...

Global Burden of Armed Violence report

by user

on
Category:

statistics

0

views

Report

Comments

Transcript

Global Burden of Armed Violence report
GENEVA
DECLARATION
G L O B A L
o f
A R M E D
B U R D E N
V I O L E N C E
www.genevadeclaration.org
T
he Global Burden of Armed Violence report is the first comprehensive assessment of the scope of human tragedy resulting
from violence around the world. More than 740,000 people
die each year as a result of conflict-related and homicidal violence—
a figure that should capture the attention of decision-makers and
activists worldwide.
The report brings into focus the wide-ranging costs of war and
crime on development and provides a solid evidence base to shape
effective policy, programming, and advocacy to prevent and reduce
armed violence. Drawing from diverse sources and approaches,
chapters focus on conflict-related, post-conflict, and criminal armed
violence, and on the enormous economic costs of armed violence.
The report also highlights some of the less visible forms of armed
violence, including sexual and gender-based violence, extrajudicial
killings, kidnappings, and forced disappearances.
Photos
Top left: Mourners at the funeral of a former gang member in Davao, the Philippines.
© Lucian Read/Small Arms Survey
Centre left: A Lebanese soldier by a burning vehicle.
© Jeroen Oerlemans/Panos Pictures
Bottom right: Rwandan refugees at a camp in Tanzania, 1994.
© Karsten Thielker/AP Photo
Centre right: An Afghan woman photographed in her village near Kandahar, 2002.
© Danielle Bernier/DGPA/J5PA Combat Camera
www.genevadeclaration.org
GENEVA
DECLARATION
ISBN 978-2-8288-0101-4
9 782828 801014
GENEVA
DECLARATION
G L O B A L
o f
B U R D E N
www.genevadeclaration.org
A R M E D
V I O L E N C E
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
ii
Copyright
Published in Switzerland by the Geneva
Declaration Secretariat
© Geneva Declaration Secretariat, Geneva 2008
First published in September 2008
All rights reserved. No part of this publication may
be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, without
the prior permission in writing of the Geneva
Declaration Secretariat, or as expressly permitted
by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries
concerning reproduction outside the scope of
the above should be sent to the Publications
Manager at the address below.
Geneva Declaration Secretariat
c/o Small Arms Survey
47 Avenue Blanc
1202 Geneva
Switzerland
Copy-edited by Michael James and Alex Potter
Photo research by Alessandra Allen
Cartography by Jillian Luff, MAPgrafix
Proofread by Donald Strachan
Design and layout by Richard Jones, Exile: Design
& Editorial Services ([email protected]).
Typeset in Meta.
Printed by Paul Green Printing in London,
United Kingdom
ISBN 978-2-8288-0101-4
iii
A
rmed violence affects all societies to
different degrees, whether they are at war,
in a post-conflict situation, or suffering
from everyday forms of criminal or political violence. Armed violence stunts human, social, and
economic development and erodes the social
capital of communities.
The evidence assembled in the Global Burden of
Armed Violence report, written by a team of experts coordinated by the Small Arms Survey for
the Geneva Declaration on Armed Violence and
Development, provides an overview of the incidence, severity, and distribution of different types
of armed violence—in both conflict and non-conflict
situations, from both large and small-scale violence,
in criminally and politically motivated contexts.
The report is an important step towards a better
understanding of—and more effective responses
to—the negative impact of armed violence.
The human toll of armed violence across various
contexts is severe. In the recent past, at least
740,000 people have died directly or indirectly
each year from armed violence. Armed violence
also has a ripple effect throughout society, creating a climate of fear, distorting investment,
disrupting markets, and closing schools, clinics,
and roads.
This report also highlights the tremendous economic impact of armed violence. The cost of lost
productivity from non-conflict or criminal violence
alone is about USD 95 billion and may reach as
FORE WORD
Foreword
high as USD 163 billion per year. War-related violence decreases the annual growth of an average
economy by around two per cent for many years.
These human and economic costs make armed
violence a development issue and explain why
the development community is increasingly motivated to promote its prevention and reduction.
Together with members of the core group of the
Geneva Declaration on Armed Violence and
Development, signed on 7 June 2006 in Geneva,
Switzerland recognizes that effective prevention
and reduction of armed violence requires strong
political commitment to enhance national and
local data collection, develop evidence-based
programmes, invest in personnel, and learn from
good practice.
In moving in this direction the Geneva Declaration
calls on all signatories to strengthen efforts to
integrate strategies for armed violence reduction
and conflict prevention into national, regional, and
multilateral development plans and programmes.
Such instruments commit countries to make good
on their promises and to back these commitments
with adequate resources and leadership.
Endorsed by more than 90 states, the Declaration
argues that ‘living free from the threat of armed
violence is a basic human need’ and sets out to
make ‘measurable reductions in the global burden of armed violence’ by 2015. Signatories to
the Geneva Declaration—including the Swiss
Confederation—have accepted the responsibility
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
iv
to make a serious and sustained effort to improving the safety and security of people and communities through armed violence prevention and
reduction initiatives.
Achieving this goal will not be easy, but by working together as governments, local authorities,
and civil society partners, we can reduce the global burden of armed violence. The Global Burden
of Armed Violence report provides important signposts that can help decision-makers and researchers
move in the right direction.
—Micheline Calmy-Rey
Head of the Swiss Federal Department of Foreign Affairs
v
Executive Summary
................................................................................................................................................................................................
CO N T E N T S
Contents
1
Dimensions of armed violence .......................................................................................................................................................................... 2
Preventing and reducing armed violence ................................................................................................................................................ 6
Chapter One
Direct Conflict Death ..................................................................................................................................................... 9
A short history of estimates of direct conflict deaths ................................................................................................................ 10
Measuring the global burden of direct conflict deaths .................................................................................................... 13
The risk of dying in armed conflict .............................................................................................................................................................. 26
Conclusion ........................................................................................................................................................................................................................ 28
Chapter Two
The Many Victims of War: Indirect Conflict Deaths .................................................. 31
What is excess mortality? .................................................................................................................................................................................... 33
Challenges to collecting and using data on indirect deaths ............................................................................................... 34
Methods for quantifying indirect conflict deaths ......................................................................................................................... 36
1
2
3
Direct versus indirect deaths in recent conflicts ........................................................................................................................... 39
4
Conclusion ........................................................................................................................................................................................................................ 42
5
Armed Violence After War:
Categories, Causes, and Consequences .................................................................................................................. 49
Chapter Three
Disaggregating post-conflict armed violence ................................................................................................................................... 50
Risk factors facing post-conflict societies ........................................................................................................................................... 58
Conclusion: promoting security after conflict .................................................................................................................................. 63
Chapter Four Lethal
Encounters: Non-conflict Armed Violence ...................................................... 67
Defining and measuring violent deaths ................................................................................................................................................. 68
Estimating global homicide levels ............................................................................................................................................................... 71
Behind the numbers: trends and distribution of violent deaths ..................................................................................... 75
Armed violence and the criminal justice system ........................................................................................................................... 82
Conclusion: knowledge gaps and policy implications ............................................................................................................. 85
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
vi
What’s in a Number?
Estimating the Economic Costs of Armed Violence ............................................................................... 89
Chapter Five
Approaches to costing armed violence ................................................................................................................................................... 91
Costing armed violence in a sample of countries ......................................................................................................................... 97
Estimating the global economic costs of non-conflict armed violence ................................................................. 99
Positive effects of armed violence? ........................................................................................................................................................ 102
Conclusion ...................................................................................................................................................................................................................... 105
Chapter Six
Armed Violence Against Women ............................................................................................................ 109
The gender dimensions of armed violence ....................................................................................................................................... 110
Violence against women in conflict settings ................................................................................................................................... 112
Non-conflict violence against women .................................................................................................................................................... 115
Intimate partner violence ................................................................................................................................................................................... 117
Sexual violence .......................................................................................................................................................................................................... 118
Honour killings ........................................................................................................................................................................................................... 120
Dowry-related violence ....................................................................................................................................................................................... 121
Acid attacks ................................................................................................................................................................................................................... 121
Female infanticide and sex-selective abortion ............................................................................................................................ 122
Conclusion ...................................................................................................................................................................................................................... 123
Other Forms of Armed Violence:
Making the Invisible Visible ..................................................................................................................................................... 125
Chapter Seven
Armed groups and gangs ................................................................................................................................................................................. 126
Extrajudicial killings, disappearances, and kidnapping ..................................................................................................... 130
Armed violence and aid workers ................................................................................................................................................................ 138
Conclusions ................................................................................................................................................................................................................... 139
Bibliography
................................................................................................................................................................................................................
141
List of Illustrations ............................................................................................................................................................................................ 157
Note to reader: Cross-references to other chapters are indicated in capital letters in parentheses.
For example, ‘The number of battle deaths estimated in the preceding chapter for the DRC in the period
2004–07 is about 9,300 (DIRECT CONFLICT DEATH).’
About the Geneva Declaration
T
he Geneva Declaration on Armed Violence
and Development, endorsed by more than
90 states, commits signatories to supporting initiatives to measure the human, social, and
economic costs of armed violence, to assess risks
and vulnerabilities, to evaluate the effectiveness
of armed violence reduction programmes, and to
disseminate knowledge of best practices. The
Declaration calls upon states to achieve measurable reductions in the global burden of armed
violence and tangible improvements in human
security by 2015. Core group members of the
Geneva Declaration include Brazil, Guatemala,
Finland, Indonesia, Kenya, Morocco, the Netherlands, Norway, the Philippines, Spain, Switzerland, Thailand, and the United Kingdom with
support from the United Nations Development
Programme (UNDP).
A B O U T T H E G E N E VA D E C L A R AT I O N
vii
For more information about the Geneva Declaration, related activities, and publications, please
visit www.genevadeclaration.org.
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
viii
Acknowledgements
M
ore than 30 specialists from a dozen
Tania Inowlocki, the Small Arms Survey’s publi-
research institutes around the world
cations manager, coordinated the production of
contributed to the Global Burden of
the Global Burden of Armed Violence. Katherine
Armed Violence report (report, hereafter), making
Aguirre Tobón, Sarah Hoban, Jasna Lazarevic,
it a truly collective undertaking. The Secretariat
and Tanya Rodriguez fact-checked the report;
of the Geneva Declaration on Armed Violence
Alex Potter and Michael James served as copy-
and Development was responsible for conceiving,
editors; Jillie Luff designed the maps; Alessandra
researching, commissioning, and editing the
Allen assisted with photo research; Richard Jones
report. The primary editors of the report are Keith
conceived of the design and provided the layout;
Krause, Robert Muggah, and Achim Wennmann,
and Donald Strachan proofread the book. Carole
all with the Small Arms Survey. They were sup-
Touraine managed the finances for the project,
ported by Katherine Aguirre Tobón, Jasna Lazarevic,
and other staff of the Small Arms Survey provided
and Ilona Szabó de Carvalho.
various forms of support.
The report reflects the contributions of a wide
While the report represents a collective effort, the
range of institutional partners. Beginning in 2007,
editors wish to recognize the following experts
a series of consultations and workshops were
and institutions. We apologize to anyone who may
held in Geneva to identify key indicators of armed
have inadvertently been omitted.
violence, knowledge gaps, and methodological
approaches. Important input was provided by
Andrew Mack, Peter Batchelor, Theodore Leggett,
Steven Malby, Irene Pavesi, Jennifer Milliken,
Direct conflict deaths
and David Taglani. Critical feedback was also
Much has been written on the extent of direct
provided by Jean-François Cuénod, Paul Eavis,
conflict deaths, though uncertainty about its
and David Atwood, as well as from core group
overall magnitude remains. Jorge Restrepo and
members of the Geneva Declaration. Preliminary
Katherine Aguirre Tobón—and their collaborators,
outputs were presented by the editors during the
Juliana Márquez Zúñiga and Héctor Galindo, at the
regional summits of the Geneva Declaration for
Centro de Recursos para el Análisis de Conflictos
Latin America in Guatemala, for Africa in Kenya,
(CERAC) in Bogotá—made significant contributions
and for the Asia–Pacific region in Thailand. The
in this area of research. They were commissioned
editors are grateful for the valuable feedback
to assess a combination of datasets in order to
gathered on these occasions.
generate an estimate of the scale and distribution
Indirect conflict deaths
Recent surveys have stimulated debate and awareness on the way indirect deaths almost always
exceed direct deaths in war zones. For this report,
research on indirect conflict deaths was led by
Debarati Guha-Sapir, Olivier Degomme, Chiara
Altare, and Ruwan Ratnayake, all members of a
team of epidemiologists at the Centre for Research
on the Epidemiology of Disasters (CRED) in Brussels.
On the basis of a systematic review of dozens of
population-based surveys undertaken in conflictaffected countries, these experts established a
series of estimates of the indirect costs of armed
conflict. Their work was supported with case studies on mortality in Sierra Leone and South Sudan
compiled by, among others, Catrien Bijleveld,
Lotte Hoex, and Shanna Mehlbaum of the VU
University Amsterdam. Jorge Restrepo and Olivier
Degomme carried out critical research for the
box on armed violence in Iraq. The editors thank
Richard Garfield, Colin Mathers, and Alex Butchart
at WHO, who also provided input into the estimates
produced in this report.
a critical contribution that formed the basis for
the chapter on post-war armed violence. Astri
Suhrke and Ingrid Samset also issued a critique
of predictive models of war onset, while Simon
Reich from the Ford Institute for Human Security
at the University of Pittsburgh supplied evidence
relating to the forms of armed violence faced by
refugees and internally displaced persons. Richard
Cincotta from the Long Range Analysis Unit of the
National Intelligence Council in Washington, D.C.,
provided insight into the ‘youth bulge’ and the
way patterns of demography can affect future
trends in armed violence. Finally, Frances Stewart
and Rachael Diprose from the Centre for Research
on Inequality, Human Security and Ethnicity at
Queen Elizabeth House, University of Oxford,
offered insight into the relationships between
horizontal inequality and armed violence.
ix
ACKNOWLEDGEMENTS
of direct conflict deaths in selected countries, as
well as a meta-review of direct conflict deaths.
Feedback on preliminary chapter drafts was also
provided by Richard Garfield from the World Health
Organization (WHO) and Robin Coupland in his
personal capacity.
1
Non-conflict armed violence
Although homicidal violence is one of the better
indicators of the extent and global distribution of
armed violence, disaggregated data and analysis
are strikingly limited. The editors based this chapter on data from and the publications of the United
Nations Office on Drugs and Crime, as well as
other publicly available sources. This evidence
base allowed for regional and subregional comparisons of homicide rates. Additional inputs were
provided by CERAC and colleagues from WHO.
Armed violence after war
Economic costs of armed violence
Public health experts and humanitarian workers
have long recognized that armed violence can
actually escalate in the aftermath of war. Astri
Suhrke and Torunn Wimpelmann Chaudhary from
the Chr. Michelsen Institute in Norway provided
The debate on the economic costs of armed violence tends to focus narrowly on war. Much less
common are econometric assessments of homicidal
or other forms of armed violence. Nevertheless,
in either case, large-scale datasets tend to be
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
x
analysed to model the relationships between
armed violence and gross domestic product (GDP)
losses over time. The chapter on the economic
costs of armed violence seeks to widen the lens
and account for the multi-dimensional impacts.
Carlos Bozzoli, Tilman Brück, Thorsten Drautzburg,
and Simon Sottsas at the Deutsches Institut für
Wirtschaftsforschung in Berlin provided an overview of approaches to measuring the economic
costs of armed violence. Likewise, James Putzel
from the Crisis States Research Centre at the London
School of Economics and Francisco Gutiérrez Sanín
of the Instituto de Estudios Políticos y Relaciones
Internacionales in Bogotá provided a background
paper on the efficiency and distribution effects
of armed violence. Jorge Restrepo, Brodie Ferguson, Juliana Márquez Zúñiga, Adriana Villamarín,
Katherine Aguirre Tobón, and Andrés Mesa of
CERAC established a robust estimate of the economic burden of conflict and criminal violence.
Armed violence against women
As a first step to generating a more nuanced understanding of what is and is not known about the
gendered impacts of armed violence—particularly
the implications for women and girls—Megan
Bastick and Karin Grimm from the Centre for the
Democratic Control of Armed Forces and Jasna
Lazarevic from the Small Arms Survey were tasked
with reviewing the state of the art on the issue. Rahel
Kunz from the University of Lucerne helped integrate
the various contributions into the chapter and ensured that the gender dimensions of armed violence
were underscored throughout the report.
Other forms of armed violence
The chapter on other forms of armed violence
combines a number of disparate strands. Thanks
are extended to Dennis Rodgers from Manchester
University, together with Chris Stevenson and
Robert Muggah from the Small Arms Survey for
contributing with their work on gangs. The section
on extrajudicial killings and enforced disappearances is based on a background study prepared
by Sabine Saliba. Eric Mongelard of the Special
Procedures Branch of the Office of the High Commissioner for Human Rights (OHCHR), as well as
Claudia de la Fuente of the OHCHR’s Working Group
on Enforced or Involuntary Disappearances, provided valuable comments. David Cingranelli of
Binghamton University and David L. Richards
from the University of Memphis kindly facilitated
the use of material from the Cingranelli-Richards
Human Rights Data Project. Control Risks, a private risk consultancy firm, provided generous
insight into their data on kidnap for ransom cases
and we thank Lara Symons for facilitating this
contribution. Finally, Larissa Fast from the Kroc
Institute for International Peace Studies at the
University of Notre Dame and Elizabeth Rowley of
Johns Hopkins Bloomberg School of Public Health
in Baltimore contributed new findings on global
trends in aid worker victimization.
Finally, thanks must be extended to the participants in the Geneva Declaration core group. Under
the guidance of its chair, Ambassador Thomas
Greminger from the Swiss Federal Department of
Foreign Affairs, the core group provided strategic
guidance for the report. Together with Anna IfkovitsHorner, Elisabeth Gilgen, and Ronald Dreyer, the
contribution of Switzerland in particular must be
specially recognized. It should be recalled that the
report is an independent contribution to the Geneva
Declaration on Armed Violence and Development
and does not necessarily reflect the views of the
Government of Switzerland or the states signatories of the Geneva Declaration. While the report is
a collective effort, the editors are responsible for
any errors and omissions of fact or judgment.
Executive Summary
A
rmed violence imposes a tremendous
human and economic burden on individuals, families, and communities. More
than 740,000 people die each year as a result of
the violence associated with armed conflicts and
large- and small-scale criminality. The majority
of these deaths—as many as 490,000—occur
outside war zones. This figure shows that war is
only one of many forms of armed violence, and in
most regions not the most important one.
Armed violence is spread across age groups but
affects certain groups and regions disproportionately. It is the fourth leading cause of death
for persons between the ages of 15 and 44 worldwide. In more than 40 countries, it is one of the
top ten causes of death. In Latin America and
Africa, armed violence is the seventh and ninth
leading cause of death, respectively (Peden,
McGee, and Krug, 2002; WHO, 2008b).1 Yet certain
demographic groups (especially young men) and
geographic regions are much more affected than
others. The full dimensions of armed violence are
often invisible unless they are closely monitored
and analysed.
Beyond the chilling calculus of deaths, armed
violence imposes huge human, social, and economic costs on states and societies. An untold
number of people each year are injured—often
suffering permanent disabilities—and many live
with profound psychological as well as physical
scars.2 The damaging effects of armed violence
include such things as physical and mental disabil-
ities, brain and internal organ injuries, bruises and
scalds, chronic pain syndrome, and a range of sexual
and reproductive health problems (WHO, 2008a).
