guerrillas in the countryside
where thirty-six insurgents were killed. The guerrillas and paramilitaries coin-
cided in the municipio for about six months until the guerrillas withdrew at the
end of 2003.122
The FARC also came to Vianí with relatively good intentions. According to
one resident, they said, “We’re coming to help the small landowners (minifun-
distas)” and were not going to ask for much from the campesinos economic-
ally.123 They would hold ideological meetings at houses about Communism and
other political topics. However, the guerrillas would soon become repressive.
They had a strong presence for only about three years, but for that period it was
“total subjugation” (“sometimiento total”).124 Guerrillas had bases in the north-
west mountainous villages of Manillas and Alto Pueblo because there was never
state presence there and Vianí was a strategic corridor. There were similar
curfews, where people could not go to the villages after 6 p.m. because of security
risks. Holding meetings and driving cars were not allowed and stores would close
early. As in Quipile, the residents were prohibited from drinking Colombian beer
or Coca-Cola, and could only drink Venezuelan beer (Polar).
Residents also reported the “Pico y Plomo” policy, under which villagers felt
threatened and were forced to collaborate with the guerrillas out of fear.125

116
For instance, when they would call farmworkers in for lunch with a bell, the guerrillas would
show up first to eat.
117
Q#9, Quipile, 3/2009. This phrase is a macabre play on the Colombian program of “Pico y
placa,” or “Rush-hour (peak time) and (license) plates,” which refers to a traffic-control
program that restricts the use of cars on different days in urban areas with license plates ending
in certain numbers. It also references “plata o plomo” (silver or lead), a phrase drug cartels
would use to coerce deals through either bribes or bullets.
118 119 120
Q#6, Quipile, 3/2009. Q#2, Quipile, 3/2009. Q#9, Quipile, 3/2009.
121 122 123
Q#11, Quipile, 3/2009. V#3, Quipile, 3/2009. V#14, Vianí, 3/2009.
124 125
V#9, Vianí, 3/2009. V#2, Vianí, 3/2009.
The Nature and Severity of Armed Conflict 245

They had to give guerrillas lunch, were forced to be messengers, had to pay
coffee “taxes,” and faced extortion. In the villages, people had to serve both the
guerrillas and army (as “employees”). Pressure was especially strong in the
villages of Manillas, Cuchira, Cañadas, and Calambata. Forced collaboration
was prevalent higher up the mountain, close to the FARC bases.126
As in Quipile, police and mayors were under severe threat and many had to
leave. The guerrillas bombed the Bancafé bank in 2001 in a robbery attempt in
the town center (again demonstrating that they were good at bombing banks,
but not as good at opening safes; “So stubborn,” one resident decried).127
There was also a separate attack on the town that the army fought off. The
greater tendency, though, was of state abandonment (no state presence, “nada,
nada”). There were only six police officers in the town and, despite some
shellings, only sporadic army presence. Only rarely would some patrols come
from Villeta to the north. The army returned only at the end of 2002 and
brought the peasant soldiers program (Soldados Campesinos).
The community saw the paramilitaries as a solution to their guerrilla prob-
lems at first, but it did not work out so well. The paramilitaries came around
2004–2005 and extorted their own taxes (vacunas), were abusive, and killed
former guerrilla collaborators, saying, “This is our town.”128 There were even
some reports of paramilitaries dressed as police.129
In Bituima, the guerrillas also arrived with some good “communicators”
who amiably talked to people about ideology, brought gifts, gave food to the
poor, and resolved social problems.130 And, like in the other towns, there were
signs of the guerrillas’ dominance. They would be in town on Sunday drinking
beer, and even the buses coming to or from the region had to pay vacuna taxes.
They recruited youths and would give 10-year-old boys bikes and cell phones to
patrol the road between Bituima and La Sierra (in Quipile) as informers.
Residents were similarly fearful of speaking.
There was little military presence and few if any army informants. The army
would patrol only occasionally and would not stay. Although the Pan-
American Highway was developed in the late 1990s, residents did not feel it
brought much more security or increased state presence.131 According to one
woman, it mainly meant more displacement as it became easier for people to go
to Bogotá to look for work and return to visit family. Paramilitaries also came
around 2003, before the army secured the area. They were drawn in part by the
prospect of running an oil racket (a pipeline runs through Albán, the municipio
to the east; Verdadabierta.com 2009). The paramilitaries were less visible, but
bodies started to appear as they killed people with guerrilla ties.

126
V#6, Vianí, 3/2009. The ELN also arrived to appropriate oil when reserves were discovered up
the mountain (but they left when it was determined the oil was not extractable).
127 128 129
V#11, Vianí, 3/2009. V#4, Vianí, 3/2009. V#1, Vianí, 3/2009.
130 131
B#5, Bituima, 3/2009. B#1, Bituima, 3/2009.
246 Discovering Civilian Autonomy in Cundinamarca

In sum, these towns faced many difficult years of repression and conflict.
Some differences are seen in the towns’ conflict dynamics, but they are not
marked ones. The guerrillas may have been more repressive in Quipile, while
Vianí and Bituima were under a slightly looser grip, but there were largely
similar armed groups and pressures throughout the region. There were attacks
on towns and forced collaboration, though only a few episodes of combat.
Even with the guerrillas’ control, some residents still feared getting caught
later by the army or paramilitary forces, which eventually did arrive. State
counterinsurgency efforts finally dislodged the guerrillas in 2003 and the secur-
ity situation gradually improved. The key question now is how did the civilians
survive the many difficult conflict years?

juntas during the conflict and mechanisms for

autonomy
The histories of the municipios have shown differences in their social relations.
Did the varying social landscapes (preexisting to the conflict) position these
communities for varying experiences and assertiveness when the armed conflict
arrived? Did Vianí or Bituima have fewer social divisions that armed actors
could exploit than Quipile? Were they better prepared to organize to deal with
armed actors and the violence and fear they imposed? How and why did armed
actors’ behavior change or produce variation in violence? Because of the
probabilistic nature of civilian autonomy theory, resistance organizations and
events were not expected to occur everywhere in the better-organized munici-
pios.132 Rather, I expected that levels of organization would make it more likely
that some instances of “resistance” would occur in at least some areas. Vianí
and, incrementally more, Bituima were expected to witness more and deeper
collective actions than Quipile. This is indeed what I found.
I first consider the relationship between juntas and armed groups and then
explore what antiviolence actions occurred. I find that, although the juntas
may have created strong social bonds, they were weakened prior to and
during the conflict for reasons both related and unrelated to the conflict. As
a result, they usually did not become the explicit vehicles of choice for
organization or resistance to violence and therefore did not directly impact
how the conflict would unfold. However, as per expectations, in the more
organized municipios, civilians did take individual and collective actions
aimed at autonomy of decision-making and freedom from violence that could
have plausibly affected the behavior of armed groups. Vianí saw some actions
and Bituima in turn saw progressively stronger and more cooperative actions.
Quipile had few responses, though not for lack of good intentions. While

132
Again, if the protest actions were very big or extremely effective, they would likely already have
been discovered and reported in the press.
Juntas during the Conflict and Mechanisms for Autonomy 247

drawing explicit connections between greater civilian activity and levels of

violence is challenging, I find support for several observable implications from
civilian autonomy theory.

The Juntas’ Day-to-Day Role during the Conflict

The juntas have a vibrant past in Vianí and Bituima, but by the time the conflict
reached the western side of Cundinamarca in the early 1990s, their profiles had
already been reduced. The juntas were deactivated due to external reasons prior
to the conflict or became dormant out of fear in many but not all parts of these
municipios. This meant that the FARC could bend the juntas to its will with
little if any resistance sprouting directly from the juntas, though there were
some notable exceptions.
As one might expect from their weak historical legacy, the juntas in Quipile
did not serve as a strong basis for civilian unity to confront the problems
brought by the armed conflict. The juntas did not meet during the conflict years
(“people didn’t like to meet”) but some residents reported that the FARC
guerrillas manipulated the juntas in strongholds such as La Botica and had
meetings with the population in places such as Costa Rica and La Argentina.133
As one man explained, most people in these meetings did not truly support the
guerrillas, but would say “Sí sí sí,” and agree with their propaganda to avoid
consequences. The guerrillas pressured successive mayors to fix roads near their
strongholds and forced campesinos to complete public works projects. They
would make people clear trails and roads with ultimatums like, “Either you do
it, or you do it.”134 There was coerced collaboration for the provision of
information (with sapo informants) because, even though Quipileños did not
want to take sides in the conflict, as the same man quoted above put it, “One
heads to the tree that gives shade – one goes to whichever side gives protection”
(i.e., is stronger; “Al arbol que sombra de”).
In Vianí the juntas were mostly inert, though there were some signs of juntas
attempting to act strategically. According to the Personero, “The juntas always
avoided taking sides and were impartial.”135 But the juntas’ place as authorities
in the villages had receded (“The armed groups – guerrillas – did not view the
juntas as strong”). While there were no collective responses through the juntas,
there are multiple characterizations of relationships between the guerrillas and
the juntas. One person said the guerrillas did not co-opt the juntas but another
said the guerrillas called some meetings with the juntas and frightened
people.136 When the FARC came to the village of Manillas, the junta fell quiet
and did not function, only following the guerrillas’ orders.137 Even with the
suppression of the juntas, there is some evidence of juntas strategically

133 134 135

Q#14, Quipile, 3/2009. Q#9, Quipile, 3/2009. V#13,Vianí, 3/2009.
136 137
V#1,14,Vianí, 3/2009. V#17, Bogotá, 3/2009.
248 Discovering Civilian Autonomy in Cundinamarca

managing information about the security situation (in one instance in the early
2000s, a junta cautioned a priest about where he traveled and gave him security
advice based on their local knowledge).138
In Bituima in the 1990s, the juntas were also coerced by the FARC and they
became quiet (“callados”) out of fear (“zozobra”).139 As in Vianí, in 2002 the
FARC forced residents of Bituima to attend a village meeting and to repair
roads, and imposed fines if people did not participate. The guerrillas also
influenced voting, candidate selection, and political decisions. Yet, in some
parts of the municipio – the same areas where the church aided land
reforms – a subtle social unity persisted. A woman spoke about how her village
maintained solidarity even during this period of war.140 The neighbors had
lived there and gotten along for many years and had a tacit policy to solve their
own intracommunal problems (there was, by contrast, relatively more gossip in
the town center).
The scant evidence for the expansion of the role of juntas to deal with armed
conflict would appear to be a strike against civilian autonomy theory. How-
ever, as shown later, the ties that brought the juntas about and the strengthened
ties that they left appear to have facilitated civilian responses.

Uncovering Collective Actions for Civilian Autonomy

Surprisingly, even though the juntas themselves did not show broad differences
in behavior or enhanced organizational capacity, the differences in social
cohesion (either created or reflected by the juntas) corresponded with the extent
of collective action against violence. These cases support the statistical measure
of juntas reflecting preexisting or persisting social capital as a basis for cooper-
ation. Residents in areas with more collective actions explicitly cited cooperation
and communication as catalysts for their efforts, and vice versa. I explore the
collective actions that occurred and the kinds of social cohesion that made them
possible.
Quipile did not see any forms of organization or collective responses to
violence. But this was not for lack of trying or because people had not thought
about it. Rather, it was due to a lack of coordination, organization, and
historically fractured social relations. As one resident told me, he and some
others in the town center considered taking action and studied experiences of
other communities in the early 2000s to stand up to guerrilla pressures, but
nothing came of it. “Here in Quipile,” he said, “we quietly looked at neutral
models of other municipios such as Samaniego (Nariño) and the Nasa Indians
(Cauca), and other places where people came out with white flags to protest
combat and violence.”141 Because of disunity, the analysis was never translated
into reality in the form of a civilian organization to overcome armed group

138 139 140

V#5,Vianí, 3/2009. B#2, Bituima, 3/2009. B#1, Bituima, 3/2009.
141
Q#2, Quipile, 3/2009.
Juntas during the Conflict and Mechanisms for Autonomy 249

pressure and fear. It came down to residents not having sufficient confidence in
their neighbors to be assured a glimmer of safety and success. There was poor
communication and little common knowledge of each other’s preferences
(for similar arguments, see Petersen 2001 and Chwe 2001).142
During one of the guerrilla attacks on the town center in the early 2000s, a
priest encountered similar troubles. He tried to persuade the guerrillas to treat
the townspeople as neutral, but found little public backing. As a couple of
residents recalled, “Here, in one of the ‘tomas’ (takings of the town), the priest
came out of the church waving a white flag, and nothing happened – no one
came to his side and the guerrillas laughed at him and ignored him.”143 The
town’s judge also tried to no avail to organize people against violence, and
eventually had to displace.144 Interestingly, the man quoted in the previous
paragraph believed that if the juntas had been stronger, they could have better
organized to deal with the armed conflict. He thinks the guerrillas saw civil
society as weak and divided (by religion, politically, and socially), and could
therefore easily influence the town through force and fear. The same was true
for the moderately organized area of La Virgen to the south, where the guer-
rillas were seen as a problem, but there were never discussions of resistance
because people did not feel they were capable of organizing against the guer-
rillas. They had to “comer callados,” or eat silently (not talk out of fear, and
instead sit there and “take it”).145
With Vianí’s heritage of prior social unity and local anti-cattle theft commit-
tees, some form of civil society resistance to violence might have been expected.
Yet, the residents I spoke with did not recall formal resistance acts. There is
some evidence of individual acts to avoid taking sides in the conflict. There were
also some attempts to “humanize” paramilitary behavior, but not much col-
lective action.146
Individuals in Vianí reported trying to avoid taking part in the conflict. As
one man said, “Most people did not take one side or the other, and just focused
on (agricultural) production and working their farms.”147 A family that had
been accustomed to providing assistance to whoever passed by decided on the
advice of a priest that, for their own protection, it was better to tell the army
they could no longer provide food or aid (which the army accepted).148

142
Q#2, 5, Quipile, 3/2009. A possible impediment was that some people perceived that the mayor
was involved with one side in the conflict – the guerrillas – and they therefore realized they
would not be able to attract widespread support. One man noted that some people tried to be
individually neutral and stopped talking to police.
143
Sandoval (2004) lists municipios with these kinds of collective actions that were supported and
succeeded.
144 145
Q#8, Bogotá, 3/2009. Q#15, Quipile, 8/2009.
146
However, ex-combatants recalled several such collective actions in the Vianí-Bituima region.
147
V#6, Vianí, 3/2009.
148
V#3, Vianí, 3/2009. She never had problems with the guerrillas and would have given food to
them too because “that’s just what one did in the countryside.”
250 Discovering Civilian Autonomy in Cundinamarca

Bullet holes in the door of Quipile’s church from a FARC attack in the early 2000s.
Photograph by Oliver Kaplan.

In another account, a former village leader told the guerrillas how they were
treating the people was wrong – stealing from the poor when they were
supposed to be protecting them and threatening suspected army collabor-
ators.149 When the guerrillas came to his father’s house one day and forced
him to provision them with food, he refused and told the guerrillas not to ask
civilians for such things on the grounds of hypocrisy and poverty.150
The civilians in Vianí did not collectively resist violence, and there were no
responses through the juntas. Even in the seemingly well-organized village of
Manillas, due to fear civilians never tried to organize. The guerrillas, rather

149
V#1, Vianí, 3/2009. The guerrillas threatened this man and he had to leave (although later he
and his family were able to resolve the issue with the guerrillas so he could return and live
without problems).
150
He said, “If we’re poor campesinos and you’re supposedly fighting for us, why are you
extorting from us?” To this the guerrillas replied, “We’re trying to reduce corruption, provide
order; we’re saviors of the motherland (salvadores de la patria).” He replied, “If that’s true,
what good are you doing?” This argument did not succeed in changing guerrilla behavior and
he became a “persona non grata,” a “stone in their shoe,” and had to leave under threat. He
brazenly told them, “If you’re going to shoot me, do it in the plaza so all can see.”
Juntas during the Conflict and Mechanisms for Autonomy 251

than the people, intervened to solve problems there. But there were some quiet
civilian efforts to humanize the war. Some leaders in the municipio did try to
dialogue with armed actors.151 The same outspoken leader would explain to
the army that the campesinos were mostly “innocent,” and that people only
aided the guerrillas because they were forced to. He asked the army to “leave us
alone so nobody [neither the army nor the FARC] screws with us.” He had
also considered starting a family-based armed group to fight off the guerrillas
(a “defensa”), but without a foundation of local organization (or armaments)
he was discouraged. Short of this, the large families in Hatillo united to reach
agreements about the problem of gossip (chisme) reaching armed groups. In the
later years of the conflict, some local leaders reportedly spoke with army
officials and the paramilitaries to communicate concerns about the killing of
civilians. According to the perception of the village leader, this worked to some
degree and the paramilitaries decreased their presence.152
In Bituima, even without many signs of enduring social organizations,
there were both individual and collective protection efforts. These actions
arose in the subregions with the highest degrees of historical organization and
cohesion.153 As one resident who traveled to many villages for his work
noted, many people feared that the army or paramilitaries would arrive later
and seek revenge, so it was better to stay independent.154 In one act of
individual resistance, a woman refused guerrilla compensation for damages
to avoid appearing to be a collaborator. The guerrillas came to her village and
took her family’s motorbike and its permits to evade scrutiny from the public
forces.155 Not long afterwards, a truck came to the house with a new motor-
bike as repayment. Instead of accepting the offer, the woman said, “I don’t
want it and I don’t want you sleeping in my barn. If the army finds out that
you gave me this moto, I’m in trouble.”
As expected, the collective actions in Bituima were more broadly based and
more forceful than in the other towns. In one key episode in 2002, a non-
collaborating and particularly united village that was tired of being afraid and
endangered by pressure to collaborate with the FARC stood up against the
guerrillas’ dominance and demands.156 Although there were contrasting

151
V#1, Vianí, 3/2009.
152
Some of these discussions occurred through official channels such as the mayor. Today, there
are more formal “consejos de seguridad” or meetings between the juntas and the police to
discuss security, but fear remains.
153
B#6, Bituima, 8/2009.
154
The man said that other villages had to help the guerrillas because they were desperately poor,
but once they did and received benefits from the guerrillas, they were in too deep and had to
keep assisting.
155
B#7, Bituima, 8/2009.
156
I was able to speak with several people who were present at the meeting (B#5,7 Bituima, 8/
2009) as well as several other people who were not but heard about the episode through friends
and relatives (B#1, Bogotá, 3/2009; B#6, Bituima, 8/2009, B#8, Bogotá, 8/2009). Residents
252 Discovering Civilian Autonomy in Cundinamarca

accounts of how events unfolded, the protest began when the guerrillas called
the residents of the village to a meeting at the polideportivo (multisport court).
The guerrillas demanded that they provision them with food and water,
saying, “He who doesn’t collaborate with us will be killed.” In response, a
woman who had been forced by the guerrillas to lend them her phone stood
up and said, “I’m not giving anything to you because my sons were put in
danger and almost killed by the army and could come under threat from the
paramilitaries for transporting goods for you. You don’t have any business
here. You’re not from this land. This has been our land all our lives.”
At this point, the guerrilla commander threatened to kill these insubordi-
nates right then and there. It is not entirely clear what happened next, though it
triggered a reaction from the community. By one account, all the villagers
present at the meeting apparently stood up in solidarity and said, “If you’re
going to kill her, you’ll have to kill us all.” One man, perhaps aided by some
aguardiente-fueled157 liquid courage, said, “Look, Commander, you are not
God; you are not the owner of life who decides who lives or who will die. You
don’t even know if you’ll outlive us [with the army gunning for you].”158 The
people were fearful but the close relationships of the villagers meant the man
had the tacit support of his neighbors. In the face of this, the commander and
other guerrillas backed off and left the meeting.
The retelling of such a hidden narrative is powerful, stunning, and consistent
with what might be expected if civilian organization and legacies of cohesion
are important. All of this occurred before the villagers knew the army or
paramilitaries would come in force. The villagers attested to changes after their
response to the guerrillas. Psychologically and organizationally, the act was
crucial because the people saw that they were capable of facing and dealing
with a threatening situation together. The residents “lost their fear” and, as one
person said, “We felt that we could defend ourselves.” Through unity, they

believed cooperation was in fact likely in this village for a number of reasons. First, its houses
were close together and all shared strong friendships and would, for example, exchange food.
Second, small groups of neighbors discussed the conflict (producing common knowledge and
informal coordination instead of succumbing to the law of silence). Third, another man singled
this village out as being more standoffish (reacio) and with the good “values” (good sense) to
stay independent. Another said residents were more educated, aware, and unified (“echada pa’
adelante”), with many stores and the ability to cooperate to run bazaars. In contrast, one
neighboring village had “too many fights,” while another village was too large for coordin-
ation. However, some also recalled that a second nearby and cohesive village resisted in a
similar fashion.
157
Aguardiente is a sugarcane-based alcohol produced in Colombia.
158
An alternative account told that this man was the only one to speak up, because others were too
fearful, but that he did so with tacit support of his tight-knit community. Even though this was a
brave man, one woman thought he would have kept quiet in a weaker village. The process of
collective memory and how people interpreted this event is perhaps equally significant as how it
actually may have occurred. His words were a signal to his neighbors about what was possible
and confirmed the feelings they all had.
Juntas during the Conflict and Mechanisms for Autonomy 253

A polideportivo, or “multi-sport court,” in Quipile, 2009. FARC guerrillas would use

these spaces to convene meetings with villagers. In Bituima, some villages voiced
resistance at such a meeting. Photograph by Oliver Kaplan.

broke the “law of silence” that pervades civil wars and increased their commu-
nication.159 One man said the residents became even more uncooperative with
the guerrillas after this episode.160
This story is certainly inspirational, but for a couple of reasons, the concrete,
measurable, lasting effects of this social mobilization on guerrilla behavior are
less obvious. First, about two weeks later, the guerrilla commander from the
meeting was killed. A man in another village, who the guerrillas threatened and
ordered to pay a tax, was summoned to meet with them but had informed the
DAS (the since-dissolved Colombian intelligence bureau) and the army arrived
and killed the commander.161 Shortly thereafter, there were reports that the
paramilitaries had arrived and filled the power vacuum. Second, it is difficult to
imagine the counterfactual of what degree of violence would have otherwise
visited the community.

159 160 161

B#1, Bituima, 8/2009. B#6, Bituima, 8/2009. B#5, Bituima, 8/2009.
254 Discovering Civilian Autonomy in Cundinamarca

Civilian Actions According to the Armed Groups

The civilian accounts of resistance to the guerrillas are confirmed by the
guerrillas themselves. Interviews with former members of the FARC fronts
from the region corroborate that civilian resistance generally occurred
where it was predicted to and even make references to roles for the junta
councils (or at least what the combatants believed to be juntas). Most of these
ex-combatants were active members from around 1996 through 2004. Key
insights come from a former member of the FARC who was originally from the
region and operated there, and had a deep knowledge of Quipile, Vianí, and
Bituima.162 As an escolta, or escort guard, he would accompany mid-level
commanders to meetings with communities (and was also privy to some com-
manders’ discussions). Additional supporting statements are drawn from other
ex-combatants that spent less time in the case-study municipios or spent time in
neighbor towns (meaning their observations were not always geographically
precise), though dissenting or more muted opinions about civilian activity are
also encountered.
The escort guard recounts that his front saw more resistance in Vianí and
Bituima than in Quipile starting around 1998 or 1999 as they attempted to
consolidate power. He could recall instances of at least four villages relatively
close to the town seats (cabeceras) in Vianí and Bituima that resisted guerrilla
influence.163 The episodes of resistance proceeded in a fairly similar manner to
the civilians’ descriptions of them. When entering a village, the guerrillas would
hold meetings with residents, which he would commonly attend alongside his
commander. The cadre would introduce themselves and announce their aims of
resolving neighborly disputes and conducting social cleansing by saying,
“We’re from Front 42 of the FARC, and we’re here to take charge of this
village. Tell us who are the thieves and people taking advantage of the
community.”
The resisting villages united peacefully and collectively against the guerrillas
and would not let them enter. The guerrillas would usually give notice when
planning meetings with communities so that residents would attend. The guard
believed that junta (or community) leaders, armed with this foreknowledge of
the guerrillas’ arrival, were able to meet with and organize their communities in
anticipation of a pending confrontation. At one meeting, the civilians
responded to the guerrillas’ demands by advocating together for what sounds
a lot like autonomy, saying, “We don’t want you guerrillas here. Kill us all if
you must. We don’t want the army either. We want to manage our affairs
amicably.” Even with the strong guerrilla influence in the region, this statement

162
Exc#2, Quipile, 8/2009. He operated in the area from around 1998–2000 and so was present at
a different time than the civilians’ report of resistance reported earlier. Thereafter, he had less
experience in the region.
163
He did not recall specific village names.
Juntas during the Conflict and Mechanisms for Autonomy 255

underlines civilians’ fear of collaborating and becoming stigmatized, and their

collective solutions.
Some forms of resistance in Quipile were reported as well, but they were
of a different nature. Around 2000–2002, a handful of some of the guer-
rillas’ strongest collaborating villages reportedly turned against them. The
residents were upset about abuses and tired of the guerrillas’ demands,
which included, for instance, the requirement to work two days for the
guerrillas for every two days they worked for themselves. This was “not
sustainable.” However, the pleas were not as broadly based as in the other
towns and the guerrilla response was supposedly an even stronger hand to
crush opposition there.
The guard believed the differences in resistance between the Vianí–Bituima
communities and the Quipile communities stemmed more from civilians’ social
differences than differing levels of pressure the guerrillas could bring to bear or
military control. Both prior to and during the years of conflict, the residents of
Vianí and Bituima were more unified and people had more contact with each
other. In these villages it was not “keep to yourself” – the same description
given by Quipile’s priest – and the juntas worked.164 He used the word “nada,”
or “nothing,” to describe how the juntas in Quipile were almost nonexistent
and did very little to unite villages.165 In contrast to the residents in Vianí and
Bituima, many Quipile civilians simply collaborated with the guerrillas out
of fear.

Armed Group Responses to Civilian Autonomy

What was the impact of these movements? Interviews provide a look inside the
guerrilla group and show that guerrilla leaders were surprised by the resistance
put up by the civilians. They expected to be able to easily take the villages of
Vianí and Bituima, the way they took the villages of Quipile. As the guard said,
“The guerrillas wanted to take all of Vianí and Bituima but couldn’t.” The
guerrillas could not order civilians like they wanted to. This caused a debate
within the front about what to do with civilians (corroborated by other soldiers
based on pushback from villages in Viotá).166 They supposedly sent more
cadres to obligate people to collaborate, but in contrast to his statements on
responses to resistance in Quipile, they did not want to force residents too hard
to collaborate because it was “bad.”
Because of the civilian resistance, the guerrillas were reported to have more
frequently desisted from killing. How and why might these collective actions
have changed their behavior? Combatants provided a complex mix of

164
“Cada uno por su lado,” or literally, “each to his own side.” He held this view of Quipile even
before he was recruited by the guerrillas.
165
Only now are Quipile’s juntas being renewed, with a junta functioning today near Botica.
166
Exc#3, Bogotá, 8/2009.
256 Discovering Civilian Autonomy in Cundinamarca

explanations and reasoning.167 Taken together, they spotlight the group’s

balancing of their need for civilian support with costs of attacking and using
heavily violent tactics – a balance that can shift based on civilians’ actions.
According to the strategic logic provided by the ex-combatants’ collage of
statements, the use of violence to cement control by coercion and fear loses
some of its effectiveness when an organized community puts up resistance. The
group goes from facing a decision of whether to kill individual resisters or
community leaders to the possibility of targeting many more residents to quash
resistance (since eliminating leaders in an organized village may not guarantee
submission).
Many of the ex-combatants referred to the centrality of the backing and
support of the entire community and being more prone to selective violence
against lone individuals or small groups who resisted. To explain, a former
FARC mid-level commander and infiltrator used the poetic saying, “Una sola
golondrina no hace verano,” or, “One swallow does not make a summer”
(unbeknownst to me, an adage from Aristotle and Shakespeare).168 This is to
say that a single individual – a single swallow – will not signify or bring much
change but a group, a flock of swallows, carries more weight. The juntas were
seen as key institutions in this regard, “providing vigilance of communities and
keeping local order.” Hearkening back to the previous “alone we’re screwed”
comment by the Vianí resident, he said, “While the juntas aren’t united, they
won’t achieve anything.” Across the region, he noted some junta leaders left or
were killed, but leaders of resisting communities were mainly safe because of
civilian support.
Not only do civilians’ chances at sustaining their communities improve if
they remain united, but they are also more taken into account by armed
groups. The intent of civilian cohesion then is to tilt the groups away from
using fear to break a community and toward respecting existing social struc-
tures.169 In these acts of collective resistance, civilians throw down the gaunt-
let to armed groups, gauging that they will not take the leap toward mass
violence. In doing so, they can activate four main concerns that affect armed
group calculus.
First, the guerrillas did not want to kill everyone and “end an entire vil-
lage.”170 Ex-combatants cited some moral considerations, with distinctions
between few killings (perhaps for some conceived greater good) versus massive

167
Evidence about these reasons are provided by ex-guerrillas that operated in the case-study
towns, as well as other members of the same fronts who operated in nearby towns but had
knowledge of the fronts’ decision-making.
168
Exc#4, Bogotá, 8/2009.
169
As the mid-commander said, “When entering a community, there are two possible strategies a
commander can choose. He can break it and use fear, or work with it. If there’s a strong
structure, a commander will ask how they can work with it?”
170
Exc#4, Bogotá, 8/2009.
Juntas during the Conflict and Mechanisms for Autonomy 257

killings (perhaps evoking cognitive dissonance of being for the people and yet
committing atrocities).171
Second, resistance by organized communities may activate sensitivities about
the guerrillas’ reputation for using violence. The guerrillas did not want the
reputation of being killers, and the distinct level of violence required to deal
with an organized community would be more greatly publicized than smaller
acts.172 Organized communities may themselves also be more effective at
publicizing atrocities and amplifying reputation costs (or perhaps appealing
to government authorities). If the guerrillas kill, people could say the guerrillas
are not the “freedom fighters” they claim to be but really are terrorists. Keeping
up appearances is worth something and they wanted to avoid falling into a trap
of rhetorical inconsistencies.173 Front commanders were even reported to have
debated how violence might worsen the group’s image.174 While these groups
of course participated in the disreputable practice of kidnapping and the narco-
economy, these revenue-generating activities may be easier to justify as the
means to a revolutionary end than killing.
The ex-combatants referred to concerns about their reputations with three
potential audiences. First, excessive violence could create political problems for
the guerrilla group in the eyes of the broader population.175 They worried
about greater difficulties in securing the support of other villages (if those
communities saw the guerrilla’s true behavior, they would be more reluctant
to help them). Second, sensitivity to their international image can also enter into
the equation. If they are killing everywhere it may become a political liability
for leaders from other countries to support them.176 Third, they feared it might
eventually bring the army to the zone. By definition, the army of a weak state
with rough terrain like Colombia has difficulty projecting its power to many
parts of the national territory at once. But that does not mean the army cannot
be mobilized to address trouble spots when necessary or politically demanded.

171
The mid-commander continued, “If one sees a strong social structure, it can change the way a
group thinks. To kill one or two [resisting] people is one thing, but to kill a whole [resisting]
community is too far” (emphasis added).
172
This may be because mass violence may become a focal point – people in the community and
neighboring areas talk about it more, the media pick up the story, etc. Killing organized resisters
would be even worse given that the civilians approached them nonviolently and for dialogue.
173
While they might also value a reputation for toughness and ruthlessness in the face of resistance,
having to resort to large-scale violence could also be interpreted as an act of desperation.
174
Exc#2, Quipile, 8/2009.
175
As a guerrilla in Peña (1997) explained, for this reason, at least in the front’s early days, they
supposedly preferred roadblocks to demanding protection taxes (vacunas), “Today we don’t
demand vacunas because it scares and nauseates the masses.” But roadblocks also have political
costs, “The retentions are politically, economically, and psychologically costly for the move-
ment, as much for the victims as for the soldiers that carry it out . . . It provokes a hostile attitude
in the community toward the movement.”
176
The FARC was documented to have relied upon support from the Venezuelan government
under Hugo Chávez (BBC 2011).
258 Discovering Civilian Autonomy in Cundinamarca

In sum, exemplifying a degree of risk aversion, the guerrillas believed they were
more likely to blow their cover by targeting a village that is united against them.
A third factor is the relatively high cost of using mass violence to obtain the
marginal benefit of the allegiance of new villages. Although the guerrillas
generally wanted to gain the support of more and more communities (they
“needed villages for power”), they already had a base and some power when, in
their expansion, they ran up against resisting villages. The guerrillas did not
attack the resisting populations in part because it was “not worth it” – it was
not worth the risk.177
Fourth and lastly, the FARC fronts would debate abuses in their internal
assemblies in response to civilian protests, at times exhibiting splits over what
course of action to take. A plausible interpretation is that civilian pushback
activated particular concerns that may have empowered more dovish com-
manders over their hard-line or abusive counterparts within the group.178
Exemplifying these tensions, one commander might declare, “I won’t work
with this other commander because he is undisciplined.”179 Some commanders
eventually faced sanctions by the group for their “errors,” and abusive prac-
tices were tempered. These changes would probably have been less likely
without civilian pushback.

an intervening explanation: clientelism

If Vianí and Bituima had historically high levels of juntas, why did the juntas
themselves not (survive to) have a clear role in the armed conflict in these
municipios? Field research indicates the opportunity for the junta councils to
act was short-circuited by political clientelism and later by the conflict itself.
Political clientelism is defined as the manipulation of local constituents and
organizations through (short-term) political favors or payoffs to gain their
support (see Martz 1997 on Colombia; Magaloni et al. forthcoming). This type
of political arrangement can affect whether people look (horizontally) to their
communities and neighbors to solve problems or (vertically) to mayors and
local politicians – whether they wait for solutions from outside the village or
neighborhood. In this study, the damaging effect of clientelism on community
networks and organization would bias against observing any collective actions
for protection from armed groups.

177
Exc#5, Bogotá, 8/2009.
178
Commanders might be dovish because of their personal values, background, or education; their
closeness to the population; or simply because they are rational calculators who tend to believe
that moderating the use of violence is a superior strategy for gaining control or extracting more
resources from the population. In Kaplan (2013b), I explore this dynamic further and develop a
theory of “nudging armed groups,” in which civilian collective actions are able to exploit intra-
armed group fissures to reform their behavior better than international humanitarian actors and
the laws of war.
179
Exc#6, Bogotá, 8/2009.
An Intervening Explanation: Clientelism 259

Clientelism creates resource dependencies on local politicians and their funds

and can sap the intrinsic drive of individuals and local organizations to take
actions by their own initiative. It becomes a system, an equilibrium, when no
politician can compete for political power without these relationships. The
perniciousness of clientelism for responding to the armed conflict in Colombia
was cited in a United Nations study (UNDP 2003). In my analysis, in all three
towns and even in areas that had strong, functioning juntas, residents com-
plained that the juntas were weakened by two main factors related to cliente-
lism: (1) instituting the direct election of mayors and (2) the dependence of
juntas on external financial support and the subsequent elimination of that
support.
In 1988, Colombia underwent a nationwide institutional change when
municipal mayors ceased to be appointed by department governors and became
popularly elected. Before the popular election of mayors, the juntas were the
main forum where local democratic politics and decision-making were chan-
neled for tasks such as public planning and the resolution of conflicts. Once
mayors began running for election, they used funds and favors to co-opt
support through the local juntas. This had the effect of juntas and junta leaders
frequently acquiescing to local politicians for short-term benefits (vote buying).
It increased the expectation that public goods provision was primarily the
concern of the mayor and his administration, and not the juntas.180 Juntas that
were already weak lost even more relevance in citizens’ daily lives.
Second, from 1965 up through 1993, the national government made some
development funds available to the local juntas through grants known as
“auxilios parlamentarios,” or congressional assistance. This was aid pro-
grammed into the national budget for local juntas by representatives or sen-
ators for their constituents (see Borrero 1989, Leal Buitrago 1990). These funds
were often matched with labor or funds from the local junta and functioned as
a stimulus for activities and public works. In some cases, the main purpose
of forming a junta was to take advantage of these funds (though once a junta
is formed even for this reason, it may take on additional tasks of local
governance).
It was a blow to the juntas when these funds were eliminated in 1993. People
partly participated in junta activities as a means of getting funds and carrying
out projects. The disappearance of national government funding also com-
pounded the problem of vote buying by local politicians and the newly electable
mayors, as the juntas became more autonomous but at the same time more
desperate. Together, these changes were devastating for many though not all
local juntas in the country. Indeed, up to the early 1990s, the junta movement
had been signaling the benefits of increasing political autonomy (Borrero 1989).
This was apparently sidelined.