Armed violence also corrodes the social fabric of
communities, sows fear and insecurity, destroys
human and social capital, and undermines development investments and aid effectiveness. The
death and destruction of war—which ebbs and
flows from year to year and is concentrated in a
few countries—reduces gross domestic product
(GDP) growth by more than two per cent annually, with effects lingering many years after the
fighting ends. The economic cost—in terms of
lost productivity—of non-conflict armed violence
(large- and small-scale criminal and political
violence) is USD 95 billion and could reach up
to USD 163 billion annually worldwide.
Undertaking research and gathering data on armed
violence is difficult and often controversial. Violence
has political implications (even when the violence
itself may not be politicized) and is seldom random.
Different groups often have an interest in understating or concealing the scope of lethal armed
violence, making the collection of reliable data
and impartial analysis particularly challenging.
The promotion of effective and practical measures
to reduce armed violence nevertheless depends
on the development of reliable information and
analysis of its causes and consequences. The
Global Burden of Armed Violence report draws on
a wide variety of sources and datasets to provide
E XECUTIVE SUMMARY
1
1
2
3
4
5
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
2
a comprehensive picture of the worldwide scope,
scale, and effects of armed violence. It contributes
to the generation of a broader evidence base on the
links between armed violence and development and
is part of the process of implementing the Geneva
Declaration on Armed Violence and Development.
dimension of the most prominent forms of armed
violence. While the overwhelming majority of
victims (and perpetrators) of armed violence are
men, there are gender-specific forms of violence
that warrant greater analysis and that are poorly
documented.
Key findings of the report are that:
Dimensions of armed violence
For the purposes of this report, armed violence is
the intentional use of illegitimate force (actual or
threatened) with arms or explosives, against a
person, group, community, or state, that undermines people-centred security and/or sustainable
development.
This definition covers many acts, ranging from
the large-scale violence associated with conflict
and war to inter-communal and collective violence,
organized criminal and economically motivated
violence, political violence by different actors or
groups competing for power, and inter-personal
and gender-based violence.3
This report provides cross-regional and international comparisons of some of the most dramatic
consequences of armed violence: direct conflict
deaths, indirect conflict deaths, post-conflict
mortality, and non-conflict deaths such as homicide, disappearances, kidnappings, and aid worker
killings. These forms of armed violence are usually
the best documented, and as leading indicators
can provide a good basis for understanding the
scope and distribution of armed violence around
the world and for exploring other, less prominent
dimensions of armed violence.
The report also explores in a separate chapter
the less-visible forms of violence against women,
and where possible considers the gendered
more than 740,000 people have died directly
or indirectly from armed violence—both conflict
and criminal violence—every year in recent years.
more than 540,000 of these deaths are violent,
with the vast majority occurring in non-conflict
settings.
at least 200,000 people—and perhaps many
thousands more—have died each year in conflict zones from non-violent causes (such as
malnutrition, dysentery, or other easily preventable diseases) that resulted from the
effects of war on populations.
between 2004 and 2007, at least 208,300
violent deaths were recorded in armed conflicts—an average of 52,000 people killed per
year. This is a conservative estimate including
only recorded deaths: the real total may be
much higher.
the annual economic cost of armed violence
in non-conflict settings, in terms of lost productivity due to violent deaths, is USD 95 billion
and could reach as high as USD 163 billion—
0.14 per cent of the annual global GDP.
These figures, which are explained in detail in
different chapters in this report, underscore that
violent deaths in non-conflict settings and indirect deaths in armed conflicts comprise a much
larger proportion of the global burden of armed
violence than the number of people dying violently
in contemporary wars.
Traditionally, these various manifestations of armed
violence have been treated separately, as if their
underlying causes and dynamics were fundamentally different. Yet the changing nature of armed
violence—including the rise of economically
motivated wars, the blurring of the line between
political and non-political violence, the growth of
trans-national criminal gangs, the expansion of
non-state armed groups, and persistently high
levels of insecurity in most post-conflict situations—
makes drawing clear distinctions between different
forms of armed violence practically and analytically impossible.
Continuing to treat these different forms of armed
violence separately also impedes the development of coherent and comprehensive violence
prevention and reduction policies at the international and local level. Since one goal of the Global
Burden of Armed Violence report is to promote a
better understanding of the negative impact of
armed violence on human, social, and economic
development, it is critical to adopt the broader
lens of armed violence rather than focusing on
only one of its many manifestations.
Figure 1 Categories of deaths
INDIRECT CONFLICT
DEATHS
3
NON-CONFLICT
DEATHS
E XECUTIVE SUMMARY
Figure 1 captures graphically the distribution of
the different categories of deaths within the global burden of armed violence. The small green
circles illustrate the direct burden of violent
death in conflict, including both civilians and
combatants. It represents roughly seven per cent
of the total global burden. The larger blue circle
represents the indirect deaths from violent conflict—some 27 per cent of the total. And violent
deaths in non-conflict settings—490,000 per
year—represent two-thirds (66 per cent) of the
total. 4 Beyond this lie the untold number of physically injured or psychologically harmed people
who also bear part of the global burden of armed
violence.
Battle-related deaths
Civilian deaths in
armed conflict
DIRECT CONFLICT DEATHS
The report also presents the geographic distribution and concentration of different forms of armed
violence. Conflict-related deaths, which appear to
have increased since 2005, are highly concentrated:
three-quarters of all reported direct conflict deaths
took place in just ten countries. Ending the armed
conflicts in Afghanistan, Iraq, Pakistan, Somalia,
and Sri Lanka in 2007 would have reduced the
total number of direct conflict deaths by more than
two-thirds. And within countries, armed violence
is usually concentrated in certain municipalities
or regions, while other areas may be relatively
untouched.
Most international attention focuses on the
numbers of recorded violent deaths in conflicts.
While such data potentially helps decision-makers
and analysts assess the intensity of a war and its
evolution over time, these relatively low figures
(in the tens of thousands) obscure the larger burden of mortality arising from indirect deaths in
1
2
3
4
5
6
7
4
G LOBAL BURDEN
of
ARMED VI O L EN CE
armed conflicts. A minimum estimate is that an
average of 200,000 people have died annually in
recent years as indirect victims during and immediately following recent wars. Most of these people,
including women, children, and the infirm, died of
largely preventable illnesses and communicable
diseases. Yet they are every bit as much victims of
armed violence as those who die violently, and an
adequate accounting of the victims of war has to
include these indirect deaths. The scale of indirect
deaths depends in part on the duration and intensity
of the war, relative access to basic care and services,
and the effectiveness of humanitarian relief efforts.
Map 4.1 Homicide rates per 100,000 population, by subregion, 2004
0)+)2(
Per 100,000
population
>30
25–30
20–25
10–20
5–10
3–5
0–3
Note: The boundaries and designations used on this map do not imply endorsement or acceptance.
Source: UN Office on Drugs and Crime (UNODC) estimates
The ratio of people killed in war to those dying
indirectly because of a conflict is explored in the
chapter on indirect deaths (INDIRECT CONFLICT
DEATHS). Studies show that between three and
15 times as many people die indirectly for every
person who dies violently. In the most dramatic
cases, such as the Democratic Republic of the
Congo, up to 400,000 excess deaths per year
have been estimated since 2002, many of which
resulted indirectly from war. Consequently, this
report’s estimate of a global average of 200,000
indirect conflict deaths per year should be taken
as a conservative figure.
5
E XECUTIVE SUMMARY
This report also finds that the aftermath of war
does not necessarily bring a dramatic reduction
in armed violence (ARMED VIOLENCE AFTER WAR).
In certain circumstances, post-conflict societies
have experienced rates of armed violence that
exceed those of the conflicts that preceded them.
They also exhibit a 20–25 per cent risk of relapsing
into war. So long as such countries must contend
with high youth bulges (exceeding 60 per cent of
the total population), soaring rates of unemployment, and protracted displacement, the risks of
renewed armed conflict remain high.
The majority of violent deaths occur in non-war
situations, as the result of small or large-scale
criminally or politically motivated armed violence
(NON-CONFLICT ARMED VIOLENCE). More than
490,000 homicides occurred in 2004 alone. This
represents twice the total number of people who die
directly and indirectly in armed conflicts. As violent
as wars can be, more people die in ‘everyday’—
and sometimes intense—armed violence around
the world than in armed conflicts. Map 4.1 (presented in Chapter 4) illustrates the distribution of
conflict and non-conflict armed violence, expressed
as the number of homicides per 100,000 persons.
The geographic and demographic dimensions of
non-conflict armed violence are significant. SubSaharan Africa and Central and South America
are the most seriously affected by armed violence,
experiencing homicide rates of more than 20 per
100,000 per year, compared to the global average
of 7.6 per 100,000 population. Countries in Southern Africa, Central America, and South America—
including Colombia, El Salvador, Guatemala,
Jamaica, South Africa, and Venezuela—report
among the highest recorded rates of violent death
in the world, ranging from 37 (Venezuela) to 59
(El Salvador) per 100,000 in 2005, as reported
by official police statistics. 5
1
2
3
4
5
The weapon matters. As much as 60 per cent of all
homicides are committed with firearms—ranging
from a high of 77 per cent in Central America to a
low of 19 per cent in Western Europe. And there is
a gendered component to armed violence: although
most victims are men, the killing of women varies
by region: in ‘high-violence’ countries, women
generally account for about 10 per cent of the
victims, while they represent up to 30 per cent in
‘low-violence’ countries. This suggests that intimate partner violence does not necessarily rise and
fall with other forms of armed violence, and may
not decline as other forms of armed violence are
reduced.
Photo ! A policeman
carries a child away
during a gun battle in
Tijuana, Mexico, 2008.
© Jorge Duenes/Reuters
6
7
6
G LOBAL BURDEN
of
ARMED VI O L EN CE
There are a host of other forms of armed violence
that, while largely invisible, undermine the real
and perceived security of people around the world.
In some regions, the state (or state agents) commit or are implicated in acts of armed violence.
At least 30 states register more than 50 reported
extrajudicial killings per year. Forced disappearances occur ‘frequently’ in more than a dozen
countries and ‘occasionally’ in 20 others. And
kidnap-for-ransom is a growing phenomenon with
approximately 1,425 cases reported in 2007 in
Latin America, Asia, Africa, and the Middle East.
Photo " People stand at
the scene of a car bomb
in the Campsara district
of Baghdad, 2008.
© Moises Saman/Panos
Pictures
Armed violence embodies literally thousands of
inter-connected events that generate negative
consequences across societies and at multiple
levels. It can result in the destruction of human
and physical capital and opportunity costs, and
its economic consequences are often felt hardest
by the poorest and most vulnerable. The economic
costs of armed violence in both conflict and nonconflict settings, and the negative impact on
development, are considerable. Using contingency
valuation approaches, the global cost of insecurity
generated by armed violence every year amounts
to roughly USD 70 per person, or a global annual
burden of USD 400 billion.
Preventing and reducing
armed violence
Armed violence is preventable. Moreover, early
interventions to save lives can significantly reduce
the overall burden of armed violence. Map 5.1
(presented in Chapter 5) reveals the significant
gains in life expectancy that could be realized—
more than one year for men in many Central and
South American countries. Although this report
does not focus on concrete strategies to reduce
armed violence, it points towards a number of
entry-points for promoting armed violence prevention and reduction (WHO, 2008a). Grounded
in up-to-date data and research, it also documents how a failure to address armed violence
can impede development and economic growth.
At a minimum, the report should help international aid donors and practitioners, government
officials, and civil society actors recognize that
promoting safety and security is central to human,
social, and economic development.
At a practical level, it is critical that relevant
national and international agencies enhance
their regular and routine monitoring of armed
violence trends. This entails making serious
investments in mechanisms to measure real and
perceived risks and impacts of armed violence,
and drawing on social science and public health
methods to quantify the effectiveness of armed
violence prevention and reduction programmes.
Reinforcing international, national, and local
data collection and surveillance is an essential
first step to planning effective interventions,
7
E XECUTIVE SUMMARY
Map 5.1 Potential gains in life expectancy in years in the absence of non-conflict armed violence, by country, 2004
0)+)2(
4SXIRXMEPKEMRWMRPMJI
I\TIGXERG]MR]IEVW
1.00–1.81
0.66–1.00
0.42–0.66
0.26–0.42
0.00–0.26
0.00
1
Male
Female
2
Not included
3
4
Source: CERAC
5
identifying priorities, evaluating activities, and
saving lives.
Investing in armed violence prevention and reduction will also mean supporting and reinforcing the
capacity of public and private actors to design,
execute, and monitor interventions. It requires
developing a sophisticated understanding of local
conditions and concerns through surveys and other
participatory research methods. It demands recognizing that armed violence has multiple and
often interacting causes, and does not ebb and
flow in a simple linear fashion. Finally, it requires
protecting the safety and security of humanitarian
and development personnel—many of whom are
killed in the line of duty. As this report observes,
the violent death rate for aid workers is roughly
60 per 100,000, a disturbing reminder of the
acute risks facing humanitarian workers around
the world.
This Global Burden of Armed Violence report is
only the first step towards the implementation of
an international armed violence prevention and
reduction agenda. This report highlights the importance of developing and enhancing the evidence
base—identifying who is vulnerable, from what
forms of armed violence, committed by whom,
6
7
and under what circumstances—as a critical step
towards achieving measurable reductions in the
global burden of armed violence and tangible
improvements in human security worldwide.
G LOBAL BURDEN
accounts for even more deaths than conflict or homicidal
violence (WHO, 2008a, p. 1). Its estimate of 1.6 million
deaths from violence includes suicide (54 per cent of the
total), and is thus broadly consistent with the figures
presented here. The definition also is meant to focus on
the physical use of violence, and to exclude such concepts as structural, cultural, and psychological violence.
Endnotes
1
of
ARMED VI O L EN CE
8
2
3
The figures are from the WHO Global Burden of Death
database and are calculated by adding the categories of
inter-personal violence and deaths from war injuries.
Armed violence is overall the 18th leading cause of death
worldwide.
4
Overlap between the red and green circles represents
the possibility of double-counting some conflict deaths
in homicide statistics (NON-CONFLICT ARMED VIOLENCE).
5
Figures are from national police sources. See: http://www.
derechos.org.ve/publicaciones/infanual/2005_06/pdf/
seguridadciudadana.pdf (Venezuela); http://www.fgr.gob.
sv/estadisticas/homicidios2005.pdf (El Salvador); www.
According to the World Health Organization (WHO), approximately ten times more people are injured than killed by
violence (WHO, 2008a, p. 4).
saps.gov.za/statistics/reports/crimestats/2007/_pdf/
This definition does not include self-directed violence
(suicide). The WHO estimates that self-directed violence
%20de%20la%20Violencia%20en%20Guatemala%20
category/murder.pdf (South Africa); http://www.undp.
org.gt/data/publicacion/Informe%20Estad%C3%ADstico
final.pdf (Guatemala); CNP (n.d.).
Direct Conflict Death
A
rmed conflict destroys lives and livelihoods. Notwithstanding the appalling
human costs of protracted conflict-related
violence in countries such as Iraq, Afghanistan,
or Sudan, however, the total number of people
dying violently during conflict is relatively low in
comparison to those dying indirectly from armed
conflict. The rate at which people experience
violent deaths in war is also low compared to
violent death rates in many countries that are
not affected by armed conflict. The average
annual number of direct conflict deaths in recent
years is likely between 10 and 20 per cent of
those violently killed in ostensibly non-conflict
environments.
Establishing credible estimates of direct conflict
deaths is central to effective strategic and public
health planning. It is also crucial for promoting
meaningful reconciliation of war-torn societies and
for reparations and other forms of transitional
justice. In spite of considerable efforts devoted to
understanding the global, regional, and national
distribution of direct conflict deaths, there are still
fundamental disagreements over basic estimates,
trends, and methods used to count the dead. In
order to build awareness of the scope and scale of
direct conflict deaths, it is critical to understand
how such figures are determined and assessed.
This chapter provides a unique comparative
analysis of several global and national conflict
datasets. Most comparative national datasets
are based on what is called ‘incident reporting’,
i.e. the monitoring of authoritative media, governmental, and non-governmental reports in order
to document the numbers of combatants reported
to have died in battle. This approach often yields
reasonably good, but incomplete, information on
conflict dynamics and patterns of victimization.
It is limited because incident reporting seldom
captures all violent conflict deaths, especially in
places where access for journalists and humanitarian workers is restricted. One way of circumventing
incomplete reporting is to integrate several datasets and consolidate estimates in order to issue
a more complete figure.
Direct conflict deaths are highly concentrated,
with the top ten deadliest conflicts accounting
for more than three-quarters of the global burden
of violent mortality in war.1 The chapter derives
its estimates on the basis of a review of more than
19 comparative and national datasets and reports.
Specifically, the chapter finds the following:
D I R E C T CO N F L I C T D E AT H
9
Chapter One
1
2
3
4
5
6
7
At least 52,000 direct conflict deaths were
recorded on average every year between 2004
and 2007, although the real direct conflict
death toll is likely much higher.
While the overall figures are low in historical
terms, direct conflict deaths increased from
42,500 to 63,900 between 2005 and 2007,
due principally to wars in Iraq, Afghanistan,
Somalia, Sri Lanka, and Sudan.
The conflicts in Afghanistan, Iraq, Pakistan,
Somalia, and Sri Lanka accounted for two-
thirds of the global burden of direct conflict
death in 2007.
2007. 4 This is principally due to the wars in
Iraq and Afghanistan.
A R M E D V I O L E N CE
The direct conflict death rate in war-affected
countries is approximately 2.0 per 100,000
population, while the worldwide homicide
rate is 7.6 per 100,000 population.2
Fewer than two per cent of all direct conflict
deaths can be attributed to international
terrorism for 2004–07.
G LO B A L B U R D E N
of
10
Photo " A crowd looks
at the body of a Taliban
soldier in Kunduz,
northern Afghanistan,
2001. © Dusan Vranic/
AP Photo
In 2007 the risk of dying violently from warfare
was highest in Iraq (78.5 per 100,000 population) and Somalia (74.2 per 100,000 population).
More people were violently killed as a result of
international and internationalized conflict 3
than due to intrastate conflict in 2006 and
This chapter first reviews the evolution of policy
and academic research on armed conflict and
the estimation of combatant and civilian violent
deaths. Drawing on eight specific conflict mortality datasets, the next section presents conflict
death estimates at the global, regional, and
national levels. The third section reviews the
associated risks of dying violently during armed
conflict. The conclusion highlights a selection of
next steps for research and policy on preventing
and reducing conflict deaths. A discussion of the
methodology is provided in an online methodological appendix.5
A short history of estimates of
direct conflict deaths
Policy-makers and military planners have long been
preoccupied with understanding the effects of
armed conflict on military personnel and civilians.
With the expanding reach of international humanitarian and human rights law in the late 19th and
20th centuries, war makers sought to minimize
so-called ‘collateral damage’ by adjusting tactics,
techniques, and reporting on armed conflict.
Attempts to quantify the human costs of war
expanded in breadth and sophistication in the
latter decades of the 20th century.
At least four distinct approaches to documenting
the incidence of conflict deaths are now widely in
use (detailed in Box 1.1). At the outset, an emphasis was placed on so-called documentation-based
There are many international conflict datasets
that feature different numbers of countries and
varying methodological approaches to gathering
data. A small number of these databases offer
country- and conflict-specific information that
can be used to calculate annual figures on direct
conflict deaths. The most prominent of these
include the Uppsala Conflict Data Programme
(UCDP) and databases of the International Peace
Research Institute in Oslo (PRIO), the Political
Instability Task Force database of the Center for
Systemic Peace of the University of Maryland, and
the International Institute for Strategic Studies
(IISS) conflict dataset in London. In recent years,
the public health and humanitarian communities
have also started to measure direct conflict deaths
on the basis of population health surveys (WHO,
forthcoming; 2004).
Reconciling different datasets and methodological approaches is difficult, but feasible. There are
still serious debates over the definition of ‘conflict’
and different rules for counting events and casualties. For example, the original COW threshold
11
D I R E C T CO N F L I C T D E AT H
approaches that featured cross-national datasets
drawing on ‘event data’ or ‘incident reporting’ of
specific conflict events. Information was typically
gathered from historical accounts and news
reports. The assumption was that by analysing
the main correlates of armed conflict, it might be
possible to predict their onset and intervene to
prevent or reduce them (Richardson and Wright,
1960). From the 1960s onwards, increasingly
sophisticated computer software packages capable of selecting events from large collections of
media information began to emerge. But it was
not until the 1970s and 1980s that systematic
datasets emerged that could be readily subjected
to analysis. For example, the Correlates of War
(COW) project at the University of Michigan was
one of the first of its kind to systematically collect
and analyse data on armed conflict (Sarkees, 2000;
Singer, 1979; Vasquez, 1987).
1
2
3
Box 1.1 Methods to estimate direct conflict deaths
The most common form of estimating direct conflict deaths is incident reporting. The effectiveness of this approach—of which the
Iraq Body Count is arguably the most prominent example today6—
depends significantly on the quality of available documentation.
The robustness of the data is therefore a function of the effectiveness of media coverage and official and NGO reporting.7
Other approaches such as victimization or epidemiological surveys
rely on statistical techniques to assess the true level of direct (and
indirect) conflict deaths. Likewise, demographic methods are also
used to assess the size of populations ‘lost’ during war. Multiple
systems estimation techniques seek to estimate the true number
of people affected, based on several overlapping and incomplete
data sources (Small Arms Survey, 2005, pp. 240–41).
Incident reporting tabulates conflict-related events selected from
news and NGO data, but also other information derived from
morgues and hospitals. These are then coded and entered into
databases in order to track temporal and spatial trends. The relatively recent availability of global news databases, such as Factiva
or LexisNexis, facilitates the capture of incidents and associated
deaths in a wide number of war zones. Moreover, the sophistication of incident reporting has increased in concert with so-called
‘parsing programmes’ that permit the electronic selection of
news stories and events. Incident reporting provides estimates
based on traceable events, useful for monitoring purposes and
for building legal cases to prosecute perpetrators of war crimes
(Small Arms Survey, 2005, pp. 233–39).
Incident reporting suffers from intrinsic limitations and frequently
undercounts the true magnitude of conflict deaths. The phenomenon of armed violence makes data collection risky and contributes
to the deterioration or destruction of population surveillance and
monitoring systems. Since high rates of war-related mortality tend
to occur in dangerous areas where eyewitnesses are less likely to
be present, under-reporting is common. In some contexts, the level
of coverage is sparse or data is censored. Factions taking part ##
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
12
in an armed conflict are also likely to apply political pressure to
distort information and encourage under-reporting in order to
minimize the scale of fighting and human suffering. In Peru, for
example, more than half of the 69,000 conflict deaths between
1980 and 2000 were not recorded in the press or other accounts.
Victimization surveys can also measure the magnitude and distribution of particular forms of conflict mortality. They tend to involve
so-called ‘verbal autopsies’ in which a random or semi-random
sample of the population is interviewed about experiences of
violence. Estimates are generated on the basis of probabilistic
sampling and are usually adjusted using different weighting
systems that account for gaps in coverage and reporting biases.
Victimization surveys have the advantage of providing a rapid
estimate of the overall level and distribution of real and perceived
armed violence at a relatively low cost. But they are frequently
difficult to administer due to safety concerns and logistical challenges. Also, respondents frequently give varying accounts of
death and victimization that, if not properly accounted for, can
unintentionally undermine the quality of the data.
Multiple systems estimations (MSEs) are widely used in the natural
sciences to estimate the magnitude and changes of wildlife populations. Also described as ‘capture-recapture’, they were pioneered
by Patrick Ball to estimate the level of human rights violations in
a given war zone, particularly in situations where information is
highly dispersed and coverage is partial (Ball et al., 2003; Small
Arms Survey, 2005, p. 240). MSEs have been attempted in Guatemala,
Timor-Leste, Bosnia and Herzegovina, Colombia, and Peru, among
other places. Taken together, they suggest that documentationbased approaches to reporting often greatly underestimate direct
conflict deaths. But MSEs are also difficult to undertake: not only
must they be extremely sensitive to the victims and their families,
but they require a minimum of two sources of data with properly
matched names in order to estimate an overall burden of direct
conflict deaths.
Epidemiological surveys were originally developed to gauge the real
incidence and likely direction of disease in a specific population. More
recently, epidemiological surveys have been employed in armed
conflicts in order to estimate the number of deaths arising from
both direct and indirect conflict-related causes (INDIRECT CONFLICT
DEATHS). As with victimization surveys, epidemiological surveys
are difficult to administer in areas that lack detailed demographic
information. Without so-called denominator data, it is extremely
difficult to make reliable projections from a small sample of the
population to the national level. Equally, the quality of the resulting
estimate depends on how carefully the sample is drawn from the
population, the questionnaire and research methods used, and
how surveyors or enumerators are trained and conduct the survey.
Policy-makers and practitioners are often confused by the bewildering range of estimates of conflict deaths in war zones around
the world. Indeed, even a cursory review of the literature indicates
that MSE approaches and victimization or epidemiological surveys
regularly record higher levels of conflict deaths than incident reporting. For example, depending on the database and methodologies
used, estimates of direct conflict deaths arising during the 1999
conflict in Kosovo range from 2,000 to 12,000 (Small Arms Survey,
2005, pp. 241–42). Such discrepancies have led to serious and
sometimes acrimonious debates between policy-makers and
researchers, and even within academia. This chapter shows that
different estimation techniques can and should be regarded as
complementary, so long as their limitations and strengths are
clearly understood.
for ‘war’ (and therefore for direct conflict deaths)
included those situations in which at least 1,000
battle deaths occurred per year. Over time, the
threshold was lowered to 25 battle deaths per
year for the UCDP dataset to account for lower
intensity, but no less important, ‘armed conflicts’
(Eriksson et al., 2002, p. 617).8 In one sense, these
debates mirror the changing nature of armed disputes from classic interstate ‘warfare’ between
states to the broader category of ‘armed conflict’.
This latter category accounts for intrastate con-
flicts of varying intensities between a state and
non-state actors or among competing armed groups.
Likewise, restrictive rules that count only ‘staterelated battle deaths’ can reduce estimates of the
overall burden of direct conflict deaths. As the
situations in Colombia, Iraq, Sri Lanka, and Afghanistan suggest, paramilitary and militia forces are
often aligned with the state and can be among
the most potent perpetrators of armed violence.
Equally, as many intrastate wars in Africa remind
While disagreements persist, there appears to be
a consensus that the level of direct conflict deaths
in contemporary armed conflicts is relatively low
when compared to the estimated 5.4–10 million
battle-related deaths occurring between 1955 and
2002 (Obermeyer et al., 2008; PITF, 2008). According
to researchers Nils Petter Gleditsch and Bethany
Lacina, for example, more than half of these violent
deaths occurred in Vietnam, Korea, the Chinese
civil war, the Iran–Iraq war, and Afghanistan
(Gleditsch and Lacina, 2005, pp. 154–55).9 Another
estimate holds that 18–25 million civilians died
in civil, international, and colonial wars between
1945 and 2000 (Huth and Valentino, 2008, p. 79).
The logistical challenges in selecting and coding
data on low-intensity events in a large number
of countries with different languages, uneven
press coverage, and variable reporting rates are
formidable. Nevertheless, the expansion of data
generation and analysis over the past decade
has enhanced the evidence base on which an
understanding of the magnitude and distribution
of direct conflict deaths may be based. As the
next section demonstrates, a composite analysis
of multiple datasets can generate a more reliable
and consolidated estimate.
13
Box 1.2 Using population surveys to estimate direct conflict deaths
There are a wide number of competing estimates of the global burden of
violent war deaths since the 1950s. A recent article in the British Medical
Journal disputes the claim, advanced by the Human Security Report Project
(Human Security Centre, 2006, p. 8), that direct conflict deaths have steadily
declined to an average annual figure of 15,000–20,000 since the Second
World War (Obermeyer et al., 2008). By offering alternative methods—
notably, population-based surveys focusing on sibling mortality and administered by the World Health Organization—Obermeyer et al. estimate that
at least twice as many people are directly killed as a result of war than
reported in the Human Security Report 2005.
Overall, population-based surveys routinely estimate many times more violent
deaths than do incident-reporting techniques. For example, Obermeyer et al.
find that an estimated 5.4 million people (3–8.7 million with a 95 per cent
confidence interval) were killed in just 13 countries from 1955 to 2002. The
spread between countries was large, with the cumulative direct death toll
ranging from 7,000 in the Republic of the Congo to 3.8 million in Vietnam.
The average between 1995 and 2002 was 36,000 deaths per year—a decline
on the previous two decades, but nevertheless rising in the last few years
(Obermeyer et al., 2008).13
Household population surveys routinely offer a more reliable estimate of
direct conflict deaths than other methods, including incident reporting,
rapid epidemiological surveys, and demographic assessments. Few
researchers dispute the claim that robust population-based surveys are
the best route to determining the scale and distribution of conflict deaths.
Undertaking high-quality population surveys in conflict zones is extremely
difficult. As a result, there may not be enough high-quality surveys on
which to base global estimates. The analysis presented by Obermeyer et al.
(2008) generalizes global trends from a modest sample of just 13 studies 14
and also features extremely wide confidence intervals. It should be recalled
that undercounting varies greatly between and within conflicts, and that
certain countries register excessively high numbers of ‘war dead’ (Garfield,
2008). Greater investment in surveys and surveillance in war zones is
therefore especially critical.
Measuring the global burden of
direct conflict deaths
This section proposes an estimate of direct conflict deaths for 2004–07 based on a new metadatabase established for the Global Burden of
Armed Violence (GBAV) report.10 The estimate
draws on a combination of conflict databases,
national datasets, and studies that capture direct
conflict death using incident-reporting methods.
A comparative analysis of multiple datasets and
the establishment of ranges for direct conflict
deaths potentially provide a more complete picture of direct conflict deaths than a narrower focus
on a single dataset.
D I R E C T CO N F L I C T D E AT H
us, non-state and intercommunal armed violence
in wartime are often significant contributors to
the overall burden of direct conflict deaths.
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
14
Photo " An Iraqi boy
runs past a car, just as it
explodes in front of
al-Nahdha High School,
Baghdad, 2005.
© Ali Jasim/Reuters
The GBAV database on direct conflict death analysed
19 cross-country databases11 capturing information on 141 conflict-affected countries since 2000.
Eight of these databases include sufficient information from the period 2004–07 and were used
in generating a global estimate. These include:
UCDP One-Sided Violence Dataset, v.1.2, covering 2004–05 (UCDP, 2006c);15
IISS, Armed Conflict Database, for data covering the period 2004–07 (IISS, 2008);
Project Ploughshares, Armed Conflicts Report,
covering 2004–07 (Project Ploughshares,
2007); and
UCDP Battle-Deaths Dataset, v.4.1, covering
2004–05 (UCDP, 2006b);12
UCDP Non-State Conflict Dataset, v.1.1, covering 2004–05 (UCDP, 2006a);
SIPRI Yearbook 2007, covering 2006 (SIPRI,
2007);
Political Instability Task Force database, covering 2004–06 (PITF, 2006);
PRIO, Battle-Deaths Data, v.2.0, covering
2004–05 (PRIO, 2008).
In 22 cases, data from country-specific databases
was used to complement figures from crosscountry datasets. While these datasets have
somewhat different definitions and methodologies, it is possible to compare and integrate them
into a homogeneous measure centred on the
threat to human life arising from conflict-related
violence. The differences in the figures reported
by different databases are linked to different
methods and definitions, but careful comparison
allows a wider range of estimates to be used.16
The analysis covered conflicts between 2004 and
2007, revealing that 41 armed conflicts accounted
for some 98 per cent of all direct conflict deaths.17
Cases selected for the meta-database conformed
to the following three criteria:
the conflict-affected country appeared in at
least 7 of the 19 databases;18
at least one of the eight databases reported
more than 100 deaths in one year; and
the conflict was ongoing, and at least one
death occurred in 2007.19
A comparative analysis of direct conflict deaths
allows for the identification of differences across
databases that are due to varying capturing tech-
A comparative analysis also allows for the verification of direct conflict death trends over time.
The similarity in trends across databases is striking for 2004 and 2005, although it is possible to
observe differences in levels across the databases (see Figure 1.1). The similarities, however,
do not apply to the period 2005–07, where large
differences are observable among databases.
These are mainly due to outliers or omissions of
major conflicts; nevertheless, a comparative
analysis calls into question the declines in conflict
deaths reported in certain sources that typically
rely on a single database (Human Security Report
Project, 2008, pp. 6, 33–34).
Another advantage is that the creation of point
estimates provides a more comprehensive understanding of the magnitude of armed conflict deaths.
Due to its comparative nature, the GBAV estimate
does not rely on the data from a single database,
but establishes the average of various databases.21
Nevertheless, as with all databases focusing on
conflict deaths, the GBAV estimates are equally
subject to undercounting.
Figure 1.1 Total direct conflict deaths by database,
main armed conflicts, 2004–07
15
D I R E C T CO N F L I C T D E AT H
niques and inclusion requirements (see Figure 1.1).
In most cases, the differences between datasets
are related to the over- or non-counting of a particular conflict. The (unusually) high direct death
toll reported by IISS in 2004, for example, is due
to an estimate of 50,000 conflict deaths in Darfur,
which are not reported by other databases. The
high figure reported by Project Ploughshares in
2006 is determined by an Iraq estimate that is
higher in comparison to other databases. In 2007
Ploughshares records a massive decrease in direct
conflict deaths, however, because it does not
capture estimates for several large conflicts (Iraq
and Colombia). These examples illustrate how a
comparative analysis of datasets allows for crosschecking and the identification of outliers that
may over- and undercount direct conflict deaths.20
100,000
90,000
80,000
70,000
60,000
50,000
40,000
30,000
20,000
10,000
0
2004
2005
2006
2007
Legend:
GBAV estimate
(2007)
IISS (2008)
PITF (2006)
UCDP non-state (2006a)
Project Ploughshares (2007)
UCDP state (2006b)
PRIO (2008)
UCDP one-sided (2006c)
Global and regional estimates
and trends
Approximately 52,000 direct conflict deaths
occurred every year between 2004 and 2007. In
the combined four-year period, at least 208,300
people died directly as a result of armed conflict.22
This figure is higher than the annual estimate provided by others, including Obermayer et al. (2008)
and the Human Security Brief 2007 (Human Security
Report Project, 2008, p. 34). Such estimates highlight that direct deaths from armed conflict are
far from negligible, even though they are remarkably low in comparison to historical levels.
Reducing the incidence of armed conflict could
reduce the global burden of armed violence by a
maximum of ten per cent. However, the impact of
the reduction of conflict violence is possibly much
higher due to the simultaneous reduction of the
disruptive indirect consequences of warfare
(INDIRECT CONFLICT DEATHS).
SIPRI
UCDP total
1
2
3
4
5
6
7
Table 1.1 Direct conflict deaths by region and subregion, and as percentage of total direct conflict deaths, 2004–07
Direct conflict deaths by region and subregion
Subregion
Africa
East Africa
4,188
2,459
2,399
9,078
18,124
4,531
9%
6%
4%
14%
9%
North Africa
7,783
1,603
2,793
2,154
14,332
3,583
17%
4%
5%
3%
7%
38
21
10
–
69
23
0%
0%
0%
0%
0%
West and
Central Africa
5,642
4,882
2,793
3,156
16,472
4,118
12%
11%
5%
5%
8%
Africa total
17,651
8,965
7,995
14,388
48,997
12,255
38%
21%
14%
23%
24%
315
150
61
4
530
133
1%
0%
0%
0%
0%
Central America
28
54
–
–
82
41
0%
0%
0%
0%
0%
North America
38
180
65
–
283
94
0%
0%
0%
0%
0%
South America
3,047
3,142
2,162
3,648
11,999
3,000
7%
8%
4%
6%
6%
America total
3,428
3,526
2,288
3,652
12,894
3,268
7%
8%
4%
6%
6%
Central Asia and
Transcaucasia
101
250
60
29
440
110
0%
1%
0%
0%
0%
East and Southeast Asia
1,556
1,574
1,037
1,244
5,410
1,353
3%
4%
2%
2%
3%
Near and
Middle East/
South-west Asia
13,096
18,380
35,369
34,863
101,708
25,427
28%
43%
63%
55%
49%
7,729
7,444
7,718
7,252
30,143
7,536
17%
18%
14%
11%
14%
22,482
27,648
44,184
43,388
137,701
34,426
49%
65%
79%
68%
66%
1,641
1,079
405
267
3,391
848
4%
3%
1%
0%
2%
South-east
Europe
223
612
256
418
1,508
377
0%
1%
0%
1%
1%
Western and
Central Europe
211
–
7
4
222
74
0%
0%
0%
0%
0%
2,075
1,691
668
689
5,121
1,299
5%
4%
1%
1%
2%
435
660
743
1,793
3,631
908
1%
2%
1%
3%
2%
46,071
42,490
55,878
63,910
208,344
52,156
100%
100%
100%
100%
100%
Southern Africa
G LO B A L B U R D E N
Americas
Asia
Caribbean
South Asia
Asia total
Europe
Eastern Europe
Europe total
International terrorism
Total
2004
2005
2006
2007
2004–
07
Annual percentage of total direct conflict deaths
Year
of
A R M E D V I O L E N CE
16
Note: dashes indicate that values are either zero or that no information is available.