180
AC#1, Bogotá, 2/2009.
260 Discovering Civilian Autonomy in Cundinamarca

The effect of clientelism on juntas is problematic for social responses to armed

conflict for several reasons. First, if local organizations are in it for “the buck” –
the short-term gain – and come to rely on local politicians, their actions lose
legitimacy among members of the community or village. If people do not feel
that their contributions are necessary to the process or feel excluded and that
their interests are not being represented in the use of funds, there is little point to
contributing to local “democratic” processes, and participation and social
bonds between community members can atrophy. Crosscutting social contacts
and exchanges of information can also be damaged if juntas or other organiza-
tions lose importance. Under these circumstances, communities can be chal-
lenged to coordinate responses to intrusions by outside forces because they have
few “standard operating procedures” and little common knowledge.
Second, mayors in a local clientelist political system may only be responsive
to a narrow group of political supporters and less broadly accountable to
community members. Mayors may then more easily cut deals with and sell
out to armed groups to save their own political skins – deals that do not benefit
the community or town as a whole but may instead privilege certain sectors or
interests, such as landowners.181 They may either fall in line with armed groups
or be forced to pay off such groups to receive permission to govern or run for
office. These deals may undercut community efforts to avoid violence and may
block other social sectors from acting.
Third, clientelism may be further problematic if it blocks effective economic
development. If clientelism derails the provision of economic opportunities,
more people may come to participate in the conflict due to poverty. In this
way, clientelism can both tip the balance of individuals’ choices toward joining
armed groups and become an impediment to a community’s effort to act
collectively to resist armed group recruitment.
Consistent with trends in many parts of Colombia, reports from interviews
show the stranglehold of clientelism was present in all three of the case muni-
cipios of western Cundinamarca. In Quipile, juntas were reportedly late to form
because residents traditionally looked to mayors to solve problems and there
existed a long-standing clique, or oligarchy, of leaders. Residents pointed to the
direct election of mayors as leading to more clientelism.182 The local and
national governments made juntas depend on them, since the juntas were used
in political campaigns. Some juntas were even formed by local political bosses
(gamonales) for personal ends.183 Some mayors look to exclude junta leaders
who are their political enemies from the policy process by, for example, inviting
them to meetings, but only at the last minute so they cannot attend.184
Although a few of Quipile’s recent mayors were clean, many left office amid
corruption scandals. In sum, clientelism is “endemic.”185

181 182
E.g., critiques by Quipile#8, Quipile, 8/2009. Quipile#4, Quipile, 8/2009.
183 184
Quipile#10, Quipile, 3/2009. Quipile#8, Quipile, 8/2009.
185
Q#15, Quipile, 8/2009.
Levels of Violence (vs. Reported Violence) 261

In Vianí, residents spoke of how it has been common for mayors and
councilmen to “buy” votes by, for instance, sending gifts to villages before
elections.186 This has moved voters to vote for “favors,” not necessarily for
good managers. A former councilman believed that when mayors became
elected instead of appointed, it aggravated local political conflicts.187 One
person also suggested that the existence of a few powerful families has fostered
a closed and monopolized political system.188
In Bituima, despite equal social relations and evidence of organization in
some villages, the influence of gamonales in local politics persisted. One person
directly attributed the weakness of juntas to the paternalist clientelism that
began with the 1988 election of mayors, saying that they had functioned up
until about that time.189 Before, there was more participation in and commit-
ment to the juntas as they were substitutes for unpopular appointed mayors.
People were also more self-sufficient in taking on public works projects, but
today they look to mayors to organize such projects. One person alleged some
mayors have even kept petty cash in their desk drawers to buy people favors
and buy votes.190 Another person aptly characterized Bituima as “a town that
never wanted to progress,” which could be taken to mean that even with local
grassroots initiatives, the town has remained politically backwards.
This analysis points to a new theory to explain how local civilian organiza-
tions interact with and affect armed conflict. Clientelism can pull the rug out
from under some local organizations, leaving them to crumble in the face of
conflict, while others are able to persist. This could explain why strong social
organizations and peace communities arise in some areas of Colombia, but are
not found countrywide.
Even with the burden of clientelism, the armed groups faced multiple con-
siderations about how to deal with the actions of different civilian communities,
with moments of restraint despite their atrocities. The next section assesses
whether civilian social processes map to patterns of violence.

levels of violence (vs. reported violence)

What types of violence occurred over time across the different municipios, and
for what reasons? Like other conflict zones of Colombia, these towns suffered
their share of killings and oppression. Field observation and interviews show
there is significant measurement error in levels of violence based on press and
police reports arising, for instance, out of civilians’ fear of reporting victimiza-
tion to authorities. Comparatively, the qualitative reports indicate greater
overall levels of violence as well as greater disparities across the towns that
are consistent with what would be expected based on differences in social

186 187 188

V#1, Vianí, 3/2009. V#11, Vianí, 3/2009. V#12, Vianí, 3/2009.
189 190
B#11, Bituima, 3/2009. B#1, Bituima, 3/2009.
262 Discovering Civilian Autonomy in Cundinamarca

organization. This triangulation with multiple sources of information adds

further confidence to the theory and the statistical model.
The Colombian government data and CINEP’s press reporting suggest that
most of the violence was committed by the guerrillas, although there are some
reports of acts by the armed forces and, later, by right-wing paramilitary
groups. The levels of violence in these municipios as measured by the police’s
homicide rate data are well-predicted (close to the regression line): Quipile has
a relatively higher average rate compared with Vianí and Bituima (though the
counts of violent events in both data sources are inordinately low). Illustrating a
common challenge in conflict zones, these sources only appear to capture the
more publicized events.
Fieldwork revealed many reporting issues. First, in some of the towns, the
police were entirely absent, or people were too afraid to risk reporting events to
them. Furthermore, press reporting was sporadic and there were few or no
formal human rights or peace organizations monitoring the situations in these
towns. Even among the civilians themselves, there was apparently little sharing
of information. Some tales of events that occurred in the countryside have only
recently come to light, while many others remain hidden and untold.
I sought to increase confidence in measurement by considering several quali-
tative and observational sources of information. These are summarized in
Table 8.2 and graphed in Figure 8.2. First, I sought to obtain counts of the
number of crosses along country roads in these towns. In instances where
people were too afraid to report cases to the authorities or hold a proper burial
in a cemetery, people were buried where they lay, their graves marked with
crosses that often simply read “NN,” or “no name.” Second, I interviewed the
personeros, or human rights ombudsmen, of each town to get estimates of the
number of claims by victims – widows and orphaned children – who began to
come forward to collect government reparation monetary benefits.191 Third,
I obtained an additional estimate by speaking when I could with submunicipal
“police inspectors” who were in charge of exhuming and registering the
dead.192 In general, less violence occurred in the more densely populated
county seats compared with the outlying villages and corregimientos.
According to the theory of the balance of armed group control, Quipile
should have expected less violence than the other towns because, if anything, it
was the most strongly controlled by guerrillas. Even with the matching on
armed group actions from Chapter 6, Quipile experienced slightly greater and

191
Q#3, Quipile, 3/2009; V#13, Vianí, 3/2009; B#9, Bituima, 8/2009. Decree 1290 of 2008
allowed victims seeking reparation “by the administrative path” an additional two years from
April 2008 to make their claims. www.fiscalia.gov.co/justiciapaz/Documentos/Decreto_1290_
Abril_22_2008.pdf. It is possible that these data could exhibit an opposite bias – of over-
reporting – if people have the incentives and ability to make false reports.
192
These are unarmed, nonuniformed local residents who do office work for the police and receive
reports and complaints from the population.
Levels of Violence (vs. Reported Violence) 263

table 8.2 Qualitative indicators of violence

Quipile Vianí Bituima

Population (1993) 10,033 4,107 2,932
Average police homicide rate (1990–2005) 42.8 42.1 44.4
Total CINEP homicides (1990–2005) 5 (5 events) 5 (1 event) 0
“No name” crosses along roads Many Some (3 est.) Some (2 est.)
Personero estimates
Number of victim claims 200 30 25
Average annual rate 1,993 (125) 730 (45.7) 852 (53.3)
(1990–2005)
(based on CNRR benefit claims)
Police inspector
Number of killings 150–400 N/A N/A
Average annual rate 100–200

figure 8.2 Estimates of violence from quantitative and qualitative sources.

longer guerrilla domination. But Quipile also suffered more violence than the
other towns, or the violence at least made a greater impression on the residents
with whom I spoke.
The years from 1993 to 2001 marked a period of high violence in Quipile
that took several forms. First, violence was used for day-to-day social control
and social cleansing. There were no robberies because guerrillas maintained
order and killed thieves. As one man said, when the guerrillas came, people
264 Discovering Civilian Autonomy in Cundinamarca

denounced each other for their own ends, sometimes resulting in killings. The
guerrillas told fathers who were drinking too much and not caring for their
children to shape up or they would be killed (though “probably many wives
were glad”).193 Second, there are some examples of guerrillas more intention-
ally targeting suspected enemy collaborators. The guerrillas threatened some
nuns who were seen walking with the police. In a letter to the priest, they said,
“We’ll make them whores of the countryside.”194 There were cases of guerrillas
“carrying out justice” against individuals for refusing to collaborate, which
they would justify by saying, “We killed this puta or perro (whore or bastard)
for being a sapo (literally frog, meaning an informant).” In contrast, civilians
were not greatly targeted during attacks on the town centers. The guerrillas
would sometimes notify the population and give them time to leave or hide and
mostly targeted the police or the banks (even the police closed themselves into
the police station). Still, three civilians were killed in cross fire.195
Quipile also suffered from the stigmatization of being guerrilla collabor-
ators. In 2003, the army and DAS (intelligence bureau) conducted a major
roundup of suspected collaborators that were fingered by a single former
guerrilla.196 As alluded to in the song at the beginning of the chapter, they
came with tanks and trucks early in the morning. They brought people – young,
old, male, female – into the plaza, still in their pajamas, and took them to prison
in Bogotá to be interrogated. In news reports they were referred to as (sus-
pected) guerrillas. Most were let free after a few days or weeks, but some were
held for up to six months. Afterwards, there was a lot of uncertainty and fear
that the army lists might fall into the hands of the paramilitaries.
When the paramilitaries came they reportedly killed some of the people who
the army had rounded up and then released. Around five civilians were killed in
2004 “by who knows who,” and some of the bodies were dumped on the
highway.197 “The paramilitaries had good intelligence. They surrounded Qui-
pile and La Sierra at night with hit lists.” In 2008, threats began anew in the
form of phone calls. About a month prior to my first visit to the town in 2009, a
man who was named in an anonymous pamphlet was killed, perhaps because
he was a “delinquent,” had previously helped the guerrillas, or had a daughter
who joined their ranks. Another man recalls at least three or four killings of
suspected enemy collaborators that could also be called social cleansing.
According to additional numerical estimates from the town Personero
obtained during fieldwork, there were about 200 women in the villages that
were widowed as a result of the conflict.198 He explained that victims were
fearful of denouncing acts and were only coming forward now for state
reparation money. He also reported three or four events he referred to as
massacres, one of which involved nine victims (he was “stunned” when he

193 194 195

Q#2, Quipile, 3/2009. Q#5, Quipile, 3/2009. Q#9, Quipile, 3/2009.
196 197 198
Q#2,3, Quipile, 3/2009. Q#9, Quipile, 3/2009. Q#3, Quipile, 3/2009.
Levels of Violence (vs. Reported Violence) 265

took the job and found out about all of this). A police inspector from one of
Quipile’s corregimientos comprising about one-quarter of the villages in the
municipio said they had to recover approximately 150 bodies during the
conflict years, and this was not even the most violent region of the munici-
pio.199 Some townspeople mentioned the figure of at least 300 deaths in the
villages attributed to illegal groups and a recent anonymous pamphlet alluded
to a similar figure.200 If these estimates are correct, it would translate into an
average annual homicide rate of 100–200 per 100,000 residents for the period
1990–2005, or about two to four times the rates reported by the police.
In Vianí, the FARC sought to rule by fear as well. According to a priest, the
FARC would bring families to meet and say, “We have a list of people and we
are not afraid to kill them.”201 As early as 1994, the guerrillas threatened a
“talkative” priest and he had to leave.202 The guerrillas would kill army
collaborators and people had to lie about whether the army (or in the army’s
case, the guerrillas) passed by. Acts of violence certainly did occur and some
interviewees said many people were killed as “accomplices.” There were also
reports of torture, disappearances, massacres, and high-violence pockets in the
villages of Alto Rosadas and Cañadas. One respondent said he was kidnapped
by the guerrillas but then let go.203 But, according to the reparation forms filed
with the Personero by victims and their family members as of 2009, there were
approximately thirty victims from the conflict years – fewer than in Quipile.
According to the Personero and other residents, paramilitary forces from
Puerto Boyacá to the north arrived around 2004 and were present for about
two years. They reportedly killed an estimated five victims, especially pressur-
ing Cañadas, where people were stigmatized as being former guerrillas or
guerrilla supporters (“tildados”).204 Still, all told, people spoke of less violence
in Vianí relative to Quipile during the conflict years.
In Bituima, there were also reports of violence, but it was less widespread.
The Personero estimated that about twenty-five people had come forward to
claim victim benefits, mostly for acts that occurred between 1998 and 2003 at
the hands of guerrillas (characterized as a fairly “comprehensive” reporting of
crimes). The Personero also reported there were not many crosses along roads.
The paramilitaries were around for two years, from 2003 to 2005, and mainly
committed violence against suspected guerrilla collaborators in the town center
and some villages, killing around five people. They targeted a store owner and
killed a teacher who had supposedly taught students about the guerrillas and
weapons.205 Residents of one of the resisting villages said they experienced only

199
Q#16, Quipile, 8/2009. The police inspector reported these deaths to the fiscalía (prosecutor)
but not necessarily to the police because it was too dangerous given fear of reprisals from the
FARC, so most of these counts are likely not reflected in the police’s statistics. The count does
not include victims who were simply “disappeared.”
200 201 202
Q#9, Quipile, 3/2009. V#1, Vianí, 3/2009. V#10, Vianí, 3/2009.
203 204 205
V#1, Vianí, 3/2009. V#10, Vianí, 3/2009. B#1, Bogotá, 3/2009.
266 Discovering Civilian Autonomy in Cundinamarca

one killing of a former guerrilla collaborator at the hands of the

paramilitaries.206
In sum, while depictions of violence can sometimes be vague, Quipile seems
to have suffered more than the other towns. This is supported by several
sources of information as well as across several motivations for and types of
violence.

conclusions
This chapter’s controlled comparison was designed to identify cross-town
differences in civilian cooperation and collective actions and their later effects
on armed group behavior. It is true that similar autonomy actions may have
indeed occurred in many towns across Colombia. But with many possible
confounding factors in comparisons of other sets of cases, their marginal effects
would be difficult to disentangle. The structured process tracing here uses
interviews to help understand the stories and motives behind different deci-
sions. The result is a rendering of the social history of three towns over sixty
years. When the conflict did reach these towns in the 1990s, it was not arriving
in a barren social landscape, but rather one characterized by varied associations
and corrugated legacies of social capital.
The statistical matching of cases proved to be accurate and helped highlight
several significant features of the social landscape during the conflict years.
There were differences in “culture,” including patterns of social interaction and
cooperation (“convivencia”/coexistence), and the junta organizations that set
the towns apart. In Bituima and Vianí, the junta organizations and cooperation
were more prevalent and more consistently recalled. While guerrilla dominance
induced acquiescence or collaboration in a number of villages across the towns,
experiences with organization proved to be a key ingredient for dealing with
threats. Lack of organization inhibited community advocacy in Quipile
whereas the presence of organizations enabled it in Bituima. The histories
further alleviate concerns about spuriousness stemming from social
organizations only thriving in peaceful conditions. The record shows the towns
had similar social and conflict-related historical starting points: all suffered
similar degrees of intercommunal and political strife during La Violencia. In
sum, the evidence supports social capital as an explanation for civilians’
responses to the conflict.
The relationship between social organization and conflict is also more com-
plicated, however, as many of the juntas were weakened by armed group
pressure. Guerrilla presence created an environment of fear, coercion, violence,
and silence. It is not that fear and danger were absent in the resisting villages of
Bituima- quite the contrary. Residents did fear the guerrillas as well as the

206
B#6, Bituima, 8/2009.
Conclusions 267

return of the “out-group” – the army or paramilitaries – and therefore were

reluctant to support the guerrillas, or in some cases, the patrolling army. Yet at
the same time, preexisting social cooperation also empowered civilians as
agents to retain their autonomy. Growing dangers created cause for action
and the residents were able to overcome their fear through a history of solidar-
ity and experiences with cooperation. The outcome was a greater “humaniza-
tion” of the war and less violence suffered.
The histories show several civilian mechanisms and strategies for autonomy
and protection, even though these areas were at times characterized by single
group dominance. First, given that the guerrillas strove to assert authority and
performed social “cleansing,” civilians’ simple unity and fewer local conflicts
meant there were fewer inroads for armed groups to use violence. Second,
civilians used some explicit organizational strategies toward the armed groups.
In Bituima, some villages collectively opposed the guerrillas in the hope of
increasing costs to using violence. This induced some reduction in guerrilla
abuses of the population. Third, the collective actions and shared preferences
for autonomy from the guerrillas meant fewer civilian alliances with them.
Though the link is less clear, this in turn meant less violence and pressure later
by the army and paramilitaries. Reports from the guerrillas themselves help
explain why their decision-making was “sensitive” to communities that held
out in unison. Even for ruthless and opportunistic armed groups, legitimacy can
serve as an important lever of influence. Violence in these cases, then, is
accounted for by pressure from the armed groups as well as variation in social
characteristics across communities.
In comparison with the ATCC, the strategies encountered in Cundinamarca
did not approach the same degrees of formality and complexity. Though
impressive in their own right, they were less premeditated, less enduring, and
less self-conscious. For example, the civilians’ harmony and conflict resolution
were less overtly managed and communities did not develop an explicit vouch-
ing system like the ATCC’s. Because the Cundinamarca towns’ social lives and
actions are hidden narratives, with fewer formal records and less social
memory, they are difficult to detect and carefully measure. Causal links
between civilian social-capital mechanisms and violence are therefore not as
clear as in the ATCC case.
Because of poor flows of press coverage or the subtle nature of the actions
identified in these towns, these episodes were largely unknown histories, even
though they were occurring just three hours from the capital. If these cases had
been much more organized, they would already have been “discovered.” They
would be broadly known and lumped with the forty-plus formal organizations
the media has covered. The distinctions between towns in this chapter’s analysis
therefore suggest that subtle and likely effective forms of civilian organization
and resistance can occur widely. For instance, in the ATCC case there were
paramilitaries and guerrillas who were amenable to civilian organization. Here,
an additional FARC front operated similarly. Beyond these local findings,
268 Discovering Civilian Autonomy in Cundinamarca

the chapter holds several more general implications for inference and the value
of these research methods.

The Advantages of a Guided Search for Civilian Autonomy

Was encountering the instructive differences in civilian autonomy across these
towns of Cundinamarca pure luck? Perhaps, but social science tools and
models were instrumental in making an educated guess about where to look.
These methodological advances are as important as the substantive findings,
since data can have weaknesses, model specifications can be sensitive, and
theories were made to be refined. Combining methods and triangulating mul-
tiple sources hold several advantages for understanding behavior in conflict
settings.
First, combining statistical case selection with qualitative case studies facili-
tates making broader inferences beyond solely productive within-pair case
comparisons. Matching neighbor cases within the same region helps rule out
wider, nonzone specific explanations that might be difficult to measure and
control for with statistics. For instance, levels of equality, variants of armed
groups, or proximity to the frontier could be omitted (or poorly measured)
variables that could produce spurious relationships between juntas and vio-
lence. Because these factors vary little within the localized region of western
Cundinamarca, they are unlikely to account for the variation among these
comparison-set towns.
Second, the qualitative explorations revealed some exogenous “twists of
fate” that helped explain variation in the formation of juntas within this region
that were not easily detectable with large-n methods. These unexpected
events (or “qualitative instrumental variables,” as they might be called)
explain organizational development in Vianí and Bituima and offer hope that
interventions in social-capital-challenged localities can change the “paths” they
are on. Juntas (or social organizations) could be considered “treatments” that
policymakers can apply to promote resilience. While organizations like the
juntas may indicate preexisting cohesion and more easily thrive where cohesion
exists, they can also be created in adverse circumstances and go on to have
benevolent effects.
Third, beyond testing theories and mechanisms, the qualitative analysis is
helpful for generating theory. Residents pointed to the explanation of
clientelism to account for weakened social structures and why the junta coun-
cils did not more strongly embrace explicit mechanisms for human rights
protection. This could be because they mainly existed to manage funds from
the government to build (clientelist) projects. Although ATCC residents also
cited the influence of clientelism, their juntas fulfilled other social functions,
such as providing order and raising funds, and therefore generally remained
stronger. This was likely out of necessity from being situated in an isolated,
neglected colonist zone. For the same reason, they were also likely less
Conclusions 269

integrated into politicians’ clientelist networks. Juntas may therefore be more

likely to thrive where states are weak and their survival does not primarily
depend on municipal politicians.
Combating clientelism was also a motivation for the Congress to enact the
national-level funding cuts in 1993. However, perversely, these cuts weakened
the junta councils just prior to the surge in the armed conflict. The funding had
made the juntas stronger and more active by giving them purpose and allowing
them to carry out community development projects. Yet these funds were cut
because they were also used for political favors.
Comparing how these cuts unfolded across my case regions underscores how
they led to the greater political autonomy of juntas, but also, in many cases, to
their dissolution.207 Because the conflict arrived early in the ATCC region –
before the funding cuts could weaken the junta councils – those communities
were advantaged in organizing for autonomy. In contrast, by the time the wave
of conflict came to the Cundinamarca municipios, national-level (external)
policy changes had already weakened local organizations and they struggled
to mount autonomy strategies to deal with armed groups. Of course, civilian
autonomy movements may also be less common in the early phases of conflicts
for ideational reasons, as communities often need time to develop repertoires of
strategies. Still, my mixed-methods approach reveals that seeking autonomy is
even harder, though not impossible, when clientelism is also present and
threatens organizational capacity.

207
AC#1,2 Bogotá, 3/2009.
9

Civilian Autonomy around the World

“The people are very unified so the Taliban failed. We are dead set against the
army, too.”
– Elder, Buner, Pakistan, 2008 (Parlez and Shah 2008b)
“Early on in this war, I met with the main religious leaders in the community: the
bishop and the mother superior of the main convent. We decided that even if
the mountains around us were exploding with fighting, we would not go to war. . . .
We decided adamantly that Maloula would not be destroyed. . . . The situation
here will not deteriorate; it’s the opposite. People support each other.”
– Sunni imam, Maloula, Syria, 2012 (Di Giovanni 2012)

Civilian efforts for autonomy reach far beyond Colombia. In this chapter,
I explore four additional out-of-sample cases of civilian autonomy from
around the globe. Although these additional cases from Colombia, the Philip-
pines, Afghanistan, and Syria were not systematically selected for comparison,
they exhibit geographical diversity, involve conflicts of diverse types and
origins, and have ample documentation of civilian strategies and armed actor
responses. They highlight unique aspects of civilian autonomy, including show-
casing the diversity and prevalence of strategies and how they function in
sectarian conflicts, Muslim-majority countries, and with the presence of inter-
national forces. They are also “tough” cases for autonomy given the intensities
of the conflicts. Their displays of civilian autonomy contrast strikingly with
conventional interpretations that emphasize the victimization of civilians in
these conflicts.
The inclusion of these additional cases and countries shows that the main
argument of this book is not culturally bounded. In fact, it shows that
local civilian organization frequently supersedes cultural differences as an
explanation for violence. Put differently, cohesion and organization explain
differences in outcomes of violence within particular cultural groups. The cases

270
Civilian Autonomy in FARClandia 271

also further showcase the diversity of innovation by civilians in war as well as

the limits of their efforts when circumstances turn dire.
I first assess the civilian autonomy theory I developed in Colombia based on
additional firsthand field research in Colombia on “tough,” out-of-sample
community cases in the highly contested demilitarized zone of the Macarena
region. I then also explore the protective autonomy strategies that communities
have pursued in both Muslim and Christian areas of the Philippines. Using
interviews with community advocates and archival research, I conduct a quan-
titative analysis of the strategies implemented by communities that declared
themselves “Zones of Peace.” In the religious-extremist conflicts in Afghanistan
and Pakistan, I document examples of community actions to avoid the NATO/
government–Taliban crossfire using secondary sources and news reports.
Lastly, I analyze secondary sources on communities caught between govern-
ment and rebel forces in Syria and show that civilian autonomy has been
attempted and proven helpful even in that extremely brutal conflict.
These cases provide helpful insights about autonomy but are not controlled
studies since the influence of additional predictors of violence, such as the
balance of control among armed actors, are not easily measured. Without
adequate data on the complicated selection issues and precisely measured
outcomes (as is available in Colombia), these cases are only a first exploration
of protection and resilience in these countries. Still, these cases show that
civilians have effectively organized for autonomy in diverse and challenging
locations across different continents.

civilian autonomy in farclandia

In addition to the Colombia case studies from Cundinamarca and Santander, I
also conducted fieldwork in the former Colombian demilitarized zone, or Zona
de Despeje (or distensión) and, even there, I found examples of autonomy.
Often informally referred to as “FARClandia,” this region in the department of
Meta, about 100 miles southeast of the capital of Bogotá, is an out-of-sample
and extremely tough case for civilian autonomy. The despeje territory of five
municipios, covering an area about the size of Switzerland, was ceded to the
FARC by President Andrés Pastrana in 1998 for the Caguán round of peace
negotiations.1 The top-down negotiated cease-fire held for a short time, but it
ended up being violated in many areas of the country by the guerrillas before
the peace talks eventually collapsed in 2002. For this reason, and because the
region became the pilot for the government’s “territorial consolidation” pro-
gram in the mid-2000s, FARClandia is perhaps the best-known conflict zone in
Colombia. However, because the negotiations were the focus of attention – to

1
The region includes the municipios of La Uribe, Mesetas, La Macarena, and Vista Hermosa in the
department of Meta and San Vicente del Caguán in the department of Caquetá.
272 Civilian Autonomy around the World

the neglect of conditions in the zone – FARClandia is also one of the conflict
zones that is most poorly understood.
The despeje is a tough case for civilian autonomy because of the FARC’s
entrenched territorial control, as well as being the setting of peace talks while
also suffering periods of shifts in control and intense conflict. The long, narrow
mountain range known as the “Lying Indian” (el Indio Acostado) that shoots
up out of the eastern plains runs through the heart of the despeje municipalities
of Meta. This mountain range was a strategic base for the FARC, linking its
coca production activities in the plains with close access to the capital and
Cundinamarca.2 Upon assuming control of this territory, FARC rule was
dominant as evidenced by the guerrillas’ regulation of most aspects of daily life
(Campbell 2000). Nevertheless, declassified U.S. State Department cables,
which provide rare insights into the conflict conditions that civilians faced
within the despeje territory, noted that “most area inhabitants are not FARC
supporters, and feel abandoned by the [Government]” (Kamman 1999,
National Security Archive).
The level of confrontation between armed actors declined for a time from
1998 to 2000 because of the peace process, although there were still reports of
various rights violations committed by the guerrillas and the paramilitaries
(Amnesty International 2002, Vicepresidencia 2003, 2003b). Then, beginning
in 2000, conflict intensified. According to a U.S. Embassy cable, while the
FARC had twelve fronts in or surrounding the despeje, paramilitaries were
reported to have expanded their presence and had three fronts in the area
(Patterson 2000, 2002). As in other areas of Colombia, armed actors fought
each other through the civilian population, using intimidation to exert their
influence and control both during the extended period of talks and afterwards
when FARC control began to erode (Vicepresidencia 2003, 2003b).
The FARC implemented forced civic action days (Kamman 1999), and there
were forced displacements, kidnappings, extortion, targeted killings, and resig-
nations by mayors (Patterson 2000). The paramilitaries, for their part, entered
the zone and threatened residents not to cooperate with the FARC (Kamman
1999).3 Short of full combat, paramilitary action was also focused on the
population, as they “appear[ed] to be looking for targeted opportunities against
the FARC, but [were] apparently not planning any large-scale operations”
(Patterson 2002). These events led one U.S. Embassy analyst to pessimistically
conclude, “The right to free political expression has effectively been limited
through fear and intimidation” (Kamman 1999). Yet even in this dangerous
environment some of the vibrant village junta councils helped insulate commu-
nities from violence and adapt to changing conflict conditions.

2
Alias Mono Jojoy, the FARC’s second-in-command and field marshal, was killed during airstrikes
against these bases in 2010.
3
The author of this U.S. Embassy cable ominously warned he “would not be surprised to see
paramilitary massacres of civilians suspected of pro-FARC sympathies in coming weeks.”
Civilian Autonomy in FARClandia 273

The Macarena mountain range, also known as the “Lying Indian,” rising up out
of the eastern plains in Meta department, within what was the demilitarized zone
(Zona de Despeje), or FARClandia, 2011.
Photograph by Oliver Kaplan.

I visited several municipalities in the Macarena region in January 2011 and

conducted interviews with a variety of residents, including several junta council
leaders (and had previously interviewed ex-combatants in 2009 who had
operated in Meta).4 In this frontier region, the colonist residents had spent
years building their small communities and strengthening social relations,
and they had a strong tradition of junta councils (see also Molano 1987). All
the juntas had legal charters (personerías), stayed informed about events in the
zone, and were recognized by residents as the authority in the villages. The
juntas also had traditions of reconciliation in the form of conciliation commit-
tees (comités de conciliadores), which consisted of five people who would
impartially resolve conflicts in the community (adjudicating compensation
when cattle would trample a neighbor’s crops, etc.). As with the junta comités
in other parts of the country, in the absence of state institutions and in condi-
tions of armed conflict, the Macarena comités adapted to mediate social

4
M#1,2, Meta, 1/2011. Specific municipios are not reported for anonymity reasons.
274 Civilian Autonomy around the World

disputes even though this was beyond their official purview of monitoring junta
governance and procedural issues.
Even during the most intense years of armed conflict surrounding the des-
peje, the juntas kept functioning. Villages maintained their community conflict
resolution processes and, where conciliation committees were strong, it was
preferable to resolve issues in the community rather than take them to the
guerrillas. The strength of a junta would depend on the quality of its leader –
whether he or she was a strong and vocal advocate for the community – and
whether the junta was recently formed or had longer experience with commu-
nity decision-making. In the context of the conflict, armed groups would say
“you’re either with us or against us,” and residents looked to the juntas for
protection and to dialogue with armed groups on the community’s behalf. The
juntas advocated for communities and were “salvation” that allowed people
who preferred neutrality to say, “My side is the [community] junta.”5 Armed
groups respected strong juntas and would not get involved in their community
meetings, although the guerrillas would at times send representatives. By con-
trast, some juntas were historically weaker or more “timid” and had greater
difficulty negotiating with tough commanders.
The strategy and philosophy of junta leaders was to “walk the line, neither
here nor there” and not take sides in the larger armed conflict. Junta leaders
participated in dialogues with armed groups and negotiated with them to avoid
threats of displacement so residents could stay in their homes. During the
despeje, if people “behaved well,” they “wouldn’t have any problems.” One
of the leaders described his relationship with the FARC guerrillas as follows,
“If they greet me, I greet them. If they speak to me, I speak to them as the leader
of the community, but have no other interaction.”
The guerrilla commander was like the “mayor” during the despeje years,
and while residents had to do what the guerrillas said, the guerrillas also
listened to them. Notably, some guerrilla commanders were more tolerant
and more willing than others to accept junta autonomy and allow communities
to run their own affairs. For instance, two commanders from the FARC’s
27th Front had better “political skills” and were “more aware” of the popula-
tion whereas other, more militant commanders (who were more concerned
with military strategy) were less concerned about the population. The juntas’
independence from the armed actors was on display when the paramilitaries
and guerrillas wanted the juntas to protest against government coca eradi-
cation programs and the juntas instead negotiated on their own for crop
substitution programs.

5
Consistent with these accounts, WOLA (2011) reports that in the early 1990s, even prior to the
despeje, civic leaders used similar dialogues to form “peace pacts” for autonomy with guerrilla
and paramilitary commanders. This tradition continued through the despeje period as the junta
leaders later negotiated to allow the juntas and the town council to continue to function.
Civilian Autonomy in FARClandia 275

After the peace talks collapsed and the demilitarized zone came to an
end, the military and then paramilitaries came to the region. It was “tough”
(“berraco”) when there were multiple armed groups, and the entry of the
paramilitaries was “terrible” for the population. Residents were stigmatized
as guerrillas, and the guerrillas also came to see the people as paramilitaries.
They were “screwed” (“jodido”) and it was “hard, not easy” (“duro, no
suave”). But one man observed that the juntas were especially active in protect-
ing people during this shift in control and sought to find remedies so that the
“population was not pressured by multiple armed groups.” As a leader, he
wanted the population to be left aside from the conflict and remain “very
neutral.” Although there are no reports of open protests against armed actors,
there is evidence of community management and negotiation with combatants.
A FARC ex-combatant that operated in Meta confirmed the general activism
of some juntas there.6 The FARC wanted the juntas to solve local problems (e.
g., domestic and property boundary disputes) and would hold “orientations”
and provide a document to communities with rules and norms to live by. If
there were murders committed among the population, the guerrillas would join
community meetings to impose order and authority. In general, however, the
guerrillas reportedly respected junta decisions. For example, if civilians told
guerrillas not to stay at their farms, they would respect the decision and leave.
Some juntas even insisted on managing their own affairs, as did two villages in
Meta that did not want army or guerrilla visits and instead preferred to
maintain peace on their own.7 As in Cundinamarca and La India, when the
FARC arrived at one of the villages at night, the entire community came out to
protest, with community leaders politely saying, “So sorry about this, but we
want peace” (“Me da pena con ustedes, pero queremos paz”).
There is unfortunately little reporting and few histories on community rela-
tions from the despeje region, making it difficult to more broadly corroborate
these accounts. One of the few additional investigations of this region is
Vásquez’s (2013) interviews of residents of the neighboring municipality of
San Vicente del Caguán in the department of Caquetá, which indicates less
civilian autonomy in the interactions between the FARC and communities. The
FARC exerted more control over the population in San Vicente, largely co-opt-
ing the juntas, and the coca economy expanded its reach. Although community
organizations were strong and “each village had a junta,”8 communities had
relatively little independence and there were not reports of resisting FARC
influence. In some cases, the FARC stimulated the creation of junta councils
to organize communities and then serve as interlocutors with state institutions
(Cubides et al. 1989). As one man said, “The development here is a result of the
guerrillas.”9 Speaking to the extent of FARC control, a woman recalled that,

6 7
Exc#8, Bogotá, 8/2009. Precise locations were not recalled in the interview.
8 9
Vásquez interview #6; translated from the Spanish. Vásquez interview #3.
276 Civilian Autonomy around the World

“When those people [FARC] ruled here, it was they who governed. One didn’t
see robberies or deaths because they maintained order. Almost everyone
depended on them because they generated employment. One simply did the
work one had to do.”10 However, there is a contrasting account from a U.S.
Embassy cable of an additional instance of activism against the FARC, where a
local priest criticized the FARC’s management of the zone and publicly asked
them, “Who authorized you to govern San Vicente? . . . Who authorized you to
occupy every corner of the town, . . . control economic activities, recruit
minors?” (Kamman 1999). This portrayal suggests that even if civilian auton-
omy was not found in every part of the despeje, limited forms of autonomy
were possible in at least some areas.
In sum, the history from the despeje, albeit incomplete, reveals a diversity of
experiences. Although civilians lived under siege in Meta during and after the
years of the despeje and the Caguán peace talks, communities were still able to
come together to mitigate the harmful consequences of instability in surprising
ways. This counternarrative begs a rethinking of conventional historiography
of the region.

civilian autonomy and the peace zones

in the philippines
Similar to the formally organized peace communities in Colombia, approxi-
mately 100 communities in the Philippines organized themselves as nonviolent
“Zones of Peace” (ZOPs). Beginning with the founding of the Zone of Peace,
Freedom, and Neutrality (ZOPFAN) in Naga City in 1988, these communities
have sought to keep out and avoid entanglements with contending government,
rebel, and paramilitary forces. These diverse Philippine ZOPs hold insights for
civilian autonomy because many have made public declarations of their collect-
ive strategies. This systematic information makes it easier to see the distribution
of civilian protection strategies used relative to the more scattered and subtle
actions by the Colombian peace communities. In this section, I analyze how
ZOP communities have dealt with the pressures of armed conflict and identify
the possible effects these actions had to protect civilians. The analysis is based
on secondary sources and interviews I conducted during fieldwork in Manila
and Davao in May 2012.
The Philippines has experienced decades of conflict since the early 1970s.
Government forces (Armed Forces of the Philippines, or AFP) and paramilitary
units such as the Citizen Armed Force Geographical Units (CAFGU) have
battled insurgent groups such as the New People’s Army (NPA), the armed
wing of the Communist People’s Party (CPP), the Moro National Liberation
Front (MNLF), and the Moro Islamic Liberation Front (MILF; the CPP was

10
Vásquez interview #2.
Civilian Autonomy and the Peace Zones of the Philippines 277

founded in 1968 and the MNLF in 1969).11 The communists’ goal has been to
topple the state while the Muslim insurgencies have pursued separatism with
the goal of creating their own state on the southern island of Mindanao. The
Muslim areas gained a degree of political autonomy through the negotiated
establishment of the Autonomous Region in Muslim Mindanao (ARMM)
in 1989 (McKenna 1998). Conflict intensified in the late 1980s through the
mid-1990s after peace talks broke down in 1987 and President Corazon
Aquino launched a “total war” policy against the NPA. Fighting spiked again
in Mindanao in the early 2000s, with a series of stalled peace talks, aborted
cease-fires, and aggressive government counterinsurgency campaigns against
the MILF under President Joseph Estrada’s “all-out war” policy. To date, the
armed conflicts in the Philippines have claimed over 160,000 lives (Project
Ploughshares 2013, Reuters 2014) and nearly four million people have been
displaced since 2000 (IDMC 2014).
The Philippine state governs a territory of more than 7,000 islands and faces
geographic and bureaucratic challenges in extending its reach to rural commu-
nities. Because of this, as one NGO staff member said, “The security situation is
complex – there are not just armed actors, but also contending families, clans,
bandits, paramilitaries, and corruption.”12 The Philippines’ social landscape is
diverse, with cultures superimposed on top of each other through waves of
colonization by Muslims, the Spanish, and the United States (as well as the
preexisting Indigenous Peoples). This landscape produced tense social relations
in some areas and fed the armed conflicts – it did not predispose the Philippines
to be an incubator of civil society peace movements. One pernicious type of
social conflict that is especially prevalent on Mindanao is rido, or clan conflicts,
as widely documented by Torres (2007). In such feuds, extended families –
entire clans – carry out retributive acts against enemies, leading to cycles of
violence that can last for decades. As a staff member of the Manila-based NGO
Community Organizers Multiversity, which works with ZOPs, noted, peer and
social pressure and the concept of “maratabat,” or hurt pride, provide incen-
tives for even peaceful people to join in ridos.13 Politicians and armed groups
have preyed upon rido conflicts to incite further violence. This escalation of
rido violence was a main reason for founding some ZOPs, such as those of
Maladeg and Carmen (Catholic Relief Services 2003).
Santos (2005) identifies two main waves of ZOPs: one in the early 1990s and
another in the early 2000s. Some peace zones were still being formed as late as
2010 (Cabreza 2010). According to qualitative research by Philippine (Lee
2000, Santos 2005) and international scholars (Hancock and Mitchell 2007),
although the ZOPs can go by many names – sanctuaries of peace, spaces for

11
The Islamist militant group Abu Sayyaf Group (ASG), which has been linked to Al Qaeda, has
operated on the small, isolated southern islands of Sulu, Jolo, Basilan, and Tawi-Tawi and is
generally more removed from the ZOP regions.
12 13
Ma#1, Manila, 5/2012. Ma#2, Manila, 5/2012.
278 Civilian Autonomy around the World

peace – they often form in response to a violent triggering event, suggesting that
at least some ZOPs did not solely arise in historically peaceful areas. They also
share strong organizational foundations, since they are launched from close
community ties and often with the support of local social entrepreneurs such
as Catholic priests, Muslim imams, or NGO workers and activists that help
coordinate collective actions under threat of violence.14 For instance, CO
Multiversity reports that in Maguindanao in 2003, a ZOP was formed by
community leaders through an agreement in which residents committed to
avoid getting involved in disputes and conflict-related activities.15 Some ZOPs
are of a single ethnic or religious group (Muslim, Christian, Indigenous Peoples)
while others are constituted by several groups to promote intersectarian
harmony. Indeed, the use of declarations themselves (as a kind of written
organizational constitution) was spread by networks of NGOs and religious
organizations, becoming a norm and practice to be emulated (i.e., institutional
isomorphism described by Meyer and Rowan 1977).
For a more systematic understanding of ZOPs, I coded existing qualitative
community, rebel group, and military documents collected by Santos (2005). I
produced a mapping of the twenty most prominent Philippine ZOPs at the level
of the barangay (village) displayed in Figure 9.1 (a finer level of geographical
detail than exists in Colombia).16 ZOPs are found in diverse areas of the
country, including in Luzon, north of Manila, as well as clusters on the
southern island of Mindanao and other scattered locations. Some encompass
entire municipalities while others consist of just one or several barangays.
What do ZOPs do? Based on the declarations, I compiled a detailed inven-
tory of nonviolent protection strategies that includes their degrees of specificity,
complexity, and contentiousness.17 These diverse types of tactics can be aggre-
gated into four general categories of clauses that correspond to the autonomy
mechanisms outlined in Chapter 2: actions to strengthen community institu-
tional arrangements, regulation of social vice (community norms), actions to
protect civilians, and more contentious rules of conduct for combatants. Insti-
tutional policies include declaring the size of ZOPs, policies on migration to
ZOPs, policies to manage fear and build trust, and incentives or penalties
to promote cooperation. Strategies to manage social vices that could lead to

14
D#1, Davao, 5/2012.
15
Ma#2, Manila, 5/2012. Also, “Peace Covenant establishing the Zone of Peace at Sitio Caga-
waran, Barangay Tugaig, Barira, Maguindanao,” June 8, 2003, in Santos (2005).
16
Some of the constituent barangays of these peace zones originally formed independently and
later agglomerated into larger zones.
17
These declared institutional practices are different from particular instances of protective actions,
protests, and conflict events, which may occur within the declared ZOPs. These events would
likely be seen if and when institutional procedures alone fail to deter violence. According to an
NGO worker, declarations are generally a good indicator of ZOP activity, but some commu-
nities that had not declared ZOPs behaved like ZOPs, while the declarations of some ZOPs were
empty statements, with little implementation. D#2, Davao, 5/2012.
Civilian Autonomy and the Peace Zones of the Philippines 279

figure 9.1 The geography of Peace Zone villages in the Philippines

outbreaks of disputes and violence include bans on gambling, alcohol, drugs,

and firearms. Actions to protect civilians include resolving local disputes, anti-
recruitment strategies (especially for youth), procedures to protect community
leaders or threatened individuals, and norms of not passing information to
armed groups. Rules of conduct for combatants include barring armed groups
from territory, barring their use of public facilities, appeals against the harass-
ment of civilians, cease-fires, and providing safety for wounded soldiers.
Figure 9.2 aggregates the number of different strategies declared by the
twenty ZOPs (and their constituent barangays) to provide information on the
prevalence and distributions of particular strategies. The analysis indicates that
the majority of the ZOPs mention multiple armed actors as targets in their
statements – a clear indicator of a preference for autonomy. However, there is
also variation across the ZOPs in the specificity and sophistication of their
strategies. Around three-quarters declared at least a basic package of general
and definitional claims, including declarations about the contours of the ZOP,
appeals against harassment of civilians, bans on firearms, and calls for
ceasefires. More detailed polices (more sophisticated declarations) are found
in about one-half of ZOPs, such as the Baras and Sagada zones, including
dispute resolution procedures; anti-recruitment and migration policies; bans on
additional vices; explicit policies to deal with trust, fear, and cooperation; and
restrictions on armed actors. Indeed, ten of the declarations contentiously call
for barring armed groups from community territories.
280

figure 9.2 Frequencies of Peace Zone strategies in the Philippines.