Source: GBAV estimates
Average
2004
2005
2006
2007
2004–
07
Figure 1.2 Estimates of the regional distribution of direct conflict deaths, 2004–07
D I R E C T CO N F L I C T D E AT H
17
Legend:
East Africa
2004
2005
2006
AFRICA
North Africa
2007
Source: GBAV estimates
Southern Africa
West and Central Africa
AMERICAS
Caribbean
Central America
North America
South America
1
2
Central Asia and Transcaucasia
3
East and South-east Asia
ASI A
4
5
$
$
$
$
$
$
Near and Middle East/South-west Asia
6
South Asia
7
E U ROP E
Eastern Europe
South-east Europe
Western and Central Europe
International terrorism
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
Table 1.2 Estimates of the regional distribution of direct conflict deaths, 2004–07
Conflict
Asia
2004
2005
2006
2007
2004–07
Central Asia and Transcaucasia
Uzbekistan25
43
180
20
8
251
Armenia–Azerbaijan
10
58
15
13
96
Georgia
39
12
25
8
84
Thailand
600
535
246
452
1,833
Philippines
326
385
388
451
1,550
Myanmar
38
463
262
268
1,031
Indonesia
583
192
90
22
887
9,803
15,788
26,910
23,765
76,266
Afghanistan
917
1,000
4,000
6,500
12,417
Pakistan
863
648
1,471
3,599
6,581
Israel (and Palestinian Territories)
899
226
673
449
2,247
Lebanon–Syria
5
36
1,708
478
2,227
Iran
–
70
57
72
199
109
330
4,126
4,500
9,065
India
2,642
2,519
1,559
1,713
8,433
Nepal
3,407
2,950
792
137
7,286
1,511
1,552
1,116
777
4,956
60
54
90
108
312
Asia main armed conflicts (18)
21,855
26,998
43,548
43,320
135,717
Asia all countries (26)
22,482
27,648
44,184
43,388
137,701
Somalia
760
285
879
6,500
8,424
Ethiopia
824
825
1,091
2,418
5,158
Uganda
1,649
859
196
111
2,815
Burundi
820
269
108
49
1,246
Kenya
40
124
125
–
289
Rwanda
75
92
–
–
167
East and South-east Asia
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
18
Near and Middle East/South-west Asia
Iraq
South Asia
Sri Lanka
India–Pakistan (Kashmir)
India (Nagaland)
Africa
East Africa
Sudan
7,284
1,098
2,603
1,734
12,719
Algeria
465
381
190
420
1,456
38
21
10
–
69
DRC
3,500
3,750
746
1,351
9,347
Nigeria
1,686
298
305
535
2,824
80
105
1,325
1,044
2,554
Southern Africa
Angola
West and Central Africa
Chad
Central African Republic
Côte d’Ivoire
Senegal
Sierra Leone
–
550
128
160
838
341
168
184
24
717
10
–
106
–
116
–
–
–
4
4
Africa main armed conflicts (16)
17,572
8,825
7,996
14,350
48,739
Africa all countries (21)
17,651
8,965
7,995
14,388
48,997
2,988
3,092
2,141
3,612
11,832
315
150
61
4
530
Americas main armed conflicts (2)
3,303
3,242
2,202
3,616
12,362
Americas all countries (7)
3,428
3,526
2,288
3,652
12,894
1,641
1,079
405
267
3,391
183
603
247
398
1,430
0
0
6
12
18
191
–
2
2
195
Europe main armed conflicts (4)
2,015
1,682
660
679
5,034
Europe all countries (7)
2,075
1,691
668
689
5,121
435
660
743
1,793
3,631
Total main armed conflicts (41)
45,180
41,407
55,149
63,758
205,483
Total all countries (62)
46,071
42,490
55,878
63,910
208,344
Americas
South America
Colombia
Caribbean
Haiti
Europe
1
Eastern Europe
Russian Federation
Turkey
Former Yugoslav Republic of Macedonia
(FYROM)
Western and Central Europe
Spain
International terrorism
2
3
4
South-east Europe
Notes: The total figure of 208,344 includes all information on direct conflict deaths available for 62 conflicts. The figure 205,483 captures the information available for the
41 main armed conflicts between 2004 and 2007. Dashes indicate that values are either zero or that no information is available.
Source: GBAV estimates
D I R E C T CO N F L I C T D E AT H
19
North Africa
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
20
Photo " A Tamil woman
receives medical treatment after being
wounded by a Claymore
blast in Batticaloa
district, Sri Lanka, 2007.
© Will Baxter/WPN
While the total annual number of direct conflict
deaths decreased between 2004 and 2005 from
46,100 to 42,500, they subsequently increased to
63,900 in 2007 (see Table 1.1).23 This increase is
due primarily to armed violence in Iraq, Afghanistan, Sri Lanka, and Somalia (see Table 1.2).
Significantly reducing armed violence in Iraq,
Afghanistan, Sri Lanka, and Somalia in 2005
would have reduced the global level of direct
conflict deaths by 30 per cent in 2006 and 64
per cent in 2007.24
The regional distribution of direct conflict deaths
warrants closer inspection. Approximately two-
thirds (66 per cent) of all direct conflict deaths
between 2004 and 2007 occurred in Asia, almost
one-quarter (24 per cent) in Africa, 6 per cent in
the Americas, and 2 per cent in Europe. Two per
cent of all direct conflict deaths can be attributed
to international terrorism (see Table 1.1).
However, there are significant differences concerning the subregional distribution of direct conflict
deaths (see Figure 1.2). In Asia, for example, it can
be seen that direct conflict deaths in the Near and
Middle East/South-west Asia increased about
threefold between 2004 and 2007 (see Table 1.1).
This was mainly due to the wars in Iraq, Afghanistan, and Pakistan. In South Asia, the stable levels
of direct deaths are a reflection of fewer direct
conflict deaths in Nepal and increases in Sri Lanka.
East and South-east Asia, as well as Central Asia
and Transcaucasia have relatively low levels of
direct conflict deaths (see Table 1.2).
In Africa, direct conflict deaths decreased in 2005
and 2006, but increased in 2007. This is mainly
due to increasing numbers of direct conflict
deaths in East Africa (Somalia and Ethiopia). In
North Africa in this period, direct conflict deaths
decreased due to the lower figures for Sudan
(see Table 1.2). The slight decrease in West and
Central Africa direct conflict deaths was mainly
due to declining levels of direct conflict deaths in
the Democratic Republic of the Congo (DRC) and
Nigeria, even though there was a slight increase
in 2007 (see Table 1.2).26
In the Americas and Europe, direct conflict deaths
stayed at relatively low levels. In the Americas,
the level of direct conflict deaths is mainly defined
by the conflict in Colombia (see Table 1.2).
Direct deaths from international terrorism increased
in the period 2004–07 but remained at low levels
(see Table 1.2).27
21
D I R E C T CO N F L I C T D E AT H
Map 1.1 The GBAV conflict-affected countries
6977-%2*)()6%8-32
%61)2-%
+)36+-%
%>)6&%-.%2
74%-2 *=631
896/)=
7=6-%
-6%5
0)&%232
-76%)0
%0+)6-%
9>&)/-78%2
-6%2
4%/-78%2
%*+,%2-78%2
2)4%0
-2(-%
,%-8-
7)2)+%0
7-)66%0)32)
'3031&-%
',%(
79(%2
2-+)6-%
'%6
'Å8)(«-:3-6)
1=%21%6
8,%-0%2(
4,-0-44-2)7
)8,-34-%
76-0%2/%
731%0-%
(6'
%2+30%
/)2=%
9+%2(%
6;%2(%
&9692(-
-2(32)7-%
1
2
0)+)2(
Conflict-affected countries
3
4
5
6
Direct conflict deaths in
individual countries
Iraq accounts for 28 per cent of all direct conflict
deaths in this period.
As the previous section observes, direct conflict
deaths tend to be highly concentrated in a limited
number of countries. Table 1.3 lists the ten countries with the highest level of direct conflict deaths
for the period 2004–07. It is led by Iraq, with a total
of around 51,400 direct conflict deaths, followed
by Sudan with 12,700, Afghanistan with 12,400,
Colombia with 11,800, and the DRC with 9,300.
These five conflicts account for around half of all
direct conflict deaths between 2004 and 2007.
The top ten countries account for three-quarters
of all direct conflict deaths. While the figures presented in Table 1.3 are sure to be undercounting
direct conflict deaths by a considerable margin,
they nevertheless highlight that targeted initiatives
against specific armed conflicts could contribute
to a significant reduction in the global burden of
armed violence.
The distribution of direct conflict deaths among
these conflicts has varied greatly between 2004
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
22
Table 1.3 Top ten direct conflict death countries, relative and cumulative total percentages, 2004–07
Conflicts
1.
Iraq
2.
Direct conflict deaths
% of total conflict deaths
Cumulative % of total
conflict deaths
76,266
36.6%
36.6%
Sudan
12,719
6.1%
42.7%
3.
Afghanistan
12,417
6.0%
48.7%
4.
Colombia
11,832
5.7%
54.4%
5.
DRC
9,346
4.5%
58.9%
6.
Sri Lanka
9,065
4.4%
63.3%
7.
India
8,433
4.0%
67.3%
8.
Somalia
8,424
4.0%
71.3%
9.
Nepal
7,286
3.5%
74.8%
10.
Pakistan
6,581
3.2%
77.9%
Source: GBAV estimates
Box 1.3 The dramatic impact of particularly violent conflicts
Specific armed conflicts with high rates of direct conflict deaths highlight
the significant under-counting in aggregate estimates of such deaths from
incident reporting systems. Epidemiological surveys in DRC estimated that
approximately 5.4 million people died as a consequence of the armed conflict between August 1998 and April 2007. While most of these deaths are
attributable to indirect causes, about ten per cent were estimated to be
direct conflict deaths. This represents an annual average of about 50,000
direct conflict deaths—more than the global total reported in incidentreporting datasets (Coghlan et al., 2008).
This figure not only underlines the potential undercounting of incidentreporting methods—upon which the GBAV estimates of direct conflict
deaths are based—but also that an alternative estimate of just one particularly severe conflict can double or triple the global estimates of direct
conflict deaths. The conditions of reporting and documenting deaths in
the DRC, where there is poor mortality monitoring by the press and NGOs
in many regions, help explain this undercounting. On the other hand, the
lack of denominator data might lead to an overcount or wide confidence
intervals for survey-based estimates. It is therefore crucial to identify
clearly the variation in estimates in those conflicts with high levels of direct
conflict deaths. Cases such as the DRC, Iraq, Sudan, and Afghanistan show
that armed violence reduction in one armed conflict can lead to substantial
reductions in the global burden of armed violence.
and 2007 (see Table 1.2). Iraq registered the
highest number of direct conflict deaths for 2004
through 2007. Overall, nine conflicts registered a
higher ranking in 2007 than in 2004 (see Map 1.2).
Iraq, Afghanistan, Somalia, Sri Lanka, and Pakistan are the five conflicts with the most dramatic
increases in direct conflict deaths. Iraq reported
the highest increase in direct conflict deaths,
from around 9,800 in 2004 to 23,800 in 2007. In
Afghanistan, direct conflict deaths increased from
about 900 in 2004 to 6,500 in 2007; in Somalia
from 800 to 6,500; in Sri Lanka from 100 to 4,500;
and in Pakistan from 900 to 3,600. The number
of conflict deaths in Lebanon peaked in 2006 and
then declined in 2007, to a level higher than in
2004 (see Table 1.2).
Ten conflicts registered a decrease in their direct
conflict death figures (see Map 1.2). The armed
conflicts in Sudan, the DRC, India, Nepal, Nigeria,
Uganda, the Russian Federation, and India–Pakistan
all reported decreases (see Map 1.2). These countries also revealed lower figures of direct conflict
6977-%2*)()6%8-32
0)&%232 -6%5 %*+,%2-78%2 /EWLQMV
2)4%0
4%/-78%2
-2(-%
',%(
79(%2
2-+)6-%
)8,-34-%
'3031&-%
23
D I R E C T CO N F L I C T D E AT H
Map 1.2 Increases and decreases in direct conflict deaths in selected armed conflicts, 2004 and 2007
76-0%2/%
9+%2(% 731%0-%
(6'
-2(32)7-%
0)+)2(
Increasing direct
1
conflict deaths
2
Decreasing direct
3
conflict deaths
4
Note: This map captures changes of more than 400 direct deaths for the years 2004 and 2007.
5
deaths in 2007 than in 2004. The most significant
single reduction in direct conflict deaths occurred
in Sudan, from 7,300 in 2004 to 1,700 in 2007,
even though there was a slight increase in 2006
(see Table 1.2). Another drastic reduction took
place in Nepal, from 3,400 direct conflict deaths
in 2004 to just about 140 in 2007. In the DRC,
direct conflict deaths declined by approximately
half, although some 1,350 people were killed
during clashes in 2007 (see Table 1.2).
Fortunately, it is possible to reduce direct conflict
deaths measurably in certain countries as peace
deals are negotiated and peacekeepers are
deployed. Even so, optimism should be tempered
with caution, since the reductions noted above
were dwarfed by the rising number of direct conflict deaths occurring in so-called ‘post-conflict’
contexts, such as Iraq, Afghanistan, and Sudan
(see Box 1.4). Ultimately, a review of direct conflict deaths provides a partial picture of the burden
of armed violence. As other chapters of this report
make clear, a comprehensive estimate of the burden of armed violence in a specific country should
also include estimates of indirect conflict deaths,
non-conflict homicides, and extrajudicial killings
(NON-CONFLICT ARMED VIOLENCE, OTHER FORMS
OF ARMED VIOLENCE).
6
7
24
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Box 1.4 Accounting for direct deaths in Sudan
In spite of decades of war, it is difficult to establish the scale and
magnitude of direct conflict deaths in Sudan. Armed conflict in
the South (1955–72 and 1983–2005) and Darfur (2003–present)
have frustrated attempts to collect reliable data, whether through
incident reporting or surveys. In the South, bloody conflicts were
waged between the Sudanese army and militia against separatist
rebel groups such as the Sudan People’s Liberation Army (SPLA).
In Darfur, the Sudanese armed forces and militia are fighting
against disparate rebel groups including the Sudan Liberation
Movement (SLM).
There are widespread disagreements concerning the human costs
of war in Sudan. Estimates of the direct and indirect death toll range
from several million in the case of Southern Sudan (between 1983
and 2005) to more than 300,000 in the case of Darfur.
Drawing on multiple datasets, the Global Burden of Armed Violence
report finds that the direct toll in recent years may have declined
considerably, although indirect mortality (INDIRECT CONFLICT
DEATHS) may have remained high. Table 1.4 reveals that fewer
than 7,400 people were probably killed directly during ‘battles’
in South Sudan in 2002–07, with sharp reductions following the
2005 peace agreement. Likewise, Table 1.5 highlights that direct
deaths arising from ‘battles’ in Darfur in 2003–07 were probably
slightly more than 15,500.
These datasets do not necessarily account for routine violence or
genocide perpetrated against civilians by armed groups. One
important dataset that accounts for a broader array of direct deaths
(including civilians) is that of the Political Instability Task Force,
or PITF. The PITF tends to report much higher rates of conflict deaths
in Sudan than do most other databases. Likewise, epidemiological
surveys tend to provide a wider accounting of direct deaths, as
Table 1.6 makes clear. An important lesson is that incident reporting
should be undertaken in unison with probabilistic survey-based
estimates, especially where conflict-related violence is perpetrated
by government proxies and remains largely hidden from view.
Photo A severely injured man suffering
from gunshot wounds is carried to a waiting
helicopter in Muhajiriya, south Darfur, 2007.
© Stuart Price/AMIS/Reuters
Table 1.4 Estimated direct conflict deaths in South Sudan, 2002–07
2002
2003
2004
2005
2006
2007
Total
IISS
1,000
1,000
200
90
750
445
3,485
PRIO
2,175
3,000
253
n/a
n/a
n/a
5,428
Project Ploughshares
1,300
100
625
250
500
n/a
3,300
n/a
n/a
n/a
n/a
25–100
n/a
63
2,254
44
142
n/a
n/a
n/a
2,440
UCDP non-state
91
186
n/a
130
n/a
n/a
407
UCDP one-sided
25
69
33
n/a
n/a
n/a
127
UCDP state and non-state and one-sided
2,370
299
175
130
n/a
n/a
2,974
GBAV
2,370
3,000
625
190
750
445
7,380
SIPRI
UCDP state
D I R E C T CO N F L I C T D E AT H
Database
25
Table 1.5 Estimated direct conflict deaths in Darfur, 2003–07
Database
2003
2004
IISS
2005
2006
2007
Total
n/a
50,000
500
987
1,289
52,776
2,175
3,000
500
n/a
n/a
5,675
5,000
350
1,000
1,250
n/a
3,100
UCDP non-state
170
81
30
n/a
n/a
281
UCDP one-sided
3,056
3,283
604
n/a
n/a
6,943
UCDP state
1,636
3,025
161
1,002
217
6,041
UCDP total
4,862
6,659
795
1,002
217
13,535
GBAV
4,862
6,659
908
1,853
1,289
15,571
PRIO
Project Ploughshares
Nabarro
(excess
deaths)
US State
Department
(excess
deaths)
Coebergh
(excess
deaths)
Coebergh
(violent
deaths)
Hagan et al.
(total
deaths)
Reeves
(excess
deaths)
CRED
(excess
deaths)
CRED
(violent
deaths)
GBAV
2003
–
15,873
119,936
78,979
310,355
152,000
12,692
7,530
4,862
2004
52,500
21,164
159,915
105,305
228,846
182,400
75,813
22,588
6,659
2005
–
1,764
–
–
91,518
45,600
42,555
10,817
908
Total
35,000–
70,000
63,000–
146,000
253,573–
306,130
172,542–
196,025
630,719
38,000
131,060
40,935
12,429
Note: Dashes indicate that values are either zero or that no information is available.
2
3
4
5
6
7
Table 1.6 Epidemiological surveys of direct conflict deaths in Sudan, 2003–05
Year
1
The risk of dying in armed conflict
26
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Wars grab headlines, but the individual risk of
dying violently in an armed conflict is today relatively low—much lower than the risk of violent
death in many countries that are not suffering
from an armed conflict. Although there is a widespread perception that war is the most dangerous
form of armed violence in the world, the average
person living in a conflict-affected country had a
risk of dying violently in the conflict of about 2.0
per 100,000 population between 2004 and 2007
(see Table 1.7).28
This can be compared to the average world homicide rate of 7.6 per 100,000 people (NON-CONFLICT
ARMED VIOLENCE). This illustration highlights
the value of accounting for all forms of armed
violence rather than an exclusive focus on conflictrelated violence.
Certainly, there are huge variations in the risk of
dying from armed conflict at the national and subnational level, and the the risk of dying violently
in a conflict in specific countries remains extremely
high. In Iraq, for example, the direct conflict death
rate for 2004–07 was 65 per 100,000 people per
year and, in Somalia, 24 per 100,000 people. This
rate even reached peaks of 91 per 100,000 in Iraq
in 2006 and 74 per 100,000 in Somalia in 2007.
Table 1.7 shows that ten countries had a direct
conflict death rate higher than 5 per 100,000
population between 2004 and 2007. Map 1.3
graphically shows the highest risk of dying from
direct conflict death per conflict in 2004–07.
Box 1.5 International vs. intrastate conflict
It is possible to distinguish between international
and intrastate armed conflicts. The former refer
to classic interstate warfare, as well as armed
conflicts in which at least one of the belligerents
is an external state party, while the latter refer to
a situation in which two (or more) parties within
a single country fight against each other. While
such characterizations become increasingly
difficult to maintain, given the complex and
globalized nature of many armed conflicts, they
nevertheless capture the main actors involved
in and the locus of armed conflict.
The GBAV database finds that, in 2006 and
2007, more people died from international and
internationalized armed conflicts than from
intrastate conflict, mainly due to the wars in
Iraq, Afghanistan, and Somalia, all of which
involved some form of intervention by other
states. The number of direct conflict deaths in
international and internationalized armed conflicts increased nearly threefold from around
14,400 in 2004 to 41,000 in 2007, mainly due to
the war in Iraq. The number of direct conflict
deaths from intrastate conflicts decreased by
about one-third from 31,600 in 2004 to 23,500
in 2007 (see Figure 1.3).