Source: Author’s tabulations, based on Santos (2005).
Civilian Autonomy and the Peace Zones of the Philippines 281

Similar to the peace communities in Colombia and civilian autonomy actions

in other countries, social regulation (e.g., barring weapons) and norms of
nonviolence to limit participation in the conflict figure prominently in ZOP
declarations. Several reference the concept of bayanihan, or communal work
and reciprocity (Catholic Relief Services 2003), evoking the convite or minga
communal work groups found in Colombia. A Philippine scholar who has
accompanied several peace zones emphasized the importance of moral persua-
sion to promote peace.18 Such persuasion is leveraged in at least two ways.
First, within communities, women and mothers frequently compel good
behavior, as the Philippines is “very matriarchal,” and dialogue and education
efforts are used to promote and sustain norms of nonviolence and social
control. For example, an early ZOP declaration calls for “Discussion groups
that will explain the situation, dissuade people from actions such as joining
paramilitary civilian defense groups [CAFGU], cooperating in intelligence oper-
ations of armed groups, [or] joining fundamentalist groups.”19
Second, in relations with rebel and government forces, communities have
invoked rhetorical traps similar to those in Bituima, Cundinamarca, or the
Colombian Indigenous communities. It is characteristic of ZOPs to attempt to
influence armed actors by making such claims as, “You [armed actors] say
you’re our protectors but why do we suffer when you’re here? You fight for
our liberation, but we get hurt.”20 As indicated in a ZOP planning docu-
ment, there is also moralistic naming and shaming for transgressions by armed
actors, “Sanctions will be imposed by the armed parties on members of their
own forces that violate their agreements with a Peace Zone . . . Moreover, the
full force of the people’s judgment and condemnation will publicly be cast
on parties.”21
There is less information available on the implementation of ZOP declara-
tions or observed collective actions to prevent violence. However, there are
some positive examples, including from the Sindaw Ko Kalilintad (Light of
Peace) cluster of Peace Zones. These zones were formed in Mindanao beginning
in 2002 to deal with fighting between MILF and government forces in areas
where some members of the population were MILF sympathizers. Local peace
zone leaders and CO Multiversity built upon the existing structure of commu-
nity organizations, including women’s and youth groups, farmer cooperatives,
and barangay councils, and trained a group of elders in mediation to resolve
ridos. At one point, the community also warned the AFP and MILF about

18
Ma#3, Manila, 5/2012.
19
“Implementing Guidelines of our PCPR Agenda for Peace (Some characteristics of our Zones of
Peace)” by the Promotion of Church Peoples’ Rights, September 18, 1989, in Santos (2005).
20
Ma#2, Manila, 5/2012.
21
“Working Paper for the Establishment of a Policy on Peace Zones: A Proposal to the Parties to
the Armed Conflict (n.d.)” by the Peace Zone Technical Committee, Multi-sectoral Peace
Advocates, in Santos (2005).
282 Civilian Autonomy around the World

A flyer promoting the ban on firearms in the Naga City Zone of Peace in the
Philippines. Solimon Santos personal archive. Photograph by Oliver Kaplan.
Civilian Autonomy and the Peace Zones of the Philippines 283

fighting near the community, saying, “Don’t do your shooting here because it
will affect our water pipes.”22 The combatants apparently acquiesced to the
community’s demands.
What have been the perspectives of armed actors toward the ZOPs? State-
ments by government forces and insurgent groups indicate that they have at
times been willing to abide by the nonviolent demands of the ZOPs and armed
actors have been signatories to various ZOP declarations. However, the armed
actors exhibit ambivalence toward ZOPs similar to that of the armed actors in
Colombia toward the peace communities. ZOPs have initially been viewed as
threatening, and armed actors have been wary of yielding perceived battlefield
advantages to the enemy. Both government and rebel forces have therefore
attempted strategies of co-optation. Guarantees of reciprocity and fairness,
which have been aided by the ZOP’s commitment to autonomy, have been
helpful for stabilizing the agreements and inducing the armed actors to respect
the population’s wishes. This has led to diverse and evolving perspectives on the
ZOPs, including toward the notion that peace should be community-based and
toward greater restraint by the military.
For example, different parts of the Philippine government have been skep-
tical of peace zones at different moments, with the military starting out as
especially skeptical for fear of ceding military advantages to insurgents. In
reference to Naga City, one of the most prominent ZOPs, a military com-
mander believed that “the outlawed Communists will enjoy the golden oppor-
tunity to propagate their prescribed doctrine,”23 while another commander
worried that it would place the National Democratic Front (NDF; an umbrella
organization for the NPA, the Communist Party of the Philippines, and other
revolutionary organizations) on “equal footing” and provide them with a
sanctuary.24 The military has also expressed concerns about the absence of
reciprocity and ceding terrain, as one commander noted that “all of the
demands and responsibilities are being made only on the side of the Military,
while no similar demand is being made on the rebels . . . the situation is unjust
and . . . in favor of the rebels and against the Military.”25 In this vein, General
Rodolfo Obaniana starkly proclaimed, “I do not believe in peace zone[s]”
(Mallari Jr. 2007).

22
Ma#2, Manila, 5/2012.
23
“Comments, Reaction and Position of the Camarines Sur Constabulary/Integrated National
Police Command to the Proposal of the Hearts of Peace (HOPE) to Make Naga City a
Permanent Zone of Peace, Freedom and Neutrality (ZOPFAN)” by Lt. Col. Rufo R. Pulido (n.
d. but March 1989), in Santos (2005).
24
Col. Marino L. Filart, PC Regional Commander, letter to Naga City Mayor Jesse M. Robredo,
February 28, 1989, in Santos (2005).
25
“Comments, Reaction and Position of the Camarines Sur Constabulary/Integrated National
Police Command to the Proposal of the Hearts of Peace (HOPE) to Make Naga City a
Permanent Zone of Peace, Freedom and Neutrality (ZOPFAN)” by Lt. Col. Rufo R. Pulido (n.
d. but March 1989), in Santos (2005).
284 Civilian Autonomy around the World

In interviews, military officers reported mixed results of ZOPs. One mid-

level officer thought the Sagada ZOP does not work and is likely “an NPA
sanctuary, a front, a ploy” and believed that, at the height of insurgency, social
cohesion and formal governance can break down while insurgent shadow
governments gain influence.26 However, the same officer also noted that strong
barangay captains, ulamas (Muslim religious leaders), and tribal chiefs can help
deal with the NPA and that the NPA is not able to dismantle all informal civil
society institutions. Another military officer observed that there is variation in
barangay captains and participation in barangay governance from one locale
to another.27 More exuberantly, in a Philippine military trade publication,
General Ariel Bernardo notes how different AFP institutions can “produce a
force multiplier effect to the whole process of establishing a PZ” and called the
peace-oriented approach “preferable [to] continued military confrontation”
(Bernardo 2010).
More political parts of government that have broader views of the conflicts
(and peace efforts) have generally held more favorable views toward the ZOPs
and have adopted more conciliatory stances with the passage of time. Under
President Estrada in the mid-1990s, some peace zones were designated as
Special Development Areas (SDA) and received additional development
support from the government.28 As a government peace commission noted,
“The development of ‘Peace Zones,’ while starting out as an attempt to manage
conflict, has become a process of empowerment.”29 Similarly, the government’s
peace advisor (OPAPP) later stated that, “Zones of Peace should not be viewed
as . . . hindering . . . a negotiated settlement” and recommended “self-imposed
restraint on the part of the Armed Forces of the Philippines from conducting
military/combat operations in the areas covered by the Peace Zone declar-
ation.”30 Philippine congressional legislation has called for ensuring the “integ-
rity and autonomy” of ZOPs.31 In interviews, an official from OPAPP noted
that “a day without the breakout of hostilities is another day of peace [so] if
more and more communities declare peace zones, so much the better.”32 Some
of the peace zones have also been favored by the government for their roles in

26
Ma#4, Manila, 5/2012.
27
Ma#5, Manila, 5/2012. For these officers, limiting comments to barangays and informal auton-
omy actions was perhaps a more politically acceptable way of addressing this subject.
28
Yet, in interviews, an official from OPAPP noted the limitations of the state-imposed SDAs by
saying “You cannot legislate peace zones.” Ma#6, Manila, 5/2012.
29
1993 Report of the National Unification Commission, section on “Respect for and Recognition
of Community Declarations of Peace Zones,” in Santos (2005).
30
“Memorandum for Her Excellency Gloria Macapagal-Arroyo from Secretary Eduardo R.
Ermita, Subject: ‘OPAPP Position on the Current Debate re Advisability of the Peace Zones of
the Country,’” May 7, 2003, in Santos (2005).
31
House Bill No. 1867, Thirteenth Congress, “An Act Declaring a National Policy on Peace
Zones,” July 28, 2004, in Santos (2005).
32
Ma#6, Manila, 5/2012.
Civilian Autonomy and the Peace Zones of the Philippines 285

ensuring the durable reintegration and protection of insurgent ex-combatants

(similar to the role of the ATCC in Colombia).
Rebel groups have also exhibited ambivalence, alternating between oppos-
ing and supporting the ZOPs. The NPA registered initial opposition in refer-
ence to the Sagada ZOP, stating, “We cannot agree outright to an indefinite
‘total troop pullout’ by NPA units from Sagada or any other locality, nor
to a permanent local truce. To do so is to abandon the local people.”33
Skepticism toward ZOPs is also seen on the part of the MILF as made clear
in a press release, which asked, “What can this so-called ‘peace sanctuary’
offer? Nothing.”34 Yet, in another letter by the NDF (NPA), the emphasis on
reciprocity comes out, “If it is true that the AFP Officers and Civilian Officials
respect the declared ‘Peace Zone,’ the NDF will also . . . respect the people’s
declaration.”35
The views of rebel groups toward ZOPs also evolved over time, as evidenced
by public statements in support of the ZOPs. The NPA has outright supported
principles of protection enshrined in peace zone declarations.36 Similarly, MILF
Chairman Salamat Hashim stated, “The assurance we can give is that we
will keep our armed presence away from these Sanctuaries of Peace.”37 The
Ginapaladtaka Space for Peace was described by a MILF leader as “a noble
contribution alongside the peace process.”38 A former mayor confirmed that
the ZOP efforts have contributed to respect for rights in at least some occasions,
“We earned the promise from the leaders [of the AFP and the NPA] that they
would respect our rules . . . There were violations [of the peace zone that had
never gone public]. But . . . We assert the agreement so [the AFP and the NPA]
were compelled to punish their own people” (Cabreza 2010).

33
“Once More on the Question of Peace: On the Proposal for the Demilitarization of Sagada.”
Joint Statement of the CPDF Provisional Council and the NPA Chadli Molintas Command,
January 30, 1989, in Santos (2005).
34
MILF Central Committee Press Release, “Gov’t resorting to dirty tactics,” June 15, 2003, in
Santos (2005).
35
“Letter of National Democratic Front Far South Region Media Liaison Officer Raul Tan to
Whom It May Concern [translated from Cebuano],” September 13, 1991, in Santos (2005).
36
One statement proclaims, “We support points 5, 10, and 11 of the proposal: First, that the
civilian population not be harassed, intimidated, or subjected to other criminal acts by the
contending forces. Second, that customs and traditional practices be respected and observed.
And third, that suspects not be tortured or killed, and be investigated with the participation of
the people.” “Once More on the Question of Peace: On the Proposal for the Demilitarization of
Sagada.” Joint Statement of the CPDF Provisional Council and the NPA Chadli Molintas
Command, January 30, 1989, in Santos (2005).
37
Message from Ustadz Ameerul Salamat Hashim, Chairman, MILF (By Mohamad Nur, Chief of
Staff, Office of the MILF Chairman, Camp Abu Bakar As Siddique), responding to request to
“give approval and safety assurance” of a peace zone, May 31, 2005, in Santos (2005).
38
Remarks delivered by Von Al Haq, MILF CCCH member, during the Declaration of the
GINAPALADTAKA Space for Peace and Children as Zones of Peace, at Takepan, Pikit, North
Cotabato, November 29, 2004, in Santos (2005).
286 Civilian Autonomy around the World

This review of Philippine ZOPs points to their richness, diversity, and

intermediate forms of success, suggesting that Philippine civilians have crafted
one of the strongest autonomy movements around the world. While some “lose
their steam or cease to exist”39 and some have suffered periods of combat
(Alipala 2008), there is evidence of successful implementation, since many
persisted long enough to convene to discuss lessons learned in 2003 (Catholic
Relief Services 2003). Another NGO staff member observed, “[ZOPs] may not
be the safest areas, but are safe compared to many other communities.”40
Further analysis that accounts for conflict conditions is required to determine
the extent of implementation of declared policies and whether and how ZOPs
have affected violence.
At least one indicator of success is that some of the ZOPs have persisted and
continue to work toward peace at the local level and new ones continued to
form (Cabreza 2010, Fernandez 2011). In one instance from 2014, villagers
from the Sagada Peace Zone mobilized to evict any rebels that might be present
in their region and verify their departure to convince the military that there was
no threat to the area and therefore no reason for them to remain either
(Quitasol 2014). The movement for civilian autonomy from conflict in the
Philippines could also be far broader than what is accounted for here, as many
communities may have employed autonomy strategies without formally declar-
ing ZOPs. At the time of this writing, as peace talks continue between the
government and the MILF, one can point to the hopeful reality that some ZOPs
have endured long enough to survive the most trying period of conflict.

civilian autonomy in afghanistan and pakistan

The Afghan and Pakistani conflicts share several similarities with the Colom-
bian context, including diverse community strategies for protection and cases of
civilian autonomy. First, like Colombia, these countries have rural peripheries
with historically weak state presence, leaving many communities to confront
pressures from multiple armed actors, be they Taliban, NATO/ISAF, or Afghan
National Army (ANA) forces,41 or other paramilitary groups and bandits.
Second, due to tribal and cultural differences, there are different histories of
social organization, cohesion, and resistance across local communities. Lastly,
the Taliban insurgency appears to fit some of the theoretical conditions for
having sensitivity to civilian actions: like the armed groups in Colombia, they
appear to have hybrid characteristics. They have an ideological-religious
aspect, are concerned with legitimacy (e.g., see Mullen 2009 on Taliban insti-
tutions), and have published rules on how to conduct warfare and interact with

39 40
Ma#2, Manila, 5/2012. Ma#7, Manila, 5/2012.
41
International Security Assistance Force.
Civilian Autonomy in Afghanistan and Pakistan 287

civilians (Rubin 2010b). However, they are also known drug traffickers, indi-
cating different strains and motivations within the group.
Afghanistan has wide subnational variation in the strength of local insti-
tutions, tribal structures, and social cohesion. There are differing traditions of
maintaining local order from one zone to another and among ethnic groups (e.
g., the Pashtun Wali codes of conduct; Malkasian 2013). These organizations
and traditions are enablers of self-rule in relation to the government as well as
the Taliban – being wary about dangerous entanglements, stigmatization as
enemy collaborators, or abuses of authority. The civilians also hold preferences
for autonomy, stemming either from tribal traditions or from more immediate
concerns for protection. For example, at a shura (tribal meeting) in the Zhari
District of Kandahar Province, a vocal elder told U.S. military officers, “I’m not
going to let the enemy or you in my village. I’m going to take care of security
myself” (Brulliard 2010).
These preferences, combined with social cohesion and organizational cap-
acity, have produced a variety of autonomy actions. In the context of the
conflict, Hazaras in Jaghori District of Ghazni Province peacefully resisted the
initial incursion of the primarily Pashtun Taliban into the district (around
1997, pre-9/11) by dialoguing with them to maintain their autonomy and
minimize violence (Suleman and Williams 2003). In Helmand Province, villa-
gers in some districts organized to take up arms against foreign troops (U.S. and
NATO) to protect their homes and voice their anger after losing relatives in air
strikes (Gall 2009). In the eastern province of Nangarhar, a village rose up
against the Taliban, apparently as a response to Taliban violence and abuses as
well as to an entreaty of development resources from the Afghan government
(Gopal and Rosenberg 2009). Shortly thereafter, the rest of the tribal brethren
of the 400,000-member Pashtun Shinwari tribe followed suit (Filkins 2010).42
The ability of these communities to organize for protection has been attributed
to their relatively close social relations.
Tribal elders and local politicians in some local towns, including the Musa
Qala district of Helmand Province, also organized to peacefully resist the
Taliban by negotiating local cease-fires. In Musa Qala in 2006, residents
nonviolently pressed for neutrality and also sought to limit the presence of
NATO forces (Gall and Wafa 2006). According to Semple’s analysis (2009,
81), the fifteen-member district tribal jirga (council) signed a written agreement

42
As Filkins reports, although this action was against the Taliban, it also exhibits autonomy
motives, “Tribal loyalties are strong and the tension between the Shinwaris and the Taliban
long-standing. The Shinwari elders did not merely declare their opposition to the Taliban.
Although they declared their allegiance to the Afghan government, they directed at it a nearly
equal measure of fury, condemning ‘all the corruption and illegal activities that threaten the
Afghan people.’ . . . ‘We are doing this for ourselves, and ourselves only,’ said one of the elders.
‘We have absolutely no faith in the Afghan government to do anything for us. We don’t trust
them at all.’”
288 Civilian Autonomy around the World

with the provincial governor stating that only the police were to be allowed
within 5 kilometers of the district center.43 As with the Colombian cases,
acceptance of the civilians’ initiative would depend on their organizational
capacity, as the governor of Helmand recounted, “They made a council of
elders and came to us saying, ‘We want to make the Taliban leave Musa Qala.’
At first we did not accept their request, and we waited to see how strong the
elders were.” NATO/ISAF forces withdrew for a time and the tribal elders
“mobilized all tribes and subtribes in the area to apply social pressure to the
Taliban” to call off their attacks (85). The fighting decreased for five months as
the Taliban respected the ban on entering the bazaar and “the jirga succeeded
in turning them back the first time that a commander tried to enter with
arms” (82). According to a UN assessment, it was the “non-ideological local
Taliban” who most respected the jirga (85).
Unfortunately, in 2007, the accord frayed, as the jirga did not have the
necessary resources to deliver tangible benefits to the community and was
unable to maintain support and legitimacy (recalling the clientelist challenges
seen in Colombia). Although some rumors suggested the bargain came to be
dominated by the Taliban, “There was no evidence of the Taliban exploiting it
strategically,” and it was instead destabilized by an ISAF airstrike against a
Taliban leader close to the protected zone, which triggered a Taliban incursion
(88). The Musa Qala movement had promising origins and proved to be well
organized. However, with strong strategic interests among the armed actors to
control the territory and shifting conflict dynamics, it was not quite able to
credibly signal its neutrality or capacity to manage local security on an endur-
ing basis. With the agreement discarded, insecurity persisted as the territory
continued to be contested militarily (Goldstein and Shah 2015).
Pakistan, in the same theater of conflict, has similar examples. In some
instances, civilians armed to form local-based Lashkar militias to resist
Taliban violence, especially in the “lawless” FATA tribal region (Perlez and
Shah 2008; Wilkinson and Marwat 2008; Taj 2011). The town of Buner (near
FATA) adopted a hybrid approach using both armed and nonviolent collective
actions (Parlez and Shah 2008b). According to news reports, the residents of
this town preferred to manage their own security affairs to avoid getting caught
in the crossfire, “The villagers in Buner say they would prefer to handle the
Taliban on their own, rather than have the heavy hand of the army come and
do it for them.” Residents first armed to punish Taliban aggressors and then
formed a “peace committee” composed of elders and politicians and passed a
resolution declaring Buner a zone free of both the army and the Taliban. The
resistance and efforts at self-governance in Buner succeeded for a time, as seen

43
According to the analysis of one community elder, “The Taliban stopped fighting because we
convinced them that fighting would not be to our benefit.” Another resident said, “The Taliban
are not allowed to enter the bazaar with their weapons. If they resist with guns, the tribal elders
will disarm them” (Gall and Wafa 2006).
Civilian Autonomy in Afghanistan and Pakistan 289

in the elder’s statement at the beginning of the chapter. However, they were
undercut the following year after the government agreed to allow the Taliban to
impose Sharia (Islamic) law in the neighboring Swat Valley and the Taliban
gained the upper hand (Perlez 2009).
In the nearby Dir District, civilians also rejected the Taliban in response to
violent transgressions (Tavernise and Ashraf 2009). When Taliban suicide
bombers attacked a mosque, killing thirty civilians, it was the “last straw,”
and the residents of the town armed to fight them off. Lastly, the elders of Landi
Kotal and Jamrud Tehsil decided to form a peace committee to negotiate with
the Taliban instead of creating an armed lashkar (the option promoted by the
government) in hopes of avoiding aggravating local conflicts or inviting retali-
ation by arming (Shinwari 2008). Overall, some of these civilian efforts appear
beneficial, including at least temporary increases in security, although other
communities have faced repression in retaliation for collective actions.
Several government policies in these countries are also at least partly predi-
cated on the importance of social organizations and civilian autonomy. While
civilians may not strongly support the government or international forces
(though this is of course the counterinsurgents’ first preference), the Afghan
government has hoped that some programs will enable communities to at least
amicably manage their own affairs and development and not defect to the
Taliban. Examples include the National Solidarity Program’s establishment of
village development councils and other initiatives to strengthen local commu-
nities (Gall 2010).44 An episode from Nawa District reflects this approach. A
key factor distinguishing the success of the case of Nawa in ejecting Taliban
insurgents versus neighboring and more intransigent Marja was the strength
of local chiefs and their resistance to the Taliban (Chandrasekaran 2010).45
In another instance, paralleling the successful efforts of the Nasa Indians’
Indigenous Guard to free FARC kidnap victims in Colombia, members of a
local development council in Ghazni Province mobilized to win the release
of government development workers that were captured by an armed group

44
Beginning in 2003, the World Bank and the Afghan government built community decision-
making capacity and promoted local-based development through this program. It has reached
over 23,000 communities in most of the country’s districts but its relevance for counterinsur-
gency and stability outcomes has not yet been evaluated.
45
As Chandrasekaran explains, “A patch of desert in Helmand province that was transformed into
farmland by canals designed by American engineers in the 1950s, Marja was populated from
scratch by the country’s late king with settlers from a variety of tribes. The rank and file moved
to Marja, but the chiefs didn’t. This decades-old experiment in Afghan social engineering has
now complicated efforts to find the same sorts of tribal leaders who influence the population in
other Afghan communities. They simply don’t exist in Marja. . . . Why, then, did the Taliban fold
in Nawa? Residents interviewed in the bazaar earlier this year said it was in part because the
insurgency enjoyed little support in the community. Locals chafed at the Taliban’s taxation, and
they grew tired of the near-constant firefights between the insurgents and a team of British police
trainers holed up in the district center. Tribal leaders made it clear they wanted the bad guys out,
in part so they could reassert themselves as the chief power brokers in the area.”
290 Civilian Autonomy around the World

(NPR 2013). These vignettes show tribal cohesion has a role in rejecting
extremists and has implications for more tactful counterinsurgency.
Civilian organization and autonomy are also cornerstones of policies to deal
with detainees that are captured on the battlefield or arrested in raids. With the
establishment of the “community-release program,” ISAF and U.S. military
policy migrated toward supporting ATCC-style investigation processes (Rubin
2010a, Bumiller 2010).46 The U.S. military’s Joint Task Force 435 began to
work with communities and local institutions to deal with suspected insurgents
because they had difficulty obtaining the necessary local information to adjudi-
cate their status or prosecute them. Further, holding prisoners involves costs,
since detentions that are perceived as illegitimate (e.g., due to false accusations)
and even legitimate detentions can anger the population.47 Under this program,
the detainees sign pledges to stay away from the insurgency, and tribal elders
then also agree to “vouch” for them with a signed contract to ensure that they
do not return to war.48 This process implicitly relies on well-organized commu-
nities in areas of conflict to guarantee suspects will not become insurgents. The
program raises the autonomy-related questions of whether fractured commu-
nities would be less able to absorb and vouch for detainees and whether the
Taliban could adopt a similar process to limit killings on their part of suspected
government collaborators?
Lastly, the U.S. military has trained several villages to fight and defend
themselves (Tyson 2008 and Rubin and Oppel 2010). Armed villages appear
to receive public goods and be allowed autonomy as long as they do not
aid the Taliban (the government does not necessarily request strong forms

46
“Now, in Afghanistan, detainees who are deemed not to be a threat are handed over to local
elders on the understanding that it is the community’s responsibility to ensure that they stay on
the right side of the law” (Rubin 2010a). A similar program has been used to deal with detainees
in Iraq.
47
Indeed, Kalyvas’s local grudges figure prominently in these cases, “In interviews, former detain-
ees and their families said the Americans were routinely misled by informants who either had
personal grudges against them or were paid by others to give information to the Americans that
would put the person in jail. In addition, many Afghans have experienced the detentions as
humiliating, and found almost unbearable the depths of poverty borne by their families during
their internment.” As an elder said in one case, “‘The information you had about these men was
wrong in the first place. We are confident they were not involved with insurgents. If they were,
we wouldn’t be here to sign for them’” (Rubin 2010a).
48
In one episode of releasing a detainee back to his community, “A United States Marine
commander who was acting as the prosecutor, told the prisoner: ‘This letter right here is a
sworn pledge from all of your elders that they’re vouching for you and that you will never
support the Taliban or fight for the Taliban ever again.’ . . . But what is preventing him from
rejoining the Taliban? The Marines say the village elders who vouched for him will help keep
him in check, as will a parole-like program. The Marines will meet with him regularly and pump
him for information about his friends” (Bumiller 2010). Democratic governments involved in
counterinsurgency campaigns (as opposed to illegal paramilitaries or guerrillas) will not typically
use threats and killing as coercive tools, but will instead arrest suspects, sometimes with slow
judicial processes.
Civilian Autonomy in Syria 291

of allegiance). Likewise, the United States considered replicating an “Anbar”

tribal strategy for the Pakistani tribal areas (Schmidt et al. 2007; interest
may have declined after the Pakistani government commenced a counter-
insurgency campaign in early 2009). Afghanistan and Pakistan have seen more
arming by local communities than has Colombia. Even so, these countries
are also home to many traditions and examples of nonviolent advocacy and
dispute resolution.

civilian autonomy in syria

The Syrian civil war began in March 2011, when the regime of Bashar al-Assad
violently cracked down on antigovernment protestors and an armed rebel
movement mobilized in response. The civil conflict in Syria is one of the most
brutal in recent memory. It has seen the use of cluster munitions, chemical
weapons, barrel bombs, mass starvation, and beheadings and mutilations on
the part of the regime as well as the use of extra-lethal violence by rebel forces.
In a country with a population of 23 million, estimates indicate more than
150,000 civilians have been killed, over 6.5 million have been forcibly displaced
within the country, and an additional 2.5 million are refugees in neighboring
countries (UNHCR 2014). The violence was so dire that the UN stopped
officially counting civilian deaths in January 2014.
Since the earliest days of the Syrian civil war, international actors have been
at a loss about how to stop the violence against civilians and decisively bring
the conflict to an end. Numerous proposals were either implemented only half-
heartedly or were deemed impractical. Failed international proposals to stop
the violence included the establishment of no-fly zones, safe zones and humani-
tarian corridors, military interventions, observers, peacekeepers, cease-fires
(e.g., for the holiday of Ramadan), providing adequate support and armaments
to the (“moderate”) rebels, and invoking the Responsibility to Protect doctrine.
Western decision-makers were paralyzed by war fatigue, casualty aversion, fear
of resistance from geopolitical rivals such as Russia and Iran, and potential
spillover of the conflict to neighboring countries. The reluctance to mount a
humanitarian intervention was underscored by U.S. President Barack Obama’s
“red line” comment, which threatened consequences for the Syrian govern-
ment’s use of chemical weapons but did not result in action. Limited inter-
vention was only threatened to secure Syria’s stockpile of chemical weapons
after the Syrian regime used Sarin gas against civilians and killed an estimated
1,500 in Hama in November 2013. By contrast, intervention was not seriously
contemplated in response to the tens of thousands of civilians that had been
killed by conventional weapons up to that point.
The focus on belligerents in these proposals meant that the actions of the
civilian population were largely neglected. The combination of violence and
paralysis of international policymakers left civilians to their own devices. As I
detail in this section, Syrians on the ground were not solely waiting for help
292 Civilian Autonomy around the World

from the outside. According to disparate and less-known subnarratives, many

Syrians found their own ways to get by and survive the violence. Indeed, while
many individuals have taken part in the conflict in some form or been affected
by it – Syrian civilians have certainly suffered a heavy toll – there are also
examples of creativity and collective actions for autonomy from the conflict. As
in Colombia and other countries, resilient Syrians were able to hold out, stay in
their homes, and avoid aligning with a particular side in the conflict.
Despite limitations of access and press reporting, I identify three types of
civilian autonomy movements in Syria rooted in preexisting social cooperation
and organization: 1) intersectarian harmony; 2) ethnic minority enclaves;
and 3) other general protests and actions against violence. When these dispar-
ate examples are collected and compared, it becomes evident that citizen
collective action for autonomy in Syria endured long into the conflict in diverse
forms. Some of these nonviolent mobilizations for autonomy blended with the
broader freedom and democracy protest movement that called for the removal
of the Assad regime (and that some rebel fronts also supported).49 Other
autonomy efforts had less close links with the broader pro-democracy move-
ment. The accounts of the civilian autonomy efforts indicate varying degrees of
success at avoiding violence, with greater challenges and suffering all around
as the conflict dragged on and intensified.
Civilians in Syria faced dangers from a mix of combatants: the Syrian
military, the Shabiha pro-regime Alawite militias, the Free Syrian Army (FSA)
and then, as the conflict evolved, jihadist and extremist Islamist rebel groups
such as the Al Qaeda-affiliated Jabhat al-Nusra (Al-Nusra Front) and the
Islamic State of Iraq and Syria (ISIS). These different belligerents worked to
build networks of cadres and supporters within different communities and
fought to a stalemate (Fahim 2012, Barnard and Saad 2013a) in which civilians
were widely targeted and extra-lethal violence was reportedly used by the
regime (Baker 2014) and some rebel fronts (Barnard 2012a, 2012b, Chivers
2013a; especially against captured regime soldiers). Given these conflict condi-
tions, many Syrians remained ambivalent about which side to support long into
the conflict, with one man saying, “This armed revolution, I refuse it as much as
I refuse the regime” (Barnard and Saad 2013a). Instead, most civilians have
been concerned with survival, worried about the regime as well as extremist
(Islamist) or undisciplined rebels (Solomon 2012). According to one woman,
“People feel it’s ‘with us or against us.’ I personally am with no one, I am
against the whole thing. It’s killing our children” (Solomon 2012). This analysis
shows that throughout the conflict – early on and even in later periods, though
to a lesser extent – some people acted upon these sentiments, with some
individuals and communities holding out and trying to avoid participating in

49
For instance, local organizations such as the Local Coordination Committees (LCCs) sought to
establish civilian governance in both contested areas and areas “liberated” by insurgents.
Civilian Autonomy in Syria 293

the conflict. However, as the conflict continued to intensify, the economy

collapsed, supplies were cut off, and resources were depleted, many (though
not all) autonomy-seeking civilian populations were strained and buckled.

Oases of Intersectarian Harmony

Prior to the onset of conflict, Syria was known as a haven of sectarian harmony
and religious freedom, and religious identity was not generally viewed as an
important social cleavage (McDonnell 2014). While the conflict did much to
tear at Syria’s social fabric (Stors 2013), some communities have been referred
to as “oases of tolerance” and remained de facto safe zones – just not the
(military-backed) ones that most international policymakers and commentators
imagined and hoped to establish. I review how two such religiously tolerant
communities actively worked to prevent being dragged into intersectarian
conflicts.
One of these oases is the town of Maloula (or Maaloula), about 30 miles
north of the capital of Damascus (England 2008). Maloula, where the ancient
language of Aramaic is still spoken, had a mix of Christian and Sunni Muslim
residents as well as important ancient Christian religious sites, such as the
Convent of St. Takla. Maloula was initially caught between the contending
forces of the Syrian government and the FSA, and later also the Islamist Jabhat
al-Nusra front. Even with these pressures, unity was seen among the population
in the reluctance of residents to classify themselves by religion, instead simply
saying, “I am from Maloula” (Di Giovanni 2012). As illustrated in the quote by
the Muslim imam from Maloula at the beginning of the chapter, there was a
commitment among religious leaders to keep the community from devolving
into intersectarian violence and engaging in the larger armed conflict that was
being fought around them.
The cohesion and good relations between sects was, at least for a time,
translated into an ability to manage the pressures of the armed conflict and
avoid attacks by armed actors. In referring to the intensification of the conflict
in the middle of 2013, one resident reported that, “For months the rebels have
been around Maloula but there has been a sort of an understanding with the
residents that they would not enter . . . they do not seem to have touched
churches or homes” (BBC 2013). In September of 2013, the rebels moved
against the town but, as described by a rebel-affiliated activist, they quickly
withdrew, citing the key role of influential community elders, “When the rebels
moved in, the elders of the town were afraid of [government] airstrikes and
shelling. They wanted us to go, so we left” (Sly 2013). Rebels stated that the
motivation of the withdrawal was to protect the community, “To ensure no
blood is spilt and that the properties of the people of Maalula are kept safe, the
Free Syrian Army announces that the town of Maalula will be kept out of the
struggle between the FSA and the regime army” (AFP 2013). According to one
of the town’s nuns, residents maintained freedom of movement between rebel
294 Civilian Autonomy around the World

and government territory (Barnard and Saad 2013b). Consistent with the
demands of reciprocal treatment among armed actors seen in other conflicts,
the rebel withdrawal was conditional on government forces and militias not
entering the town (AFP 2013).
Eventually, more intense fighting did come to Maloula in the second half of
2013, as the more extreme al-Nusra front came to the area and the Syrian
government mounted a counteroffensive. Although the contestation eventually
ended in heavy damage to the town and mass displacement, al-Nusra and the
FSA did not want to be seen as targeting Christians, since this could have made
the United States more reluctant to intervene militarily on the side of the rebels
(although the Mother Superior of the local convent was held by the rebels for
several months, she confirmed she was not mistreated; Sly 2013, Barnard and
Saad 2014). Similar to concerns of the FARC rebels in Colombia, FSA rebels
worried about the potential consequences for their reputation of being associ-
ated with targeting cohesive nonviolent communities. Nevertheless, the incur-
sion by al-Nusra led to retaliatory shelling by government forces, leaving
Maloula a ghost town. With the incursion by al-Nusra (and social cohesion
also apparently finally breaking down), a Christian resident reported having
been betrayed by the Muslim residents (Bowen 2013, Fisk 2013).
Similar sentiments for autonomy and maintaining local order among diverse
sects are found in the neighboring town of Yabrud (Yabroud). As a Christian
resident said, “The regime wants us to fear Muslims, but I don’t fear my
brothers” (Tice 2012). Another man from this community extended this senti-
ment to the armed actors in the conflict, saying, “I don’t care if you are a
loyalist or a dissident, respect yourself by respecting me” (Tice 2012). In
Yabrud, there was a civilian council to negotiate with the different armed actors
so that neither side would fight over the town (Barnard 2014). There were also
conflict management procedures to deal with disputes and cool tempers. One
resident cited restraint in response to a Shabiha militia incursion, “They [the
Shabiha] broke into many houses, my father’s house, stealing and breaking
things. We did not react strongly. We did not want to bring the war here
(Tice 2012).”
Other news reports confirmed the ability of Yabrud to endure even several
years into the conflict, into 2014, noting that when “Islamist insurgents tried to
make inroads, they were largely rebuffed or ignored” (Barnard and Saad 2014,
Barnard 2014).50 A rebel noted respect for the neutrality of the Christian
residents of the town, saying, “We don’t have any problem with Christians,
they are living among us for thousands of years. Before, with and after Assad”
(Barnard and Saad 2013b). According to one resident, governance by the
town’s local council was also working to limit government incursions and

50
According to one report, “Mediation, rebels say, helped keep Yabroud relatively untouched by
fighting until recently. Yabroud has essentially governed itself, with some local Christians
remaining, even as the war turned sectarian elsewhere” (Barnard and Saad 2014).
Civilian Autonomy in Syria 295

manage violence, “I personally consider Yabrood to be an area free of govern-

ment control. . . . We have not suffered a lot compared to other cities in Syria”
(Saeed 2012). The security situation remained stable until the end of 2013
when, as with Maloula, rebels finally entered Yabrud. Government forces then
moved on the town in 2014 and eventually gained control, provoking mass
displacements of the population (Barnard 2014).
These cases of intersectarian oases held out for nearly three years by pursu-
ing neutrality and promoting intersectarian norms so that disputes would not
provide openings for belligerents, even becoming receptors for displaced
Syrians from other parts of the country (Barnard 2014). They are noteworthy
since it is not obvious that such harmony can be maintained in the context of a
war that pits family against family, where the regime explicitly used the strat-
egies of dividing the sects and polarizing the opposition by framing them as
terrorists and extremists, and where even children in some communities voiced
desires for retribution against members of other sects (Kirkpatrick 2012,
Karam 2012). However, with the intense and enduring conflict, a politicized
environment, and cut supply lines, the cases also show the limitations of
autonomy under extended periods of siege.

Autonomy Among Ethnic and Religious Minorities

Similar to some of the minority populations in Colombia and other countries,
many of the ethnic and religious minority groups in Syria did not historically
have strong interests in national-level political conflicts. In addition to the
instances of intersectarian bridging found in some diverse communities, more
ethnically and religiously homogenous communities also organized around
their identity-based social ties to stay out of the conflict and avoid the effects
of the war. Non-Sunni ethnic and religious minorities made up about 33
percent of Syria’s prewar population, with about 10 percent of the population
belonging to the ruling Alawite sect, a Shia offshoot. The non-Sunni (and non-
Alawite) populations, especially Christians, historically aligned with the gov-
ernment and, fearing an Islamist takeover, were anxious about how the rebels
might treat them were they to come to power. But they were also repulsed by
the regime’s brutal tactics and did not want to weaken the pro-democracy
movement or antagonize the rebels by overtly siding with the regime.51
The behavior of Syrian Christian groups, including the Armenians and
Assyrians (Syriacs), exhibits clear motivations and actions for autonomy.
Armenian residents of Aleppo armed for self-defense while their religious
leaders publicly called for the group’s neutrality. On September 14, 2012, the

51
Although many of the Alawite minority form the Assad regime’s base of political support, there
were also some splits as some Alawites favored a transition to democracy (DePetris 2013, Oweis
2013). Some but not all members of the other minority groups have preferred and pursued
neutrality and autonomy.
296 Civilian Autonomy around the World

leaders of the three Armenian churches in Aleppo issued a joint statement

underscoring the neutral positioning of the community:

What adds to our anguish are the unsuccessful attempts of presenting the Syrian
Armenians as taking part in the armed battles of the current Syrian crisis or trying
to actually drag them into such a conflict . . . We reiterate today, that the peaceful
co-existence that the Syrian Armenians have cultivated throughout the decades con-
tinues . . . and it will definitely stay against all kinds of violence and armed collisions . . .
We are not worried. We fear the situation for the whole country, for all the people in
Syria. But we are not taking sides in this crisis. (Armstrong and Williams 2012)

The religious leaders of the Assyrian Christians also made similar calls for
neutrality and respect of their populations (Jawad Al-Tamimi 2012, Cheikho-
mar and Austin 2013). These Christian groups were able to remain intact for a
short period of time, but many Armenians eventually ended up leaving the
country for greater safety in Armenia (Malek 2012), while many Assyrians fled
to Turkey.
The Palestinians living in Syria were welcomed to the country in the 1950s as
refugees and, like the Christians, became wary of opposing the Assad regime
because their rights had historically been protected by the Syrian government.
Yet the Palestinians also did not want to be seen as opposing the revolution and
become subject to rebel attacks. As expressed by one man from the Yarmouk
district (originally a refugee camp) on the outskirts of Damascus, there was a
strong preference for neutrality, “The Assad regime wants us to express our
support for his regime, and the opposition wants us to demonstrate against the
Assad regime” (Sands 2011, Nordland and Mawad 2012). During the first few
years of the conflict, Yarmouk was left largely undamaged. As the conflict
progressed, however, some rebel fighters took refuge within the district, pro-
voking government retaliation. After an initial displacement due to fighting,
community leaders were able to negotiate local cease-fires to allow them return
and again asserted neutrality (AFP 2012). The autonomy actions yielded some
small gains but were ultimately to little avail, as insurgents continued to infil-
trate the neighborhood, prompting the government to respond with airstrikes
(Barnard 2012c) and a siege that caused starvation among the residents. When
Islamic State militants entered the fray in Yarmouk in late 2014, a mass
displacement of the remaining population ensued.
The Kurdish minority of about 2.5 million people also took actions for
autonomy from the conflict between the Assad regime and largely non-Kurdish
Sunni rebels. They have sought to create an enclave from the conflict in the
northern part of the country, a move that also hews toward seeking greater
political autonomy (Solomon 2014). As a Kurdish man from the town of
Ras al-Ain, on the border with Turkey, said, “I don’t want the rebels in my
town. Why would I want Assad’s planes to come and bomb us? I don’t want
Assad, nor do I want the rebels” (Burch 2012). Kurdish populations aimed to
avoid taking sides in the conflict, but some Kurds have also armed and formed
Civilian Autonomy in Syria 297

militias to protect their towns and, in some cases, have fought on both sides of
the conflict (Chivers 2013b). Similar to the other minority populations, the
reclusive Druze population took a neutral stance, largely refusing military
service and refraining from joining the rebels (Sands 2012, Naylor 2015). The
Sunni tribes in eastern regions have also been hesitant about taking sides
(Khalek 2012).
With community organizations, preexisting social cohesion, leaders, and
no strong preferences for either side of the conflict, minority groups in Syria
positioned themselves early on to stay out of the conflict. The initial efforts
consisting of negotiation and public statements held promise for many groups.
Unfortunately, their security deteriorated in the long run as more Islamist rebels
gained strength and the regime resorted to stronger tactics.

Additional Autonomy Examples

Syria also has additional instances of civilian autonomy not directly related to
intersectarian unity or minority enclaves. In various regions there were non-
violent protests for community autonomy and against the conflict (distinct from
protests against the regime or larger political goals). In the face of FSA trans-
gressions against civilians, protestors have called out, “The people want the
reform of the Free Syrian Army. We love you. Correct your path” (Barnard
2012b). Nonviolent pro-democracy activists mounted similar protests in the
town of Saraqeb in Idlib Province against rebel excesses (NOW 2013) and
in Mayadeen to demand that Jabhat al-Nusra fighters leave town (Khalek
2013). Later protests in Idlib against al-Nusra’s abuses and rigid interpreta-
tions of Islam successfully forced the group to withdraw from the town of
Maarat al-Numan and won the release of kidnap victims (members of the FSA;
Naylor 2016).
Activists also engaged in public protests against abuses of both Islamic State
militants and government forces. In Aleppo and the village of Kafranbel,
citizens used witty signs and slogans to draw attention to their conditions and
shame the armed actors (Mackety 2014). Similar protests were mounted
against ISIS in Mosul, Iraq, to halt the destruction of cultural sites, including
the city’s famous Minaret (Arango 2014). There is even evidence of civilians
“nudging” armed groups to reform (Kaplan 2013b) in Mosul by exploiting
fissures between ISIS’s domestic and foreign fighters. When an old man cursed a
foreign fighter for denouncing his beard as being too short, local ISIS fighters
stood up for the man and turned on the foreigner, beating and detaining him
(Bradley 2016).
There have also been hunger strikes and negotiation efforts for “local
autonomy” to break starvation sieges by government forces and ensure food
deliveries to embattled communities (Hassan 2014). Small-scale cease-fires
were brokered in several Damascus suburbs, and pressure by residents from
the towns of Moadamiya and Barza got rebels to agree to a cease-fire so food
298 Civilian Autonomy around the World

could enter (although there were also some reports of rebel reneging). The
Assad regime accepted some of these cease-fires and they even became a prong
of U.S. and Russian diplomatic strategies (Barnard 2014).
Finally, there are several examples of other autonomy strategies oriented
toward community governance. The activist-run pirate TV station AleppoTo-
dayTV provided early warning of battles to residents so they could avoid
getting caught in the crossfire (Amos 2012). In rural areas, given the unpredict-
able nature of the fighting and targeting of communities, some populations
took to hiding in caves for shelter (Chivers 2013c), similar to the Communities
of Populations in Resistance in Guatemala (Falla 1994). Lastly, in areas that fell
under rebel control and even contested areas, security remains a problem and
civilians have frequently taken it upon themselves to organize junta-like Local
Coordination Committees (LCCs) to make community decisions and perform
government functions such as policing and managing disputes (Kirkpatrick
2013, Amos 2012b). Activists in Raqqa and outside of Syria formed the group
“Raqqa Is Being Slaughtered Silently” to monitor the conflict’s civilian toll and
report on and protest Islamic State abuses (Abdulrahim 2014, Sly 2013).
Civilians in Syria have suffered greatly from the effects of the conflict, yet
their actions also highlight many examples of resilience. They have pursued a
variety of autonomy strategies, including neutral positioning, negotiation,
norms and managing disputes, early warning, and protests and shaming. While
the examples discussed are not controlled comparisons, they show the potential
for civilians to act on their own behalf even in extremely challenging condi-
tions, including the rapid onset of conflict and Syrians’ limited experience with
civil society organization and self-governance, having historically lived under
strict authoritarianism. The record indicates that few civilian efforts for auton-
omy in Syria have ultimately succeeded or endured, so the examples also point
to the limitations of civilians when facing ideologically violent groups and
government forces pursuing brutal tactics and “final solutions” over an
extended period of conflict. Yet, these communities may have held out longer
than they would have with less organization and no community strategies. If
the war had ended sooner, perhaps they could have endured long enough to
remain intact.
International policymakers have fixated on various diplomatic and inter-
vention scenarios but, with the failure of the doctrine of “Responsibility to
Protect” (R2P), they were left with a blind spot and did not consider the role
that local institutions could play to keep people safe. Given the rapid onset and
intensity of the conflict and insufficient external support, it was a challenge
for the disparate local autonomy efforts to connect with each other to stitch
together a broader movement to end the war. These diverse autonomy efforts
therefore also raise the question of whether outcomes would have been differ-
ent if these models had been identified, supported, and extended by western
governments, international organizations, or local actors earlier in the conflict.
In sum, although Syria may seem worlds away from Colombia, the Colombian
Conclusions 299

campesinos hold lessons for Syria and Syria, conversely, is also instructive for
other conflicts around the globe.

conclusions
The cases reviewed in this chapter are tough cases where reports of wide-
spread violence might lead one to discount the odds of encountering civilian
agency and autonomy. Yet evidence from these conflicts supports aspects of the
civilian autonomy theory. The presence and capabilities of social organizations
varied across the landscapes but were still found in war-torn regions. These
organizational bases served as helpful platforms for implementing community
autonomy strategies, from community management and promotion of pacifist
norms to more contentious protests and shaming. There were even cases of the
sophisticated investigations mechanism that was innovated and perfected by
the ATCC in Santander. Finally, the collective strategies influenced armed
actors and helped avoid violence in at least some circumstances and for some
amounts of time. Although these cases are some of the most well-known
conflicts and have figured centrally in debates of global security and received
broad press coverage as they occurred, there was little awareness of local
civilian autonomy activities due to biases in reporting and our collective focus
of attention. This chapter therefore serves as a needed corrective to our myopia.
Still, these cases point to many outstanding questions to be studied for a
more rigorous analysis. An implication for inference from these cases is that
effective civilian protective strategies are likely to be found in other countries
and conflict situations that resemble the moderate (and even harsh) conditions
found in some places and periods in Colombia (but are embattled as conflict
intensifies).52 Some conflicts may resemble these conditions more than others
depending on armed actors’ preferences and modes of operating. Some coun-
tries may also have greater preexisting bases for collective action across their
social landscapes to facilitate civilian strategies than others. It will become
easier to assess the causes and effects of civilian strategies and organizations
through controlled comparisons as more data becomes available at the local
level in these settings.

52
A discussion after a presentation of this research in 2009 with a former U.S. military officer who
served in Iraq revealed that, even there, community organization parlayed protective benefits. In
the Sadr City district of Baghdad, elder sheiks of organized communities were better able to tell
Al Qaeda (AQI) militants to leave them alone and keep them out relative to unorganized
communities.
10

Conclusions and Policy Implications

To you that want violence, this ended in shit; for the right to life, peace, and work.
Ustedes que quieren violencia y esto acabó carajo; por el derecho a la vida a la paz
y el trabajo.
– ATCC leader’s diary (1990s)

This book brings a new perspective to civilian agency and civilian responses to
the dangers and uncertainty of armed conflict. I focus on the civilians’ collective
action problem – opposite that of armed group recruitment – of how to keep
from participating in the conflict and avoid the tyranny of the relatively fewer
combatants and militant extremists. It is a puzzle that some communities
apparently solved this collective action problem and were able to protect
themselves while others did not. This book addresses this puzzle by taking
civilian institutions seriously as an explanation for violence. In doing so, it both
contrasts with and complements state-based, structural explanations of civil
wars and refines our understanding of the production of violence.
The bottom-up approach shows that collective action for peace, and not just for
violence, is possible even in settings of armed conflict. This is more than wishful
thinking. Social organizations and cohesion function as an important buffer
between communities and armed actors, enabling strategies to limit the fighting
to only the combatants and the aggrieved. The nonviolent strategies that commu-
nities use to avoid getting caught in the crossfire are diverse and work because they
are adapted to different threats of violence. However, these strategies can be
difficult to observe because civilian autonomy is often a hidden behavior in rural,
isolated areas. It does not always get reported as “news” – it is the dog that did not
bark. This is in part because autonomy is not easy to define or identify when it
occurs. The theory of civilian autonomy I developed therefore first involved the
conceptual ground clearing of supplying a definition of autonomy in civil wars.

300
Conclusions and Policy Implications 301

The theory then posits that armed groups’ motivations and abilities to use violence
vary based on the ability of civilians to impose costs on them and avoid entangle-
ments with them, as captured by levels of civilian organization. The challenge thus
became analyzing whether armed groups would have used more violence if not for
civilian organizations.
Some formal autonomy organizations, such as the ATCC, are capable of inno-
vating remarkably sophisticated and effective civilian autonomy strategies to help
deal with violence. These include “weapons of the not-so-weak” that directly
engage armed actors. Even under changing conflict conditions and amid denunci-
ations of suspected armed group collaborators, the ATCC’s investigation insti-
tution often proved effective for limiting violence against civilians who opted to not
participate in the conflict. Even some juntas, such as those in Belén de los Anda-
quies, can carry out these kinds of procedures. As the ATCC is fond of pointing out
and as the ATCC leader’s diary at the beginning of the chapter attests, they have
outlived many of the members of the armed groups of their region. The ATCC case
demonstrates that social networks and information flows can be traced empirically
and that identities can be sorted out in the “fog of war” of counterinsurgency.
Some informal social organizations across rural Colombia even proved
capable of reaching for civilian autonomy as well. The junta councils and the
social cohesion among ethnic minority populations were found to most effect-
ively limit violence because they are likely less political and more community-
oriented than other organizations, namely cooperatives and land reform councils.
The role of the junta councils in the case towns of Bituima, Quipile, and Vianí in
Cundinamarca confirmed that protection through social cohesion is a broader
phenomenon than previously believed. With relatively many juntas, Bituima
and Vianí saw more forms of civilian cooperation (“weapons of the weak”) and
even some forms of protest, were viewed differently by armed actors, and
suffered less violence than Quipile. Civilians in these and other towns mobilized
to resist war in a variety of ways. They reoriented existing institutions, net-
works, and shared histories, and channelled collective tragedies and experiences
with violence into cooperation. These were at times catalyzed by technical and
epistemic support from external actors, including government promoters, the
Peace Corps, NGOs, and churches.
Civilian autonomy was documented across multiple locations and armed
actors. In Colombia, it was observed in relation to the FARC guerrillas, the
paramilitaries, and in some cases, the public forces. The compilation of experi-
ences from other countries further highlights the true breadth of civilian auton-
omy. Across the many contexts, there is a remarkable consistency not only in
autonomy strategies themselves, but also the rhetoric surrounding them. This is
seen among communities as well as in the responses of armed actors, including
their ambivalence toward and acceptance of civilian autonomy.
In the rest of this chapter, I review the scholarly implications and policy
implications of this research. I then outline a future research agenda. I conclude
with some final thoughts on the broader significance of the book.
302 Conclusions and Policy Implications

the nuances of civilian autonomy

This study does not make a blanket statement about the effects of civilian
organization on violence. Rather, it has a more nuanced view. I found that
the effects of civilian autonomy organizations are conditional, depending on
conflict intensity (with limits at intense levels of fighting) and varying across
armed groups, time periods, and types of violence. Some kinds of organizations
may provide protection but other organizations may be seen as threatening and
be targeted. Some organizations may respond to and deal with violence but
others may also be weakened by violence. The juntas appeared more effective in
managing selective violence under moderate conflict conditions while qualita-
tive evidence showed that strategies of deeper cooperation (e.g., protest or
arming) can deal with even graver threats of violence.
By contemplating circular relationships and the conditions for effective
civilian agency, the theory and findings also explain several puzzles. They help
clarify some of the contrasting claims among the existing works about when
autonomy strategies will be effective. For instance, while Valentino observes
few civilian actions in response to mass killings and Kalyvas attributes civilian
self-rule only to armed groups’ permissiveness, the civilian autonomy results
point to a middle ground of conflict conditions and hybrid armed actors where
civilians can contribute to their own security. Incorporating the circular rela-
tionship also clarifies why in some settings we hear of many acts of resistance
and in others only acquiescence, and why social leaders are targeted in some
cases and, elsewhere, leaders and organizations endure.
The success of civilians’ collective actions is conditional because, while
civilians have levers of influence over armed groups, these levers are finite.
Civilians’ strategy choices depend on types of violent threats, which in turn
may determine the overall effectiveness of protection through solving particular
problems of violence. Certain strategies, such as community management (to
limit armed group inroads) and coordinated protests (to leverage reputations)
appear to be the most prevalent strategies. By contrast, investigations of sus-
pected collaborators and early warning systems for combat require more
capacity and are therefore found to be less common, though these have not
yet been well systematized.
The canvassing of armed groups’ viewpoints highlighted their preferences
and how their choices are affected by civilian strategies – their “sensitivities.”
First, ex-combatants reported perceiving differences in the level of social organ-
ization across communities and treating organized communities differently.
They viewed these communities as better managed, more trustworthy, and also
more threatening in their potential to mobilize against them. Second, some
armed groups can be induced to reduce their use of violence if, for instance,
they can save face when they back down, or benefit from interacting with
civilian organizations by reducing manpower costs for managing communities.
Sometimes collective appeals to morals can nudge groups to change their
The Nuances of Civilian Autonomy 303

default positions on using violence. These strategic interactions explain why

civilians appeared most frequently able to affect selective violence and attempts
at coercion. A broader implication, though, is that the behavior of armed
groups is not destined for an inexorable deterioration as they, for instance,
become more resource-based (Weinstein 2006). Instead, their behavior can be
reformed over time in response to civilians.
Additional qualitative evidence indicates which types of armed groups are
and are not amenable to civilian strategies. Two cases of armed group
demobilization show that civilians have less influence as armed groups become
less political and turn more toward pure banditry. In the latter half of the 2000s
in the ATCC region, certain paramilitary units either did not demobilize or
remobilized after the official paramilitary demobilization. With mainly the
hard-core criminal and drug-trafficking elements left, the ATCC encountered
more problems dialoguing with this group and having their investigations
honored.1 Similarly, Jehovah’s Witness groups in Mozambique that had suc-
cessfully held out against Renamo rebels faced greater difficulties in having
their pacifism respected by bandit groups that arrived after Renamo disbanded
(Wilson 1992).
Junta councils may fail to protect civilians in many places for several
reasons. One is that they can be weakened by clientelist dependencies. At least
some communities with formal organizations like the ATCC and some juntas
can still act. But juntas may also either decay or fail to provide protection if they
become too politicized. And, even when juntas achieve civilian autonomy, they
may need to balance their neutrality by censoring their political aspirations.
Survival and managing risk in the midst of war may therefore sometimes carry
the price of forfeiting advocacy for broader political and economic rights (e.g.,
labor rights or land for peasants).
Civilians were not found to have much influence over severe forms of
violence or broader conflict dynamics through either organizing or trying to
opt out of the conflict. However, this does not mean civilians are never
able to affect conflict dynamics or indiscriminate violence. For example, the
ATCC and other organized communities assisted with combatant demobili-
zations, worked to reduce participation in the coca economy, and were able to
help victims of forced displacement remain in the area so they could later return
to their homes or resettle nearby (interviews; CNRR 2009, 2011). Further, the
findings of some massacre-proofing effects and an interaction between the
juntas and anti-Patriotic Union party violence suggest that the juntas may
have limited the killing by countering political stigma against residents. To
the extent that the violence constituted a politicide, the juntas could also have
implications for genocide and mass atrocities prevention. Further study will
help clarify whether “peace communities” can bring a broader peace.

1
ATCC#3,4, La India, August 2008.
304 Conclusions and Policy Implications

the historical and state-building foundations of

civilian autonomy
Civilian autonomy in Colombia’s conflict did not appear out of nowhere.
Colombia’s peace movements were rooted in the country’s historical social
and political context and arose as a response to the conflict. The country’s
political tensions surrounding the urban–rural divide and contrasting state-
building philosophies pre-positioned certain organizations like the juntas for
later autonomy in the conflict (accounting for why the juntas, more than other
organizations, merit attention for this topic). The same historical half-neglect of
the juntas also explains why little is known about how they behave in conflict
settings. This disengagement reflects the larger urban–rural disconnect of the
highly unequal Colombian society. Other countries with instances of civilian
autonomy exhibit similar historical cycles of social capital formation in the
shadow of state neglect.
This research adds to the knowledge base on the little-studied period of
Colombian history following the La Violencia period of the 1950s through
the present day, during which time there were attempts to reshape the social
landscape. There are many studies on La Violencia and the later expansion of
guerrilla fronts, paramilitaries, and narco-trafficking. Yet, apart from some
national level histories of this interim period (e.g., Zamosc 1986, Bagley
1989, and Sánchez et al. 2007), there has been little unpacking of the societal
changes and interventions that were designed to prevent a return to conflict.
The prior conflict was so devastating that there was a tendency to wear
historical blinders during this era, to remain in denial about the possibility that
the conflict would again metastasize and instead focus attention on the promise
of progress that lay ahead.
At a national level, this history highlights the more than fifty years of
experience Colombia has in what is known today as community-driven
development (CDD), with the juntas embodying an alternative, decentralized
state-building strategy. Of course, the junta councils and other programs did
not keep conflict from resurging entirely. As a result, some communities organ-
ized under suspicions of their autonomy being political, a fear cast by the
shadow of the early guerrillas’ Independent Republics. They grappled with
how to credibly commit to and signal the apolitical nature of their wartime
autonomy.
This study’s local-level analysis of counterfactual scenarios therefore
answers the deeper question of where and why interventions succeeded or
failed to dampen the later effects of conflict. In the process, it catalogues the
histories of neglected and marginalized populations and the Acción Comunal
social movement. The ground-level case studies reconstruct the social histories
of small communities whose experiences tend to go unrecorded. These histories
whisper that community social capital can be sustained over time against steep
odds and contribute to local order.
Methodological Contributions 305

methodological contributions
Civilian autonomy is a question that begs for methodological rigor and a
careful research design. This study’s attention to case selection and data
collection was indispensible for producing a transparent evaluation of civil-
ians’ strategies. The analysis involved the competitive testing of hypotheses
across a broad spectrum of cases. By migrating to new and more tractable
organizations (like juntas) and standard geographic units (like municipios),
I was able to study cases with divergent values on key variables to construct
plausible counterfactuals. This helped avert possible bias in the selection
of cases.
This analysis of civilian autonomy integrates multiple methods: a quantita-
tive overview of these cases and process-tracing case studies and fieldwork.
These methods took advantage of multiple datasets and levels of analysis,
including the substantial generation of new data. For instance, interviews in
the ATCC zone yielded a unique within-case database of threats and ex-
combatant interviews elicited their views of civilian social movements. The
combination of methods helped deal with the threats to inference of measure-
ment error and bias. Quantitative control variables (in Chapter 5) and multiple
sources of more precise qualitative measurements (in Chapter 8) were suggest-
ive though not conclusive about organized areas being more rather than less
likely to report acts of violence to official sources. These efforts assuage con-
cerns of reporting bias undermining the relationships encountered between
social cohesion, strategies, and violence.
A key part of the research design involves case selection procedures. I used
the quantitative data to select cases for fieldwork and qualitative study with
statistical matching techniques. This set up a quasi-experiment on the juntas for
selected towns in Cundinamarca, accounting for observed and possible unob-
served variables and balancing the concerns of researcher security and mobility.
The matching of cases helped control for difficult-to-measure global shocks to
communities (nonregion-specific factors) to better parse the effects of juntas
from other variables. This can be especially advantageous when confounding
variables are correlated with a key causal variable or data is missing for parts of
the sample. For instance, inequality (equality) washed out the juntas’ effect in
the large-n tests, likely because it also reflects social cohesion – one of the
possible causal pathways through which the juntas are thought to affect vio-
lence. But inequality is not a factor that varied locally among the matched
towns, suggesting that the violence-limiting effects are due to the additional
capacity and strategies of the juntas and not conditions of equality. The
dialogue between quantitative and qualitative methods also helped structure
the data collection, producing more easily interpretable and generalizable
findings. For instance, the histories of the Cundinamarca towns are not
just any histories, but histories with clear expectations about similarities and
differences and where they fit in the global distribution of cases. Similarly, the
306 Conclusions and Policy Implications

ex-combatant perspectives on the selected towns are not random but are also
structured according to differing values of the quantitative data.
The research methods were also helpful for untangling the two-way relation-
ship between civilian populations and armed actors. Conflict can destroy
organization but also spur it on, and organization can then work to limit
conflict. A look at the history of La Violencia and the following years tells that,
on average, juntas are not found in historically more peaceful locations. The
similarities of the quasi-experimental cases of Cundinamarca in their early
patterns of conflict – having all experienced La Violencia – also helped rule
out reverse causality for the effect of juntas and cohesion. The matching of
cases also guided the search for exogenous causes to explain differences in the
junta councils across towns, or what could be called “qualitative instruments.”
In Bituima and Vianí, these included random events such as the collapse of
a church roof, the arrival of an enthusiastic priest, or a particular type of
land reform.
The various methods and tests across the different chapters tended to
reinforce each other. No single method or test provided a complete picture.
Some are better at testing implications of civilian strategies while others are
better at providing an overview of organizations. However, the small contribu-
tions of each approach add up to a larger portrait, or gestalt, of civilian
behavior. The inconsistencies across methods also suggested new conditions,
questions, and unresolved issues, reinforcing this study’s explicit interaction
between theory building and theory testing.
A general and enduring contribution of the research design and methods is
the framework for evaluating civil society and conflict resolution programs
to protect human rights. Civilians’ institutional efforts to solve their prob-
lems were known prior to this study, but there was no good framework to
evaluate them. Perhaps these protective solutions have not yet had a broader
impact on policymaker consciousness because their mechanisms have not been
well articulated or measured, making it hard to show what they do or that they
work. My approach incorporates competing structural explanations for vio-
lence to establish plausible counterfactual scenarios where civilian collective
action is absent. This pushes past simply studying the resilience to bounce back
from harm to identify civilians’ independent and proactive efforts to prevent
violence. This analysis moves in this direction but still necessarily glosses over
the details and richness of the experiences of many communities.

policy implications
This study has a number of policy implications for security in weak states. If
armies in such states were stronger or more legitimate, there would be little
need for civilian organizations and autonomy in conflicts and, for that matter,
no civil wars to begin with. Bringing about grandiose changes in such state
institutions is not an easy task (e.g., if territory could be easily controlled,
Policy Implications 307

resources denied to insurgent groups, or international interventions decisively

mounted, conflicts would already be over). By contrast, the adoption of civil-
ian-oriented policies is relatively more incentive-compatible and the key limita-
tions are usually not interest or resources, but knowledge and capacity. I review
implications for civilians, NGOs, and governments.

Civilian Communities
For communities, this study provides a roadmap to evaluate their particular
situations relative to the experiences and conditions in other towns. For civil-
ians, collective action in wartime continues to be a risky choice. However,
civilians can take steps to strengthen community bonds in advance of the arrival
of conflict. Communication and interaction can build common knowledge and
trust, and have been helpful for assessing the conflict environment and mitigat-
ing uncertainty. They can learn about strategies and gain knowledge about
their effectiveness. The more communities become aware of their options, their
conditions, and their support bases, the more able they will be to make
informed choices and act. Communities should further consider adapting their
strategies to address particular motivations for violence.
Other actors should pay attention to community autonomy efforts since the
“micro” actions of individual communities may spread and have broader
effects. As a by-product of their organizations, civilians can augment
Hobbesian notions of order in “ungoverned” spaces. The junta councils,
for example, provide lessons about local alternative forms of justice and
conflict resolution and the long-term effects of decentralized state-building
programs. Civilian actions may also affect the onset, intensity, and duration
of conflicts, especially if civilian behaviors deter or impede other actors. Some
civilian mechanisms may reduce the potential amount of recruits, informa-
tion, material resources, or combat territory available to both rebel and state
actors, and thus influence how fighting in civil wars unfolds. Civilians’
collective strategies to limit participation in illicit crop cultivation and traf-
ficking can similarly inform counternarcotics programs (e.g., Kaplan 2012).
Structures like juntas also provide alternatives for youths to get involved with
pro-social projects that are bigger than themselves, to promote peace instead
of getting involved in war.
Though this analysis is at the local community level, if resistance to conflict
by civilians can be replicated across localities, the cumulative effect of these
efforts may promote negotiations among state and rebel leaders and solutions
at the macro level of the conflict. Indeed, these insights are potentially widely
applicable, as most civilians would like to avoid conflicts even though they may
not always be able to do so. For instance, some newer rural civilian organiza-
tions in Colombia have suggested they are considering various models for
organizing resistance to armed actors.
308 Conclusions and Policy Implications

The author (right) sharing research findings with the ATCC in La India, Santander,
Colombia, 2013.
Photograph by Oliver Kaplan.

External Actors
External actors such as IOs and NGOs have historically been eager to
support local processes. Civilians, compared to foreign governments and
their interventions or peacekeeping forces, have strong interests in conflict
resolution (and little patience for top-down negotiations) but may lack
adequate resources or knowledge which international actors may be eager
to supply. Past outside interventions such as the Peace Corps and Acción
Comunal programs have broadly reshaped the social landscape. In Colombia
alone, the UN’s Redes program, USAID (for example, the ADAM and CSDI
programs as part of Plan Colombia funding), the European Union’s Peace
Laboratories, Germany’s GTZ, Switzerland’s Suippcol program, and World
Bank analyses (2000, 2004) are all predicated on the hypothesis that social
cohesion and local capacity can effectively protect civilians and promote
peace. The Colombian government’s efforts include the Social Action Agency’s
Peace and Development Program, the National Peace Prize, and alternative
justice processes such as casas de justicia (houses of justice), among other
programs. International NGOs such as Peace Brigades International, Fellowship
Policy Implications 309

of Reconciliation, and many Colombian NGOs accompany communities that

are pressured by armed groups.
These programs hold general notions about how “social capital” promotes
“peace,” but they are not always explicit in specifying protection mechanisms
or aware of which mechanisms are operating.2 The careful measurements in
this analysis help clarify which activities may work to reduce violence, and
when. A first promising way communities can be assisted is through the provi-
sion of alternative funding sources for community projects to circumvent
clientelism. Second, external actors can help transport general knowledge of
“best practices” and detailed knowledge of how certain mechanisms function.
For instance, the capacity, procedures, and information flows needed for the
ATCC investigation system to function can be digested and explained to other
communities. When tailoring programs to communities, it will be helpful to
assess local armed group dynamics and preferences.
The involvement of international actors must also be considered with cau-
tion. Although international development programs can bring various benefits,
they are not without controversy (e.g., Easterly 2007). INGOs can have positive
human security effects, but only if they are not motivated by “rent-seeking”
(Murdie 2014). A particular challenge for supporting civilian autonomy move-
ments is that external aid projects can be perceived by nonstate armed actors as
part of a counterinsurgency strategy. In these circumstances, the implementa-
tion of aid projects can stigmatize communities and make them targets, leading
to more, rather than less, violence (Crost et al. 2014). Depending on the
national political dynamics, democracy promotion projects may incur similar
risks. International actors must also be aware of the incentives their actions
create and the messages they may relay (either advertently or inadvertently) to
civilians and activists through their engagement. They must be wary that
encouraging civilian autonomy could create a moral hazard if it leads civilians
to confront armed actors or remain in situations where they may have low odds
of success. For these reasons, it may be wise to work to unite communities
preventively, prior to the onset of conflict, rather than in the thick of the fight.