Figure 1.3 Direct conflict deaths by armed conflict type, 2004–07
Legend:
International and internationalized internal
armed conflict
2004
2005
2006
Intrastate conflict
2007
Source: GBAV estimates
0
10,000
20,000
30,000
40,000
50,000
Table 1.7 Direct conflict death rate by country, 2004–07 (per 100,000)
2005
2006
2007
Average, 2004–07
34.94
54.81
91.06
78.46
64.82
Somalia
9.54
3.46
10.34
74.15
24.37
Sri Lanka
0.53
1.59
19.73
21.35
10.80
Afghanistan
3.21
3.35
12.87
20.15
9.89
Sudan
20.51
3.03
7.04
4.59
8.79
Israel (and Palestinian Territories)
13.62
3.36
9.83
6.44
8.31
Central African Republic
0.00
13.62
3.13
3.85
6.87
Nepal
12.81
10.87
2.86
0.49
6.76
Colombia
6.65
6.78
4.63
7.69
6.44
Chad
0.85
1.08
13.21
10.13
6.32
Burundi
11.26
3.56
1.37
0.60
4.20
DRC
6.27
6.52
1.26
2.21
4.06
Uganda
5.93
2.98
0.66
0.36
2.48
Ethiopia
1.09
1.07
1.38
2.98
1.63
Haiti
3.75
1.76
0.71
0.05
1.56
Algeria
1.44
1.16
0.57
1.24
1.10
Pakistan
0.56
0.41
0.91
2.19
1.02
Côte d’Ivoire
1.91
0.93
1.00
0.13
0.99
D I R E C T CO N F L I C T D E AT H
2004
Iraq
27
1
2
Rwanda
0.84
1.02
0.00
0.00
0.93
Thailand
0.94
0.83
0.38
0.69
0.71
Russian Federation
1.14
0.75
0.28
0.19
0.59
Nigeria
1.31
0.23
0.23
0.39
0.54
4
5
3
Myanmar
0.07
0.92
0.51
0.52
0.51
Senegal
0.09
0.00
0.89
0.00
0.49
Turkey
0.25
0.82
0.33
0.53
0.48
6
7
Georgia
0.85
0.27
0.56
0.18
0.47
Philippines
0.40
0.46
0.46
0.52
0.46
FYROM
0.00
0.00
0.29
0.59
0.44
Kenya
0.12
0.36
0.36
0.00
0.28
India
0.24
0.23
0.14
0.15
0.19
Spain
0.45
0.00
0.00
0.00
0.15
Angola
0.24
0.13
0.06
0.00
0.14
Indonesia
0.26
0.09
0.04
0.01
0.10
Iran
0.00
0.10
0.08
0.10
0.09
Sierra Leone
0.00
0.00
0.00
0.07
0.07
1.77
1.59
2.04
2.38
1.95
Rate (all countries listed above)
Source: GBAV estimates
Map 1.3 The risk of dying violently in armed conflict per 100,000 population per year, average, 2004–07
A R M E D V I O L E N CE
28
6977-%2*)()6%8-32
+)36+-%
74%-2
%*+,%2-78%2
896/)=
*=631
-6%5 -6%2
-76%)0
2)4%0
of
%0+)6-%
G LO B A L B U R D E N
-2(-%
,%-8-
7-)66%0)32)
'3031&-%
',%(
79(%2
7)2)+%0
2-+)6-%
'%6
'Å8)(«-:3-6)
(6'
%2+30%
8,%-0%2(
4%/-78%2
)8,-34-%
731%0-%
/)2=%
9+%2(%
6;%2(%
&9692(-
1=%21%6
4,-0-44-2)7
76-0%2/%
-2(32)7-%
0)+)2(
4IV
TSTYPEXMSR
"
¦
¦
¦
Conclusion
This chapter features a global estimate of contemporary direct conflict deaths. In synthesizing a
number of datasets, the chapter presents a somewhat higher, more pessimistic figure than those
presented by widely cited sources such as the
Human Security Report or the Human Security
Brief (Human Security Centre, 2005; 2006; Human
Security Report Project, 2008). Equally, the chapter
clearly demonstrates that the number of direct
conflict deaths has increased since 2005, thus
departing from a decline in direct conflict deaths
observed since the end of the cold war (Human
Security Report Project, 2008, pp. 6, 32–34). Even
so, the numbers of direct deaths in battle are still
relatively modest in comparison to historical figures and other forms of conflict and non-conflict
mortality. Subsequent chapters emphasize the
importance of moving beyond a single focus on
direct conflict deaths and including all types of
deaths from violence, including those dying indirectly from armed conflict and from homicide in
non-conflict settings.
Certain armed conflicts are much more deadly in
terms of their direct death toll than others. As the
chapter amply shows, a small number of countries
The risk of dying violently in armed conflict is
considerably lower than of being a victim of nonconflict homicide. Furthermore, the risk of dying
from armed conflict is not evenly distributed
among or within countries.
Enhancing our understanding of the spatial and
temporal distribution of direct conflict deaths is
critical. It is likely, for example, that the risk of
dying differs at the subnational level and among
different social groups. A more robust evidence
base—particularly more complete and better
disaggregated data—could contribute to the
strategic planning of humanitarian or peacekeeping missions in support of armed violence
reduction and prevention.
Abbreviations
COW
Correlates of War
DRC
Democratic Republic of the Congo
FYROM
Former Yugoslav Republic of Macedonia
GBAV
Global Burden of Armed Violence
IISS
International Institute for Strategic Studies
MSE
multiple systems estimation
PITF
Political Instability Task Force
PRIO
International Peace Research Institute, Oslo
UCDP
Uppsala Conflict Data Programme
Endnotes
1
For example, Iraq was the deadliest armed conflict between
2004 and 2007, with 76,266 direct conflict deaths, or 36
per cent of the total direct conflict deaths burden (see also
Table 1.3).
2
These estimates are based on figures for 2004.
3
The number of direct conflict deaths occurring in international
and internationalized armed conflicts increased threefold
from 14,462 in 2004 to 40,391 in 2007 (see Figure 1.3).
4
The number of direct conflict deaths from intrastate conflict decreased by almost one-third from 31,607 in 2004
to 23,517 in 2007.
5
See methodological appendix at:
<http://www.genevadeclaration.org>.
6
See Iraq Body Count (2008).
7
Data from specific conflicts—including Peru, Guatemala,
Kosovo, Afghanistan, Iraq, and the Democratic Republic
of the Congo—all indicate the systematic undercounting
of incident reporting datasets (Small Arms Survey, 2005,
pp. 241–48).
8
The UCDP identifies three levels of violence: ‘minor conflicts’ cause at least 25 battle-related deaths in a year,
but fewer than 1,000 overall; ‘intermediate conflicts’
cause more than 1,000 battle-related deaths overall, but
fewer than 1,000 in any single year; and ‘wars’ cause at
least 1,000 battle-related deaths in a single year (Eriksson
et al., 2008, p. 617).
9
The five deadliest wars in terms of total deaths since the
Second World War were Vietnam (2,097,705 deaths between
1955 and 1975), Korea (1,254,811 deaths between 1950 and
1953), the Chinese civil war (1,200,000 deaths between
1946 and 1949), the Iran–Iraq war (644,500 deaths between
1980 and 1988), and the conflicts in Afghanistan (562,995
deaths between 1978 and 2002) (Gleditsch and Lacina,
2005, p. 154).
10
A meta-database is an integrated database made of
comparable and equivalent records taken from several
databases.
11
The databases and reports are: 1. SIPRI (2007); 2. Human
Security Centre (2006); 3. ICG (2008); 4. UCDP (2006a);
5. UCDP (2006b); 6. UCDP (2006c); 7. UCDP and Centre for
the Study of Civil War (2007); 8. Project Ploughshares (2007);
9. COW (2007); 10. IISS (2008); 11. Center for International
Development and Conflict Management (2005); 12. CRED
(2008a); 13. CRED (2008b); 14. Center for Systemic Peace
(2004); 15. Center for Systemic Peace (2007); 16. PITF (2006);
17. PTS (2008); 18. Gleditsch (2007); 19. Country documentation from a comprehensive bibliographic search.
29
D I R E C T CO N F L I C T D E AT H
accounts overwhelmingly for the global toll of
conflict deaths. Consequently, carefully targeted
armed violence reduction in a few selected countries could lead to measurable reductions in the
global burden of armed violence. In addition to
reducing the violent death toll, efforts to reduce
armed violence could generate additional dividends for human security, including declines in
refugee and internal displacement movements,
gross human rights violations, and indirect conflict mortality (INDIRECT CONFLICT DEATHS).
1
2
3
4
5
6
7
12
In order to complete and check the information of the
UCDP Battle-Deaths Dataset, the analysis includes information from the UCDP Web site (UCDP, 2008).
13
By contrast, the Small Arms Survey (2005, p. 247) estimates
the direct conflict death toll as at least three times higher
per year than the Human Security Report (more than 54,000).
14
There were an estimated 112 conflicts during the reporting
period, suggesting that Obermeyer et al. (2008) captured
just ten per cent of the entire sample.
15
This report adds the UCDP Battle-Deaths Dataset (UCDP,
2006b) to the UCDP Non-State Conflict Dataset (UCDP, 2006a),
and UCDP One-Sided Violence Dataset (UCDP, 2006c) in
order to produce a total UCDP figure (see Figure 1.1). This
is only available for 2004–05. All of UCDP’s categories—
interstate conflict, intrastate conflict, non-state conflict,
and one-sided violence—are separate and mutually exclusive (Eck, 2005).
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
30
16
This does not imply that figures have been overestimated.
The use of different sources allows for the generation of
more accurate values.
17
Conflict deaths in most datasets are defined as battle
deaths of official combatants, or (in some cases) noncombatant deaths, where the perpetrator is identified as
a combatant.
18
The cases included by these criteria are: Afghanistan,
Algeria, Angola, Burundi, Central African Republic, Chad,
Colombia, Democratic Republic of the Congo, Côte d’Ivoire,
Ethiopia, Georgia, Guinea, Haiti, India, Indonesia, Iran,
Iraq, Israel (and Palestinian Territories), Kenya, Liberia,
the Former Yugoslav Republic of Macedonia, Myanmar,
Nepal, Nigeria, Pakistan, the Philippines, Republic of the
Congo, Russian Federation, Rwanda, Senegal, Sierra Leone,
Somalia, Sri Lanka, Sudan, Thailand, Turkey, Uganda,
Armenia–Azerbaijan, International terrorism, Ethiopia–
Eritrea, and India–Pakistan (Kashmir).
19
For details about the main armed conflicts selection, see
the online methodological appendix (see endnote 5).
20
The GBAV database corrected these outliers and differences
between cross-country databases by including alternative
figures from sources such as national datasets or reports.
21
For countries with very large differences in figures, the
information from national sources or reports was used to
establish the most plausible figure of direct conflict deaths.
22
These figures are based on data for the main 41 armed
conflicts between 2004 and 2007, as well as 2,861 (208,344
minus 205,483; see totals in Table 1.2) direct conflict
deaths from 21 smaller armed conflicts.
23
The figures have been rounded up and down to the closest
hundred in comparison to Table 1.1 in order to emphasise
that these are not precise figures, but merely estimates.
24
These figures are based on the death figures of four conflicts (Iraq, Afghanistan, Sri Lanka, and Somalia) with high
changes in direct conflict deaths between 2005 and 2007.
In 2006 these four conflicts accounted for 35,915 deaths
of a total of 55,877, and in 2007 for 41,265 out of 63,908
(see Table 1.2).
25
The conflict in Uzbekistan is an internationalized armed
conflict involving the Islamic Movement of Uzbekistan,
which conducts operations mainly in Uzbekistan, but also
in Kazakhstan, Kyrgyzstan, and Turkmenistan.
26
The increase in 2007 was due to the conflicts in the DRC,
Nigeria, and Chad.
27
International terrorism includes the direct conflict deaths
in the conflict between the US government and the multinational coalition (state parties) vs. al Qaeda and Jemaah
Islamiah.
28
The rate of the risk of dying is 2.0 per 100,000 population if
it is only related to the population of conflict-affected countries. If it is related to world population, this risk of dying
in armed conflict decreases to 0.8 per 100,000 population.
T
he lethal impact of modern war extends
far beyond the number of soldiers and
civilians who die violently in armed combat
or clashes.1 As some analysts have pointed out,
‘the number of battle deaths . . . does not provide
a remotely adequate account of the true human
costs of conflict. War kills people in less direct (but
highly predictable) ways’ (Lacina and Gleditsch,
2005, p. 148; Garfield and Neugut, 1991).
Armed conflict generates a series of lethal but
indirect impacts on communities beyond the
number of people killed in battle or combat. In
the short term, indirect victims of armed conflict
die from a variety of specific causes, usually from
easily preventable diseases such as dysentery or
measles, or from hunger and malnutrition. These
deaths are a result of the loss of access to basic
health care, adequate food and shelter, clean
water, or other necessities of life. In the long run,
armed conflict affects mortality by its destructive
impact on the national economy and infrastructure
(including health facilities), on social cohesion,
and on psychological health and well-being (Li
and Wen, 2005, pp. 473–75; Murray et al., 2002;
Ghobarah, Huth, and Russett, 2003). All of these
factors can negatively affect the prospects for
post-conflict peace-building.
These indirect victims of war do not die violently.
But, from a human, moral, and political point of
view, the distinction between a violent and nonviolent death is irrelevant. All that matters is that
a number of people died who would otherwise
have lived if armed violence had not ravaged
their communities. An adequate account of the
direct and indirect impact of armed conflict is
also important for assessing whether international
humanitarian law and human rights law have
been violated, and whether groups in combat are
preying on civilian populations (Daponte, 2008).
In almost all contemporary conflicts, the number
of indirect victims of armed violence is many times
larger than the number of battle deaths. For example,
the International Rescue Committee’s series of
mortality surveys in the Democratic Republic of
the Congo (DRC) found that 5.4 million excess deaths
occurred between August 1998 and April 2007,
with 2.1 million occurring since the formal end of
war in 2002 (Coghlan et al., 2008). Of these 5.4
million excess deaths since 1998, fewer than ten
per cent died ‘directly’ or violently. Nearly all
deaths (90 per cent)—approximately 4.8 million
people—were indirect and caused mainly by
preventable infectious diseases, malnutrition, and
neonatal- and pregnancy-related conditions that
emerged in the resource-poor post-conflict environment. The number of battle deaths estimated in the
preceding chapter for the DRC in the period 2004–07
is about 9,300 (DIRECT CONFLICT DEATH).2
While the DRC may be an extreme case, since the
end of the cold war the overwhelming majority
of conflicts (95 per cent) are now fought within
national borders in poor countries, often reflecting
communal and political disputes that trap civilians
in insecure situations (Harbom, 2007; HSC, 2005).
31
I N D I R E C T CO N F L I C T D E AT H S
The Many Victims of War:
Indirect Conflict Deaths
Chapter Two
1
2
3
4
5
6
7
32
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
This chapter discusses what we know about
‘excess mortality’ and ‘indirect deaths’ in armed
conflict. It first overviews the epidemiological
and demographic methods for estimating excess
mortality, current knowledge gaps, and the scientific challenges. The second section summarizes
data from a variety of cases to arrive at some
benchmarks to evaluate the level of indirect victimization in contemporary conflicts. The chapter
closes with three brief case studies estimating
indirect deaths in South Sudan, Sierra Leone,
and Iraq.
The main findings of the chapter are the following:
In the majority of conflicts since the early
1990s for which good data is available, the
burden of indirect deaths was between three
and 15 times the number of direct deaths.
Photo ! A mother
holds up her severely
malnourished baby in the
refugee camp of Xjosa
Sabz Poosh, Afghanistan.
© Tim Dirven/Panos
Pictures
Most conflicts are either low-intensity civil wars
that involve poorly trained armies who target
civilians, or asymmetric wars that pit a wellequipped army against a militarily weaker opponent
(Harbom, 2007). Both scenarios inflict violent
(‘direct’) and non-violent (‘indirect’) deaths on
civilians. Contemporary armed conflicts involve
organized and disorganized armed forces inflicting
violence on both soldiers and civilians, with widespread consequences for the health and economic
infrastructures of whole countries. While violent
death is an indicator of armed conflict, disease
and malnutrition have been the main causes of
death among civilians in most major conflicts
since the late 1980s (Guha-Sapir and Degomme,
2005a).
Variation in the ratio of direct to indirect
deaths depends on the pre-conflict level of
development of the country, the duration of
the fighting, the intensity of combat, access
to basic care and services, and humanitarian
relief efforts.
The lethal burden of armed conflict in 2004–
07 was many times greater than the number
of direct conflict deaths. A reasonable average estimate would be a ratio of four indirect
deaths to one direct death in contemporary
conflicts, which would represent at least
200,000 indirect conflict deaths per year,
and possibly many more.3 There may have
been up to 400,000 indirect conflict deaths
per year in the DRC alone since 2002.
Appropriate methods exist to arrive at a more
accurate account of the number of indirect
deaths in conflict zones; these should be
applied systematically wherever possible to
individual conflicts.
What is excess mortality?
Epidemiologists use mortality rates to assess
the severity of the impact of conflict on civilian
populations affected by complex humanitarian
emergencies (Toole and Waldman, 1997; GuhaSapir et al., 2005; Checchi and Roberts, 2005).
Standardized mortality calculations make possible
comparisons between populations and judgements on the severity of a crisis.
Excess mortality captures the difference between
the death rates (‘crude mortality’) in a non-conflict
situation and in a conflict or crisis situation. It
includes those dying both from the direct and
the indirect consequences of armed conflict.
However, its accuracy depends on the reliability
of baseline mortality data. In many protracted
conflict areas the establishment of this baseline
is complicated by the absence of reliable data.
The crude mortality rate (CMR) is informative
only when compared with a national or regional
baseline CMR (the ‘expected’ mortality in a country
in a normal situation) or with alert level thresholds
which signify a crisis situation4 . The numerical
difference between the ‘crisis CMR’ and the ‘baseline CMR’ is termed the ‘excess mortality’. This
value represents the mortality that can be attributed to the crisis and is used to estimate the
magnitude of the emergency and to monitor
the humanitarian response. Excess mortality
is traditionally broken down into two types of
death—direct and indirect—according to whether
or not the cause of death was violence (see
Figure 2.1).
Direct deaths are caused by war-related injuries
and attacks (such as those inflicted by a bullet,
bomb, mine, machete, or assault) (SMART, 2005,
p. 81).5 Indirect deaths are caused by the worsening of social, economic, and health conditions in
the conflict-affected area. They can result from a
variety of different factors including (but not limited to) inability to access health care, damage to
health systems and public health infrastructure,
changes in behaviour that increase the incidence
of diseases, malnutrition, unsanitary living conditions, food insecurity, and loss of livelihood and
agricultural land (Guha-Sapir and van Panhuis,
2002; Gayer et al., 2007).
33
I N D I R E C T CO N F L I C T D E AT H S
Sexual violence in armed conflict accounts
for a sizable, albeit hidden, proportion of
indirect conflict deaths with the majority of
victims being women and girls.
1
2
3
4
5
6
7
Figure 2.1 Typology of conflict mortality
Indirect deaths
Excess mortality
Crude mortality
Direct deaths
Baseline mortality
Source: Ratnayake et al. (2008)
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
34
The magnitude of indirect deaths is difficult to
quantify and verify; however, its assessment—in
addition to direct deaths—is essential to understanding the true human impact of a conflict or crisis.
Although the concept of indirect death is relatively
new, it is also possible that quantifying indirect
deaths may contribute to holding legally accountable political and military leaders who are ultimately
responsible for these deaths (Thoms and Ron,
2007). Estimates of indirect deaths have been
neglected by human rights organizations, which
have traditionally aimed to document the direct
deaths due to violence. But improved collaboration between epidemiologists, statisticians, and
human rights organizations has been encouraged
in order to address the larger picture of the indirect costs of conflict (Thoms and Ron, 2007; Asher,
Banks, and Scheuren, 2008).
From a public health perspective, the concept of
indirect deaths is useful because it captures
deaths that might have been preventable through
Box 2.1 Crude mortality rates
Crude mortality rates (CMRs) can be expressed in different ways which are
useful for various purposes. Demographers and researchers for the UN’s
annual statistical yearbooks often use deaths per 1,000 persons per year,
as annual rates are most useful in this context. In conflicts and other complex
emergencies, deaths per 10,000 persons per day is the standard unit since
it is most practical for monitoring a humanitarian situation over a short period
of time. A humanitarian emergency is considered to be any situation where
the CMR is double the baseline rate (Sphere, 2004, p. 261). Various organizations place the emergency threshold at a CMR of 1.0 deaths/10,000/day.
This is roughly in line with the Sphere approach for sub-Saharan Africa,
which is 0.9 deaths/10,000/day.
The units for CMRs can easily be converted using basic equations:
1 death/10,000/day = 3.04 deaths/1,000/month
= 36.5 deaths/1,000/year
Note: For comparison purposes in this report, most figures in this report have been expressed
in deaths per 100,000 per year.