Counterinsurgents and Warfighters

Lastly, this research pertains to government counterinsurgency programs.
Governments face the problem of how to implement public policies and provide
security in regions where they have little presence. Protecting the popula-
tion is the ethical choice and also central to gaining civilian support in

2
The UNDP 2003 study recommended the following policies: “i) Systematization of experiences;
ii) strategies of social communication; iii) training of moderators and mediators; iv) training of
local leaders . . . ; and v) execution of agreements with organizations that give credit to experiences
along the lines already mentioned.”
310 Conclusions and Policy Implications

counterinsurgency (Sewall et al. 2007), and armies have improved at gauging

bases of support and addressing civilians’ concerns to “win hearts and minds.”
Yet as military strategies of counterinsurgency increasingly result in stalemates,
states may look for new ways to deal with local security risks without alienating
civilians. The debate today among counterinsurgency specialists over the merits
of decentralization to extend government reach is a repeat of Colombia’s
experiment that began over fifty years ago. Autonomy and its decentralization
(or delegation) of power can be perceived as hostile to governments, though it
does not have to be. Insights from civilian autonomy theory point to how
government operations can protect civilians without abandoning their goals.
Before considering the implications that civilian autonomy theory holds for
counterinsurgency, it is important to address whether there is a place for them
in a book that puts nonviolent civilian agency at the forefront. This is a delicate
issue, especially because the goal of civilian autonomy is to deescalate war, and
distilling implications for warfighters could possibly do the opposite. I never-
theless believe that, to promote respect of autonomous communities by armed
actors, it is appropriate and necessary to address these implications here. This is
not to say, however, that this text encourages counterinsurgency campaigns or
any kind of belligerency. This project was forged out of the ashes of the
Afghanistan and Iraq wars – two wars in which armed actors (including
international, state, and nonstate forces) had poor understandings of the pref-
erences and social dynamics of civilian communities, and civilians suffered
extensive harm as a result. Unfortunately, the Colombian conflict tells a similar
story, but it also offers an opportunity to study and learn about these dynamics.
This discussion recognizes that counterinsurgency campaigns have been a fact
of modern warfare (and also historical warfare). Even so, they can still be
informed by new and more humane approaches to resolving conflicts, and
civilian autonomy may represent a nonviolent alternative to insurgency and
counterinsurgency.
If counterinsurgents cannot control an area, they should be wary of insuffi-
cient half measures that may end up putting local institutions in greater danger.
Moving intermittently in and out of zones is tempting when resources are
limited, but it entraps and entangles localities without being able to protect
them, which is not good for anybody. If communities are punished for this by
insurgents, some portion of the blame may ultimately fall on government forces
as well (see Oxfam 2009).3 Arming communities may contribute to security in

3
Similar critiques have also been raised by a consortium of NGOs, including Oxfam, about the
United States’ proposed Afghanistan Social Outreach Program (ASOP) to arm villagers to provide
for their own protection, claiming that the effort may not be sustained and may invite retaliation
against communities. The still-young experience of Canadian forces with development projects in
the Afghan village of Deh-e-Bagh suggests that international forces should embrace towns that
want to reject extremists, but only if their protection can be assured (Pearson 2009).
Policy Implications 311

some cases, but can ultimately be risky if civilians are not sufficiently powerful
or accountable or do not have adequate support, making them open targets
(e.g., Khan 2009).
Counterinsurgents should consider how to work with local institutions and
allow them to function and police themselves as a way to avoid heavy-handed
measures that could stigmatize them or cause unintended damage and push
civilians toward the enemy (Jaffe 2009).4 This includes sensitivity with detain-
ees and suspects. Samuel Huntingon highlighted the benefits of such an
approach in his assessment of U.S. military strategy in Vietnam in 1967,
observing that, “Efforts to arrive at such political accommodations [over
communities] with the VC [Vietcong] are preferable to intensification of the
war in the Delta” (Huntington 1967).
It becomes the responsibility of the state to provide sufficient security guar-
antees to win the allegiance of communities. Many accounts of civil conflict
observe that most civilians are not extremists and instead largely seek to survive
periods of instability and protect their livelihoods (e.g., Kriger 1992, Nord-
strom 1992). If international forces are sufficiently cautious and eventually
make good appeals to local communities, they have better chances of winning
or discouraging them from supporting insurgents. This logic is consistent with
new principles of restraint articulated by Admiral Mullen and the Powell
Doctrine’s decisive use of force when force is to be used (Shanker 2010; in
other words, pursue counterterrorism unless counterinsurgency can be compre-
hensively implemented). These kinds of community-based alternatives to mili-
tary strategies may allow government forces to operate with a smaller footprint
or redeploy to more vulnerable areas. The ramifications for international rela-
tions are clear since the United States has given over $10 billion in foreign aid to
Colombia alone over more than a decade under Plan Colombia, primarily for
military equipment and training.
Planning ahead, we should think about how to strengthen communities
for both current conflict settings and for future peace and stability. However,
while counterinsurgency strategists have become concerned with understanding
the “human terrain,” in many cases, the deep analysis of communities is still
sorely missing (McChrystal 2009). There is still a poor understanding of what
civilians want, how their communities are organized, and how their institutions
function. My ethnographies from Colombia and elsewhere suggest, however,
that armed actors are capable of learning how to respect and interact product-
ively and accountably with autonomous communities. Indeed, shifting the
focus from military to civilian alternatives can provide payoffs, including in
areas such as supporting local cease-fires, reducing illicit crop cultivation, and

4
Indeed, the United States appears to have experimented with this concept in areas it cannot
completely control during a pullback of forces in Afghanistan from remote villages to more
populated areas. Commanders appealed to village elders and even an insurgent leader to develop
a local security plan for after U.S. forces withdraw from the village of Damesh in Nurestan.
312 Conclusions and Policy Implications

fostering development. Since communities may be targeted if they are seen as

too political, development aid and capacity building should be done cautiously
so as not to make local institutions appear to insurgents as part of the oppos-
ition. Local and national elites can play crucial roles to either help or hinder the
development of horizontal social relations in local communities.

a research agenda on civilians in war

This section maps out remaining challenges for studying civilian behavior
and conflict outcomes and outlines a research agenda. While I found some
support for civilian autonomy theory, the research agenda to understand the
nuances of civilian behavior in different conflict settings is wide open. Because
this study touches on many interrelated topics, it points to many new questions
in diverse areas.
First, there is much room to improve the specification and measurement of
civilians’ strategies for protection and their alignments and preferences vis-à-vis
armed groups. Civilian processes of strategy selection (as well as armed groups’
responses) are complex and hinge on information, expectations, and social and
cognitive processes. My theoretical treatment of strategy here is a simplifica-
tion, and there are many facets that could be examined further.
In the realm of the measurement of strategies, several additional techniques
may become feasible as more data is collected. First, additional case studies can
provide measures of organizations and strategies across additional units, and
careful village comparisons can contribute to an accumulation of findings.
Second, survey questions could be helpful for identifying civilians’ preferences
and decisions, if truthful responses can be elicited. Third, data on observational
measures of political preferences can help separate the influence of organization
and cohesion from political leanings. Data on protests could also be used to at
least test revealed preferences and visible strategies of alignment and autonomy.
In sum, there is a need for deeper analysis of how additional autonomy mecha-
nisms function and influence armed actors beyond the few analyzed herein.
There may yet be strategies that remain unknown to the broader world and
have not been systematized.
Second, there are many unknowns about civilians’ decisions to use violent
versus nonviolent strategies. Why do civilians tip from pacifism to arming or
back from arming to nonviolence? When do armed efforts succeed and when
do they fail? What are armed groups’ views of civilians under arms? When and
why do armed civilians ally with macro-actors?
Third, there are many remaining questions about the capacity of the junta
councils and a need to better disentangle the mechanisms through which such
organizations can affect violence. This is a complex issue because there are
many concurrent pathways and limited available data. For instance, levels of
funds and resources, projects, social cohesion, existing social capital, political
alignments, and clientelism are all ways in which the juntas might themselves
A Research Agenda on Civilians in War 313

affect outcomes of interest or stand in for other factors that do. New insights
about these factors should inform whether the juntas can be viewed as a
“treatment” in experimental terms or whether societal patterns are path
dependent and hard to shape (external stimuli appear to effectively promote
community cooperation in at least some post-conflict circumstances; see Fearon
et al. 2009 on Liberia). This will also inform how external organizations can
best aid communities, including further probing whether bolstering juntas
enhances security.
More broadly, the challenges of researching the junta councils show the
need for capacity to collect and maintain data at the micro level on the
social landscapes and organizations of developing countries. Data collection
will be most useful if it samples the full spectrum of units of analysis – both
“treated” and “untreated,” organized and unorganized communities. Addi-
tionally, repeated measurement over time will help account for unit-specific,
time-invariant characteristics such as social histories or preexisting capacities.
Lastly, research could benefit from further clarification about possible reporting
biases since the availability of information may depend on both the levels of
community organization and levels of security.
Fourth, social capital and the internal dynamics of peaceful civilian organ-
izations merit further study. How do civilians mobilize and sustain collective
action and participation in violent environments? Are selective incentives or
other appeals most common? How do people coordinate and communicate?
How can they be supported? What are the limits of organizations? Why do
breakdowns and defections occur? What are the day-to-day management tasks
and strategies that leaders of these organizations use to both meet their goals
and protect themselves in the process?
Fifth, this research represents only a first look into the reverse causal rela-
tionship of the effect of armed conflict on social organizations – when conflict
weakens or stimulates social organization. This points to open questions about
the sequencing of the creation of councils: Can they be formed in the midst
of conflict to survive and aid in protection, or must they be established
prior to conflict as a preventative measure? Do they require sustained support
(e.g., by the government or external actors) under threat, or can they survive on
their own?
Sixth, there is much more to learn about how armed groups view and
respond to civilians. Why do some armed groups at times set up civilian
councils and rely upon them but other times do not? If armed groups can easily
create civilian organizations, what does this imply for the importance of pre-
existing civilian structures? What determines when armed groups can co-opt
civilian organizations? In Colombia, there are still questions about why
paramilitary groups vary in their attitudes and strategies toward the junta
councils, as some were respectful while others were much more brutal. There
is also more to learn about the conditions and types of groups that will be
influenced by civilians. For instance, what guarantees (or consequences) will
314 Conclusions and Policy Implications

move armed groups to not target organizations? This calls for more extensive
study of the different types of armed groups and their sensitivities and motives
than what were found in the case studies. As seen here, one productive avenue
for studying these questions is to collect information from ex-combatants.

final thoughts
Looking ahead, what are the prospects for Colombia and for autonomy move-
ments in the near future? What are the prospects for civilian autonomy around
the world? With the help of the initiatives of Plan Colombia, the conflict in
Colombia has abated and state presence has expanded to reach many more
towns, albeit at a nonnegligible human cost. Paramilitary groups have been
disbanded and rebel groups have been weakened and repelled. Even as fighting
continued, the government negotiated and signed a peace agreement with the
FARC in 2016 and announced talks with the ELN, putting peace within reach.
Like the years after La Violencia, the present conditions would seem an oppor-
tune moment to rebuild civil society and consolidate accountable state presence
and security. However, while violence against civilian populations affects fewer
communities, it still continues.
Emergent armed “criminal bands” (BACRIM) have spread to over 150
municipios and narco-trafficking, though slightly diminished, remains a con-
cern (MAPP/OEA 2010; Corporación Arco Iris 2008 estimates criminal bands
are present in over 250 municipios, or one-quarter of the country). These
emergent groups are less organized and centralized, less political, and more
criminal than other armed groups. These characteristics may make them
depend relatively less on the population and harder to negotiate with. This
may mean civilians have less leverage with them to gain protection than they
have had with other armed groups. But it may also mean that civilians may be
able to avoid transgressions by minimizing entanglements with these groups
and staying out of their way. According to civilian autonomy theory, commu-
nity management strategies would seem to be most useful, though other strat-
egies may still be effective in some circumstances.
In a wider, global view, while civilian autonomy may not broadly occur in all
conflicts, existing examples may reflect the tip of an iceberg. A lesson of the
juntas is that there may be the potential for civilian autonomy and protection
rooted in underlying social processes, perhaps not everywhere, but at least in
more countries and communities than commonly believed. Indeed, new cases of
civilian autonomy (at least in its primordial form) continue to be discovered,
even in contexts where they are least expected (e.g., Libya, Syria, Mali, Iraq,
and Ukraine).
As much as organizational strengthening is a key catalyst for civilian protec-
tion, careful analysis and understanding of examples is also necessary for
awakening cooperation. This was seen in a number of my case studies, where
civilians were not able to stand up for themselves due to a lack of reassurance
Final Thoughts 315

and a meager understanding of organizing. Armed with information and ideas,

communities can see what is possible, which limitations have been encountered
in the past, and how to assess their own situations and capabilities. This speaks
to the importance of sharing best practices, which some networks of com-
munities are already doing, both within Colombia and internationally. To
minimize harm (Anderson 1999) in acting collectively to deal with threats,
communities and their supporters must carefully assess the security situation
and organizational capacity, and gain the consent of residents, which all require
good local information networks. This research can be informative for these
tasks with its mapping of community history, protection mechanisms, context-
ual conditions, and armed group psychology.
I sought to answer questions about civilian autonomy, but the reality is that
conflict conditions, communities, and available resources to manage conflict
differ from one place to another and are ever evolving. So, more importantly,
I asked difficult questions and put forward an approach for seeking answers.
With interest in community-driven development, counterinsurgency, and, now,
“countering violent extremism” on the rise again, the topics of civilian behavior
and violence are also growing in attention and importance. As this conversation
continues, we should look to civilians as a source of peace.
Appendix A

Archives Consulted

ATCC Community Archives, La India, Santander, Colombia

Ministry of Interior Archives, Bogotá, Colombia
Solimon Santos Personal Archives, Manila, the Philippines
The National Security Archive, Colombia Project, Washington, D.C. http://
nsarchive.gwu.edu/colombia/

317
Appendix B

Supplementary Documentation on the ATCC

a note on variables and “cases”

The complex explanatory framework in Chapter 7 runs the risk of the number
of independent variables exceeding the number of cases (the “n-k problem”).
Under these circumstances, there is not ample variation to test and rule in (or
out) different factors. I argue that the ATCC’s investigation institution inter-
acts with civilian “pacifist” norms to limit violence against civilians. This
arrangement is further primarily supported when a third condition is present –
that armed actor preferences are not excessively hostile to the institution’s
existence.1 When variables representing alternative hypotheses such as Mili-
tary Balance of Control and Armed Actor Resources are considered, even
more cases are required for testing.
As a solution, I tap three kinds of variation within the ATCC meta-case.
First, I look at temporal variation in the dependent variable of violence against
civilians over time. In the ATCC case, the changes in violence can be roughly
classified into the three different time periods. Second, I examine cross-sectional
variation in violence between the ATCC region and communities in neighbor-
ing regions (lumped together). These cases are relevant “controls” since they
likely share similar characteristics, including armed group fronts, geography,
and perhaps demography.2 Third, I consider counterfactual worlds as

1
This presumes the counterfactual argument that if armed actors were more hostile than they
actually were toward civilians, the institutional arrangement would have been unstable. This
claim can be assessed by asking civilians and ex-combatants why the armed group agreed to and
benefited by the civilian arrangement.
2
I do not delve into counterfactuals for neighboring cases since they are not interesting or
analyzable – one can only assume that if they had a civilian process, civilians would have been
more protected (although the “thought experiment” of these counterfactuals makes one wonder
why civilian institutions did not arise there). This would however add an additional three “cases.”

319
320 Supplementary Documentation on the ATCC

additional cases (Fearon 1996). In theory, counterfactual cases can be derived

for each variable-time-period (e.g., what would have happened in case c
during year y if independent variable x had been present, or not present?).
The Table B.1 presents the various relevant factual and counterfactual cases.3
I identify variation across eleven real and counterfactual sub-cases within the
ATCC meta-case, which should help add confidence to civilian autonomy
theory with more cases than the seven independent variables (counterfactuals
are highlighted).

qualitative discussion of additional explanations

for trends in violence
This section assesses how various extant explanations account for the observed
trends in violence over time. Specifically, it seeks to account for the surprising
era of the absence of violence during the 1990s. Beyond shifts in territorial
control, these explanations are: general (national) trends in violence and peace
overtures, changing rebel organization and discipline, and increased inter-
national support (see Table 7.1). I assess these hypotheses with various “cases”
within the ATCC meta-case.
General Trends in Violence and Negotiations? One explanation for a reduc-
tion in violence in the Carare region during the 1990s could be that either the
country or the wider Magdalena Medio region saw general trends toward less
violence against civilians in the conflict. This could be due to some kinds of
macro-level changes in rebel strategies, government counterinsurgency strategy,
or the initiation of macro-level peace negotiations or ceasefires unrelated to the
local decisions made by the ATCC or their armed actor counterparts. However,
none of these explanations fully account for the sustained reduction of violence.
Although violence decreased nation- and region-wide at some points in the
1990s, it fluctuated during this period. National homicide rate data (Figure 3.1)
show that violence against civilians was generally stable throughout the 1990s
until it began to surge even higher in 1998. Kline (2003) reaches a similar
conclusion. This trend likely corresponds with the regrouping of the paramili-
tary forces into the AUC and ACCU umbrella organizations in late 1990s.
Recorded trends in violence from the Magdalena Medio region and the
department of Santander also do not correspond with the ATCC’s history.
Statistics from the Colombian government show that the Magdalena Medio
region and the department of Santander were no less violent and civilian
deaths were steady (or even increasing) throughout the 1990s. This is further
confirmed by CINEP data, which uses an alternate source for counts of

3
I consider cases as “relevant” according to whether the variation is interesting. Cases with values on
the military control variable that predict a low-threat environment are not “interesting” because it
becomes impossible to distinguish any independent effect of the civilian process since the values of
both variables make the same prediction of low violence – a case of “equifinality.”
Supplementary Documentation on the ATCC 321

table b.1 Real and counterfactual subcases

Period/Case ATCC Control Outcome Observed Reality or

(institution, variable prediction outcome4 counterfactual
norms) value*
1975–1991/ Absent Contested/ Violence Violence Reality
ATCC dominant
1975–1991/ Absent Contested/ Violence Violence Reality
Neighbors dominant
1975–1991/ Present5 Contested/ Low violence Counterfactual
ATCC dominant
1991–2000/ Present Contested/ Low violence Low Reality
ATCC dominant violence
1991–2000/ Absent Complete/ Violence Violence Reality
Neighbors dominant
1991–2000/ Present (weak Contested/ Violence against Counterfactual
ATCC norms) dominant opportunists
1991–2000/ Absent Contested/ Violence Counterfactual
ATCC dominant
2000–2007/ Present Contested/ Violence against Rising Reality
ATCC (degrading dominant opportunists violence
norms)
2000–2007/ Absent Contested/ Violence Violence Reality
Neighbors dominant
2000–2007/ Present (strong Contested/ Low violence Counterfactual
ATCC norms) dominant
2000–2007/ Absent Contested/ Violence Counterfactual
ATCC dominant
Present = norms and institutions present
Present (weak norms) = institutions present but weak pacifist norms among residents
Absent = institutions and norms absent
*Military balance is estimated by period, even though there are more fined-grained variations of
balance within periods. Predictions of violence in each period are presented as helpful
simplifications.

violence (OPI 2005; Holmes et al. 2007). AUC paramilitary violence was
notably ruthless in the neighborhoods of the regional capital of the oil town

4
Variation in the dependent variable of violence can further be disaggregated according to variation
in the “pacifist norms” of the ATCC process. Although not all potential counterfactual cases are
discussed, this variable accounts for differences in violence against pacifist-norm-abiding (ATCC
member) civilians versus violence against nonnorm-abiding civilians in the region. The investigation
institution theoretically only provides a protective benefit to civilians who abide by pacifist norms of
noninvolvement in the armed conflict (not participating in the coca economy, for example). In cases
where norms are weak, the institution will protect civilians that do not participate in conflict
activities but not those that do, leading to low to moderate levels of violence.
5
This could be further disaggregated into whether or not norms are present in addition to the
investigatory institution, but I desist since it is not an “interesting” comparison.
322 Supplementary Documentation on the ATCC

of Barrancabermeja. According to García, in the first half of 1989, there were

seven massacres within 100 kilometers of La India (264), while the ATCC area
remained a relative oasis of peace. An interview subject also told of a massacre
of six people found in a mass grave in the nearby town of Puerto Pinzón, in the
department of Boyacá to the south, around 1996 (the case was under investi-
gation by the Attorney General’s office). Other ATCC residents who migrated
from neighboring regions also testify that violence continued in their towns of
origins, and some residents had family members who were victims.
Peace negotiations are fairly ubiquitous in recent Colombian history, but
they do not seem to have been effective at tempering the conflict (García Durán
2005, 2006).6 Although the government negotiated with the FARC and estab-
lished the demilitarized zone in the Macarena region, this did not occur until
1998 and does not explain reductions in violence that occurred prior to that
date. Overall, decisions at higher levels of politics do not seem to explain what
occurred locally in the ATCC.
Resources of rebels. Weinstein (2006) suggests that rebel organization can
explain abusive behavior against civilians. The resources available to rebel
groups are seen as a main determinant of organizational structure. Since
resource-rich groups attract economically motivated recruits and do not depend
as much on civilians for support, they are more likely to have more abusive
soldiers and be more lax in disciplining bad behavior (allowing soldiers “pil-
lage” rewards). This theory would predict civilian violence to decrease when
either illegal armed groups lose resources or become better organized.
In the case of the Carare region near Cimitarra, the level of resources
available to illegal armed groups did not decline during the 1990s, and, if
anything, actually increased. The Magdalena Medio region, including its south-
ern edge, is known as the most oil-rich part of the country. Oil rackets and
siphoning from oil pipelines by rebel and paramilitary groups was fairly
common, according to interview accounts. Further, although Colombia is not
known to have many diamond deposits, it does have emeralds, and emeralds
and the emerald are trade prevalent in Magdalena Medio and the Carare region
(Hernández Delgado 2004). As some of the campesinos have said, emeralds
could be spotted with the naked eye in the riverbeds.7 There are several reports
of rebel and paramilitary groups emerging as protection rackets for

6
As García Durán characterizes the period from 1993 to 1999, “After the failure of the peace
process with the largest guerrilla groups in Caracas and Tlaxcala [in Gaviria’s term], the armed
forces . . . declared ‘total war’ on all guerrilla groups. Ironically, this produced only mediocre
results and left the guerrillas militarily stronger than ever. . . . Although the Samper government
(1994–1998) . . . did try to construct a new model for negotiation with the guerrilla groups, all of
its efforts were torpedoed by the weight of the political crisis produced by the investigation of the
Attorney General’s Office regarding the receipt of funds from drug trafficking in the election
campaign which brought Ernesto Samper to the Presidency.” Kline (2003) also notes that
negotiations occurred throughout the 1990s, but to no avail.
7
ATCC#24, La India, 10/2007.
Supplementary Documentation on the ATCC 323

esmeralderos, or emerald prospectors, as well as for local wealthy ganadero

cattle ranchers (UNDP 2003). However, in the analysis of an interview
respondent who was knowledgeable about the emerald trade in the region,
while armed groups certainly took advantage of emerald rents when they could,
it was not the principal source of their financing (and in fact, the FARC’s 23rd
Front may have been relatively resource-poor). Instead, a greater part of the
guerrilla groups’ operations was supported by resource transfers from other
nearby FARC fronts in the department of Bolívar to the north, who were
involved in gold mining in the Serranía de San Lucas mountain range, or by
kidnapping.
In addition to oil and emeralds, the coca crop from which cocaine is made
provided another source of income for armed groups. Many of the paramilitary
groups have their origins in protecting drug traffickers (groups that emerged
in Cimitarra were a union of drug traffickers and esmeralderos; Observatorio
del Programa Presidencial de Derechos Humanos y DIH 2001: 6). Although
there are reports of coca cultivation from the early 1990s in the region, coca
production did not flourish until the end of the decade (interviews; Echandía
1999). However, even if little coca were produced in Santander in the 1990s,
the FARC central committee collected coca profits and distributed the funds to
its regional fronts.8
Surrounded by coca fields, emeralds, and oil, resources were certainly avail-
able to the armed groups and evidence suggests that, as time progressed, the
rebel groups in Colombia only became more economically motivated, rather
than less (e.g., a “war system”; Richani 2002). Since the theorized connection
between resources and organization implies that violence should increase over
time as groups’ resource bases increase, resources per se do not convincingly
explain the low levels of violence and abuse towards civilians during the late
1980s and 1990s (but are perhaps consistent with the abuse of civilians
observed in the 1980s). The theory is further undermined when considering
that, according to Weinstein, once groups tend toward opportunism and abuse,
they are set on a path-dependent process and their behaviors become difficult to
reverse. Despite stable or increasing resources, the organization of illegal armed
groups in the area generally improved over time.
The local FARC fronts reformed by supposedly purging rogue commanders
who committed abuses. This was not done out of pure good-heartedness, but
rather in response to demands made in initial meetings with the ATCC. The
FARC realized it was losing the support of civilians, and in response to the
ATCC’s entreaty, the FARC commander apologized for the past behavior and
said he would hold abusive commanders accountable. Braulio Herrera, a leader
in the political party associated with the FARC, the Union Patriótica, came to the
region as part of the dialogue with civilians to regain support. He ordered that

8
ATCC#3, La India, 8/2008.
324 Supplementary Documentation on the ATCC

abusive commanders be eliminated, and it was later reported that perhaps as

many as seventy-eight men were “executed” or “thrown in the river,” never to be
seen again (interview; García 1996, 207, 269; Restrepo 2005, 2006). This
process may have played a role in the FARC’s improved treatment of civilians.
Yet, it is not clear that this purge would have taken place without the initiation of
dialogue by and pressure from the civilian leaders.
The increasingly dominant paramilitary groups in the region appear to have
also improved their organization during this period, though not until the end of
the decade.9 In 1997, the various independent paramilitary groups merged to
form the AUC (Autodefensas Unidas de Colombia) umbrella organization.
Reports suggest that as a result of this merge, they became more effective and
coordinated. But, interviews and statistics indicate that the paramilitaries were
responsible for a larger portion of the threats than the guerrillas during this
period. Paramilitary organizational consolidation did not coincide with the
earlier decrease in violence seen by the ATCC.
Patterns of violence against ATCC civilians do not appear to be solely a
function of armed actor resources and organizational structures. First, reforms
did not coincide with changing resource endowments or reductions in violence.
Second, although abuses, torture, and impertinent threats did occur in the
region, much of the violence and denunciations were also characterized as
purposive and coercive, and not a result of poor troop discipline. Third, some
of the organizational reforms resulted from civilian pressures.
International Support? National and international actors seeking to support
the ATCC could have raised the reputational costs to armed actors of trans-
gressing against the ATCC by raising the organization’s profile. However, this
is not a likely explanation for the armed actors’ apparent benign treatment of
the ATCC since it only gradually gained international attention.
The ATCC did garner some early attention in the country shortly after the
time of its founding, including by winning the Alternative Peace Prize in 1991,
which generated some international press (e.g., Yarbro 1990). But as one of the
first movements of its kind, it did not receive the kind of public profile and
attention that other similar civilian communities receive today (especially given
the early state of the internet for communication at that time). There is little
evidence that either international governmental organizations (IGOs, such as
the UN) or nongovernmental organizations (NGOs) were involved in the
organization’s founding (it arose indigenously) or with any kind of human
rights monitoring programs or “accompaniment,” such as those of Peace
Brigades International and Fellowship of Reconciliation (e.g., Mahony and
Eguren 1997).10 The ATCC has only had part-time Peace Brigades

9
Commander Botalón’s rise in 1994 may have improved paramilitary control and discipline.
10
Nor is there much evidence of any kind of “boomerang effect,” where NGOs lobbied foreign
governments to put pressure on the Colombian government to protect the ATCC (Keck and
Sikkink 1998).
Supplementary Documentation on the ATCC 325

figure b.1 Communication from the paramilitaries to the ATCC.

326 Supplementary Documentation on the ATCC

figure b.2 Communication from the FARC to the ATCC.

accompaniment since 2000. The Colombian government’s “peace ambas-

sador” lent only tepid support to the process (García 1996).11 In addition,
much of the positive press garnered by the ATCC came after its long
period without violence. The ATCC also tempered its public stance toward
the armed groups in a way that limited international attention to abuses. It
has had a policy of not implicating specific armed groups in their public

11
Rafael Pardo, the Presidential Peace Advisor, said, “The national government is not interested in
regional peace accords . . . because . . . they cause the guerrillas and violence to displace from one
region to another . . . but he also offered ‘all the institutional support possible’” (García
1996, 283).
Supplementary Documentation on the ATCC 327

figure b.3 Warning from the FARC to the ATCC.

denunciations of threats or human rights violations, only referring to the

culprits as “the enemies of peace.”
In sum, most accounts would have otherwise predicted that, during the
1990s, violence against civilians would have prevailed just as it had in the
preceding fifteen years.
328 Supplementary Documentation on the ATCC

figure b.4 ATCC authorization to seek dialogues with the FARC.

“We the signers below, adult members of the region of La India, area of influence
of the ATCC, authorize _____ to organize a commission to look into the possibility
that the FARC groups found in our area will allow us to dialogue with them.”
La India, 2004.
Supplementary Documentation on the ATCC 329

Verbatim transcript of meeting between the ATCC leader, the junta

leaders of San Tropel and Santa Rosa (on border of the ATCC),
and a Paramilitary Subcommander, 2001, in Santa Rosa,
Cimitarra (from the ATCC archive).

atcc leader : The association is always willing to help people of a good heart [but] . . .
there was some misinformation because they [some paramilitaries] said that the leaders of
the ATCC had come to supposedly create an ATCC here in Santa Rosa and regarding this
there was a disagreement in the high command and that the commander stated he would
not permit that they create an ATCC in Santa Rosa.
paramilitary subcommander : If tomorrow we come to spread terror, the community
will quickly be the one to react (form opinions) and organize itself and say that, “We don’t
want your presence.” I’ve spoken with some members of this community and they have
told why the ATCC before didn’t take Santa Rosa into account and why the ATCC does so
now, so I want you to explain this to me.
santa rosa junta leader : We can’t blame the ATCC leaders for this. . . . Unfortunately
here in Santa Rosa we’ve been a totally un-united community and while we are not united
and organized we won’t get anywhere.
atcc leader : Every armed group is jealous with the people. This region is delicate because
here there does not exist the total domination of any of the armed groups and there exist
split territories; but there is a dividing line and it’s the association . . .
santa rosa junta leader : Our problem is, if we’re going to have the Autodefensa here,
then we’re going to continue with the comment that Santa Rosa is land of the paras; so we
should look for another horizon, because tomorrow they’ll have us marked (tildados) as
paramilitaries and before long the guerrillas will come and they’ll take-out some innocent
people. So, I want to keep looking for a way to keep living at the margin of whichever
group. By that I don’t mean that I’m against the paras or the guerrillas.
atcc leader : If this war were directed at the guilty, there wouldn’t be problems and we
wouldn’t have to worry. . . . If it were really a directed war, more innocent people would not die.
paramilitary subcommander : If all the community said that they didn’t want our
presence here we would respect the decision.
san tropel junta leader : We’ve had a lot of contact with the ATCC and the
paramilitaries. I want to tell (the community) that there are many times that we are afraid
to express ourselves before the commander for the fear of what they’ll say after the meeting.
I want to tell you that they have changed their philosophy a little and have ended their
assault a little against the community . . . For us it hasn’t gone very well, since we’re 100%
dominated by the Autodefensas, whereas the situation is different here (in Santa Rosa).
santa rosa woman : Will you (paras) still be here regardless of whether the community
wants you here?
paramilitary subcommander : Yes, we’re going to be here whether they want us
or not.
santa rosa woman : And if the community organizes itself and says no?
paramilitary subcommander : Regarding this it would have to be that the entire
community decides it.
santa rosa woman : So it’s the community that decides it.
paramilitary subcommander : But if there are only two or three people that don’t want
our presence, then we’ll continue to be here.
330 Supplementary Documentation on the ATCC

figure b.5a ATCC Membership Document circa 2007.

Supplementary Documentation on the ATCC 331

figure b.5b ATCC Membership Document.

Glossary

This study uses a specialized set of Colombian vocabulary in Spanish and

English, as well as acronyms related to Colombia’s armed conflict and social
setting.
Abigeato Cattle theft.
ACCU Autodefensas Campesinas de Córdoba y Urabá
(United Self-Defense Forces of Córdoba and Urabá); a
private army mobilized by the Castaño family to
combat the FARC in northwest Colombia, which later
grew into the United Self-Defense Forces of
Colombia (AUC).
ACR Agencia Colombiana para la Reintegración
(Colombian Agency for Reintegration); the
Colombian government entity, created in 2011
(previously the High Advisory for Reintegration), that
oversees efforts to reintegrate demobilized fighters
into Colombian society.
ANUC Asociación Nacional de Usuarios Campesinos de
Colombia (National Association of Peasants);
community land reform councils in rural areas.
ATCC Asociación de Trabajadores Campesinos del Carare
(Peasant Workers Association of the Carare River); a
civilian community organization in the department of
Santander that created its own mechanisms for
mediating and mitigating the effects of the armed conflict.
AUC/Autodefensas United Self-Defense Forces of Colombia; a
Unidas de paramilitary umbrella organization formed in 1997 to
Colombia consolidate the country’s disparate paramilitary

333
334 Glossary

groups. At its peak, the AUC had an estimated 31,000

members. The organization demobilized from 2003
to 2006.
Autonomy Self-rule; the goal of seeking independence from
armed actors as a response to conditions of civil
conflict and state absence. In conflict settings, de
jure civilian autonomy involves independence in
decision-making. De facto civilian autonomy
reflects freedom from threats or violence achieved
through the implementation of community autonomy
strategies.
BACRIM/Bandas Criminal bands or neo-paramilitary groups that
Emergentes/Aguilas appeared in the post-paramilitary demobilization era.
Negras
Bazaar Party (fiesta) or fair, usually to raise funds for
community needs or projects.
Berraco Tough, or a tough guy.
Bestia A mule, used for transportation.
Cabecera The county seat of a municipio (urban center).
Cabildo The political council that governs an Indigenous
community.
Campesino A farmer or peasant; usually a mestizo of mixed
Spanish and Indigenous descent.
Chévere, Bacano Awesome, great.
Chisme Gossip; unreliable information circulated within
communities or shared with armed actors as part of
false denunciations.
Chusma/contra- Bandit groups during La Violencia.
chusma
Coca The plant from which cocaine is derived.
Cocalero A social protest movement led by coca growers
Movement in the Amazon region of Colombia, primarily the
department of Putumayo, that received support from
the FARC and negotiated for benefits with the
Colombian government.
Comité de The local committee of coffee growers in a
Cafeteros municipality.
Consejo The governing political council of an Afro-Colombian
Comunitario community.
Convite/Minga Community service and shared labor; the act of
working toward a common goal (“minga” is used
among Indigenous groups).
Convivencia Peace; coexistence; harmony.
Glossary 335

Corregimiento A subsection of a municipio with a small urban center;

larger than a village but smaller than the county seat.
DANE Colombian Census Bureau.
DAS Departamento Administrativo de Seguridad
(Administrative Department of Security); Colombia’s
former intelligence agency, which was dissolved in
2011 following a series of scandals, including evidence
of links between the agency and paramilitary groups
and illicit surveillance of leftist politicians, judges,
journalists, human rights defenders, and other
civilians.
DIGIDEC Dirección General de Integración y Desarrollo de la
Comunidad (General Directorate of Community
Development and Integration); the bureau of the
Ministry of Government that managed the junta
council portfolio; the office was absorbed into the
Ministry of Interior in 1996.
Eje Cafetero/Coffee Colombia’s main coffee-growing region, comprising
Axis the departments of Risaralda, Caldas, Quindío, and
the southern part of Antioquia.
ELN Ejército de Liberación Nacional (National Liberation
Front) guerrilla group; Colombia’s second-largest
guerrilla group, formed in 1962.
FARC Fuerzas Armadas Revolucionarias de Colombia
(Revolutionary Armed Forces of Colombia);
Colombia’s largest guerrilla group, founded
in 1964.
Indigenous Guard A nonviolent community self-defense movement
established in 2001 by the Nasa Indigenous
community in Cauca, southwestern Colombia.
Inspector de Policía Police inspector; an unarmed, nonuniformed civilian
who does administrative work for local police and
registers complaints from the population.
Junta de Acción A community action board of a village or
Comunal (JAC) neighborhood; a decision-making body whose duties
include planning public goods projects.
Law of Silence An unspoken rule whereby members of a community
cease to communicate with each other for fear of
reprisals from the controlling armed group,
undermining the potential for community-based
resistance.
Liberation A Roman Catholic ideological movement with roots
Theology in Latin America that places poverty at the center of
336 Glossary

Biblical interpretation, calling for social change and

critiquing the structural causes of poverty.
Limpieza “Social cleansing”; the tactics used by paramilitary
groups to eliminate individuals or groups identified as
criminals, drug users, social deviants, or military
objectives.
Machetera A fight with a machete (garden sword).
Magdalena Medio A large valley in central Colombia spanning five
departments and surrounding the Magdalena River,
the country’s primary waterway; the resource-rich
region has historically been one of the most violent
and contested parts of the country.
Medellín Cartel The drug-trafficking network led by Pablo Escobar,
responsible for hundreds of targeted assassinations
and widespread violence during the 1970 and 1980s;
it contributed to the founding of the early paramilitary
group Death to Kidnapers (MAS) in the Middle
Magdalena region.
Ministry of The national government ministry in charge of the
Government junta council program.
(Gobierno)
Municipio A municipality; a town or county.
National Front A coalition formed by the Liberal and Conservative
(Frente Nacional) political parties in 1958 to bring an end to La
Violencia. It consisted of a power-sharing agreement
to alternate the presidency between the two parties
each electoral cycle and was maintained until 1974.
NN “No nombre”, or “No name”; found on umarked
graves or crosses to mark the dead.
Peace zones/Zones Self-declared autonomous communities in conflict
of peace areas in the Philippines that seek to be free of the
influence or presence of armed actors.
Personero A human rights ombudsman for a municipio.
Plan Colombia U.S. anti-drug and counterinsurgency aid
package that has given approximately $10 billion in
aid to Colombia’s government and military
institutions since it began in 2000.
Plata o Plomo/ “Bribe or a bullet” or, variably, “Obey or a bullet”; a
Pico y guerrilla policy of intimidation and fear to coerce
Plomo civilians to cooperate; a play on “pico y placa,” an
urban traffic control initiative.
Procuraduría Inspector General; the national government branch
responsible for overseeing the conduct and operations
of politicians and political institutions; it also
Glossary 337

investigates and prosecutes corruption by government

officials.
Raso A foot soldier in an armed group.
Sapo An informant; literally, a frog.
Secretariat The leadership council of the FARC, comprising seven
commanders.
Socio A member of an organization.
Soldados Peasant soldiers recruited and deployed as part of a
Campesinos strategic program to increase Colombian military
presence in municipios that lacked state presence and
were seen as vulnerable to armed actor control.
Tildado/Señalado “Marked”; literally, “signaled” or “accented”; to be
identified as a supporter, sympathizer, or informant of
an armed group.
Toma A guerrilla assault on or taking of a town.
Unión Patriótica The Patriotic Union party; the leftist political party
founded by members of the FARC in 1985. Over
3,000 members of the party were systematically
assassinated by right-wing forces, leading to its
eventual dissolution.
Vacuna A “vaccination payment” or protection tax extorted
by an armed group.
Vereda Village; a subsection of a municipio; smaller than a
corregimiento.
La Violencia The large internecine political conflict that occurred
from 1948–1958, pitting members of the Liberal and
Conservative parties against each other in bloody
battles, primarily in rural areas. An estimated 200,000
people were killed over the course of conflict.
Zona de despeje The former demilitarized zone located in the
department of Meta, and known colloquially as
FARClandia. This territory, covering about 16,216
square miles (42,000 square kilometers) in southern
Colombia, was granted to the FARC by President
Andrés Pastrana in 1998, before the failed 1998–2002
Caguán peace talks.
References

Abdulrahim, Raja. 2014. “In Syria, Activists in Raqqa Try to Confront Militant Islamist
Group.” Los Angeles Times, May 5, 2014.
Abouzeid, Rania. 2012. “Who Will the Tribes Back in Syria’s Civil War?” Time.com,
October 10, 2012.
Acemoglu, Daron, Simon Johnson, and James A. Robinson. 2002. “Reversal of Fortune:
Geography and Institutions in the Making of the Modern World Income Distribu-
tion.” The Quarterly Journal of Economics 117(4): 1231–94.
ACIA (Asociación Campesina Integral del Atrato, Chocó), Consejo Comunitario
Mayor, and Red de Solidaridad Social Presidencia de la República. 2002. Medio
Atrato: Territorio de Vida. Quibdó: Red de Solidaridad Social; ACIA.
AFP. 2012. “Palestinian Refugees Reclaim Syrian Camp.” AFP, December 21, 2012.
AFP. 2013. “Syria Rebels Announce Withdrawal from Christian Town.” AFP, Septem-
ber 10, 2013.
Al-Ansary, Khalid and Ali Adeeb. 2006. “Most Tribes in Anbar Agree to Unite Against
Insurgents,” The New York Times, September 18, 2006.
Albertus, Michael and Oliver Kaplan. 2013. “Land Reform as a Counterinsurgency
Policy: Evidence from Colombia.” Journal of Conflict Resolution 57(2): 198–231.
Alipala, Julie. 2008. “Army Air Strikes Defeat Peace Zone Agreement Signed in Jolo.”
Philippine Daily Inquirer, April 30, 2008.
Almond, Gabriel A. and Sidney Verba. 1963. The Civic Culture: Political Attitudes and
Democracy in Five Nations. Princeton, NJ: Princeton University Press.
Amnesty International. 2000. Colombia: Return to Hope: Forcibly Displaced Commu-
nities of Urabá and Medio Atrato Region.
Amnesty International. 2001. Annual Report Colombia.
Amnesty International. 2002. “Colombia: San Vicente Del Caguán after the Breakdown
of the Peace Talks: A Community Abandoned.” AMR23/098/2002, October 16, 2002.
Amos, Deborah. 2012a. “In Syria, Aleppo Today Is Must-See TV For Survival.”
National Public Radio, November 28, 2012.
Amos, Deborah. 2012b. “Syrian Villagers Hope Their Example Will Be A Model.”
National Public Radio, December 10, 2012.

339
340 References

Anderson, Mary B. 1999. Do No Harm: How Aid Can Support Peace - or War. Boulder:
Lynne Rienner Publishers.
Anderson, Mary B. and Marshall Wallace. 2012. Opting Out of War: Strategies to
Prevent Violent Conflict. Boulder: Lynne Rienner Publishers.
Arango, Tim. 2014. “Tears, and Anger, as Militants Destroy Iraq City’s Relics.” The
New York Times, July 30, 2014.
Arjona, Ana. 2014. “Wartime Institutions a Research Agenda.” Journal of Conflict
Resolution 58(8): 1360–89.
Armstrong, Martin and Lauren Williams. 2012. “Armenian Christians Torn in Syria’s
Civil War.” The Daily Star (Beirut), October 1, 2012.
Asociación de Trabajadores Campesinos del Carare. 2006. “Estatutos.” ATCC Archive,
September 2006.
ASOPROA. 2006. ASOPROA: Fortalecimiento de la organización campesina para la
prevención del desplazamiento y el impulso al desarrollo local.
Axelrod, Robert. 1984. The Evolution of Cooperation. New York: Basic Books.
Azam, Jean-Paul. 2006. “On Thugs and Heroes: Why Warlords Victimize Their Own
Civilians.” Economics of Governance 7(1): 53–73.
Bagley, Bruce. 1989. “The State and the Peasantry in Contemporary Colombia.” Latin
American Issues 6.
Bagley, Bruce and Matthew Edel. 1980. “Popular Mobilization Programs of the
National Front: Cooptation and Radicalization.” In Politics of Compromise: Coali-
tion Government in Colombia, eds. Ronald G. Hellman and R. Albert Berry. New
Brunswick, NJ: Transaction Books.
Baker, Aryn. 2014. “‘Systematic Torture and Killing’: A New Report Points Fingers at
Syria’s Assad.” Time, January 20, 2014.
Bakke, Kristin M. 2014. “Help Wanted? The Mixed Record of Foreign Fighters in
Domestic Insurgencies.” International Security 38(4): 150–87.
Ball, Patrick. 1998. Liberal Hypocrisy and Totalitarian Sincerity: The Social and Ideo-
logical Origins of the National Non-Government Human Rights Movements in El
Salvador, Pakistan, and Ethiopia, Ph.D. thesis, University of Michigan.
Barnard, Anne. 2012a. “Syrian Insurgents Accused of Rights Abuses.” The New York
Times, March 21, 2012.
Barnard, Anne. 2012b. “In Syria, Missteps by Rebels Erode Their Support.” The New
York Times, November 9, 2012.
Barnard, Anne. 2012c. “Syrian Airstrike Kills Palestinian Refugees.” The New York
Times, December 17, 2012.
Barnard, Anne. 2014a. “Syrian Government Forces Seize Town in a Deep Blow to
Opposition.” The New York Times, March 16, 2014.
Barnard, Anne. 2014b. “Syrian Rebels Say Cease-Fire Deals Prove Deceptive.” The New
York Times, January 16, 2014.
Barnard, Anne and Hwaida Saad. 2013a. “Undecided Syrians Could Tip Balance of
Rebellion.” The New York Times, January 5, 2013.
Barnard, Anne and Hwaida Saad. 2013b. “Assault on Christian Town in Syria Adds to
Fears Over Rebels.” The New York Times, September 11, 2013.
Barnard, Anne and Hwaida Saad. 2014. “Nuns Released by Syrians After Three-Month
Ordeal.” The New York Times, March 10, 2014.
Barnett, Donald and Karari Njama. 1966. Mau Mau from Within: Autobiography and
Analysis of Kenya’s Peasant Revolt. Letchworth: MacGibbon and Kee.
References 341

Bates, Robert H., Avner Greif, Margaret Levi, Jean-Laurent Rosenthal, and Barry
Weingast. 2000. Analytic Narratives. Princeton: Princeton University Press.
Bavier, Joe. 2009. “Congo Villagers Take Up Arms against LRA Rebels,” Reuters UK,
January 16, 2009.
BBC. 2011. “Venezuela Links to Farc Detailed.” BBC, May 10, 2011.
BBC. 2013. “Syria Rebels Withdraw from Ancient Christian Town of Maaloula.” BBC,
September 6, 2013.
Bejarano, Ana Maria and Eduardo Pizarro. 2001. “The Coming Anarchy: The Partial
Collapse of the State and the Emergence of Aspiring State Makers in Colombia.”
Manuscript.
Bernardo, Ariel B. 2010. “Spaces for Peace: Societal Influence through Grassroots Peace
Building.” Army Troopers Newsmagazine, July–August 2010: 7–13.
Bobbio, Norberto. 1988. “Gramsci and the Concept of Civil Society.” In Civil Society
and the State, ed. John Keane. London; New York: Verso.
Borrero García, Camilo. 1989. “Acción comunal y política estatal: un matrimonio
indisoluble?” Documentos ocasionales CINEP, No. 57, Centro de Investigación y
Educación Popular (CINEP).
Bouvier, Virginia M. 2006. “Harbingers of Hope: Peace Initiatives in Colombia.” USIP
Special Report (#169).
Bouvier, Virginia M. 2009. Colombia: Building Peace in a Time of War. Washington,
D.C.: USIP Press.
Bowen, Jeremy. 2013. “Syrians ‘No Longer Trust Neighbours.’” BBC News, September
15, 2013.
Bradley, Matt. 2016. “Rift Grows in Islamic State Between Foreign, Local Fighters.”
Wall Street Journal, March 25, 2016.
Brulliard, Karin. 2010. “In Targeting Taliban Stronghold, U.S. Depends on Afghans’
Reluctant Support.” The Washington Post, July 16, 2010.
Bumiller, Elisabeth. 2010. “U.S. Tries Luring Taliban Foot Soldiers Back to Society.”
The New York Times, May 24, 2010.
Buonanno, Paolo, Paolo Vanin, and Daniel Montolio. 2009. “Does Social Capital
Reduce Crime?” Journal of Law and Economics 52 (February 2009): 145–170.
Burch, Jonathon. 2012. “On Syrian Border, Mixed Feelings for Rebel Liberators.”
Reuters, November 15, 2012.
Cabreza, Vincent. 2010. “Another Peace Zone in Mt. Province?’” Philippine Daily
Inquirer, September 21, 2010.
Campbell, Duncan. 2000. “Heart of Darkness.” The Guardian, June 19, 2000.
Cante, Freddy and Luisa Ortiz, eds. 2005. Acción Política No-violenta, Una Opción
para Colombia. Bógota: Centro Editorial Universidad del Rosario.
Cárdenas, Juan Camilo. 2008. Social Preferences among the People of Sanquianga in
Colombia. CEDE, Universidad de Los Andes.
Catholic Relief Services. 2003. “Reflections on Creating and Sustaining Zones of Peace:
Lessons from Mindanao, Philippines.” Catholic Relief Services Peace & Reconcili-
ation Program, Davao City, Philippines.
Caviedes, Mauricio, ed. 2007. Paz y Resistencia: Experiencias Indígenas desde la
Autonomía. Bogotá: CECOIN.
Ceballos, Miguel and Gerard Martin. 2001. Participación y Fortalecimiento Institu-
cional a Nivel Local en Colombia. Bogotá: Centro Editorial Javeriano, Pontificia
Universidad Javeriana; Washington, D.C.: Center for Latin American Studies.
342 References

Chandrasekaran, Rajiv. 2010. “In Afghanistan, Why Does Counterinsurgency Work in

Some Places but Not Others?” The Washington Post, July 25, 2010.
Cheikhomar, Ammar and Henry Austin. 2013. “Caught in the Middle: Christians Suffer
amid Syria’s Civil War.” NBC News, August 11, 2013.
Chenoweth, Erica and Maria J. Stephan. 2011. Why Civil Resistance Works: The
Strategic Logic of Nonviolent Conflict. New York: Columbia University Press.
Chivers, C. J. 2013a. “Brutality of Syrian Rebels Posing Dilemma in West.” The New
York Times, September 5, 2013.
Chivers, C. J. 2013b. “Some Syrian Kurds Resist Assad, Defying Conventional Views.”
The New York Times, January 23, 2013.
Chivers, C. J. 2013c. “Syrians, Fleeing Home, Crowd in Roman Caves.” The New York
Times, March 23, 2013.
Chwe, Michael Suk-Young. 2001. Rational Ritual: Culture, Coordination, and
Common Knowledge. Princeton, NJ: Princeton University Press.
CINEP. 2003. Comunidades de Paz de San Francisco de Asís, Nuestra Señora del
Carmen y Natividad de María: Chocó, Colombia. Bógota: CINEP.
CINEP. 2007. “Presentación - Falsos Positivos.” Noche y Niebla 34/35.
Cohen, Jean L. and Andrew Arato. 1994. Civil Society and Political Theory. Cambridge,
MA: MIT Press.
Collier, Paul and Anke Hoeffler. 2004. “Greed and Grievance in Civil War.” Oxford
Economic Papers 56(4): 563–595.
Colombia Ministerio de Gobierno. 1980. Dirección General de Integración y Desarrollo
de la Comunidad 20 Años de Desarrollo de la Comunidad 1959 - 1979. Bogotá:
DIGEDEC.
Comisión Nacional de Reparación y Reconciliación (CNRR). 2007. Disidentes, Rear-
mados y Emergentes: ¿Bandas Criminales o Tercera Generación Paramilitar? Área de
Desmovilización, Desarme y Reintegración. Bogotá: CNRR.
Comisión Nacional de Reparación y Reconciliación (CNRR). 2009. Una Historia de
Paz para Contar, Recontar y No Olvidar: Cartilla Sobre la Historia de la ATCC.
Equipo de Memoria Histórica. Bogotá: CNRR.
Comisión Nacional de Reparación y Reconciliación (CNRR). 2011. El Orden Desar-
mado: La Resistencia de la Asociación de Trabajadores Campesinos del Carare
(ATCC). Grupo de Memoria Histórica. Bogotá: CNRR.
Coronel-Ferrer, Miriam. 2005. “Institutional Response: Civil Society.” Background
paper, Human Development Network Foundation, Inc.
Corporación Nuevo Arco Iris. 2008. “Preocupante Aumento de Bandas Armadas en
Colombia.” Available at: www.nuevoarcoiris.org.co/sac/?q=node/231.
Crost, Benjamin, Joseph Felter and Patrick Johnston. 2014. “Aid under Fire: Develop-
ment Projects and Civil Conflict.” American Economic Review, 104(6): 1833–1856.
Cubides, F., Jaramillo, J. E. and Mora, L. 1989. Colonización, Coca y Guerrilla. Bogotá:
Editorial Alianza.
Cubides C., Fernando. 2006. “La Participación Política del Campesinado en el Contexto
de la Guerra: El Caso Colombiano.” In La Construcción de la Democracia en el
Campo Latinoamericano, ed. Hubert C Grammont. Buenos Aires: CLACSO, Consejo
Latinoamericano de Ciencias Sociales.
DANE (Departamento Administrativo Nacional de Estadística). 1962. Directorio
Nacional de Explotaciones Agropecuarias (Censo Agropecuario) 1960. Bogotá:
DANE.
References 343

DANE (Departamento Administrativo Nacional de Estadística). 1971. Censo Agrope-

cuario, 1970–1971: Datos Preliminares. Bogotá: DANE.
DANE (Departamento Administrativo Nacional de Estadística). 1978. Directorio
Nacional De Entidades Cooperativas 1975–1976. Bogotá: Departamento Adminis-
trativo Nacional de Estadística.
DANE (Departamento Administrativo Nacional de Estadística). 1985. Divipola.
Bogotá: DANE.
DANE (Departamento Administrativo Nacional de Estadística). 1987. Colombia Esta-
dística. Bogotá: DANE.
DANE (Departamento Administrativo Nacional de Estadística). 1997. División Político-
Administrativa de Colombia / República de Colombia. Bogotá: Departamento
Administrativo Nacional de Estadística, DANE.
DANE (Departamento Administrativo Nacional de Estadística). 2000. Divipola.
Bogotá: DANE.
DANE (Departamento Administrativo Nacional de Estadística). 2007. Divipola.
Bogotá: DANE.
DANE (Departamento Administrativo Nacional de Estadística). 2008. GIST Data
Repository [cited 2008]. Available from https://gist.itos.uga.edu.
Daniel, James M. 1965. Rural Violence in Colombia since 1946. Washington, D.C.:
Special Operations Research Office.
De Friedemann, Nina S. 1979. “Ma Ngombe: Guerreros y Ganaderos en Palenque.”
Available at: www.lablaa.org/blaavirtual/antropologia/magnom/nina0.htm.
DePetris, Daniel. 2013. “Alawite Group against Assad.” The Epoch Times, October 28,
2012.
Di Giovanni, Janine. 2012. “Mountaintop Town Is a Diverse Haven From Syria’s
Horrors.” The New York Times, November 22, 2012.
Doughty, Kristin and David Moussa Ntambara. 2005. “Resistance and Protection:
Muslim Community Actions during the Rwandan Genocide.” Available at: www.cda
inc.com/cdawww/pdf/casestudy/steps_rwanda_case_study_Pdf.pdf.
Dudley, Steven. 2003. Walking Ghosts: Murder and Guerilla Politics in Colombia.
Taylor & Francis, Inc.
Duncan, Gustavo. 2006. Los Señores de la Guerra: De Paramilitares, Mafiosos y
Autodefensas en Colombia. 1st ed. Bogotá.
East View Cartographic, Inc. 2002. VMAP Level 1 GIS Vector Data. East View
Cartographic, Inc.
Easterly, William. 2007. The White Man’s Burden: Why the West’s Efforts to Aid
the Rest Have Done So Much Ill and So Little Good. First Edition. Penguin
Books.
Eastman, Jorge Mario. 1982. Seis Reformas Estructurales al Régimen Político: Resulta-
dos Electorales de 1930 a 1982. Bogotá: Ministerio de Gobierno.
Echandía Castilla, Camilo. 1999. El Conflicto Armado y las Manifestaciones de Vio-
lencia en las Regiones de Colombia. Santa Fé de Bogotá: Presidencia de la República
de Colombia, Oficina del Alto Comisionado para la Paz, Oberservatorio de Violencia;
Imprenta Nacional.
Eckstein, Harry. 1975. “Case-Study and Theory in Macro-Politics.” In Handbook of
Political Science, eds. F. Greenstein and N. Polsby. Reading: Addison-Wesley.
Edel, Matthew. 1969. “The Colombian Community Action Program: Costs and Bene-
fits.” Yale Economic Essays 9(Fall 1969): 3–58.
344 References

Edel, Matthew. 1971. “Determinants of Investments by Colombian Community Action

Boards.” Journal of Developing Areas 5(2): 207–221.
El Espectador. 2015. “Colombia, El Segundo País Con Más Desplazados: Ya Son Seis
Millones.” December 2, 2015.
El Tiempo. 1996. “Campesinos Impiden Toma Guerrillera En Santander” (Campesinos
Impede Guerrilla Attack in Santander). El Tiempo, April 12, 1996.
El Tiempo. 2001a. “Paz: 500 Diálogos Clandestinos” (Peace: 500 Clandestine Dia-
logues). El Tiempo, November 11, 2001.
El Tiempo. 2001b. “Paeces Impiden Ataque De Las Farc” (Paez Impede FARC Attack).
El Tiempo, November 14, 2001.
El Tiempo. 2002a. “Lo De Nosotros Son Las Veredas” (What’s Ours Are the Villages).
El Tiempo, September 11, 2002.
El Tiempo. 2002b. “No Les Serviremos Ni a Uribe Ni a las Farc” (We Won’t Serve Uribe
nor the FARC). El Tiempo, July 29, 2002.
El Tiempo. 2002c. “FARC Acechan a Juntas Comunales.” El Tiempo, July 25, 2002.
El Tiempo. 2004. “Comunales Preocupados Por Politización Y Amenazas.” El Tiempo,
December 7, 2004.
El Tiempo. 2005. “A Prueba, Neutralidad de Ocho Comunidades de Paz en el Urabá.”
El Tiempo, March 9, 2005.
El Tiempo. 2008a. “Por ser de izquierda, asesinaron a líder comunitario en San Vicente
del Caguán.” El Tiempo, December 15, 2008.
El Tiempo. 2008b. “Acusan a ‘Botalón’ por Masacre en Cimitarra (Santander).”
El Tiempo, August 8, 2008.
El Tiempo. 2009. “Por masacre de 12 taladores en Cimitarra, aseguran a ‘Botalón’”
(For Massacre of 12 Woodcutters in Cimitarra, ‘Botalón’ Is Imprisoned). El Tiempo,
January 7, 2009.
El Tiempo. “Justicia y Paz: 5 años” (editorial). 2010. El Tiempo, July 25, 2010.
Ellickson, Robert C. 1991. Order without Law: How Neighbors Settle Disputes.
Cambridge: Harvard University Press.
Elster, Jon. 2000. “Analytic Narratives by Bates, Greif, Levi, Rosenthal, and Weingast:
A Review and Response.” American Political Science Review 94(3): 685–702.
England, Andrew. 2008. “Syria’s Religious Tolerance Belies Critics.” Financial Times,
September 15, 2008.
Equipo Nizkor. 2001. “Colombia Nunca Más: Crímenes de Lesa Humanidad (Zona 14ª
1966, Tomo I).” Proyecto Nunca Más, Colombia.
Estado Mayor del Bloque Oriental FARC-EP. 2002. “Contra las Instituciones del
Estado.”
Fahim, Kareem. 2012. “Stalemate Deals Grief and Fury in Syria.” The New York Times,
January 23, 2012.
Falla, Ricardo. 1994. Massacres in the Jungle: Ixcán, Guatemala, 1975–1982. Boulder:
Westview Press.
Fals Borda, Orlando. 1960. Acción Comunal en Una Vereda Colombiana: Su Aplica-
ción, Sus Resultados y Su Interpretación. Bogotá: Universidad Nacional de Colombia,
Departamento de Sociología.
FARC-EP. Date unknown. “Boletín informativo del Comité Temático de las FARC-EP
N 11.”
Fearon, James. 1994. “Domestic Political Audiences and the Escalation of International
Disputes.” American Political Science Review 88(3): 577–592.
References 345

Fearon, James D. 1995. “Rationalist Explanations for War.” International Organiza-

tion 49(3): 379–414.
Fearon, James D. 1996. “Causes and Counterfactuals in Social Science: Exploring and
Analogy between Celular Automata and Historical Processes.” In Counterfactual
Thought Experiments in World Politics: Logical, Methodological, and Psychological
Perspectives, eds. Philip E. Tetlock and Aaron Belkin. Princeton: Princeton University
Press.
Fearon, James D. and David D. Laitin. 1996. “Explaining Interethnic Cooperation.”
American Political Science Review 90(4): 715–735.
Fearon, James D. and David D. Laitin. 2003. “Ethnicity, Insurgency, and Civil War.”
American Political Science Review 97(1): 75–90.
Fearon, James D. and David D. Laitin. 2008. “Integrating Qualitative and Quantitative
Methods.” In The Oxford Handbook of Political Methodology.
Fearon, James D., Macartan Humphreys, and Jeremy M. Weinstein. 2009. “Can Devel-
opment Aid Contribute to Social Cohesion after Civil War? Evidence from a Field
Experiment in Post-Conflict Liberia.” The American Economic Review 99(2): 287–291.
Fernandez, Edwin. 2011. “Officials Want Harassed Village Declared Peace Zone.”
Philippine Daily Inquirer, February 9, 2011.
Ferro, Juan Guillermo and Graciela Uribe Ramón. 2002. El Orden de la Guerra: Las
FARC-EP entre la Organización y la Política. Bogotá: CEJA.
Filkins, Dexter. 2010. “Afghan Tribe Vows to Fight Taliban in Return for U.S. Aid.”
The New York Times, January 28, 2010.
Fisk, Robert. 2013. “Syria Crisis: In Sacred Maaloula, Where They Speak the Language
of Christ, War Leads Neighbours into Betrayal.” The Independent, September 25,
2013.
Forero, Juan. 2005. “Colombia War Spills Into Indians’ Peaceful World.” The New
York Times, May 2, 2005.
Fumerton, Mario. 2001. “Rondas Campesinas in the Peruvian Civil War: Peasant Self-
Defence Organisations in Ayacucho.” Bulletin of Latin American Research 20(4):
470–497.
Funes, Maria J. 1998. “Social Responses to Political Violence in the Basque Country:
Peace Movements and Their Audience.” The Journal of Conflict Resolution 42(4):
493–510.
Gall, Carlotta. 2009. “U.S. Faces Resentment in Afghan Region.” The New York Times,
July 3, 2009.
Gall, Carlotta. 2010. “U.S. Hopes Afghan Councils Will Undermine Taliban.” The New
York Times, June 19, 2010.
Gall, Carlotta and Abdul Wahhed Wafa. 2006. “Taliban Truce in District of Afghani-
stan Sets Off Debate.” The New York Times, December 2, 2006.
Galula, David. 1965. Counter-Insurgency Warfare Theory and Practice. Second Printing
edition. Praeger.
García, Alejandro. 1996. Hijos de la Violencia: Campesinos de Colombia Sobreviven a
“Golpes” de Paz. Madrid: Libros de la Catarata.
García Durán, S. J., Mauricio 2005. “To What Extent Is There a Peace Movement in
Colombia? An Assessment of the Country’s Peace Mobilization, 1978–2003.” Ph.D.
dissertation, University of Bradford.
García Durán, Mauricio. 2006a. Movimiento por la paz en Colombia. 1978–2003.
Bogotá: UNDP Colombia, CINEP, COLCIENCIAS Colombia.
346 References

García Durán, S. J., Mauricio. 2006b. “De Turbay a Uribe: Sin Política de Paz pero con
Conflicto Armado.” In En la Encrucijada: Colombia en el Siglo XXI, ed. Francisco
Leal Buitrago. Bogotá: Grupo Editorial Norma.
Geddes, Barbara. 2003. Paradigms and Sand Castles: Theory Building and Research
Design in Comparative Politics. Ann Arbor: The University of Michigan Press.
George, Alexander L. 1979. “Case Studies and Theory Development: The Method of
Structured, Focused Comparison.” In Diplomacy: New Approaches in History,
Theory and Policy, ed. P. G Lauren. New York: The Free Press.
Gettleman, Jeffrey. 2007. “Chaos in Darfur Rises as Arabs Fight with Arabs,” The New
York Times, September 3, 2007.
Gettleman, Jeffrey. 2009a. “Armed with Little but Resolve, and Defending a Hollowed
Village,” The New York Times, February 18, 2009.
Gettleman, Jeffrey. 2009b. “For Somalia, Chaos Breeds Religious War,” The New York
Times, May 24, 2009.
Gettleman, Jeffrey. 2009c. “Radical Islamists Slipping Easily Into Kenya,” The New
York Times, July 22, 2009.
Gettleman, Jeffrey and Eric Schmitt. 2009. “US Aided a Failed Plan to Rout Ugandan
Rebels,” The New York Times, February 6, 2009.
Gitlitz, John S. and Telmo Rojas. 1983. “Peasant Vigilante Committees in Northern
Peru.” Journal of Latin American Studies 15(01): 163–197.
GMH. 2013. ¡Basta Ya! Colombia: Memorias de Guerra y Dignidad. Bogotá: Imprenta
Nacional.
Goldstein, Joseph and Taimoor Shah. 2015. “Taliban Strike Crucial District in Afghani-
stan.” The New York Times, June 19, 2015.
Gómez-Suárez, Andrei. 2007. “Perpetrator Blocs, Genocidal Mentalities and Geograph-
ies: The Destruction of the Unión Patriótica in Colombia and Its Lessons for Genocide
Studies.” Journal of Genocide Research 9(4): 637–660.
Gopal, Anand and Matthew Rosenberg. 2009. “Afghan Villagers Attack Taliban, Sever
Ties with Militants.” The Wall Street Journal, July 19, 2009.
Greene, David. 2010. “A Village Clings to Hope Amid Dagestan’s Dangers.” National
Public Radio, March 18, 2010.
Guerra Curvelo, Weildler. 2004. La Disputa Y La Palabra: La Ley En La Sociedad
Wayuu. Bogotá: Ministerio de Cultura.
Guevara, Che. 1961. Guerilla Warfare. (Stone, I. F., The Authorized Translation).
Vintage Books.
Gutiérrez Sanín, Francisco. 2008. “Telling the Difference: Guerrillas and Paramiliataries
in the Colombian Civil War.” Politics & Society 36(3).
Gutiérrez Sanín, Francisco and Mauricio Barón. 2005. “Re-Stating the State: Paramili-
tary Territorial Control and Political Order in Colombia (1978–2004).” Crisis States
Program Working Paper no. 66. Bogotá.
Guzmán Campos, Germán, Orlando Fals Borda, and Eduardo Umaña Luna. 1963.
La Violencia en Colombia, Estudio de un Proceso Social. Bogotá: Ediciones Tercer
Mundo.
Hafner-Burton, Emilie. 2008. “Sticks and Stones: Naming and Shaming the Human
Rights Enforcement Problem.” International Organization 62(4): 689–716.
Hafner-Burton, Emilie. 2009. Forced to Be Good: Why Trade Agreements Boost
Human Rights. Cornell, NY: Cornell University Press.
References 347

Hancock, Landon E. and Christopher Mitchell, eds. 2007. Zones of Peace. Bloomfield,
CT: Kumarian Press, Inc.
Hartlyn, Jonathan. 1988. The Politics of Coalition Rule in Colombia. Cambridge:
Cambridge University Press.
Hassan, Hassan. 2014. “Hope Springs in Syria? How Local Cease-fires Have Brought
Some Respite to Damascus.” Foreign Affairs Snapshot, January 22, 2014.
Heaton, Laura and Maggie Fick. 2010. “Field Dispatch: The Arrow Boys of Southern
Sudan – An Army of the Willing.” Available at: www.enoughproject.org/ publica
tions/arrow-boys-sudan.
Henao Delgado, Hernán and María Teresa Arcila Estrada. 1993. Nariño, Antioquia.
Medellín: CORNARE; Universidad de Antioquia (INER).
Henderson, James D. 1985. When Colombia Bled: A History of the Violence in Tolima.
Tuscaloosa, AL: University of Alabama Press.
Hernández Delgado, Esperanza. 2004. Resistencia Civil Artesana de Paz: Experiencias
Indígenas, Afrodescendientes y Campesinas. Bógota: Editorial Pontificia Universidad
Javeriana: SUIPPCOL.
Hirschman, Albert O. 1970. Exit, Voice, and Loyalty: Responses to Decline in Firms,
Organizations, and States. Cambridge, MA: Harvard University Press.
Holmes, Jennifer S., Sheila Amin Gutiérrez De Piñeres, and Kevin M. Curtin. 2007.
“A Subnational Study of Insurgency: FARC Violence in the 1990s.” Studies in
Conflict & Terrorism 30(3): 249–65.
Houghton, Juan and William Villa. 2005. Los territorios Indígenas Colombianos:
Teorías y Prácticas. Bogotá: Centro de Cooperación al Indígena, CECOIN.
Human Rights Watch. 2003. “You’ll Learn Not to Cry: Child Combatants in Colom-
bia.” Human Rights Watch Report.
Human Rights Watch. 2015. “Colombia: Top Brass Linked to Extrajudicial Execu-
tions.” June 24, 2015.
Humphreys, Macartan and Jeremy Weinstein. 2006a. “Handling and Manhandling
Civilians in Civil War.” American Political Science Review 100(3): 429–447.
Humphreys, Macartan and Jeremy Weinstein. 2006b. “Handling and Manhandling
Civilians: Derivation of Hypotheses.” Appendix.
Humphreys, Macartan and Jeremy M. Weinstein. 2008. “Who Fights? The Determinants
of Participation in Civil War.” American Journal of Political Science 52(2): 436–55.
Huntingon, Samuel P. 1967. “Political Stability and Security in South Vietnam.” Report
to USAID, December 1967.
Inglehart, Ronald, Miguel Basáñez, and Alejandro Menéndez Moreno. 1998. Human Values
and Beliefs: a Cross-Cultural Sourcebook. Ann Arbor: University of Michigan Press.
Internal Displacement Monitoring Centre (IDMC). 2010. “Colombia Current IDP
Figures.” Available at: www.internal-displacement.org/idmc/website/countries.nsf/
(httpEnvelopes)/A7E1B7BD7528B329C12575E500525165?
OpenDocument#sources.
Internal Displacement Monitoring Centre (IDMC). 2014. “Philippines IDP Figures
Analysis.” February 2014. www.internal-displacement.org/south-and-south-east-asia/
philippines/figures-analysis.
Internal Displacement Monitoring Centre (IDMC). 2015. “Syria IDP Figures Analysis.”
December 31, 2015. www.internal-displacement.org/middle-east-and-north-africa/
syria/figures-analysis.
348 References

Ivan García, Jorge. 1994. “Asociaciones De Seguridad Rural Son Un Invento Viejo”
(Rural Security Associations are an Old Invention). El Tiempo, December 16, 1994.
Jaffe, Greg. 2009. “U.S. Plans to Shift Forces to Populated Areas of Afghanistan.” The
Washington Post, September 22, 2009.
Jawad Al-Tamimi, Aymenn. 2012. “Syria’s Assyrians, Caught in the Middle.” The Daily
Star (Beirut), December 7, 2012.
Justicia y Paz (Comisión Intereclesial de Justicia y Paz). 2003. “Colombia: Desplazadas
forzadamente familias del Cacarica.” September 12, 2003.
Kalyvas, Stathis. 2006. The Logic of Violence in Civil War. Cambridge: Cambridge
University Press.
Kalyvas, Stathis and Matthew Kocher. 2009. “The Dynamics of Violence in Vietnam:
An Analysis of the Hamlet Evaluation System (HES).” Journal of Peace Research
46(3): 335–55.
Kamman, Curtis W. 1999. “FARC Abusing Civilians in ‘Despeje.’” Confidential Cable
004726. United States Embassy, Colombia; National Security Archive, May 10, 1999.
Kant, Immanuel. 1970. Kant’s Political Writings. Hans Reiss, ed. translated by H.B.
Nisbet. Oxford: Oxford University Press.
Kaplan, Oliver. 2010. Civilian Autonomy in Civil War. Ph.D. thesis, Department of
Political Science, Stanford University.
Kaplan, Oliver. 2012. “A New Approach to the Drug War” (A novel model offers hope
for communities battling narcos and militants). National Interest, July 26, 2012.
Kaplan, Oliver. 2013a. “Protecting Civilians in Civil War: The Institution of the ATCC
in Colombia.” Journal of Peace Research 50(3): 351–367.
Kaplan, Oliver. 2013b. “Nudging Armed Groups: How Civilians Transmit Norms of
Protection.” Stability: International Journal of Security and Development 2(3): 62.
Kaplan, Oliver. 2013c. “Shootings and Shamans: Indigenous Group Authority Struc-
tures and Civil War Violence in Colombia,” working paper.
Kaplan, Oliver and Enzo Nussio. 2015. “Community Counts: The Social Reintegration
of Ex-combatants in Colombia.” Conflict Management and Peace Science. DOI:
10.1177/0738894215614506
Kaplan, Oliver and Enzo Nussio. 2016. “Explaining Recidivism of Ex-combatants
in Colombia.” Journal of Conflict Resolution. DOI: 10.1177/0022002716644326
Karam, Zeina. 2012. “War Rips Apart Families, Neighbors in Syria as Political, Sectar-
ian Loyalties Take Over.” Associated Press, November 30, 2012.
Keck, Margaret and Kathryn Sikkink. 1998. Activists Beyond Borders. Ithaca, NY and
London: Cornell University Press.
Keohane, Robert O. 1984. After Hegemony: Cooperation and Discord in the World
Political Economy. Princeton, NJ: Princeton University Press.
Kernell, Samuel. 1986. Going Public: New Strategies of Presidential Leadership.
Washington, D.C.: CQ Press.
Khalek, Rania. 2013. “Syria’s Nonviolent Resistance Is Dying to Be Heard.” Al Jazeera
America, September 9, 2013.
Khan, Riaz. 2009. “Tribal Elders Gunned Down by Taliban in Pakistan.” The Associ-
ated Press, September 24, 2009.
King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry:
Scientific Inference in Qualitative Research. Princeton, NJ: Princeton University Press.
Kirkpatrick, David D. 2012. “Syrian Children Speak of Revenge against Alawites.”
The New York Times, September 3, 2012.
References 349

Kirkpatrick, David D. 2013. “Syrian Civilians Take Reins in Test of Self-Government.”