Source: Guha-Sapir, Degomme, and Altare (2007)
a bolstering of the public health system. Such
figures provide strong evidence for prioritizing
basic public health interventions (such as infectious disease surveillance, immunization, disease
control programmes, and water and sanitation
projects) in conflict and post-conflict situations.
Challenges to collecting and using
data on indirect deaths
Indirect deaths are inherently difficult to quantify
and attribute to conflict-related causes. There are
three reasons for this:
ongoing data collection is weak and speciallytargeted methods must be used;
the attribution of indirect deaths to the conflict is difficult; and
it is difficult to determine baseline mortality
rates in endemic conflict zones.
In conflict situations the ongoing collection of
health information is difficult due to the breakdown of information systems, the loss of human
resources, and restricted freedom of movement.
Health information systems (HIS), which encompass vital registration, epidemiological surveillance,
and health service data systems, traditionally
aggregate data to provide key information on morbidity, mortality, and early warning and response.
However, as health systems break down during
conflicts, information systems similarly deteriorate (Working Group for Mortality Estimation in
Emergencies, 2007). Even before a conflict becomes
violent, information systems may already be underresourced and underdeveloped.
There are numerous examples of the consequences
of poor information gathering during conflicts. In
South Sudan in 1998, a relapsing fever outbreak
35
I N D I R E C T CO N F L I C T D E AT H S
continued for six months due to the lack of an
effective early warning system (Gayer et al., 2007).
A similar lapse occurred in Angola in 2005, where
health authorities were unable to identify a large,
deadly outbreak of Marburg haemorrhagic fever
in its early stages due to the reduced ability to
detect the disease (Ndayimirije and Kindhauser,
2005; Guha-Sapir and Le Polain de Waroux, n.d.).
Without working information systems, standard
practices for verifying causes of death are useless. Objective indicators that are normally used
(including death certificates and hospital records)
are frequently missing or inaccessible (Checchi
and Roberts, 2005).
One means of validating non-violent causes of
death during conflicts would be verbal autopsy
techniques. These interview-based protocols have
been developed for community workers in lowresource contexts to obtain information about a
single cause of death (Setel et al., 2006). However,
the length of time required for interviews and the
intensiveness of training impede their use in conflict situations, and greater research into their use
in conflict settings is needed (Utzinger and Weiss,
2007; Working Group for Mortality Estimation in
Emergencies, 2007).
Second, attributing indirect deaths to the impacts
of conflict remains difficult (Checchi and Roberts,
2005). Loss of livelihood, poor diets, lack of food,
displacement, poor sanitation, and countless
other factors are often treated as the underlying
determinants of mortality within a conflict. However, some of these deaths would ‘normally’ occur
under the adverse environmental and economic
conditions, such as drought and poor diet, that
prevail in most developing countries where armed
conflicts occur. While seemingly distant conflict
factors may still have an impact on deaths due to
disease and malnutrition, attributing these conditions to the conflict remains difficult.
1
2
3
Third, and perhaps even more daunting, there is
no straightforward method for determining baseline mortality rates in order to assess the severity
of a conflict (and calculate excess mortality) in
areas where for decades there have been no
public services and little accurate data collection
(Guha-Sapir and van Panhuis, 2004; Utzinger and
Weiss, 2007). Currently, there is no consensus
among researchers on how to derive and compare
baseline mortality rates.
In several conflict areas, such as the DRC and Sierra
Leone, there has been poor coverage by vital
registration for decades. There is therefore little
accurate data that can be used to estimate the
demographic profile of a population. In addition,
it is difficult to designate a point in time at which
Photo ! A child is treated
at hospital during an outbreak of cholera, which is
believed to be present in
the Yei River, Sudan.
© Sean Sutton/MAG/
Panos Pictures
4
5
6
7
36
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
to compare countries that exist in a cycle of
chronic conflict and/or emergency. For example,
as Somalia has been war-torn since the early
1980s, it may not be useful to compare current
mortality rates with the out-of-date mortality
baseline statistics for the country, which are
affected by normal demographic factors. There
are, however, currently initiatives to collect
routine demographic and mortality data in some
areas affected by conflict (e.g. the Bandim Health
Project in Guinea-Bissau) (Nielsen et al., 2006).
Photo " These young
children live in a wrecked
armoured personnel
carrier left over from the
civil war, Somalia, 1992.
© Paul Lowe/Panos
Pictures
Notwithstanding the data collection challenges,
the most widely used datasets that include baseline statistics for most countries are collected by
the United Nations Population Division and often
referenced in UNICEF’s annual State of the World’s
Children report. This data is derived from the last
census and is therefore limited by the quality of
data collection and time of collection. Mortality
rates are also compared with UNICEF’s regional
baseline rates rather than those of single countries.
This approach is useful where no country-level or
sub-national-level baseline data exists, and has
been recommended by the Sphere Project (Sphere
Project, 2004).
An important conclusion is that in some places
the ‘normal’ peacetime baseline mortality rate may
be extremely high. The baseline mortality rate
may thus not be an ideal or acceptable benchmark
for the health of the population of concern (GuhaSapir and van Panhuis, 2004).
Methods for quantifying indirect
conflict deaths
There are three main approaches to quantifying
indirect deaths: retrospective mortality surveys,
prospective surveillance, and the analysis of
multiple data sources. 6 These methods are best
used together as ‘building blocks’ to derive the
best estimates of mortality in a conflict situation
(see Table 2.1).
A retrospective mortality survey (RMS) is used to
determine past mortality rates in situations where
the direct collection of mortality data was or is
not possible. An RMS collects mortality information for a previous period from a representative
sample of a population. Surveyors administer a
standard questionnaire to households to collect
information on deaths. The advantage of an RMS
is the rapid assessment of mortality in areas
where prospective surveillance does not exist.
However, RMSs are problematic in capturing the
true medical causes of death because the information collected cannot be independently verified.
It is also difficult to establish whether deaths
occurred due to violent or non-violent causes.
Logistical problems or security risks make RMSs
challenging to implement, especially since the data
generated is politically sensitive. Nevertheless,
RMSs remain a useful tool in conflict situations
with little or no previous mortality information, and
Table 2.1 Comparison of methods for measuring excess mortality
Method
Appropriate setting
Advantage
Disadvantage
Retrospective mortality survey
!"During conflict
!"Useful for rapid assessment
!"May be difficult to carry out due
!"Post-conflict
where prospective surveillance
is not in place
!"Does not require population
denominator
!"Practical for use in disorganized
settlements
to logistical needs and
insecurity
!"Recall bias, response bias,
survivor bias
!"Measures past death, so not in
real time
!"Statistical analysis is relatively
complicated
!"During conflict
!"Occurs in real time and has
!"National information systems
!"Post-conflict
strong operational usage
!"Relatively simple analytical
procedures involved
to track health and mortality are
usually weak in conflict settings
so an ad hoc system is required
!"Requires regular updating of
data and population size to be
useful
!"Possible only in camps and
stable populations
!"Mainly post-conflict (as it is
!"Used to assess the quality and
!"Dependent on the quality and
strengths of multiple sources
of data
!"Statistical techniques are
available to employ the best
aspects of data sources (i.e.
Multiple Systems Estimation)
type of primary data sources
(i.e. data source such as a
graveyard database may not
have clear information on type
of death)
!"Dependent on the availability
and timeliness of primary data
sources
Prospective surveillance
Analysis of multiple data sources
dependent on other primary
data sources)
Sources: Checchi and Roberts (2005); Guha-Sapir, Degomme, and Altare (2007)
I N D I R E C T CO N F L I C T D E AT H S
37
1
2
3
4
5
6
RMS methods have been standardized through an
inter-agency humanitarian initiative (Working Group
for Mortality Estimation in Emergencies, 2007).
The prospective surveillance of mortality through
a health information system (HIS) is a better
method to document and verify mortality in stable
environments. By targeting health facilities and
death registries, these systems can provide accurate and timely mortality data. However, HISs are
almost universally weak in conflict-affected areas,
and between two-thirds and three-quarters of the
world’s population are not covered by any type
of health surveillance (Fottrell, 2008, p. 4). But
mortality detection can be integrated through ad
hoc surveillance within humanitarian operations
and in refugee camps even though it may be prone
to under-reporting due to the lack of accurate
demographic information (Thieren, 2005; CRED,
2006). The problem of verification and reporting
of death in conflict situations is symptomatic of
the general lack of standard sources on the causes
of deaths.
The analysis of multiple data sources permits the
reconstruction of mortality profiles using sources
7
38
Box 2.2 Sexual violence in armed conflict
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
During armed conflict, women and girls are specifically targeted by sexual violence that occurs
in homes, detention places, military sites, and
camps for refugees and displaced persons. Brutal
rapes, sexual assaults, sexual slavery, and mutilation are systematically used in many armed
conflicts. Survivors suffer grave psychological
trauma, permanent physical injury, unwanted
pregnancy, and long-term health risks including
HIV/AIDS and serious complications in reproductive health.
Data on the scope and magnitude of sexual violence, especially rape, in armed conflicts worldwide is scarce, making it impossible to estimate
its extent. In addition to the usual obstacles to
data collection, sexual violence is surrounded
by social taboos and stigmatization, resulting in
a lack of (and under-) reporting even in peacetime. Table 2.2 illustrates the wide range and
imprecision of estimated incidents of rape in
selected armed conflicts.
Photo ! A 13-year-old
and her three-month-old
baby, born as a result of
her rape, in hospital in
Goma, DRC.
© Robin Hammond/
Panos Pictures
‘Sometimes, when they said that you were
the most beautiful woman, it was a disaster!
They put you in the middle of everyone, on a
cross, with your head down and your legs
spread and they raped you in that position.
And the others had to cheer them on and
dance around you [. . .] I was everybody’s
woman and nobody’s woman. Whoever wanted
to satisfy his sexual needs came on us.
Sometimes they would shout “Food! Food!”
We thought maybe they were bringing us
food. But unfortunately, it was not food. It
was us, the women, who were their “food”.’
— Onarata Kazende, 55 years old, DRC (BBC, 2008)
A clear example of widespread sexual violence
is in the DRC. Victims report that all armed groups,
including state security forces, are responsible
for rapes and high levels of sexual violence. The
majority of the perpetrators remain unpunished,
however, especially when belonging to the state
security forces.
Rape is becoming more violent and more common
in the DRC. It seems that male relatives are forced
at gunpoint by militias or paid security forces to
rape their mothers, sisters, or daughters. Often
women are shot or stabbed in their genital organs
after being raped (Wakabi, 2008). According to
the UN special rapporteur on violence against
women, 31,500 rapes were recorded in South
Kivu province between 2005 and the first half of
2007, with probably many more going unreported.
The Provincial Synergy for South Kivu estimates
that 22 per cent of rape victims are HIV-positive
due to the incidents (HRC, 2008).
Civil society organizations (CSOs) have expressed
widespread concern for the pervasive nature ##
Findings from surveys in different countries and
among refugee and internally displaced persons
(IDPs) camps show varying prevalence of sexual
violence. While in some camps women and girls
are especially at risk when they leave the camp
to collect wood and fetch water, in others the
majority of assaults happen within the camp. A
2006 UNHCR report on sexual and gender-based
violence notes than more than 20 of 104 camps
that supplied data reported rates of sexual and
gender-based violence of between 250 and 500
per 100,000 persons, with approximately ten
camps reporting rates of between 500 and 1,000
per 100,000, and 20 camps reporting rates greater
than 1,000 per 100,000 (UNHCR, 2007, p. 65).
This means that 50 per cent of camps reporting
data had rates of sexual and gender-based violence greater than 250 per 100,000.
The 2008 United Nations Security Council (UNSC)
Resolution 1820 on sexual violence, adopted
unanimously on 19 June, classifies rape and
other forms of sexual violence as a weapon of
war. It can constitute a war crime, a crime
against humanity, or a constitutive act with
respect to genocide. Resolution 1820 stresses
that perpetrators of crimes of sexual violence
should be excluded from amnesty provisions
and should be prosecuted (UNSC, 2008). The
responsibility for perpetrators of sexual violence
is now collective. Some NGOs are concerned
that the new resolution on sexual violence does
not strengthen the provisions of UNSC Resolution
1325 on Women, Peace and Security and that it
does not offer clear measures to end impunity
for acts of sexual violence.
Armed conflict
Estimated number of incidents
Sierra Leone (1991–2001)
More than 215,000
Rwanda (1994)
250,000–500,000
Bosnia and Herzegovina (1992–95)
14,000–50,000
Liberia (1989–2003)
Approximately 500,0009
Kosovo (1998–99)
23,200–45,60010
Sources: UNICEF (2005, p. 4); UNECOSOC, Commission on Human Rights (1996, §16); OCHA/IRIN
(2005, p. 178); Refugees International (2004); AI (2004, p.10); Hynes and Cardozo (2000)
of mortality statistics collected before, during, and
after conflict. Demographers and statisticians
offer several approaches based on the availability of data sources and the derivation of the best
estimates. Multiple systems estimation (MSE)
techniques can, for example, assess databases
of human rights violations, a census of public
graves, and an RMS to estimate mortality. The
clear advantage of such an analysis is the assessment of quality among different data sources to
derive a best estimate. However, the approach
could also aggregate potentially flawed sources
of secondary data, which may result in inaccurate results.
39
I N D I R E C T CO N F L I C T D E AT H S
Table 2.2 Estimated incidents of rape in selected armed conflicts
of sexual violence in the country.7 The International
Rescue Committee reported assistance to more
than 40,000 rape survivors in DRC since 2003. A
United Nations Populations Fund survey among
half of the health centres in the country showed
that 50,000 rape cases were reported in 2007
(Wakabi, 2008). The ceasefire of January 2008 did
not stop the incidence of sexual violence. In North
Kivu province, 880 cases of rape were documented
by NGOs and UN agencies in April 2008 alone. 8
1
2
3
4
5
6
Direct versus indirect deaths in
recent conflicts
Given the challenges to arriving at an assessment
of the burden of indirect deaths in armed conflict,
it is difficult to provide a precise assessment of
the annual burden of indirect conflict deaths. Based
on the figure of 208,300 conflict deaths between
2004 and 2007 (an average of around 52,000 per
year) presented in the chapter on conflict deaths,
it is possible to provide some indication of the
likely indirect burden in recent years.
7
periods. Several points should be noted from this
table. First, in all but one case (Kosovo, 1998–99),
indirect deaths were greater than direct deaths,
and usually by a wide margin. The Kosovo case
The first step is to examine the available evidence
on indirect versus direct deaths in recent conflicts.
Table 2.3 below does this for 13 conflicts, from
different continents and covering different time
Table 2.3 Direct vs. indirect deaths in several recent armed conflicts
Indirect deaths as percentage
of total excess deaths
Kosovo, 1998–99a
of
A R M E D V I O L E N CE
40
b
G LO B A L B U R D E N
Iraq, 2003–07
c
Northern Uganda, 2005
Democratic Republic of the
Congo, 1998–2002d
Congo-Brazzaville, Pool
Region, 2003e
Burundi, 1993–2003f
Ratio of indirect to
direct deaths
Conflict mortality rate (per
100,000 per year, average)
Total conflict deaths
(direct and indirect)
011
–
334
12,000
63
3.0
246
347,000
85
5.6
476
26,000
90+
9.0
1,316
3,300,000
83
4.8
n/a
n/a
78
3.5
500
300,000
g
94
15.7
1,101
462,000
69
2.3
730
142,000
South Sudan, 1999–2005i
90+
9.0
1,178
427,000
89
8.1
676
1,500,000
86
6.1
889
175,000
82
4.6
638
103,000
77
3.3
784
144,500
Sierra Leone, 1991–2002
h
Darfur, Sudan, 2003–05
Angola, 1975–2002j
Liberia, 1989–96k
l
East Timor, 1974–99
Iraq, 1991 war
m
Sources:
a
b
Based on Spiegel and Salama (2000, p. 2204). Detailed calculation in Small Arms Survey (2005, p. 259).
There is considerably uncertainty around both direct and indirect conflict deaths in Iraq. Figures used here (87,185 direct and 259,000 indirect conflict deaths) should
be considered conservative; it is possible that up to 150,000 direct deaths and as many as 326,000 indirect deaths have occurred. This would yield a total of 476,000
conflict deaths, a conflict mortality rate of 337 per 100,000. Based on data in Box 2.5.
c
Based on WHO (2005). Total deaths is + or - 4,000; UBOS (2006).
d
Based on IRC (2000, pp. 1, 3); IRC (2003b, pp. 5–6); IRC (2001a. pp. 6, 8–11); IRC (2004a, pp. 11, 13, and 17); Coghlan et al. (2008), p. 13. Total death figure from Coghlan
e
Based on a survey in the Pool region (IRC, 2004b, p. 7). Details in Small Arms Survey (2005, p. 259).
f
Indirect death ratios for 2002–03, based on IRC (2002a; 2002b; 2002c; 2003a). Details in Small Arms Survey (2005, pp. 258–59). Total deaths are for the entire conflict
et al. (2006).
(1993–2003) from IRC (n.d).
g
See Box 2.4.
h
Based on Guha-Sapir and Degomme (2005a; 2005b). This is a meta-analysis of more than 24 different surveys in the region.
i
See Box 2.3.
j
Based on Lacina and Gleditsch (2005, p. 159). The 11 per cent ‘battle deaths’ estimate appears to include both civilian and combatant violent deaths.
k
Based on Lacina and Gleditsch (2005, p. 159). The 12–16 per cent ‘battle deaths’ estimate appears to include both civilian and combatant violent deaths. Total deaths are +
l
Based on Silva and Ball (2006). Death total is + or - 12,000.
m
Based on Daponte (2008, p. 59).
or - 25,000.
41
I N D I R E C T CO N F L I C T D E AT H S
can be explained by the relatively well-developed
pre-war basic health and service infrastructure,
the rapid and effective humanitarian response to
the population displacement that occurred during
the fighting, and the relatively short and intense
nature of the armed conflict.
Second, the conflict mortality rates that these
figures suggest are very high, ranging from 334
to 1,316 per 100,000 per year. These are considerably greater than the highest direct conflict and
non-conflict death rate, underlining that the risk
of dying in warfare can be much higher if accounting for indirect conflict deaths.
Although there is a wide variation in the relationship, in only two cases other than Kosovo did the
ratio fall below three indirect deaths for every
direct death. Both the Iraq 2003–07 and Darfur,
Sudan, 2003–05 cases have been the subject of
numerous analyses. The low ratio in the Iraqi
case is partly due to the intensity of the violence
and the relatively well-developed infrastructure
(compared to other conflict zones), and is discussed
in Box 2.5. The lower ratio for Darfur is partly due
to the fact that studies focused on conflict-affected
populations, groups among which the violent
deaths were concentrated. It is based on an estimated 142,000 total deaths in 2003–05, of which
43,935 are estimated to be violence-related (GuhaSapir and Degomme, 2005a; 2005b). Whatever
the ratios, the conflicts in Iraq and Darfur exacted
a huge human toll.
Three main factors explain the differences in proportion between direct and indirect conflict deaths:
the quality of pre-existing health care systems and
patterns of disease; the speed and extent of the
humanitarian response; and the intensity and duration of battle. Relatively healthy populations with
prior access to good health care are much less vulnerable to rapid increases in mortality, whereas
vulnerable and weak populations quickly fall victim.
A vigorous humanitarian response—food, water,
protection, shelter, and basic health care—and
good access to affected or displaced populations
can also reduce mortality. Conventional battles
between regular armed forces in limited areas—
which characterizes few contemporary wars—also
reduces the burden of indirect deaths on the civilian population, and can (if fighting is intense) also
increase the proportion of battle deaths. These
three factors taken together can help explain the
relatively low ratio for the 1991 Iraq war, compared
with the conflicts in Africa.