The New York Times, March 1, 2013.
Kline, Harvey F. 2003. “Colombia: Lawlessness, Drug Trafficking, and Carving Up the
State.” In State Failure and State Weakness in a Time of Terror, ed. Robert I. Rotberg.
Washington, D.C.: Brookings Institution Press.
Krasner, Stephen D. 1982. “Structural Causes and Regime Consequences: Regimes as
Intervening Variables.” International Organization 36(2): 15–205.
Kriger, Norma J. 1992. Zimbabwe’s Guerrilla War: Peasant Voices. Cambridge, New
York: Cambridge University Press.
Kuperman, Alan J. 2000. “Rwanda in Retrospect.” Foreign Affairs, January/February
2000.
Kuran, Timur. 1991. “Now Out of Never: The Element of Surprise in the East European
Revolution of 1989.” World Politics 44(October): 7–48.
Lacina, Bethany. 2006. “Explaining the Severity of Civil Wars.” Journal of Conflict
Resolution 50(2): 276–289.
Ladrón de Guevara and Andrés Dávila. 1998. “El Ejército Colombiano: Un Actor Más
de la Violencia.” In Las violencias: inclusión creciente, ed. Jaime Arocha Rodríguez.
Santafé de Bogotá: Centro de Estudios Sociales Fac. de Ciencias Humanas Univ.
Nacional.
Laitin, David D. and Avner Greif. 2004. “A Theory of Endogenous Institutional
Change.” American Political Science Review 98(4): 633–652.
Leal Buitrago, Francisco. 1990. Clientelismo: El Sistema Político Y Su Expresión
Regional. 1st ed. Bogotá: Instituto de Estudios Políticos y Relaciones Internacionales.
Leal Buitrago, Francisco, ed. 2006. En la Encrucijada: Colombia en el siglo XXI.
Bogotá: Editorial Norma.
Lederach, John Paul. 1997. Building Peace: Sustainable Reconciliation in Divided
Societies. Washington, D.C.: United States Institute of Peace Press.
Lederach, John Paul. 2003. “The Mystery of Transformative Times and Spaces: Explor-
ing a Theology of Grassroots Peacemaking.” In Artisans of Peace: Grassroots Peace-
making Among Christian Communities, eds. Mary Ann Cejka and Thomas Bamat.
Maryknoll, NY: Orbis Books.
Lederach, John Paul. 2005. The Moral Imagination: The Art and Soul of Building Peace.
Oxford: Oxford University Press.
Lee, Zosimo E. 2000. “Peace Zones as Special Development Areas: A Preliminary
Assessment.” In Building Peace: Essays on Psychology and the Culture of Peace,
eds. A.B.I. Bernardo and C.D. Ortigas. Manila: De la Salle University Press.
LeGrand, Catherine. 1986. Frontier Expansion and Peasant Protest in Colombia,
1850–1936. Albuquerque, NM: University of New Mexico Press.
León, Juanita. 2004. No somos machos, pero somos muchos. Bogotá: Grupo Editorial
Norma.
Levitsky, Steven and Gretchen Helmke. 2006. Informal Institutions and Demo-
cracy: Lessons from Latin America. Baltimore, MD: The Johns Hopkins University
Press.
Lichbach, Mark. 1994. “What Makes Rational Peasants Revolutionary: Dilemma,
Paradox, and Irony in Peasant Rebellion.” World Politics 46: 383–418.
Lichbach, Mark. 1998. “Contending Theories of Contentious Politics and the Structure-
Action Problem of Social Order.” Annual Review of Political Science 1998(1):
401–424.
350 References

Lieberman, Evan S. 2005. “Nested Analysis as a Mixed-Method Strategy for Compara-

tive Research.” The American Political Science Review 99(3): 435–452.
Llano Cano, John Jairo. 2006. El Movimiento Comunal en Colombia.
Lohmann, Susanne. 1994. “The Dynamics of Informational Cascades: The Monday
Demonstrations in Leipzig, East Germany, 1989–91.” World Politics 47(October):
42–101.
Lorente, Luis. 1985. Distribución De La Propiedad Rural En Colombia, 1960–1984.
Bogotá: Ministerio de Agricultura; Corporación de Estudios Ganaderos y Agrícolas.
Luttwak, Edward N. 1999. “Give War a Chance.” Foreign Affairs 78: 36–44.
Mackety, Robert. 2014. “Syrians Protest Assad and Islamist Militants.” NYTimes.com,
The Lede Blog, January 3, 2014.
Magaloni, Beatriz, Alberto Díaz-Cayeros and Federico Estevez. Forthcoming. Strategies
of Vote-Buying: Poverty, Democracy, and Social Transfers in Mexico.
Mahony, Liam and Luis Enrique Eguren. 1997. Unarmed Bodyguards: International
Accompaniment for the Protection of Human Rights. West Hartford, CT: Kumarian
Press.
Malek, Alia. 2012. “Syrian Armenians Seek Shelter in Armenia.” The New York Times,
December 11, 2012.
Malkasian, Carter. 2013. War Comes to Garmser: Thirty Years of Conflict on the
Afghan Frontier. London: Oxford University Press.
Mallari Jr., Delfin. 2007. “Solcom Chief Slams ‘Peace Zone.’” Philippine Daily Inquirer,
October 4, 2007.
Mao, Zedong. 1961. On Guerrilla Warfare. New York: Praeger.
MAPP/OEA. 2010. “Décimo Cuarto Informe Trimestral del Secretario General al
Consejo Permanente sobre la Misión de Apoyo al Proceso de Paz en Colombia.”
OEA/Ser.G CP/doc. 4486/10, April 26, 2010.
Martz, John D. 1997. The Politics of Clientelism: Democracy and State in Colombia.
New Brunswick, CT and London: Transaction Publishers.
Marwell, Gerald and Pamela Oliver. 1993. The Critical Mass in Collective Action.
Cambridge: Cambridge University Press.
Mason, David T. 1996. “Insurgency, Counterinsurgency, and the Rational Peasant.”
Public Choice 86(1–2): 63–83.
Masood, Salman. 2009. “Taliban Ambush in Pakistan Kills 8 Militiamen.” The New
York Times, September 25, 2009.
McAdam, Doug, Sidney Tarrow and Charles Tilly 2001. Dynamics of Contention.
Cambridge: Cambridge University Press.
McCarthy, John D. and Mayer N. Zald. 1977. “Resource Mobilization and Social
Movements: A Partial Theory.” The American Journal of Sociology 82(6):
1212–1241.
McChrystal, Stanley A. 2009. "COMISAF Initial Assessment (Unclassified).” U.S.
Department of Defense, September 21, 2009. www.washingtonpost.com/wp-dyn/
content/article/2009/09/21/AR2009092100110.html
McDonnell, Patrick J. 2014. “In Syria’s Capital, Residents Recall a Sectarian Tolerance
Gone By.” Los Angeles Times, April 19, 2014.
McKenna, T. 1998. Muslim Rulers and Rebels: Everyday Politics and Armed Separatism
in the Southern Philippines. Berkeley, CA: University of California Press.
Meyer, John W. and Brian Rowan. 1977. “Institutionalized Organizations: Formal
Structure as Myth and Ceremony.” American Journal of Sociology 83(2): 340–363.
References 351

Milgrom, Paul R. and Weingast Barry R. 1990. “The Role of Institutions in the Revival
of Trade: The Law Merchant, Private Judges, and the Champagne Fairs.” Economics
and Politics (2): 1–23.
Ministerio de Gobierno. 1993. Censo Nacional Calificado de Juntas de Acción Comu-
nal, 1993. Santa Fé de Bogotá: Ministerio de Gobierno, Dirección General de Integra-
ción y Desarrollo de la Comunidad (DIGIDEC).
Molano, Alfredo. 1987. Selva Adentro: Una Historia Orral de la Colonización del
Guaviare. Bogotá: Ancora Editores.
Molano, Alfredo and Alejandro Reyes. 1978. “Los Bombardeos en el Pato.” Contro-
versia 89.
Molano, Alfredo, Daniel Bland, and Lance Selfa. 1994. “The Dispossessed.” In La
Historia al Final del Milenio: Ensayos de Historiografía Colombiana y Latinoamer-
icana. Bogotá, Colombia: Editorial Universidad Nacional, Facultad de Ciencias
Humanas, Departamento de Historia.
Moreno, Guillermo Cardona. 2008. Acción Comunal: Cincuenta Años Vista a Través de
los Congresos Nacionales. Manuscript.
Muana, Patrick K. 1997. “The Kamajoi Militia: Civil War, Internal Displacement and
the Politics of Counter-Insurgency.” Africa Development 22(3/4): 77–99.
Mullen, Michael. 2009. “Strategic Communication: Getting Back to Basics.” Joint
Forces Quarterly (55 October): 2–4.
Murdie, Amanda. 2014. Help or Harm: The Human Security Effects of International
NGOs. Stanford, CA: Stanford University Press.
National Public Radio. 2013. “Grants to Rural Afghan Villages Pay Off.” March 18,
2013.
Navarrete-Frías, Carolina and Francisco E. Thoumi. 2005. “Illegal Drugs and Human
Rights of Peasants and Indigenous Communities: The Case of Colombia.” Manage-
ment of Social Transformations Policy Papers (15), United Nations Educational,
Scientific and Cultural Organization (UNESCO).
Naylor, Hugh. 2015. “In New Sign of Assad’s Troubles, Syria’s Druze Turn Away from
President.” The Washington Post, July 17, 2015.
Naylor, Hugh. 2016. “Al-Qaeda Affiliate Faces Unusual Backlash from Fed-up
Syrians.” The Washington Post, April 13, 2016.
Nechama Tec. 1993. Defiance: The Bielski Partisans. 1st ed. New York: Oxford
University Press.
Nordland, Rod and Dalal Mawad. 2012. “Palestinians in Syria Drawn into the Vio-
lence.” The New York Times, July 30, 2012.
Nordstrom, Carolyn. 1992. “The Backyard Front.” In The Paths to Domination,
Resistance, and Terror, eds. JoAnn Martin and Carolyn Nordstrom. Berkeley:
University of California Press.
Nordstrom, Carolyn. 1997. “From War to Peace – Examples from Sri Lanka and
Mozambique.” In Cultural Variation in Conflict Resolution: Alternatives to Violence,
eds. Douglas P. Fry and Kaj Björkqvist. Psychology Press.
NOW News. 2013. “Syrians Protest against Islamist Group Violations in Idlib.” NOW
News (Beirut), January 25, 2013.
O’Donnell, Guillermo. 1999. “Polyarchies and the (Un)Rule of Law in Latin America: A
Partial Conclusion.” In The (Un)Rule of Law and the Underprivileged in Latin
America, eds. Guillermo O’Donnell Juan E. Méndez and Paulo Sergio Pinheiro. Notre
Dame, IN: University of Notre Dame Press.
352 References

Ober, Josiah. 2008. Democracy and Knowledge. Princeton, NJ: Princeton University
Press.
Observatorio de Paz Integral. 2005. “La Población Civil y el Conflicto Armado en el
Magdalena Medio, 1996–2004.” Barrancabermeja.
Observatorio del Programa Presidencial de Derechos Humanos y DIH. 2001. “Panorama
Actual del Magdalena Medio.” Bogotá: Vicepresidencia de la República de Colombia.
Offstein, Norman. 2005. National, Departmental, and Municipal Rural Agricultural
Land Distribution in Colombia: Analyzing the Web of Inequality, Poverty and
Violence. CEDE Working Paper. Bogotá: Universidad de Los Andes.
Oficina del Alto Comisionado para la Paz. 2006. Proceso de Paz con las Autodefensas.
Bogotá: Presidencia de la República de Colombia.
Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of
Groups. Cambridge, MA: Harvard University Press.
Oquist, Paul H. 1980. Violence, Conflict, and Politics in Colombia. New York: Aca-
demic Press.
Ortiz Sarmiento, Carlos Miguel. 1985. Estado y Subversión en Colombia: La Violencia
en el Quindío, Años 50. Bogotá, Colombia: CIDER Uniandes, Fondo Editorial
CEREC.
Otis, John. 2010. “What’s Andy Warhol Doing in Rural Colombia?” Global Post,
March 10, 2010. Available at: www.globalpost.com/print/5529759.
Oweis, Khaled Yacoub. 2013. “Fearing Stark Future, Syrian Alawites Meet in Cairo.”
Reuters, March 23, 2013.
OXFAM and other NGOs. 2009. “Caught in the Conflict: Civilians and the Inter-
national Security Strategy in Afghanistan.” Available at: www.humansecuritygateway
.com/documents/OXFAM_Civilians_InternationalSecurityStrategy_Afghanistan.pdf.
Páez Segura, José Antonio. 2005. Los Espacios Humanitarios: Una Pedagogía en la
Vida y Para la Vida. Bucaramanga: Programa de Paz y Desarrollo de Magdalena
Medio (PPDMM).
Palacios, Marco. 2006. Between Legitimacy and Violence. Durham, NC: Duke Univer-
sity Press.
Patterson, Anne W. 2000. “From the Outside Looking In: Meta/Caquetá Residents View
the Despeje.” Confidential Cable, 010660. United States Embassy, Colombia;
National Security Archive, December 4, 2000.
Patterson, Anne W. 2002. “Colombia’s Meta Department–FARC Engaged in Hostile
Activities/Paramilitaries Not a Factor.’” Confidential Cable 001202. United States
Embassy, Colombia; National Security Archive, February 7, 2002.
Pearce, Jenny. 1997. “Sustainable Peace-Building in the South: Experiences from Latin
America.” Development in Practice 7(4): 438.
Pearson, Craig. 2009. “New Afghan Strategy Tackles One Village at a Time.” CanWest
News Service. Available at: www.canada.com.
Peña, Carina. 1997. “La Guerrilla Resiste Muchas Miradas: El Crecimiento de las FARC
en los Municipios Cercanos a Bogotá: Caso del Frente 22 en Cundinamarca.” Análisis
Político SEP/DIC 1997(No. 32): 83–99.
Peñaranda, Ricardo. 2006. “Resistencia Civil y Tradiciones de Resistencia en el Sur-
occidente Colombiano.” In Nuestra Guerra Sin Nombre: Transformaciones del Con-
flicto en Colombia, eds. María Emma Wills, Francisco Gutiérrez and Gonzalo Sánchez
Gómez. Bogotá: Grupo Editorial Norma, Instituto de Estudios Políticos y Relaciones
Internacionales (IEPRI).
References 353

Perlez, Jane and Pir Zubair Shah. 2008a. “Pakistan Uses Tribal Militias in Taliban
War.” The New York Times, October 24, 2008.
Perlez, Jane and Pir Zubair Shah. 2008b. “As Taliban Overwhelm Police, Pakistanis Hit
Back.” The New York Times, November 2, 2008.
Perlez, Jane. 2009. “Taliban Seize Vital Pakistan Area Closer to the Capital.” The New
York Times, April 23, 2009.
Petersen, Roger. 2001. Resistance and Rebellion: Lessons from Eastern Europe.
Cambridge: Cambridge University Press.
Pizarro León-Gómez, Eduardo, Ricardo Peñaranda and Pierre Gilhodes. 1991. Las Farc
(1949–1966), de La Autodefensa a La Combinación de Todas Las Formas de Lucha.
Bogotá: Tercer Mundo Editores.
Popkin, Samuel L. 1979. The Rational Peasant: The Political Economy of Rural Society
in Vietnam. Berkeley, CA: University of California Press.
Presidencia de la República de Colombia. 2013. “Presidente Santos, en la entrega del
documento ‘De la violencia a la sociedad de los derechos.’” December 10, 2013.
Project Ploughshares. 2013. “Philippines-CPP/NPA.” Project Ploughshares. http://plough
shares.ca/pl_armedconflict/philippines-cppnpa-1969-first-combat-deaths.
Putnam, Robert D. 1993. Making Democracy Work: Civic Traditions in Modern Italy.
Princeton, NJ: Princeton University Press.
Quitasol, Kimberlie. 2014. “Villagers Help Search for Rebel Camp in Sagada.” Philip-
pine Daily Inquirer, May 16, 2014.
Raghavan, Sudarsan. 2010. “In Somalia’s War, a New Challenger Is Pushing Back
Radical al-Shabab Militia.” The Washington Post, May 27, 2010.
Ramírez, María Clemencia. 2001. Entre el Estado y la Guerrilla: Identidad y Ciudada-
nía en el Movimiento de los Campesinos Cocaleros del Putumayo. Bogotá, Colombia:
Instituto Colombiano de Antropología e Historia, Colciencias.
Ramírez, Sara and Christopher Mitchell. 2009. “Local Peace Communities in Colombia:
An Initial Comparison of Three Cases.” In Colombia: Building Peace in a Time of
War, ed. Virginia Bouvier. Washington, D.C.: USIP Press.
RECORRE. 2007. Red de Comunidades en Ruptura y Resistencia (RECORRE)
www.prensarural.org/recorre, 2007 [cited 2007].
Rempe, Dennis M. 1999. “The Origin of Internal Security in Colombia: Part I — A CIA
Special Team Surveys La Violencia, 1959–60.” Small Wars & Insurgencies 10(3):
24–61.
Restrepo, Gloria Inés. 2005. “Dinámicas e Interrrelaciones en los Procesos de Resis-
tencia Civil: Estudio de Caso Comparado de los Procesos de Resistencia Civil
Organizada de la Asociación de Trabajadores Campesinos del Carare y la Comu-
nidad de Paz de San José de Apartadó.” B.A. thesis, Universidad Nacional de
Colombia.
Restrepo, Gloria Inés. 2006. “Dinámicas e Interacciones en los Procesos de Resistencia
Civil. Estudio de Caso Comparado entre la Comunidad de Paz de San José de
Apartadó y la Asociación de Trabajadores Campesinos del Carare.” Revista Colombi-
ana de Sociología 27(2).
Restrepo, Jorge Alberto and Michael Spagat. 2006. “El Conflicto en Colombia: Quién
Hizo Qué a Quién? Un Enfoque Cuantitativo (1988–2003).” In Nuestra Guerra sin
Nombre: Transformaciones del Conflicto en Colombia, eds. María Emma Wills
Francisco Gutiérrez and Gonzalo Sánchez Gómez. Bogotá: Grupo Editorial Norma,
Instituto de Estudios Políticos y Relaciones Internacionales (IEPRI).
354 References

Rettberg, Angelika. 2006. Buscar la paz en medio del conflicto, un propósito que no da
tregua: un estudio de las iniciativas de paz en Colombia (Desde los años 90 hasta hoy).
Bogotá: Universidad de los Andes.
Reuters. 2014. “Philippines, Muslim Rebels Sign Final Peace Deal to End Conflict.”
Reuters, March 27, 2014.
Richani, Nazih. 2002. Systems of Violence: The Political Economy of War and Peace in
Colombia. Albany, NY: SUNY Press.
Roldán, Mary. 2002. Blood and Fire: La Violencia in Antioquia, Colombia, 1946–
1953. Durham, NC: Duke University Press.
Romero, Mauricio. 2003. Paramilitares y Autodefensas, 1982–2003. 1st ed. Bogotá, D.
C.: Instituto de Estudios Políticos y Relaciones Internacionales, Universidad Nacional
de Colombia: Editorial Planeta Colombiana.
Romero, Simon. 2008. “Two Colombias, at War and at Peace.” The New York Times,
September 7, 2009.
Romero, Simon. 2009. “An Isolated Village Finds the Energy to Keep Going.” The New
York Times, October, 16, 2009.
Rubin, Alissa J. 2010a. “As U.S. Frees Detainees, Afghans Ask Why They Were Held.”
The New York Times, March 19, 2010.
Rubin, Alissa J. 2010b. “Taliban Overhaul Image to Win Allies.” The New York Times,
January 21, 2010.
Rubin, Alissa J. and Richard A. Oppel Jr. 2010. “U.S. and Afghanistan Debate More
Village Forces.” The New York Times, July 12, 2010.
Rubio, Mauricio. 1997. “Perverse Social Capital – Some Evidence From Colombia.”
Journal of Economic Issues 31(3).
Ruhl, J. Mark. 1980. “The Military.” In Politics of Compromise: Coalition Government
in Colombia, eds. Ronald G. Hellman R. Albert Berry. New Brunswick, NJ: Transac-
tion Books.
Saeed, Abdirahim. 2012. “Syria: An Attempt to Live a Normal Life.” BBC, December
24, 2012.
Salazar, Boris and María del Pilar Castillo. 2001. La Hora de Los Dinosaurios: Con-
flicto y Depredación en Colombia. Bogotá: CEREC, Universidad del Valle.
Sánchez Torres, Fabio. 2007. Las Cuentas de la Violencia: Ensayos Económicos Sobre el
Conflicto y el Crimen en Colombia. Bogotá: Grupo Editorial Norma: Universidad de
los Andes, Facultad de Economía.
Sánchez, Fabio and María Del Mar Palau. 2006. “Conflict, Decentralisation, and Local
Governance in Colombia, 1974–2004.” CEDE Working Paper. Bogotá: Universidad
de Los Andes.
Sánchez, Gonzalo, and Donny Meertens. 2001. Bandits, Peasants, and Politics: The
Case of “La Violencia” in Colombia. Austin: University of Texas Press, Institute of
Latin American Studies.
Sandoval, Luis. 2004. La Paz en Movimiento. Vol. 1 and 2. Bogotá: ISMAC.
Sands, Phil. 2011. “Syria’s Palestinians Try to Stay Neutral amid Turmoil.” The
National (UAE), May 5, 2011.
Sands, Phil. 2012. “Syria’s Druze community: A Silent Minority in No Rush to Take
Sides.” The National (UAE), February 22, 2012.
Sanford, Victoria. 2003. “Peacebuilding in a War Zone: The Case of Colombian Peace
Communities.” International Peacekeeping 10(2): 107–118.
References 355

Sanford, Victoria. 2004. “Contesting Displacement in Colombia: Citizenship and State

Sovereignty at the Margins.” In Anthropololgy in the Margins of the State, eds. Veena
Das and Deborah Poole. Santa Fe: School of American Research.
Santos, Jr., Solimon. 2005. Peace Zones in the Philippines: Concept, Policy and Instru-
ments. Quezon City: Gaston Z. Ortigas Peace Institute.
Sarmiento Anzola, Libardo. 1998. Municipios y Regiones de Colombia: Una Mirada
Desde la Sociedad Civil. Bogotá: Fundación Social, Vicepresidencia de Planeación.
Schmidt, Eric, Mark Mazzetti, and Carlotta Gall. 2007. “U.S. Hopes to Use Pakistani
Tribes against Al Qaeda.” The New York Times, November 19, 2007.
Schock, Kurt. 2005. Unarmed Insurrections: People Power Movements In Nondemoc-
racies. Minneapolis, MN: University of Minnesota Press.
Scott, James C. 1985. Weapons of the Weak: Everyday Forms of Peasant Resistance.
New Haven, CT: Yale University Press.
Scott, James C. 1992. Domination and the Arts of Resistance: Hidden Transcripts. New
Haven, CT: Yale University Press.
Seawright, Jason and John Gerring. 2008. “Case Selection Techniques in Case Study
Research: A Menu of Qualitative and Quantitative Options.” Political Research
Quarterly 61(2): 294–308.
Semelin, Jacques. 1993. Unarmed against Hitler: Civilian Resistance in Europe, 1939–
1943. Westport, CT: Praeger.
Semple, Michael. 2009. Reconciliation in Afghanistan. Washington, D.C.: US Institute
of Peace Press.
Sepúlveda Roldán, David. 2004. Saiza: Esplendor y ocaso, un pueblo fantasma del
Nudo de Paramillo. Tierralta (Córdoba): Corpocodesa, ACNUR, Opción Legal.
Sewall, Sarah et al. 2007. The U.S. Army/Marine Corps Counterinsurgency Field
Manual. University of Chicago Press Ed. University of Chicago Press.
Shanker, Thom. 2010. “Joint Chiefs Chairman Readjusts Principles on Use of Force.”
The New York Times, March 3, 2010.
Sharp, Gene. 1973. The Politics of Nonviolent Action. Boston: P. Sargent Publisher.
Sharp, Gene and Joshua Paulson. 2005. Waging Nonviolent Struggle: 20th Century
Practice and 21st Century Potential. Manchester, NH: Extending Horizons Books.
Shinwari, Ibrahim. 2008. “Elders Refuse to Form Lashkar in Landi Kotal.” Available at:
www.dawn.com/2008/10/07/local6.htm.
Simmons, Beth A. 2009. Mobilizing for Human Rights: International Law in Domstic
Politics. 1st ed. Cambridge University Press.
Sivard, Ruth L. 1993. World Military and Social Expenditures 1993. Washington, D.C.:
World Priorities.
Sly, Liz. 2013. “Syrian War Makes Sudden Appearance at Convent in Historic Christian
Town.” The Washington Post, September 9, 2013.
Snapp, Trevor. 2010. “Sudan’s ‘Arrow Boys’ Challenge Militants.” National Public
Radio, August 2, 2010.
Solomon, Erika. 2012. “In War-Torn Syria, Secrets and Double Lives.” Reuters,
November 14, 2012.
Solomon, Erika. 2014. “Special Report: Amid Syria’s Violence, Kurds Carve out Auton-
omy.” Reuters, January 22, 2014.
de Sousa Santos, Boaventura and Mauricio García Villegas. 2004. Emancipación
Social y Violencia en Colombia. 1st ed. Bogotá: Grupo Editorial Norma.
356 References

Starn, Orin. 1999. Nightwatch: The Politics of Protest in the Andes. Durham, NC: Duke
University Press.
Stoll, David. 1993. Between Two Armies in the Ixil Towns of Guatemala. New York:
Columbia University Press.
Stors, Michel. 2013. “‘Their Agendas Have Ruined Us’: Religious Tolerance the Latest
Casualty of Syrian Conflict.” National Post, February 20, 2013.
Sudarsky R., John. 2007. La Evolución del Capital Social en Colombia, 1997 - 2005.
Bogotá: Fundación Antonio Restrepo Barco.
Suleman, Mohammad and Sue Williams. 2003. “Strategies and Structures in Preventing
Conflict and Resisting Pressure: A Study of Jaghori District, Afghanistan, under
Taliban control.” Steps Toward Conflict Prevention Project (STEPS) working paper,
Collaborative for Development Action (CDA), March 2003.
Taj, Farhat. 2011. Taliban and Anti-Taliban. Newcastle upon Tyne: Cambridge
Scholars Publishing.
Tarrow, Sidney. 1994. Power in Movement: Social Movements, Collective Action and
Politics. New York: Cambridge University Press.
Tarrow, Sidney. 2007. “Inside Insurgencies: Politics and Violence in an Age of Civil
War.” Perspectives on Politics 5(3): 587–600.
Tavanti, Marco. 2003. Las Abejas: Pacifist Resistance and Syncretic Identities in a
Globalizing Chiapas. New York: Routledge.
Tavernise, Sabrina and Irfan Ashraf. 2009. “Attacked, Pakistani Villagers Take on
Taliban.” The New York Times, June 10, 2009.
Taylor, Michael. 1988. “Rationality and Revolutionary Collective Action.” In Ration-
ality and Revolution, ed. Michael Taylor. New York: Cambridge University Press.
Thoumi, Francisco E. 1997. Drogas Ilícitas en Colombia: Su Impacto Económico,
Político y Social. Santa Fé de Bogotá: Planeta Colombia Editorial, S.A.
Tice, Austin. 2012. “In Syria, an Oasis from the War.” The Washington Post, July 15,
2012.
Tilly, Charles. 1992. Coercion, Capital and European States: AD 990–1992. Cam-
bridge, MA: Wiley-Blackwell.
Time Magazine. 1964. “Colombia: Death of Black Blood.” Time, May 8, 1964.
Tobler, W. R. 1970. “A Computer Movie Simulating Urban Growth in the Detroit
Region.” Economic Geography 46: 234–240.
Todd, Molly. 2010. Beyond Displacement: Campesinos, Refugees, and Collective
Action in the Salvadoran Civil War. Madison, WI: University of Wisconsin Press.
Torres Restrepo, Camilo. 1963. “La Violencia y los Cambios Socio-Culturales en las
Areas Rurales Colombianas.” Asociación Colombiana de Sociologia: Memoria del
Primer Congreso Nacional de Sociologia: 95–152.
Torres Sánchez, Jaime and Fabio Barrera Téllez. 1982. Colombia Represión, 1970–
1981. 2 vols. Bogotá: CINEP.
Torres III, Wilfredo Magno, ed. 2007. Rido: Clan Feuding and Conflict Management in
Mindanao. Makati City: The Asia Foundation.
Triana y Antorveza, Humberto. 1966. La Acción Comunal en Colombia: Resultados de
una Evaluación en 107 Municipios. Bogotá: Ministerio de Gobierno, Direccion Gen-
eral de Integracion y Desarrollo de la Comunidad.
Tulio Rodríguez, Marco. 1982. Los Municipios Olvidados. Bogotá: El Tiempo.
Tyson, Ann Scott. 2008. “New Joint Effort Aims to Empower Afghan Tribes to Guard
Themselves.” The Washington Post, March 31, 2008.
References 357

Tzu, Sun. 2005. The Art of War. Boston, MA; London: Shambhala.
U.S. Department of State. 2000. “1999 Country Reports on Human Rights.” Bureau of
Democracy, Human Rights and Labor, February 25, 2000. Available at: www.state
.gov/www/global/human_rights/1999_hrp_report/colombia.html.
United Nations. 1998–2006. Cultivos Ilícitos de Coca: Estadísticas Municipales. SIMCI,
United Nations.
United Nations Development Program (UNDP). 2003. El Conflicto, Callejón con Salida:
Informe Nacional de Desarrollo Humano para Colombia – 2003. United Nations
Development Program, ed. Hernando Gómez Buendía.
United Nations High Commissioner for Refugees (UNHCR). 2010. “UNHCR Global
Report 2010.”
United Nations High Commissioner for Refugees (UNHCR). 2014. “Syria Tops World
List for Forcibly Displaced after Three Years of Conflict.” March 14, 2014.
USAID. 2009. “Religion, Conflict and Peacebuilding Toolkit.” Available at: www.usaid
.gov/our_work/cross-cutting_programs/conflict/publications/docs/Religion_Conflict_
and_Peacebuilding_Toolkit.pdf
Valentino, Benjamin. 2004. Final Solutions: Mass Killing and Genocide in the Twentieth
Century. Ithaca, NY: Cornell University Press.
Vargas Velásquez, Alejo. 1992. Magdalena Medio Santandereano. Bogotá, D.C.,
Colombia: CINEP.
Varshney, Ashutosh. 2002. Ethnic Conflict and Civic Life: Hindus and Muslims in
India. 2nd ed. New Haven, CT: Yale University Press.
Vásquez Delgado, Teófilo. 2013. Territorios, conflicto armado y política en el Caguán:
1900–2010. Universidad de los Andes, Master’s thesis, Department of Geography,
January 22, 2013.
Verdadabierta.com. 2009. “El Pájaro’ señala a capitán de la Policía de nexos con
‘paras’ y ‘narcos.’” August 25, 2009.
Vicepresidencia de la República de Colombia. 2001. “Iniciativas Comunitarias de
Paz en Colombia: Semillas que Abren el Camino de la Paz. Fondo de Impresiones
para la Paz.” Observatory on Human Rights in Colombia Monthly Newsletter
May 2001(15).
Vicepresidencia de la República de Colombia. 2003a. “Los Derechos Humanos en el
Departamento de Meta.” Programa Presidencial de Derechos Humanos.
Vicepresidencia de la República de Colombia. 2003b. “Panorama Actual de los Muni-
cipios que Conformaron la Zona de Distensión.” Programa Presidencial de Derechos
Humanos, December 2003.
Villarraga Sarmiento, Álvaro. 2003. Vida, Dignidad y Territorio: Comunidades de Paz y
Zonas Humanitarias en el Urabá y Atrato. 1st ed. Bogotá D.C.: Fundación Cultura
Democrática; Ministerio del Interior; Programa de las Naciones Unidas para el
Desarrollo; Programa por la Paz, Compañía de Jesús; Fondo de las Naciones Unidas
para la Infancia; Instituto de Estudios para el Desarrollo de la Paz.
Washington Office on Latin America (WOLA). 2011. “In La Macarena, Colombia, a
Program on ‘Autopilot.’” June 10, 2011, video interview. www.wola.org/commen
tary/in_la_macarena_a_program_on_autopilot.
Weinstein, Jeremy. 2006. Inside Rebellion: The Politics of Insurgent Violence.
Cambridge: Cambridge University Press.
Wilkinson, Isambard and Lakki Marwat. 2008. “Pakistan Chieftain Hunts Down
Taliban.” Telegraph (UK), September 22, 2008.
358 References

Wills, María Emma, Francisco Gutiérrez and Gonzalo Sánchez Gómez, eds. 2006.
Nuestra Guerra Sin Nombre: Transformaciones del Conflicto en Colombia. Bogotá:
Grupo Editorial Norma, Instituto de Estudios Políticos y Relaciones Internacionales
(IEPRI).
Wilson, K. B. 1992. “Cults of Violence and Counter-Violence in Mozambique.” Journal
of Southern African Studies 18(3): 527–582.
Wirpsa, Leslie, David Rothschild and Catalina Garzón. 2009. “The Power of the
Bastón: Indigenous Resistance and Peace Building in Colombia.” In Colombia: Build-
ing Peace in a Time of War, ed. Virginia Bouvier. Washington, D.C.: USIP Press.
Wood, Elisabeth. 2003. Insurgent Collective Action and Civil War in El Salvador. New
York: Cambridge University Press.
World Bank. 2000. Violence in Colombia: Building Sustainable Peace and Social
Capital. World Bank Publications.
World Bank. 2004. Voices of the Poor in Colombia: Strengthening Livelihoods, Fam-
ilies, and Communities, eds. Jairo A. Arboleda and Patti L. Petesch. Washington D.C.:
World Bank.
Yarbro, Stan. 1990. “Village in Colombia Battles Violence, Wins Peace Prize.” Christian
Science Monitor, December 4, 1990.
Zamora, Gloria Lucy. 1983. Los Moradores de la Represión: en el Magdalena Medio.
Bogotá: Centro de Investigaciones y Educación Popular.
Zamosc, Léon. 1982. Los usuarios campesinos y las luchas por la tierra en los años 70.
Bogotá: CINEP.
Zamosc, Léon. 1986. The Agrarian Question and the Peasant Movement in Colombia:
Struggles of the National Peasant Association, 1967–1981. Cambridge: Cambridge
University Press.
Zamosc, Leon. 2001. “Peasant Struggles of the 1970s in Colombia.” In Power and
Popular Protest: Latin American Social Movements, 2nd ed., eds. Susan Eckstein
Merino and Manuel Antonio Garretón. Berkeley: University of California Press.
Index

ACIA, 74 Anbar Province, 28

ACR, 103 Anolaima, 170–171, 180, 220, 225
Afghan National Army (ANA), 286 Antioquia, 53, 74, 82, 104, 123, 138
Afghanistan, 26–27, 59 ANUC, 13, 131
armed self-defense, 290 activism, 132
autonomy actions, 287 effects, 71
civilian resistance, 287–288 measurement, 132
community-release program, 290 militancy, 132
conflict, 271, 286–287 stigma, 132
government, 287, 289 Arauca, 138
organizational capacity, 288 Armed Forces of the Philippines. See AFP
social cohesion, 287 armed groups
territorial control, 288 attitudes toward violence, 36, 40, 55, 57, 60,
tribal jirga, 285 116–117, 256
tribal structures, 287 behavior, 8, 10, 18, 23, 26, 34, 60, 127, 255,
U.S. war, 310 266, 303
AFP, 276, 281, 284–285, 296 abusive, 54
Afro-Colombian norms, 54–55
autonomy, 74 violent behavior, 18, 57, 184, 256
collective territories, 74 characteristics, 55–60
conflict resolution, 74 choice set, 26, 35
consejos comunitarios, 74, 130, 157 civilian collaboration, 48
organizations, 13, 25, 112 civilian response, 16
population, 74, 130 coercion, 9, 11, 19–20, 35, 43, 49, 60, 71,
agricultural cooperatives, 13, 39, 107, 110, 134, 208
131, 281 competition, 34
al-Assad, Pres. Bashar, 291 control, 4, 18, 23
Alawite, 292, 295 territorial, 55–56, 114, 189
Aleppo, 295 urban, 15
al-Nusra Front, 292–294, 297 costs, 11, 57
Al-Qaeda, 292, 299 military costs, 51
Amazon region, 138 of using violence, 11, 47, 59, 120, 258
Amazonas (department), 138 reputation, 257

359
360 Index

armed groups (cont.) ASOPROA (Association of Small- and

credibility, 60 Medium-Scale Producers of Eastern
decision-making, 18, 24–25, 107, 256 Antioquia ), 82
economics, 58 ATCC, 3, 13, 24–25, 75, 100, 168, 173
extortion, 56 archives, 33, 92, 108, 185
homicide rate, 63–64, 128 armed group presence, 190–191
hybrid groups, 55, 126, 129 autonomy, 191, 269
ideology, 50, 55–58 civilian preferences, 207
incentives, 10–11, 22, 35, 48, 54, 114, 123, clientelism, 268
129, 208 collaboration with armed groups, 25, 195,
for civilians, 36 197, 207
to commit violence, 59 collective action, 75
internal debates, 258 conciliation, 33, 186, 195–199, 206, 217
kidnapping, 47, 56, 67, 137, 167, 257 denunciations, 206
legitimacy, 50, 54, 56, 59 dispute resolution, 184, 194, 208
macro-politics, 23 effect on violence, 184, 187, 189, 194, 206,
motives, 11, 22, 57, 199, 242 301
organizational capacity, 56 false accusations, 184, 189, 195, 198,
preferences, 11, 17, 43, 54–57, 61, 174, 217, 208
299, 302 founding, 73, 183–184
pressure on civilians, 4, 12, 43, 48, 53, 217, informants, 206
242, 266 information systems, 75, 194, 198–199,
protection taxes, 45, 67, 212, 245 210
recruitment, 18, 23, 36 institutional procedures, 25, 184–185,
strategies, 56 194–209, 212
youth, 47 intervention, 207, 215
recruitment of civilians, 48, 56, 260 interviews, 183, 190, 196
youth recruitment, 3 investigations, 24, 75, 194, 196, 199,
relationship with civilians, 2, 4, 8, 18–20, 35, 206–208, 217, 301, 309
42, 45, 50, 54–60, 106, 128, 242, 251, junta councils, 184, 268
255, 306, 311, 313 junta directiva, 185, 194, 198, 216
dependence, 11, 56, 58, 256 leadership, 187, 193
repression, 50, 79, 203, 210, 217, 221, 246, levels of violence, 189
289 meetings, 207–208, 210, 215
military, 73 negotiation with armed groups, 184, 208
reputation concerns, 11, 49–50, 56, 60, 257 neighboring regions, 185, 210
resource bases, 11, 17, 55–57, 116, 188, neutrality, 187, 206–207, 212–213
322–323 norms of nonviolence, 207
response to autonomy organizations, 77 population, 213
selective violence, 14, 18, 20, 57, 112, 114, protection taxes, 193
189, 191, 194, 213 resurgence in violence, 212–216
sensitivity, 11, 185, 209, 257, 302–303, 314 role of junta councils, 14
shaming of, 10, 50 strategies, 184, 187, 208, 217, 267
strategic dilemma, 59 territorial control, 190, 193
targeting of civilians, 17 threats, 194–207
threats, 212 victims, 203, 212–213, 216
violation of autonomy, 79 weakening, 215
armed resistance, 7, 16–17, 42, 46, 51, 53, 312 youth, 215
risks, 51 ATCC territory, 185–186, 188, 203, 215
ARMM, 277 armed group control, 188
Asociación de Trabajadores Campesinos del armed group preferences, 206, 210
Carare. See ATCC AUC presence, 186, 210
Index 361

coca cultivation, 191, 199, 207, 212–213, effects on armed group behavior, 45, 54–60
216 formal organizations, 23, 71–77, 90
control, 189 characteristics, 74
displacement. See displacement gender, 75–76
drug trafficking, 303 history, 73
ELN presence, 186 implications for foreign policy, 298, 311
FARC presence, 186, 190–192, 211–212 in Afghanistan. See Afghanistan
migration, 213 in Colombian conflict, 304
military control, 189 in conflict settings, 5, 8–9, 17, 46, 266,
violence, 187, 189–194 270
Atlántico, 138 in Philippines. See Philippines
Atrato River, 73–74 in Syria. See Syria
AUC, 1, 67, 198 international case studies, 7–8, 270, 276
blocs, 104 limitations, 295
counterinsurgency operations, 67, 211 literature, 9
defection concerns, 211 mechanisms, 4, 11, 14–16, 25, 34–35, 42–60,
demobilization, 67–68, 89, 303 112, 118, 120, 134, 156, 171, 217, 267,
displacement 269, 271, 278, 301, 312
organized, 71 NGO support, 73
drug trafficking, 67 organizations, 4, 22
expansion, 89 policies, 76
extra-lethal violence, 67 political, 19, 78, 259, 269, 277, 296
Colombia, 67 indigenous groups, 73
formation, 128, 324 research, 23–24, 108, 271
incentives, 210 rural, 15
interviews, 210 self-rule, 4, 34, 45
massacres, 67 theory, 9–10, 25, 35, 40, 118, 161, 174, 221,
motivations, 210–212 246, 271, 299–300, 310, 312
origins, 67 vs. neutrality, 46
relationship to junta councils, 3
relationship with civilians, 71, 210–212, 313 BACRIM, 68, 314
treatment of junta councils, 127–129, formation, 68
210–211, 244–245, 251, 266 balance of control, 45
violence, 210 theory, 114, 117, 189, 193, 217, 262
Autonomous Region in Muslim Mindanao bazaar, 99, 110, 132–133
(ARMM). See ARMM beer, 93, 97, 111, 238, 243–245
autonomy Belén de Los Andaquies, 132–133, 301
“de facto” autonomy, 34, 118, 129 interviews, 132
definition, 54 junta councils, 132
and decentralization, 310 Bituima, 24, 26, 93, 220
armed strategies, 17 armed group presence, 245–246, 266
autonomy organizations, 15, 71, 112 AUC, 265
origins, 89 Catholic Church, 7, 239
relationship with armed groups, 77–81 cattle theft, 227, 241
benefits, 259 clientelism, 261
challenges, 4 collaboration with armed groups, 265
civilian autonomy, 21–24, 299 collective resistance, 251, 267
conditions for, 11, 15, 23, 35, 45, 63, 84, 242 conflict dynamics, 251–254
decision-making, 43, 246 conflict resolution, 241
declarations, 45, 74, 76, 254 cooperation, 239
definition, 34, 45–46, 300 culture, 223, 240, 261
effect on violence, 9, 16, 47, 302–304 FARC presence, 251
362 Index