The persistence of high levels of indirect conflict
death after the end of the violent phase of a conflict is an important problem for policy-makers
concerned with humanitarian aid and reconstruction. It is often far more time-consuming to restore
health infrastructure, services, and security than
to negotiate a ceasefire, or even demobilize combatants. States that have been weakened by longterm violent conflicts generally lack the resources
and capacity to address these challenges, and
Photo ! A line forms
outside an information
1
tent in Stenkovec 2 camp,
Macedonia, for refugees
2
fleeing Kosovo.
© Andy Johnstone/
3
Panos Pictures
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
42
progress is not made until long after a conflict has
ended. The disruption and increased mortality that
persist at the end of a violent conflict need to be
taken seriously into account in the planning of longterm reconstruction and development programmes.
Without detailed data on mortality for all the
contemporary conflicts discussed in the preceding chapter (DIRECT CONFLICT DEATH), it is not
possible to give a precise estimate of the indirect
burden of armed violence. But an order of magnitude can be offered for the purposes of comparison with other aspects of the global burden of
armed violence, based on the following data and
assumptions:
The direct death burden in conflicts for 2004–
07 from incident reporting was 208,300, or
about 52,000 per year.12 These reported
deaths clearly undercount the actual total of
direct conflict deaths, although the degree of
undercounting varies by conflict.
A previous study of undercounting in specific
conflicts demonstrated that it could be between
two and four times the level captured in incident reports (Obermeyer et al., 2008; Small
Arms Survey, 2005, p. 230). In the DRC alone,
an estimated average of 51,000 people have
died violently per year since 1998, although
the annual totals have been lower since 2002.
A conservative ratio of 4:1 indirect to direct
deaths would mean that the burden of indirect
deaths for an average year between 2004 and
2007 would be at least 200,000 and probably
higher.13
The total number of indirect deaths would vary
considerably from year to year, depending on the
number and intensity of conflicts, the nature of
the fighting, the provision of humanitarian assistance, and the condition of the affected population.
In order to avoid the impression of excessive pre-
cision in what is simply an order of magnitude,
this report concludes that on average, at least
200,000 persons have died each year as an indirect result of conflict since 2004.
The pages following the end of this chapter provide detailed discussions of three long wars—in
Iraq, Sierra Leone, and South Sudan—to provide
concrete illustrations of how field-based surveys
can provide a more adequate picture of the burden
of violence in armed conflicts.
Conclusion
Quantifying excess mortality and indirect deaths
is a difficult task. But the expert consensus is
that in almost all contemporary armed conflicts,
indirect deaths are often more numerous than
mortality arising from violence. Non-violent deaths
that can be directly linked to conflict should count
as part of the burden of armed violence, since
from a human perspective it matters little if a
parent or child dies from a bullet or from dysentery soon after an armed clash.
Several scientifically rigorous methods have been
developed and improved in recent years, by epidemiologists, demographers, and statisticians, to
provide reliable estimates. These methods continue
to be refined and standardized, as evidenced by the
SMART (Standarized Monitoring and Assessment
of Relief and Transitions) initiative and the general increase in the quality of data collection and
analysis in humanitarian research.
Continued innovation in measuring indirect mortality in conflicts will be crucial to understand
the true human impact of mortality in conflicts,
to help set priority public health goals for the
prevention of disease and malnutrition, and to
provide the evidence base to hold perpetrators
of violent acts against innocent populations
legally accountable.
Box 2.3 A very dark number: direct and indirect
mortality in southern Sudan, 1999–2005
Since Sudanese independence in 1956, civil wars have raged in
the south, with a lull between 1972 and 1983. The period 1983–
2005 was the longest and, in all likelihood, the deadliest spell. In
January 2005 the Comprehensive Peace Agreement formally ended
the fighting, and relative calm has since returned.
Large-scale human rights violations were committed during the
1983–2005 civil war, in particular against the civilian population
of southern Sudan. Massive population movements took place;
famines were chronic. Food aid to the affected population was in
numerous instances denied or purposely obstructed.
An estimated 427,337 people died (excess mortality) during the
second phase of the armed civil conflict in the period 1999–2005
in the three states of southern Sudan: 339,342 in Upper Nile,
58,663 in Bahr el Ghazal, and 29,332 in Equatoria (these three
regions have become the ten states of South Sudan).
Of these excess deaths, the percentage of direct (violent) deaths
is only 0.3, although it appears there was relatively higher direct
mortality in Bahr el Ghazal (one per cent) during this period. The
total number of direct deaths in southern Sudan between 1999
and 2005 was 1,381 (594 in Bahr el Ghazal, 520 in Upper Nile, and
167 in Equatoria). This is in addition to the previously estimated
1.7 million victims between 1983 and 1998 (Burr, 1993; 1998).
How are these figures arrived at? Direct mortality is estimated
from data on killings in all documents that could be found on the
Internet, through fellow researchers, and in libraries. Documents
were selected if they provided independent information on mortality during the conflict. Incidents and casualties were collected
in one file, and identified by location and date, to prevent double
counting. Verbal descriptions (‘many’, ‘numerous’, ‘few’) were
quantified (see Bijleveld, Degomme, and Mehlbaum, 2008).
To estimate total excess mortality, the crude (CMR) and under-five
(U5MR) mortality rates in all the surveys in the CE-DAT database
have been plotted against the years studied and the trends in
mortality have been investigated (CRED, 2008).14 Any outliers are
removed in order to arrive at a conservative estimate, and mortality
rates are applied to time frames and regions to develop a differentiated estimate.
For estimating total mortality, 78 surveys that gave either a CMR
or U5MR were found. Only 37 of these gave a recall period, but as
Virtually all surveys that reported CMRs and U5MRs above emergency level were conducted between June 2001 and August 2003
in the Upper Nile and Jonglei states. These rates are problematic,
however, as they are excessively high and would have to have
been reflected in massive starvation, which was not reported
during those years. In addition, the surveys were methodologically different from subsequent measurements. The median of
the CMR from the surveys (2.1) was used as a more conservative
estimate. With these elevated rates excluded, the average CMR
was 0.58.
For the Bahr el Ghazal and Equatoria regions the average nonelevated CMR of 0.58 was used for the entire period. For 1999,
the 2000 mortality rates were assumed to hold. For the Upper
Nile region the 0.58 CMR was used for 1999, 2000, and 2004. As
the surveys show elevated mortality for Upper Nile and Jonglei
from only mid-2001 and onwards, 2.1 was used for 2002 and
2003 for the entire Upper Nile region.
To determine excess mortality, expected mortality was subtracted
and set conservatively at 0.5. Applying these mortality rates to
estimated population sizes, the total excess mortality is 427,337
(339,342 for Upper Nile, 58,663 for Bahr el Ghazal, and 29,332 for
Equatoria).
These estimates are dependent on assumptions, and, in the case
of direct deaths, in part on a quantification of verbal statements
that may be inaccurate. However, even if 90 per cent of all direct
mortality was missed, or if total excess mortality were only 50 per
cent of what is estimated here, almost all excess mortality would
still be indirect, and only a fraction (less than five per cent) the
immediate consequence of violence.
By far the largest contribution to mortality in southern Sudan in
1999–2005 was indirect deaths. On a more general note, our calculations are on the edge of feasibility, since they have been
made from scarce data and should be used with caution.
I N D I R E C T CO N F L I C T D E AT H S
43
the largest recall period was three months, and as population
estimates for southern Sudan are fairly coarse anyway, all surveys
were used, whether or not they reported a recall period, and to
peg the mortality rate to the time that the survey was administered. Most surveys were conducted by NGOs active in southern
Sudan, both in towns like Aweil and Bentiu and in the rural areas.
No surveys were found for 1999. One outlier with an U5MR of 33
was removed (Ratnayake et al., 2008, p. 16).
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
44
Box 2.4 Direct and indirect mortality in
Sierra Leone, 1991–2002
Figure 2.2 Distribution of killings in
Sierra Leone, 1991–2001
Massive human rights violations took place during the civil war in
Sierra Leone from 1991 to 2002. During almost 11 years of conflict,
many thousands of people were displaced from their homes or fled
the country. As the conflict moved across the country, population
moved in its wake.
800
700
600
500
With infrastructure destroyed and/or facilities looted in most
conflict zones, parts of the population were unable to plant their
crops, and had severely reduced access to health care. In addition
400
300
to being caught up in the fighting, the civilian population was also
200
actively targeted. Among the crimes committed were widespread
100
and systematic sexual violence, sexual slavery, abduction, use of
0
child soldiers, murder, robbery, destruction, amputations, displacement of people, and starvation (PHR, 2002).
1990
1992
1994
1996
1998
2000
Source: Benetech report to the TRC (Conibere et al., 2004)
Different estimates of civilian deaths from these gross human
War-related sexual violence in Sierra Leone
During more than a decade of armed conflict in Sierra Leone
sexual violence and associated abuse against women and
girls was characterized by extreme brutality. As many as
215,000–257,000 women and girls were affected by sexual
rights violations do exist, ranging from 35,000 to 200,000 deaths
(cf. Bijleveld and Hoex, 2008). These estimates are, however,
barely substantiated. Also, it is unclear what part of mortality is
direct (violent) and what part is indirect (consequence of disease,
starvation, exhaustion, injuries, etc.).
violence (PHR, 2002, p. 4). According to the Truth and Recon-
To estimate direct mortality, the distribution of direct deaths as
ciliation Commission (TRC) all armed factions, in particular
reported by the Sierra Leonean Truth and Reconciliation Commis-
the Revolutionary United Front (RUF) and the Armed Forces
sion (TRC) is used. Next we assume that all killings in Sierra Leone
Revolutionary Council (AFRC), systematically and deliberately
in the period under investigation did follow the trend as given by
raped women and girls (TRC, 2005, p. 162).
the TRC report. Finally, the level of this trend curve was set to match
In addition to rape, other human rights violations, such as
the available data (mainly from UN and Amnesty International
abductions, beatings, killings, torture, forced labour, firearms
sources) on direct killings from 1996–99.
and other injuries, and amputations were committed on a reg-
Figure 2.2 describes the distribution over time of the total esti-
ular basis. One survey found that 94 per cent of 991 randomly
mated direct mortality of 26,704.
surveyed households reported at least one of the above listed
abuses during the course of the war. Of those who experienced
This number should be regarded as conservative when compared
sexual violence, 89 per cent were raped, 33 per cent were gang
with other sources. TRC data is an underestimate: for instance, in
raped, 33 per cent were abducted, 14 per cent were molested,
January 1999 around 5,000 persons were killed in Freetown, while
15 per cent experienced sexual slavery, and 9 per cent were
the total TRC number adds up to approximately 4,500.
forced into marriage. The majority of incidents occurred between
1997 and 1999 (PHR, 2002, pp. 2–4).
Similarly, the Amnesty International deaths are also an underestimate, since they cover only six months in 1996, only five months
Violence against the civilian population and especially against
in 1997, and eight months in 1998; as well, not all districts were
women and girls perpetrated by combatants in Sierra Leone was
covered and some periods and areas were too dangerous to
widespread, representing a significant long-term health burden.
survey.
##
45
Photo Victims of violent rebels: a family in the
I N D I R E C T CO N F L I C T D E AT H S
Murray Town amputee camp, Freetown, Sierra Leone.
© Stuart Freedman/Panos Pictures
1
2
3
4
5
Total excess mortality was calculated using a hypothetical population size for 2002 and assuming uninterrupted non-conflict
population growth from the 1990 population for Sierra Leone of
4,087,000. Using a conservative growth rate of 1.96 and correcting for migration leads to a hypothetical population size by 2002
of 4,979,321.
The actual population size in 2002 was estimated by calculating
back from the 2004 census, and again correcting for migration, to
estimate actual Sierra Leonean population size in early 2002 at
approximately 4,517,330.
Total war-related mortality, estimated as the difference between
the hypothesized and the actual population, is then approximately
461,990—meaning that an estimated 460,000 Sierra Leoneans
15
lost their lives as a result of the conflict between 1991 and 2002.
Approximately 26,704 of these deaths—or six per cent—were
most probably directly due to violence. Roughly 94 per cent of the
total excess mortality was thus indirect, mostly attributable to
causes other than violence.
These estimates all depend on assumptions. It may have been
that the Sierra Leonean population would, without the conflict,
not have grown at the assumed rate, but at a much slower rate.
In that case, the percentage of direct deaths becomes higher.
However, even if the growth rate were set at the lowest rate ever
measured (1.4 per cent, which is unrealistic and too low), still
around a quarter of total excess mortality is direct, and threequarters is indirect. By far the greater part of the mortality in the
Sierra Leone war was indirect.
6
7
Violent (direct) deaths
A R M E D V I O L E N CE
Estimates of violent deaths (both direct and indirect) in Iraq since 2003 have
generated extreme controversy, in part because of the wide variation in
the number of deaths, in part because of lack of clarity regarding what
different techniques measure or count. Sources may focus on combatants
(battle deaths), civilians who die violently, or on changes in overall mortality rates since 2003.16 But as shown below, it is reasonable to conclude
that armed violence has claimed more than 200,000—and perhaps up to
400,000—lives since 2003.
Two main techniques have been used to collect
data and estimate levels of violent deaths: incident
reporting and mortality data from surveys. The
Iraq case is one of the few in which a comparison
between different methods can be made.
The situation in Iraq also shows how difficult it is to draw a line between
‘conflict’ and ‘post-conflict’ violence, or between conflict and ‘non-conflict’
or criminal violence. In many cases, the identity or motive of the perpetrator
of violence is unknown, making it difficult to establish why particular killings
occur. Furthermore, the ebb and flow of armed violence since 2003 call into
question the very notion that violent deaths decrease after a conflict has
##
been declared over.
G LO B A L B U R D E N
Box 2.5 Armed violence in Iraq: what’s in a number?
of
46
Table 2.4 Violent deaths reported in Iraq, 2003–07
Database
2003
2004
2005
2006
2007
Total*
Cross-country databases
IISS
10,000
15,000
12,900
23,000
31,560
92,460
Ploughshares
12,500
6,500
10,500
35,000
n/a
64,500
PITF
100–
1,000
1,000–
5,000
10,000+
10,000+
n/a
23,500**
PRIO
10,000
9,500
8,100
n/a
n/a
27,600
SIPRI
18,600
n/a
5,500
n/a
n/a
24,100
UCDP state
8,313
1,987
2,299
3,537
n/a
16,136
UCDP state and
non-state
8,494
2,304
3,418
3,537
n/a
17,753
11,672
9,843
13,816
26,659
23,427
85,417
Iraq Coalition
Casualty Count
598
1,093
3,542
3,042
2,833
11,108
GBAV estimate
10,919
9,803
15,788
26,910
23,765
87,185
National databases
Iraq Body Count
* For available years only.
** This includes averages for the ranges for 2003 and 2004 and the lowest figures for 2005 and 2006.
Source from this chapter: Iraq Coalition Casualty Count (2008).
Sources from Chapter 1 (DIRECT CONFLICT DEATH): IISS (2008); Iraq Body Count (2008); PITF (2006);
PRIO (2008); Project Ploughshares (2007); SIPRI (2007); UCDP (2006a)
Table 2.4 shows data from incident-based databases, both from cross-country databases and
country-based studies for Iraq. The differences
are mostly due to different techniques and (more
importantly) different rules for counting. The
UCDP dataset, for example, measures only battlerelated deaths; Iraqi Body Count measures civilian casualties including morgue reports; and the
Iraqi Coalition Casualty Count measures casualties among combatants and civilian contractors.
The last row in Table 2.4 provides the consolidated
estimate for 2003–07 used in this report. It is
based on combining figures for country-based
studies and accounting for the different counting
methodologies used. This report thus estimates
that since the start of the war at least 87,000 direct
conflict deaths have occurred, of which only 15
per cent are identified as state or coalition combatants. Translated into mortality rates, this would
equal approximately 65 violent deaths per 100,000
people per year—a high rate. All the sources used
note that undercounting of the real burden is likely
because of difficulties encountered in gathering
reliable information on all violent deaths.
The GBAV estimate is calculated by pooling a
variety of incident-based datasets. In order to
control for overlap across sources, this technique
includes civilian data from Iraq Body Count after
discounting morgue data, which cannot be tied
to conflict actions with any certainty. While it
also excludes accidents and civilian data, the
estimate includes figures for military and contractor casualties as well as Iraqi armed forces
generated by the Iraq Coalition Casualties Count.
The GBAV estimates track the perceived intensity
of the war over time and are similar to the trends
documented in most other data sources.17
##
Period covered
Violent deaths
Roberts et al.
(2004)
March 2003–
Sept. 2004
(18 months)
14,700–49,980*
Burnham et al.
(2006)
March 2003–
July 2006
(40 months)
601,027
(426,369–
793,663)
Alkhuzai et al.
(2008)
March 2003–
June 2006 (40
months)
151,000
(104,000–
223,000)
* This estimate is based on the percentage of the recorded deaths
that were violent deaths (15 per cent if the deaths from Falluja are
excluded; 51 per cent if they are included), multiplied by the midpoint estimate of 98,000 excess deaths. It should be noted that
there is a wide confidence interval for the estimate of 98,000
deaths, so these figures should be taken as indicative only.
Table 2.6 Overview of indirect death estimates
from three mortality surveys
Period covered
Excess deaths
estimate
March 2003–
Sept. 2004
(18 months)
83,300*
Burnham et al.
(2006)
March 2003–
July 2006
(40 months)
53,938**
Alkhuzai et al.
(2008)
March 2003–
June 2006
(40 months)
259,000***
Roberts et al.
(2004)
* The figure is the total of 98,000 excess deaths minus the violent
deaths (14,700), excluding violent deaths recorded in the Falluja
cluster, which was itself excluded from the estimates given for
excess deaths.
** The figure is low because of the very high rate of violent deaths
reported (see Table 2.5).
*** The range for this estimate is 213,000–327,000. Figure based
on WHO calculations from the original dataset. Mills and Burkle
(2008) suggest a higher figure of 282,000 non-violent indirect
deaths.
47
Several recent epidemiological studies provide further information on the
scale and scope of direct and indirect conflict deaths. Two studies were
published in the medical journal The Lancet in 2004 and 2006 (Roberts et
al., 2004; Burnham et al., 2006) and a third in the New England Journal of
Medicine in 2008 (Alkhuzai, 2008), all based on sampling survey techniques
used to calculate an estimate for the entire population. At least one of these
estimates stirred a controversy by revealing an extremely high level of violent deaths (conflict and non-conflict), much larger than the one estimated
by incident reporting or other studies. The results of all three epidemiological
studies for violent deaths are summarized in Table 2.5.
At first glance, such a wide range seems to imply that the exact number of
deaths due to violence remains unknown. But the quality and reliability of
these surveys is not equal. The most recent study (2008) surveyed 9,345
households, and was conducted under the auspices of the World Health
Organization. The previous two studies, both conducted under difficult
circumstances and with limited resources, surveyed 990 (2004) and 1,849
(2006) households. The gain in precision with greater numbers of households surveyed in the 2008 study is obvious, and some concerns have been
raised about the accuracy of the estimates in the 2006 study.
The estimate of 151,000 violent deaths for the 40-month period from March
2003 to June 2006—an average of 45,300 deaths per year (Alkhuzai, 2008)—
is approximately three times higher than the equivalent period in the incident
reporting data. The figure can in part be explained by the under-reporting
that characterizes all incident reporting systems, especially where media
coverage is patchy and conflict is intense. It also underscores the main
message of the conflict deaths chapter—that the figures of 52,000 conflict
deaths per year for all conflicts in recent years, based on incident reporting,
is certainly an undercount of the burden of direct deaths (CONFLICT DEATHS).