Bituima (cont.) Santander, 271

junta councils, 170, 223, 239, 248, 266, 301 social conditions, 220–221, 226
formation, 239 Castaño, Carlos, 67
role, 240 Catholic Church, 97, See priests
La Violencia, 226 accompaniment, 75
land ownership, 240 in Philippines, 7
macheteras, 240 relationship with junta councils, 111, 157
organizations, 240, 251, 266, 268 role in Colombian conflict, 63
poverty, 240 cattle theft, 99, 234, 238, 249
protection, 251 Cauca, 28, 65, 73–74, 123, 136, 138, 218, 248
social relations, 223, 240, 248, 251–252, 267 cease-fires
ties to Vianí, 223 local, 32, 60, 209
victims, 265 Musa Qala, 279
violence, 262, 265–267 Philippines, 279
Bogotá, 24, 26, 63, 94, 97, 99, 102, 110, 171, Syria, 279
185, 199, 219–220, 222, 234, 236, 242, Central Andean region, 138
245, 264, 271 Cesar, 123, 138
Bolívar, 104, 123, 138, 213 Chicoral, Pact of, 73
Bolivarian Movement for a New Colombia child soldiers, 69
(MBNC), 65 chisme. See gossip
Boyacá, 104, 123, 136, 138 Chocó, 53, 73–74, 136, 138
Buner, 29, 270, 288 Chusma, 224–225
Chwe, Michael Suk-Young, 37
Cacarica, 53 Cimitarra, 94, 186, 210
CAFGU, 276, 281 CINEP, 68, 71, 113, 116–117, 120–121,
Caldas, 104, 123, 138, 156 126–127, 139, 142–143, 154, 176,
campesino, 1–2, 13, 24, 64, 75, 93, 95, 103, 191–192, 203, 262–263
111–112, 130, 174, 191, 211–212, 240, support for peace communities, 73
243–244, 247, 251, 299 Citizen Armed Force Geographical Units
autonomy, 73–74 (CAFGU). See CAFGU
in case study towns, 220 civil war, 2, 7
organizations, 74 civilian preferences, 43
Caquetá, 65, 132, 138, 275 civilian strategies, 270
Carare River, 3, 183, 186, 200 civilians in civil war context, 111, 307, 311
Caribbean Coast region, 138 conflict conditions, 3, 11, 18, 40, 54–55, 60,
CARP-SPOT framework, 52 270
Casanare, 104, 123, 136, 138, 154 dynamics, 8, 12, 17, 22, 44, 100
case studies, 312 literature, 18–21, 35, 55
case study towns, 14, 84, 91, 93, 103, 169 studies, 18, 23, 91, 166, 300
armed group presence, 26, 91, 171, 219, 242, victims, 3
266 violence, 8
characteristics, 169 Civilian Mandate for Peace, 71
colombian military, 221 civilian organizations, 5, 8, 18
comparison, 266 apolitical, 42
conditions, 220 autonomy organizations, 5
conflict patterns, 86, 171 behavior, 23, 84
Cundinamarca, 86, 93, 169, 220, 268, 271 benefits, 40
case selection, 24 capacity to limit violence, 17, 35, 49, 54,
selection, 25 59–60, 111, 134, 168, 299–300, 307,
geography, 221 312
history, 267, 305 challenges, 8, 36, 109, 217, 223
levels of violence, 261 characteristics, 34, 40–42
Index 363

civilian agency, 42 alignment, 35

collective action, 8–9, 15, 19, 21, 23, 35–36, contentious politics, 10, 51–54, 60, 278
39, 61, 216, 266, 299, 307 limitations, 52
challenges, 36–38 peace, 71
implementation, 38, 40–41 protection, 276
over time, 71 selection, 9, 42, 78, 312
strategies, 37 strength, 21, 36, 40, 270
collective processes, 15, 18–19, 23 structures, 130
benefits, 17 targeting of, 40, 42
communal action program, 110 threats, 15
conditions, 12, 17, 60, 90, 267, 299 types, 16, 128, 130–132, 155–156
conflict resolution, 27 civilian’s dilemma, 4, 36, 38, 183
cooperation, 301 civilians
decision-making, 9, 25, 40, 52 agency, 3, 11, 19, 21, 23–24, 35, 85, 109,
detainees, 290 217, 299–300, 302, 310
dialogues with armed groups, 5, 50 alignment strategies, 43
early warning systems. See early warning behavior, 18
systems theory of, 60
effect of armed conflict, 313 choice set, 15, 34, 52, 86, 302, 307
effect on violence, 253 collaboration with armed groups, 19
formal organizations, 301 daily life, 93
formation, 8, 112 innovation, 271
ideology, 43 political affiliations, 15
information systems, 19, 41, 47, 49, 206 preferences, 15, 36, 43, 311–312
internal dynamics, 313 resistance, 106, 112, 254
investigation capacity, 49, 60, 185, 218, role in conflict, 19
239 social capital, 35
leadership, 37, 40–41, 48, 313 social divisions, 47, 60, 255
legitimacy, 42 stereotypes, 20, 34
limitations, 8, 18, 61, 127, 215, 217, 313 clientelism, 26, 64, 86, 223, 258–261, 268,
memory, 41 288, 303, See ATCC, Bituima, Quipile,
negotiation with armed groups, 41, 53, 60 Vianí
organizational capacity, 11, 16, 37, 127, 269, effect on junta councils, 258, 260, 269
311 coca, 3, 56, 323
origins, 12, 71 and violence, 117, 213, 215
peace organizations, 13 cultivation, 116–117, 129, 207
political neutrality, 16 economy, 185, 213, 275, 303
preexisting conditions, 39–41, 83, 155, eradication, 216, 274
266 cocaine, 323
pre-La Violencia, 80 production, 67, 116
protection, 15, 24, 60, 73, 79, 118, 309 trade, 59, 162
strategies, 34, 218, 299 cocalero (coca grower), 45, 53, 117
relationship with armed groups, 306 COCOMACIA, 53, 75
resistance, 19, 22, 217, 242, 246, 256, 307 coffee, 26, 93, 171, 219, 229, 234, 237,
responses to violence, 10, 36, 40, 43, 111, 245
266 collaboration with armed groups, 189
risks, 36 accusations, 60, 185
self-defense, 71 denunciations, 301
social cohesion, 86 fear of, 255
social relations, 13, 48 prevention, 48
stigmatization, 16 stigma, 46, 255
strategies, 8–9, 11–12, 21, 52, 302, 307 threats, 212
364 Index

collaboration with armed groups (cont.) role in conflict, 67

violent consequences, 128 violence against civilians, 92
collective action, 23, 248 Colombian National Police, 78, 113
challenges, 4 police inspector, 238, 241, 265
problem, 300 reforms, 68
theories, 12 relationship with junta councils, 121
Colombia colonization
as failed state, 68 Colombia, 70
as unique case, 87 LeGrand, Catherine, 70
history, 24–25, 156, 161, 304 Philippines, 277
social, 61, 84 Comité de Cafeteros (Committee of Coffee
political parties, 63, See Conservative, Growers), 229, 236–237
Liberal, Unión Patriótica Communist People’s Party (CPP). See CPP
politics Communities of Population in Resistance, 28
institutional change, 259 community-driven development, 304, 315
terrain, 63, 70 conciliation
Colombian Reintegration Agency (ACR): ATCC, 186, 195–199
102–103, 105 junta councils, 82, 134, 273
Colombian Clandestine Communist Party techniques, 206, 241
(PCCC). See PCCC Vianí, 238
Colombian Communist Party, 64, 71, 126 confirmation bias, 107
Colombian conflict, 25, 62, 310 conflict resolution, 9, 306
case studies, 8 Conservative Party, 1, 63–64, 225–226
civilian responses, 63 Constitution of 1991, 67, 73
consequences for civilians, 69 Contra-Chusma, 224
de-escalation in 2000s, 68, 89 control, zones of, 242
expansion, 24, 63, 128, 162 complete, 18, 116, 189
history, 44, 62–84, 156, 222 contested, 20, 55–56, 116, 137, 189
international context, 87–88 counterinsurgency, 310
patterns, 14, 61, 63, 70, 87, 161, 168, 174, dominant, 189–190, 193, 202, 210, 242
215, 218, 303, 314 incomplete, 116, 189, 242
in case study towns, 242 shifting, 58
peak violence, 5, 68, 89 control-collaboration theory. See balance of
research on, 86 control
studies, 18 convivencia, 22, 226, 266
victims, 262 cooptation, 83, 247, 259, 275
waves of violence, 63 attempts, 283
Colombian government Córdoba, 45, 53, 67, 138
response to autonomy organizations, countering violent extremism (CVE), 60, 315
324–325 counterinsurgency, 18, 246, 290, 301,
Colombian military, 62 309–312
capacity, 67 civilian support, 310
counterinsurgency operations. Colombian military, 171
See counterinsurgency community-based alternatives, 311
human rights violations, 75, See false positives in Peru, 31
in case study towns, 245 in Philippines, 277
in FARClandia, 275 problems with, 67, 309–310
interviews, 103 CPP, 276
political homicides, 113 CRIC, 73
presence, 68 crime, 56, 59, 67, 116–117, 162, 185, 191, 199,
relationship with civilians, 78 207, 212–213, 243, 254, 263
relationship with paramilitaries, 68 criminal bands. See BACRIM
responses to autonomy organizations, 77, 251 sensitivity of criminal groups, 58, 254
Index 365

criminal bands. See BACRIM individual, 44

crossfire, 3, 20, 23, 26, 51, 54, 83, 186, 218, internally displaced persons, 69
264, 271, 288, 298, 300 organized, 44, 298
Cuban Revolution, 65 resistance to, 44, 298
Cubides, Fernando, 80, 83, 121 victims, 99, 109
culture of peace, 39, 43, 49, 208, 265–266 weakening social cohesion, 44
Philippines: 279–280 do no harm, 26, 36, 315
Syria, 279 drug cartels, 58, 67, 116–117, 162
ATCC, 207 in Mexico, 58
costs for armed groups, 49 Medellín cartel, 162
creation, 48 drug trafficking, 70, 257, 323
definition, 22, 48 role in Colombian conflict, 62
limitations of, 49
culture of resistance, 39 early warning systems, 10, 51, 54, 194, 298, 302
Cundinamarca, 14, 24–26, 64–65, 104, 110, Eastern Andean region, 138, 157
122–123, 138, 154, 156, 169, 172, Eastern region, 138
220–221, 226, 242, 247, 260, 267–269, economic development, 17, 63, 79–80, 132,
272, 275, 281, 301, 305–306 155, 161, 194, 207, 212, 260, See
curfew, 243–244 community-driven development
aid, 312
Dagestan, 31 alternative, 216–217
Damascus, 293, 296–297 international programs, 309
DANE, 131, 136 policy, 80
DAS, 253 El Salvador, 43
data, 305 elevation, 117, 137, 143, 157, 171, 237
availability, 92, 112 ELN, 1, 104
CINEP, 262 history, 162
collection, 100, 305 kidnapping, 67
government, 262 origins, 65–66, 162
historical, 88 political homicides, 113
observational, 89 EPL, 1, 67
on civilian strategies, 312 Estrada, Pres. Joseph, 277, 284
sources, 90 ethnic cleansing, 58
data collection, 313 ethnic minority groups, 301, 331–333
Davao, 276 autonomy, 44
Declaration of Ámbalo, 73 effect on violence, 130, 301
Defensoría, 78, 96 legal autonomy, 130
demilitarized zone, 275, See FARClandia:zona population, 130
de despeje social organizations, 174
demobilization, 281, 289–290, 303 territorial rights, 130
community-based, 60, 209, 303 external actors
Democratic Republic of Congo, 28 best practices, 309
Democratic Security, 83 support for communities, 313
denunciations, 116, 189, 194, 198, 212, 217, extra-lethal violence 67, 69, 126
243 Colombia, 265, 321
DIGIDEC, 80 Syria, 291–292
Dir District, 30, 289
displacement, 3, 19, 22, 96, 137, 223, Fals Borda, Orlando, 80, 110–111
245 false positives, 68
ATCC, 303 FARC, 1
causes, 58 22nd Front, 104, 242
civilian responses, 71 27th Front, 104, 274
data, 126 42nd Front, 104, 242, 254
366 Index

FARC (cont.) Free Syrian Army, 292–293, 297

drug trafficking, 59, 272 Frente Nacional. See National Front
expansion, 65, 162
extortion, 272 Gaitán, Jorge Eliécer, 63
FARC-EP, 65 García Durán, Mauricio, 21, 52, 63, 71, 73
fronts, 104 García Márquez, Gabriel, 62, 93
ideology, 245 Gaviria, Pres. César, 67
interviews, 254–256, 275, See interviews, gender
ex-combatants women as peace activists, 281
kidnapping, 74, 243, 272, 323 genocide, 55
militias, 65 draining the sea, 11, 18, 208
origins, 64–65, 162 Holocaust, 11
political activities, 65 ideology, 58
political homicides, 113 politicide, 126, 303
political strategy, 126, 128–129 resistance to, 17, 57, 126
relationship with junta councils, 129 Rwanda, 31
secretariat, 243 Unión Patriotica. See Unión Patriotica
FARClandia, 26, 271–276 geography, 135
armed group control, 275 of case study towns, 221, 319
autonomy, 271 political, 135, 156
conflict, 272 rough terrain, 157
conflict resolution, 274 Tobler’s Law, 166
cooptation of junta councils, 275 Ghazni Province, 27, 289
interviews, 273 GINI, 123, 128, 177
junta councils, 272–275 gossip, 93, 191, 211, 243, 248, 251
relationship with FARC, 274 as dangerous, 215
paramilitary presence, 275 Greece, 28, 36, 166
political structure, 274 Guainía, 138
territorial control, 272 Guajira, 123, 136, 138
violence, 272 guardabosques (forest rangers), 216
zona de despeje, 271, 322 Guatemala, 7, 28
FATA tribal region, 29, 288 CPRs (Communities of Populations in
Fearon, James, 3, 21, 163, 206 Resistance), 43, 298
fieldwork. See research methodology Guaviare, 138
in conflict zones, 89 guerrilla groups
security concerns, 86, 89–91, 93, 156, 163, characteristics, 79
167 drug trafficking, 67
fog of war expansion, 68, 127
clarification, 189 extortion, 67
coping strategies, 59, 184, 301 forced recruitment
dangers, 18, 49, 59 youth, 244
definition, 3 in Colombian conflict, 62
investigation mechanisms, 25, 185 in El Salvador, 19
foreign fighters, 56, 197 origins, 64–66, 79, 162
foreign policy presence, 162
countering violent extremism. See countering violence against civilians, 128
violent extremism Guevara, Che, 2, 11, 58, 68
in Syria, 291 Guzmán Campos, Germán, 64, 123, 160–161
international aid, 78, 87, 309, 311
interventions, 308 Helmand Province, 27, 287
Plan Colombia. See Plan Colombia homicide rate 63–64
U.S. in Colombia, 68, 311 ATCC territory, 203
Index 367

CINEP, 113 Inter-American Court of Human Rights, 75

civilian, 113 Inter-Church Justice and Peace Commission
in Colombian conflict, 63 (Justicia y Paz), 71
police data, 113, 120, 262 interethnic conflict, 21
variable, 117 International Humanitarian Law, 258
Huila, 65, 123, 133, 138, 162 interviews
human rights, 17–18, 22–24, 26, 68, 78, 82–83, “elite” subjects, 96
129, 163, 189, 191, 207, 216, 223, 262, ATCC, 185, 187
268, 306 case study towns, 221
abuses, 50 ex-combatants, 100, 169, 210, 221, 242,
organizations, 92, 262 254, 256, 101–107
protection, 26 FARC, 104, 110, 275
violations, 3, 18 recruitment, 103
Human Rights Defender. See Defensoría structure, 104
Human Rights Observatory, Colombian junta leaders, 96, 169
Vice-presidency, 78, 116 recruitment of subjects, 97
humanitarian zones, 76 structure, 98–99
techniques, 199
Idlib Province, 297 with ex-combatants, 24
IGOs, 308–309 women, 75
illicit crops, 58, 67, 87, 213, 307, 311, See coca, Iraq, 28
poppy, cultivation Anbar awakening, 8
illicit economy, 55, 63, 257, 275 counterinsurgency, 310
Independent Republics, 304 Mosul, 297
India, 21 U.S. war (Operation Iraqi Freedom), 310
indigenous groups ISIS. See Islamic State
authorities, 74 Islam, 270, 277–278, 284, 297
autonomy, 73–74 Islamic State, 292, 296–298
state response, 78
cabildo councils, 74, 130, 157 Jabhat al-Nusra. See al-Nusra Front
effects of conflict, 74 Jehovah’s Witness
elders, 48 in Mozambique, 7, 29, 303
exclusion, 73 junta councils
justice, 74, 130 ability to protect civilians, 120, 127,
organizations, 25, 39, 112, 218 129
population, 74, 130 activities, 259
protective actions, 74 analysis of, 14
protests, 74 autonomy, 111
resistance, 28, 47, 51, 53, 73 capacity to protect civilians,
armed, 16 81–83
rights movement, 73 clientelism, 14, 167, 312
shamans, 74 collective action, 81
territory, 74 conditions, 268
inequality, 69, 304 cooptation, 83, 247, 259
and social organization, 173 data, 25, 112
and violence, 63, 123, 160, 305 decision-making, 81
land distribution, 123, 157 definition, 13, 80
measurement, 122, 305 dispute resolution, 82
relationship with juntas, 125, 157 distribution, 81, 91, 159, 231
informal institutions, 21, 23 effect on violence, 14, 81, 89, 112, 118,
informants, 20, 49, 60, 103, 116, 167, 127–128, 133–134, 156, 158, 161, 163,
188–191, 200, 210, 216, 245, 247, 264 167, 173, 302, 312
368 Index

junta councils (cont.) daily life, 2, 33–34

formation, 14, 110, 125, 134, 155–158, 161, interviews, 33
173 paramilitary violence, 3
in case study towns, 268 population, 2–3, 12, 33
government support, 79–81, 155, 159, 230, La Violencia, 14, 24–25
235, 237, 259, See promoters bandits, 123, 225–226
growth, 80 casualties, 64
history, 14, 25, 79–83 Chusma. See Chusma
independence, 16 effects, 63, 123, 158–159, 173, 224, 304
information sharing, 121 end, 161, 224
leadership, 237, 241 history, 63, 304
networks of, 14, 156, 159 in case study towns, 224–226, 266
neutrality, 303 levels of violence, 168
organizational capacity, 248 memories of, 99, 224
origins, 25, 79–80, 173, 241 origins, 63
participation, 81 relationship with junta councils, 79, 155,
and Peace Corps. See Peace Corps 158–161, 163, 174
presence, 14, 174 research, 304
public good provision, 14, 39, 80, 112, 122, resistance to, 71
174 waves of violence, 123
relationship with armed groups, 81, 111, Laitin, David, 3, 21, 163, 206
246–247, 256 land reform, 42, 73, 123, 235
relationship with Colombian state, 83 agrarian reform laws, 236, 239
reporting on violence, 84, 121 Bituima, 240, 248, 306
research challenges, 313 councils, 13, 16, 71, 131–132, See ANUC
role in communities, 71, 75, 79, 82–83, 86, and insurgency, 137
111, 133, 248, 256, 259 and social cohesion, 235
role in community, 14 Vianí, 236, 306
role of women, 75 Landázuri, 191
strategies, 82, 167, 301 lashkar (militia), 29, 288–289
strength, 121 latifundio, 227, 236, 239
targeting of leaders, 127, 129, 133, 256 law of silence, 9, 37, 244, 253
variation, 81, 84, 118, 156, 161, 306 Law of the Black Communities (Law 70 of
violence against, 127 1993), 74
weakening, 266, 269 Laws of War. See International Humanitarian
juntas hypothesis, 118, 120, 130, 168 Law
juntas variable, 112, 118, 126, 226 Liberal Party, 1, 62–64, 158, 225–226,
analysis of, 14 237
juntas per capita, 136 Liberation Theology, 43
testing, 89, 120–128, 168 limpieza. See social cleansing
Lithuania, 29
Kalyvas, Stathis, 3, 19–20, 22–23, 29, 35, 49, Lord’s Resistance Army, 28
56, 114, 116, 166, 189, 208, 242, 290
civilian self-rule, 302 M-19 (Movement of April 19), 67
logic of violence, 190 Maaloula, 293–294
selective violence, 18 Macarena region. See FARClandia
theory, 11 macheteras, 92, 187
Kandahar Province, 287 Magdalena, 123, 136, 138
Kenya, 29 Magdalena Medio (Middle Magdalena) region,
65, 67, 73, 173, 320, 322
La India, 2–3, 12, 73, 100, 186, 191, 213, 275 Manila, 276–278
conflict, 2 Mao, Zedong, 2, 58
Index 369

Marquetalia, 64 New People’s Army. See NPA

mass atrocities prevention NGOs, 10, 44, 97
limitations of international community, 126, accompaniment, 40, 75, 308
303 interaction with civilians, 107
massacres, 69, 113, 126, 137, 187, 265 monitoring, 60
La India, 322 presence in communities, 100
megaprojects, 58 role of, 26
Meta, 65, 104, 136, 138, 271–273, 275–276 shaming of armed groups, 50
Mexico, 29 support for civilian processes, 308
Middle Magdalena Program for Peace and technical assistance, 9, 35, 40, 73
Development (PDPMM). See PDPMM no name graves, 262
MILF, 31, 276, 281, 285–286 noncombatant, 17
Mindanao, 31, 277–278, 281 nongovernmental organizations. See NGOs
minifundios, 123, 125, 157, 239 nonviolent strategies, 5, 16, 35, 42–43, 74, 216,
minimize harm. See do no harm 300, 310, 312
MNLF, 31, 276 in conflict, 16
mobilization mass mobilization, 51
for peace, 62 norms of nonviolence, 48, 76, 185, 207–208,
Moro Islamic Liberation Front. See MILF 217
Moro National Liberation Front. See MNLF nudging armed groups, 56, 258, 197, 303
Movimiento Armado Quintín Lame, 16, 28, 53, pro-democracy, 16
67 resistance, 16, 42
Mozambique, 29, 303 Norte de Santander, 138
multinational corporations (MNCs), 143 NPA, 31, 276, 284–285
municipios, 13
analysis, 113, 135–136 Obama, Pres. Barack, 291
Musa Qala, 27, 288 oil, 56
infrastructure, 117
Naga City, 276, 283
Naprama, 7 Pablo Escobar, 162
Nariño (Antioquia), 155 Pacific Coast region, 138
Nariño (department), 74, 104, 123, 136, 138, pacifism, 14, 48, 75, 185, 195, 208, 303, 312,
248 319
Nasa (indigenous group), 73–74, 218, 248 norms, 299
Indigenous Guard, 51, 53, 74–75, 289 Pakistan, 26, 29
National Association of Peasants. See ANUC civilian resistance, 288–291
National Democratic Front (NDF). See NPA conflict, 271, 286
National Front, 64, 79, 81, 123, 125, 158, counterinsurgency, 291
224 government, 289, 291
National Juntas Confederation (CNAC), 129 militias. See lashkar (militias)
National Liberation Army (ELN). See ELN peace committees, 288
national peace movement, 71–74 Swat, 30
NATO, 26–27, 271, 286–287 tribal areas, 291
NDF, 285 Pan-American Highway, 245
Negro Alfonso, 242 paramilitaries, 59, 62, See AUC, See CAFGU
Negro Antonio, 243 (Philippines)
neighborhood watch groups, 71, 238 and ATCC, 210
neutrality characteristics, 79
active, 189 counterinsurgency operations, 17, 67
vs. autonomy, 3, 46, 50–51, 77, 288 formation, 127
declaration of, 16, 50, 274, 296 MAS (Death to Kidnappers), 162
definition, 46 in Mexico, 29
370 Index

paramilitaries (cont.) autonomy mechanisms, 271

neo-paramilitary groups. See BACRIM barangay, 278, 281, 284
origins, 67 CO Multiversity, 277, 281
political homicides, 113 conflict, 286
politics, 17 culture, 277
relationship with civilians, 128 fieldwork, 93
relationship with junta councils, 79, interviews, 276
127–129, 210, 267 military, 7, 283
Shabiha (Syria), 292 Muslim regions, 277
targeting of civilians. See AUC Office of the Presidential Adviser on the Peace
targeting of leaders, 129 Process (OPAPP), 284
territorial control, 67 peace talks, 277, 286
violence against civilians, 128 rido, 281
Pashtun social landscape, 277
Shinwari, 287 Pico y plomo, 244
Taliban, 287 Plan Colombia, 68, 101, 308, 311, 314
Pashtun Wali, 287 Plan Patriota, 68, 89, 91, 127, 221
Pastoral Social, 71 polarization
Pastrana, Pres. Andrés, 68, 73, 271 political, 122, 125, 158
Patriotic Union party. See Unión Patriótica policy implications, 26, 306–312
PCCC, 65 political elites, 21, 63–64, 84, 240, 312
PDPMM, 73 political homicides, 113, 120, 137
peace communities, 13, 25, 29, 34, 51, 53, political ideology
71–77, 86, 171, 221, 261, 276, 281, 303 of armed groups, 50, 58
alternate names, 76 politicide, 67, 126, See Unión Patriótica
definition, 76 poppy
vs. juntas, 82, 112 cultivation, 56
origins, 73 Popular Liberation Army (EPL). See EPL
perceptions of, 78 positivism, 18
relationship with armed groups, 283 poverty, 43, 69, 87, 123, 157, 260
strategies, 21, 276 precautionary measures, 75
Peace Corps priests, 248–249, 306
and junta councils, 80 and armed groups, 264–265, 276
interventions, 308 Belén de los Andaquies, 132
volunteers, 235 Philippines, 278
peace negotiations, 307 Quipile, 228–229, 249
Caguán, 67–68, 271, 276 role in junta formation, 234
Havana, 78, 314 Vianí, 225, 234–236, 240
national-level, 9, 63, 73, 189, 217–218, 320, promoters (of juntas), 80, 160, 235
322 property
Philippines, 277, 284, 286 disputes, 39, 49, 133, 238
relationship to juntas, 134 impact of land reform laws, 236
Syria, 297 land titles, 39
Peace Zones. See Zones of Peace private, 136
peacebuilding, 22, 109, 134, 207, 217 rights, 39
peasant soldier program, 53, 245 protests, 50–51, 53, 71–72, 74–75, 133,
Peasant Workers Association of the Carare 248–250, 274–275, 292, 298–299
River (ATCC). See ATCC civilian strategies, 50
Peru, 31 public protests, 11, 50
Philippines, 7, 31, 276–286 Puerto Araujo, 202
archival records, 271 Puerto Boyacá, 265
armed conflict, 276–277 Putumayo, 45, 73, 138
Index 371

qualitative analysis, 24, 88, 156, 163, 220, 226, intersectarian harmony, 293–295
241, 268 minority groups, 295
quantitative analysis, 18, 24–25, 85, 88 organizations, 278
Quindío, 138 Seventh-Day Adventist, 33, 207
Quintín Lame Armed Movement. Syria, 293
See Movimiento Armado Quintín Lame research methodology, 8, 84, 88–109,
Quipile, 24, 26, 93, 220 305–306
armed conflict, 247 archival, 100, 107–108
armed group presence, 231, 243 benefits, 85, 88, 109, 268–269
AUC, 264 biases, 109
FARC, 243, 262 case studies, 156
cattle theft, 227, 233 benefits, 88
clientelism, 260–261 case selection, 12, 85, 87–88, 90–93, 109,
collaboration with armed groups, 264 156, 163–174, 268, 305
collective action, 228, 230 case study towns, 24
culture, 231 challenges, 87–88, 262
homicide rate, 265 in conflict zones, 108–109, 164
junta councils, 170, 228–233, 247, design, 305–306
301 evaluation framework, 306
activity, 228 external validity, 90, 108
history, 229 fieldwork, 25, 84–86, 91–95, 132, 163, 221,
weakness, 228–229, 231, 249 223, 261
La Violencia, 225 interviews, 24–25, 169, 305
leadership, 230 participant observation, 107
macheteras, 233 policy implications, 301
military raid, 244 process-tracing, 88, 109, 266, 305
organizational failure, 230, 266 propensity-score (statistical) matching,
poverty, 223 24–25, 92, 159, 165, 168–169, 242,
public works projects, 247 266, 305
resistance, 255 protection of subjects, 92, 99
response to violence, 248 quasi-experiment, 24, 91, 165, 168,
social capital, 228 305
social divisions, 223, 233, 248–249 reporting biases, 108, 299, 313
threats, 264 research design, 14, 24–25, 85, 92, 100, 156,
victims, 264 163
violence, 262–265 reverse causality, 8, 85, 89–91, 109, 111,
155, 163, 166, 173, 306
Raqqa, 298 triangulation, 109, 262
Regional Indian Council of Cauca (CRIC). resilience, 8, 217, 268, 271
See CRIC vs. autonomy, 298
reintegration, 38, 103–105, 290, 303 Syria, 298
community-based, 281, 289–290, 304 resources, Oil: 67, 116, 143, 244–245
Joint Task Force 435 (in Afghanistan), armed group discipline, 116, 322–323
290 for armed groups, 322
religion, 44, 48, 87, 107, 207, 249 diamonds, 56
Catholic Church. See Catholic Church drugs, 243, 323
churches, 9, 63, 71, 75, 94, 133, 173, 248, emeralds, 322
296, 306 mining, 323
ATCC, 33 oil, 322
role in community life, 40, 157, 238–239 Responsibility to Protect (R2P), 291,
Evangelical, 207 298
extremism, 271 Restrepo, Gloria Inés, 18
372 Index

Revolutionary Armed Forces of Colombia Sharia law, 289

(FARC). See FARC Shining Path, 30
rhetorical traps, 50, 257, 281 Sierra Leone, 31
Risaralda, 138 Civilian Defense Forces (CDF), 8, 53,
Roldán, Mary, 123, 160 58
rondas campesinas, 8, 31, 53 Revolutionary United Front (RUF), 58
rural sector, 71 social capital, 4, 8, 14, 25, 38–40, 45, 49, 60,
development programs, 73 70, 112, 131, 135, 143, 156–157, 159,
exclusion, 63, 73 167, 224, 226, 248, 266–268, 304, 309,
inclusion, 63 312
organizing, 83, 301 and violence, 159
social ties, 38 definition, 38
state absence, 15, 78, 80, 84, 304 sources, 38
urban-rural divide, 84, 113, 304 social cleansing, 206, 254, 263–264, 267
Russia, 7, 31 social cohesion, 4, 9–10, 111, 248
Rwanda, 21, 31 in case study towns, 221, 241
Muslim communities, 31 intersectarian, 295
junta councils as indicator, 156
San José de Apartadó, 47 social cooperation, 3, 8, 12, 87, 159, 161, 241
San Tropel, 3, 173, 202, 210 preexisting bases, 4
San Vicente del Caguán, 275–276 variation, 86
Sánchez Torres, Fabio, 18, 161 social landscapes, 34, 83–84, 87–88, 173, 226,
Sangre Negra, 225 246, 266, 299, 304, 308, 313
Santander, 24, 73, 75, 85, 104, 123, 136, 138, social mobilization, 37, 71, 125, 253
173, 190 Somalia, 31
Santos, Pres. Juan Manuel, 78 sovereignty, 77
Santos, Solimon, 277–278 stability, 11, 134, 186, 209, 217, 311
sapo. See informants state neglect, 133, 171, 245, 304
Saucío, 110–111 state-building, 304
Scott, James, 9, 21 decentralized, 80, 304, 307
security policies, 63
social contract, 79 stateless areas, 8
in urban zones, 84 statistical analysis, 12, 24–25, 111, 156, 163,
security dilemma, 48, 60 167, 184, 220, 223, 226, 242
selection bias, 8, 23, 88, 90–93, 109, stigmatization, 255, 264
173–174 acknowledgment by Pres. Santos, 78
self-defense groups. See civilian organizations: ideological, 126
self-defense of civilians, 3, 46
sensitivity to civilians, 55–60, See armed of organizations, 42
groups:sensitivity of peace communities, 78
and ideology, 55 political, 16, 42, 126
and norms, 209 risks, 16
and resources, 59 Sudan, 32
and territorial control, 56 Sunni, 8, 293, 296–297
application of theory, 185 Swat Valley, 289
BACRIM, 314 Syria: 69, 270
definition, 54 armed groups, 292–293
foreign fighters, 294 armed self-defense, 295–296
relationship to violence, 57 Armenians, 295–296
sources of, 59 autonomy movements, 291–299
Taliban, 286 limitations, 298
Sepúlveda Roldán, David, 45 casualties, 291
Index 373

Christians, 293–296 United Self-Defense Forces of Colombia (AUC).

civil war, 271, 291 See AUC, paramilitaries
escalation, 291, 293 Urabá region, 65, 67, 73, 75, 78, 154
collective action, 292 Uribe, Jorge Alberto, 77
cooperation, 292 Uribe, Pres. Álvaro, 68, 244
displacement, 294–295
Druze, 297 vacuna. See armed groups:protection taxes
intersectarian oases, 293–295 Valle del Cauca, 74, 138
Local Coordination Committees (LCCs), 298 Vaupés, 138
military, 292, 294–295 Vianí, 24, 26, 93, 220
minority groups, 295–297 armed group presence, 244
Kurds, 296 AUC, 265
Muslim population, 293–294 FARC, 250, 265
negotiation with armed groups, 294, 296 cattle theft, 227, 238
neutrality, 295 church, 234
nonviolent mobilization, 292 clientelism, 261
Palestinians, 296 collaboration with armed groups,
sectarianism, 294–295 265
social cohesion, 293–294, 297 collective action, 236, 249
violence against civilians, 291–292, conciliation, 238
298 conflict dynamics, 249–251
conflict resolution, 238
Taliban, 26–27, 270–271, 286, 290 cooperation, 234, 237
drug trafficking, 59, 287 culture, 223, 234–235, 237
ideology, 286 informal justice, 238
in Afghanistan, 27, 59, 287, junta councils, 170, 234, 236–237, 247–248,
289 266, 301
in Pakistan, 29–30, 289 activity, 234
resistance to, 287–288 formation, 226, 234
sensitivity to civilians, 286 weakness, 236
technical assistance, 9, 157 La Violencia, 225–226, 234
tejo, 93, 133 land ownership, 235
threat environment, 10 macheteras, 237
effect on social capital, 40 organizations, 268
Tobler’s law, 166 resistance, 249
Tolima, 64–65, 123, 136, 138, 162 social relations, 223, 234, 237, 249
Toribío, 74 ties to Bituima, 170–171, 220, 226, 239–240,
tribes, 39 See Bituima, ties to Vianí
Afghanistan, 287 victims, 265
Iraq, 28 violence, 245, 262, 265
Syria, 297 Vichada, 138
tribal leaders, 284, 287, 290 Villeta, 220, 225, 242, 245
violence
U.S. Embassy, 272, 276 and cultural differences, 26
U.S. military, 287, 290, 294, 311 as selective reward, 58
Uganda, 28 civil war, 18, 25
Unión Patriótica, 65, 67, 71, 126, 128, 303 cycles, 48, 63, 277
correlation with violence, 126 historical, 158
politicide, 67, 126, 303 joint production, 19, 59, 217, 300
stigma, 126, 303 patterns, 223
unit of analysis, 113, 135, 313 political, 303
United Nations, 116, 213, 259 selective, 19, 35
374 Index

violence (cont.) Zones of Peace, 7, 26, 31, 76, 271, 276–286

studies (violentology), 18 and Catholic Church, 278
types, 266 characteristics, 278
variation, 18 collective actions, 281
cooptation, 283
weapons of the not-so-weak, 9, 46 declarations, 281
weapons of the weak, 9, 21, 46 dispute resolution, 279
Weinstein, Jeremy, 3, 11, 18–20, 31, 35, 50, effect on violence, 281, 286
55–56, 58–59, 116, 143 ethnic identity, 278
women formation, 276–277, 281
and peace, 75–76, 281 government opposition, 283
Wood, Elisabeth, 16, 19, 52, 166 government support, 284
World Values Survey, 157 history, 277–279
institutional policies, 278
Yabrud, 294–295 norms of nonviolence, 281
youth, 48, 67, 215–216, 239, 244, 245, 307 reintegration of ex-combatants, 285
opportunities, 216 relationship with armed groups,
recruitment, 48, 133 283–286
religious beliefs, 278
Zimbabwe, 32 research, 277
zona de despeje. See FARClandia role of women, 281
Zone of Peace, Freedom, and Neutrality strategies, 276, 278–283
(ZOPFAN), 276 violence against, 278