Indirect deaths
The Iraqi conflict also potentially produced indirect deaths—persons who
have died from such preventable causes as disease and malnutrition, due
to loss of access to basic health care, water and sanitation, or other basic
services. The three mortality surveys discussed above estimate both violent and non-violent mortality; consequently, they can also estimate the
burden of indirect conflict deaths in Iraq. Table 2.6 presents an overview
of the results of the non-violent mortality rates.
The figures in Table 2.6 provide a very wide range of estimates: between
1,348 and 3,900 per month. Nevertheless, based on these figures, which
calculate the difference between the post-invasion and pre-invasion mortality rates in Iraq, one can arrive at an estimate of indirect deaths from March
2003 to March 2008 (five years) for the Iraq conflict: more than 150,000
indirect deaths, with a wide possible range between 80,000 and 234,000.
These figures illustrate that the estimate for excess indirect mortality in
Iraq remains as imprecise as the estimate for direct deaths.
Regardless of the final figure, the total number of direct and indirect victims
of the Iraq war since 2003 is very large, almost certainly exceeding 200,000
and perhaps as high as 400,000.
I N D I R E C T CO N F L I C T D E AT H S
Table 2.5 Violent death estimates from three
mortality surveys
1
2
3
4
5
6
7
Abbreviations
AFRC
Armed Forces Revolutionary Council
CE-DAT
Complex Emergency Database
Crude mortality rate
CSO
Civil society organization
DRC
Democratic Republic of the Congo
HIS
Health information system
IDP
Internally displaced person
MSE
Multiple systems estimation
of
CMR
RMS
Retrospective mortality survey
G LO B A L B U R D E N
A R M E D V I O L E N CE
48
RUF
Revolutionary United Front
TRC
Truth and Reconciliation Commission
UNSC
United Nations Security Council
U5DR
Under-5 death rate
U5MR
Under-5 mortality rate
which was commissioned for the Global Burden of Armed
Violence report.
7
Letter to the UNSC from 71 Congolese organizations representing the women of DRC. 12 June 2008.
8
Letter to the UNSC from 71 Congolese organizations representing the women of DRC. 12 June 2008.
9
Estimated 40 per cent of the female population, averaged
over 15 years.
10
Population-based survey of 1,358 Kosovo Albanians
(who had been internally displaced or who had recently
returned to Kosovo) conducted in August and September
1999 by the Centres for Disease Control and Prevention
(CDC). Extrapolation to an estimated 800,000 Kosovo
Albanian women over 15 years of age (Hynes and
Cardozo, 2000).
11
In Kosovo the number of violent deaths recorded in the
sample population actually exceeded the number of
calculated excess deaths (both direct and indirect) in the
conflict. This may be a statistical artefact due to the small
numbers used to calculate ratios, but it also reflects the
fact that intentional injury was a cause of death in Kosovo
even before the most intense phase of the conflict. Some
direct deaths may therefore have been included in the
number of expected deaths for the population.
12
This figure includes civilian victims of violence in conflict;
the number of combatant deaths is lower.
13
A qualitative assessment of the most important ongoing
conflicts would support this assumption of a 4:1 indirect
to direct death ratio as a minimum average.
14
The Complex Emergency Database (CE-DAT) is an online,
publicly accessible, searchable database of global
humanitarian emergencies. It contains more than 1,800
surveys previously collected in complex emergencies
occurring since the year 2000. <http://www.cedat.be>
Endnotes
1
This chapter draws extensively upon Ratnayake et al.
(2008), which was commissioned for the Global Burden
of Armed Violence report.
2
Of the 2.1 million reported indirect deaths since 2004, only
0.4 per cent—or 8,400—were calculated as violent deaths,
a figure that accords well with the direct conflict death
estimates for the same four years (Coghlan et al., 2008).
3
This ‘reasonable estimate’ is based on the assumed undercounting of combat deaths, and conservative assumptions
about indirect deaths. The figure is explained in more
detail below.
4
The use of alert thresholds is explained further in Checchi
and Roberts (2005, p. 7).
15
It should be stressed that these are a conservative estimates; Bijleveld and Hoex (2008) give a range.
5
Accidents are sometimes grouped under direct deaths as
they specify a grey area where deaths may have indeed
been due to violence.
16
It is impossible to summarize all the relevant contributions
to these debates. For some examples, see Dobbs (2007);
Fischer (2007); Ahuja (2007); and Tapp et al. (2008).
6
For a more detailed account of the methods of quantifying
indirect deaths, see Ratnayake et al. (2008, pp. 6–12)
17
See the online annexe at www.genevadeclaration.org for a
detailed explanation of the methodology.
T
he end of war does not necessarily herald
a return to security. Ceasefires, peace agreements, arms control activities, or even elections—important as they are—do not necessarily
guarantee tangible improvements in the safety—
real or perceived—of individuals and communities.
In fact, many so-called post-conflict theatres presented more direct and indirect threats to civilians
than the armed conflicts that preceded them.
processes of development and democratization
Since many armed conflicts end without a strong
commitment to the peace agreement or ceasefire,
efforts to impose a ‘victors’ justice’ can actually
escalate armed violence (Kreutz, Marsh, and Torre,
2007; Licklider, 1995). Similarly, some armed
groups may be dissatisfied with the terms of the
‘peace’, providing a source for instability (Muggah,
2008; Darby, 2001). Pre-existing networks and
structures associated with the war economy may
remain intact. Post-conflict armed violence may
thus be perpetrated by a fluid constellation of
state agents and armed groups with competing
(and often changing) motivations and interests.
Armed violence that may previously have been
concentrated in specific geographic areas in the
hinterland may shift to new spaces—from war
zones and border areas to urban slums.
mal justice, and post-war displacement and dis-
Post-conflict armed violence is a policy concern,
for two reasons: because it often contains the
‘spoiler’ potential to disrupt a peace process or
contribute to a relapse into war, and in its own
right as a condition that can undermine longer-term
(Chaudhary and Suhrke, 2008).
This chapter focuses on the character and shape
of post-conflict armed violence. Post-war contexts
are as complex and varied as war-affected environments, and several different types of post-
49
A R M E D V I O L E N C E A F T E R WA R
Armed Violence After War:
Categories, Causes, and Consequences
Chapter Three
conflict violence can be distinguished, including
political violence, routine state violence, economic
and crime-related violence, community and inforputes. A number of important patterns emerge
from an analysis of post-conflict environments:
1
2
Some post-conflict situations have rates of
armed violence comparable to (or even higher
3
than) the conflicts that preceded them.
4
Indirect (non-violent) deaths can remain high
5
in post-conflict societies, long after the fighting stops.
Post-conflict countries are at greater risk of
war recurrence than those that have not experienced armed conflict.
Structural risks in post-conflict environments—
youth bulges, high rates of male unemployment,
and concentrations of displaced populations—
can contribute to armed violence.
In post-conflict situations, violence against
women often continues, and in some cases
increased incidence of such violence has been
reported.
6
7
50
A R M E D V I O L E N CE
Refugee and internally displaced populations
in camps and settlements are often exposed
to high levels of armed violence.
G LO B A L B U R D E N
of
The chapter concludes by noting that there are a
range of security promotion strategies to quell
the effects of armed violence that can be useful
in post-war (as well as non-war) contexts. These
range from post-conflict disarmament, demobilization, and reintegration (DDR) to security sector
reform (SSR) and activities focused on armed
violence prevention and reduction.
Photo " An old woman
and child in the remains of
a bombed-out warehouse.
© Teun Voeten/Panos
Pictures
These interventions may be useful if targeted at
specific groups at risk for, or vulnerable to, violence, and at potential ‘spoilers’ (individual
combatants and groups) of peace transitions.
But these programmes often lack clear measures
of effectiveness particularly when they contend
with the criminal and quasi-political violence that
often overtakes politically oriented violence in
the post-conflict period. Medium- and long-term
strategies that are not pursued in isolation may
be more useful to reduce the risks of high levels
of post-conflict armed violence.
Disaggregating post-conflict
armed violence
A common belief is that when armed conflicts
come to an end improved safety and security will
soon follow. While direct conflict deaths rapidly
decline when war ends, new forms of armed violence can emerge, and the level of indirect deaths
can remain comparatively high until access to
basic services is re-established.1 Conflict and postconflict armed violence substantially increases
the exposure of civilians, particularly women and
children, the elderly, and the displaced, to a higher
risk of mortality and morbidity (WHO, 2008a;
2008b; Ghoborah, Huth, and Russett, 2003). For
example, in the wake of the 1990–91 Gulf War,
one expert remarked that ‘far more persons died
from postwar health effects than from direct war
effects’ (Daponte, 1993). Where wars are especially long and severe, post-conflict mortality and
morbidity can escalate further still.
The persistence of above-average rates of mortality and morbidity in the post-conflict period is
linked to reduced financial investment and human
resources in public infrastructure, including health
care. Depending on the length and severity of the
conflict, the professional health workforce may
be seriously depleted, often taking generations
to recover (Hoeffler and Reynal-Querol, 2003).
But because surveillance and monitoring systems
may also collapse, there are considerable challenges to defining and measuring the global burden
of post-conflict armed violence.2 Another challenge
is linked to disagreements over how to define
‘post conflict’: as with definitions of ‘war’, ‘armed
conflict’ and ‘violent crime’, there is no internationally agreed definition of when a country is officially
pre- or post-conflict.
A post-conflict situation is here described as a
situation following an armed conflict, character-
Table 3.1 Selected post-conflict countries: 1995–2005
cessation of war (i.e. peace agreement or ceasefire), a stalemate, or a significant reduction in
End date
Outcome
51
A R M E D V I O L E N C E A F T E R WA R
ized by a clear victory of one party, a declared
armed violence. Post-conflict environments are
Afghanistan*
2001
Victory
more easily described than defined. Table 3.1,
Angola
2002
Peace agreement
which lists several recent ‘post-conflict’ countries,
Bosnia and Herzegovina*
1995
Peace agreement
highlights the nature of the challenge. Afghanistan
Burundi
2003
Peace agreement
is ‘post-conflict’ in the sense that the Taliban gov-
Cambodia
2000
Peace agreement
Cameroon*
1996
Reduced conflict
Central African Republic*
2002
Reduced conflict
remaining rebel group was not brought into the fold
Comoros*
1997
Ceasefire
until 2008. Other conflicts have similar complexities.
Congo, Democratic Republic
1999, 2002
Peace agreement
These semantic disagreements generate contra-
Congo, Republic of
2000
Peace agreement
dictions and challenges. For example, there are
Côte d’Ivoire*
2004
Peace agreement
routine disagreements over how to ‘count’ vio-
Ecuador-Peru *
1995
Ceasefire
lent deaths, human rights violations, and criminal
Eritrea*
1997, 2000
Peace agreement
violence during and after wars. Certain govern-
Ethiopia*
1997, 2000
Peace agreement
Guinea-Bissau*
1999
Victory
1
Indonesia/Timor-Leste
1999
Peace agreement
2
atic or synthetic analysis of post-conflict violence,
Indonesia/Aceh
2005
Peace agreement
3
and few comprehensive datasets exist to explain
Israel*
1999, 2006
Reduced conflict
patterns and trends before and after war.
Lesotho*
1998
Victory
While it may be difficult to define post-conflict
Macedonia*
2001
Peace agreement
circumstances precisely, certain broad generaliza-
Myanmar*
1997
Ceasefire
6
tions can be made about different post-conflict
Nepal
2005
Peace agreement
contexts. According to Chaudhary and Suhrke
7
Niger*
1997
Ceasefire
Nigeriaa,*
2004
Victory/ceasefire
Russia (Chechnya)*
1996
Ceasefire
emerged from war, such as Nicaragua, Guatemala,
Rwanda
2002
Peace agreement
and El Salvador, continue to exhibit acute levels
Sierra Leone*
2000
Peace agreement
of armed violence—sometimes at rates higher
Solomon Islands
2003
Intervention
than during periods of their armed conflicts.
Sri Lanka*
2001
Ceasefire
ernment was overthrown in 2001, but significant
fighting continues in many areas. Burundi witnessed
a power-sharing arrangement in 2003, but the last
ments may feel they have legitimate reasons to
shield the true magnitude of armed violence from
public scrutiny.3 As a result, there is little system-
(2008) post-conflict countries can be differentiated according to how they experience armed
violence. Some countries that have long since
Other countries, such as Peru, Mozambique, the
a
Solomon Islands, and Sierra Leone, successfully
fire agreement).
transitioned into more peaceful societies.
Sources: * UCDP, Conflict Termination dataset v. 2.0, 1946–2006. Other entries by editors.
There were two conflicts that ended in 2004: northern Nigeria (victory) and Niger Delta (cease-
4
5
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
52
Box 3.1 Post-conflict violence in the Democratic
Republic of the Congo
As conflict subsides and violence is brought under control, direct
mortality rates decline rapidly. Indirect mortality rates also decline,
but somewhat more slowly, and they remain elevated for an unspecified time (Ghobarah, Huth, and Russett, 2001). These trends have
been documented in Sierra Leone, Angola, Liberia, and South
Sudan, among other places. 4
The persistence of above-average rates of indirect conflict deaths
in the aftermath of war is a critical challenge facing humanitarian
and recovery operations. Far more time and resources are expended
on reconstructing basic health infrastructure than in negotiating
ceasefires and disarming and demobilizing former combatants.
The relative vulnerability of populations combined with the inability of states to rehabilitate and resume basic service delivery can
contribute to an increase in mortality that persists well after armed
conflicts come to an end.
The Democratic Republic of the Congo (DRC) was affected by systemic armed conflict for more than a decade, with devastating
implications for population health. The acute armed violence
(1998–2002) contributed to a massive upsurge in violent deaths,
a serious deterioration in health services, food shortages, displacement, and ultimately spiralling rates of excess mortality.
Despite the signing of a formal peace accord in late 2002 and a
reduction in levels of armed violence, persistent conflicts in several eastern provinces continued to exact a monumental human
toll long after the shooting stopped. Although a reduction in the
risk of violent death and more robust UN peacekeeping efforts by
United Nations Mission in the Democratic Republic of the Congo
(MONUC) have shored up the security situation since 2004, the
situation for the Congolese remains precarious.
On the basis of five surveys conducted between 2000 and 2007,
the International Rescue Committee (IRC) estimates that more
than 5.4 million excess deaths occurred after 1998. An estimated
2.1 million of these excess deaths—more than one-third—have
occurred since the formal end of war in 2002. Six years after the
signing of the formal peace agreement, the country’s national crude
mortality rate (CMR) remains stubbornly high at 2.2 deaths per
1,000 per month—more than 50 per cent higher than the subSaharan African average. As Table 3.2 shows, CMRs are higher in
the volatile eastern provinces, at some 2.6 deaths per 1,000 in 2007.
The IRC claims that DRC represents the ‘world’s deadliest crisis
since World War II’ (IRC, 2007, p. ii). Crucially, fewer than 10 per cent
of all these deaths were attributed to armed violence. The vast
majority of the victims died as a result of easily preventable diseases such as malaria, diarrhoea, pneumonia, and malnutrition.
Table 3.2 Approximate crude mortality rates (CMRs) in east and west DRC, 1999–2007
Percentage of crude mortality rates (CMRs)
due to violent deaths
CMR in east DRC (per 1,000 population)
CMR in west DRC (per 1,000 population)
1999
11.1
5.4
–
2000
–
5.4
–
2001
9.4
5.4
–
2002
–
3.5
2.0
2003
1.6
2.9
1.8
2004
–
2.9
1.8
2006
–
2.6
2.0
2007
0.6
2.6
2.0
Note: 2005 was a period that was not surveyed.
Source: IRC (2007, pp. 9, 13)
Why would the incidence of post-conflict violence
remain high, and why would its form change?
One reason is that the domestic balance of power
is usually fundamentally realigned after an armed
conflict. Whether as a result of concessions made
during peace negotiations, the disarmament and
Table 3.3 Typology of post-conflict armed violence
Type of violence
Indicators
Political violence
Assassinations, bomb
attacks, kidnappings,
torture, genocide, mass
displacements, riots
Routine state violence
Violent law enforcement
activities, encounter
killings, social cleansing
operations, routine torture
Economic and crimerelated violence
Armed robbery, extortions,
kidnappings for ransom,
control of markets through
violence
Community and informal
justice and policing
Lynching, vigilante action,
mob justice
Post-war displacements
and disputes
Clashes over land, revenge
killings, small-scale ‘ethnic
cleansing’
Source: Chaudhary and Suhrke (2008)
53
Box 3.2 Sexual violence in the aftermath of war
Higher levels of rape and domestic violence have been reported in many
post-conflict situations, such as in the Democratic Republic of the Congo,
in the former Yugoslavia, in Afghanistan, Burundi, and Liberia, but also in
Guatemala, Nicaragua, and Peru (Bastick, Grimm, and Kunz, 2007). Postconflict sexual violence has been explained by a multitude of factors including the influx of returnees to their communities, high unemployment rates,
lack of economic opportunities, widespread availability of arms, breakdown of social norms, post-conflict masculinity crisis, and high prevalence
of single female-headed households. Weak justice and police institutions,
general lawlessness, and a climate of impunity further increase the risk of
violence and the victimization of groups vulnerable to sexual violence,
such as women and children.
In this environment, the culture of violence and lack of respect for human
rights persists. In some post-conflict countries, it has been observed that,
while during conflict the majority of perpetrators of violence and sexual
violence were identified as members of armed groups and security forces,
an increasing number of perpetrators during the post-conflict period seem
to be neighbours and community members.
In Sierra Leone, experts estimate that between 215,000 and 257,000 women
and girls were affected by sexual violence (PHR, 2002, p. 4). The legacy of
widespread sexual violence during armed conflict continues into post-conflict
society. Half a decade after the end of the conflict, women and girls were
not safe from sexual assault. The International Rescue Committee together
with the Government of Sierra Leone established Sexual Assault Referral
Centres, also referred to as ‘Rainbo’ centres, offering free medical, psychosocial, and legal support to victims (Kellah, 2007). In 2007, 1,176 women and
girls were treated at the centres. Victims of sexual assault and rape were
very young: 65 per cent of reported cases were girls younger than 15 years.
In 149 cases women and girls were gang-raped. Most of the cases came from
areas with large numbers of ex-combatants. This number represents only
a fraction of all incidents. Most police stations received at least one complaint of rape every day. But the unreported cases remain very high because
victims are very reluctant to report what happened to them (IRIN, 2008).
Many DDR programmes established in the aftermath of war still observe traditional gender roles and focus disproportionately on male combatants. Thus,
women and girl combatants are often excluded or their special needs are not
taken into account. This increases the risk of social exclusion and poverty
for women and children ex-combatants, making them more vulnerable to
trafficking and prostitution, perpetuating a cycle of sexual violence. Thus,
excluding women and girls from DDR has important implications for the
victims themselves, but also for development more generally. Some DDR
programmes, such as the United Nations Mission in Liberia‘s DDR Action
Plan, have started to include an explicit gender focus and special arrangements for female combatants.
Source: Bastick, Grimm, and Kunz (2007, pp. 183–86)
A R M E D V I O L E N C E A F T E R WA R
On the basis of Chaudhary and Suhrke (2008), it
is possible to discern several overlapping postconflict scenarios. These include political violence, routine state violence, economic and
crime-related violence, community and informal
justice, and post-war property-related disputes.5
Post-conflict environments imperfectly reflect
the conflicts that precede them. They may continue to feature government-supported militia,
the emergence of organized crime relying on new
forms of capital, and the progressive militarization of society, including in the service of economic and political elites, and high levels of sexual violence (see Box 3.2).
1
2
3
4
5
6
7
54
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
remnants of the enemy and its affiliates, as was
the case in Rwanda following the 1994 genocide.6
By contrast, if a war ends with a clear settlement
overseen by international forces, there may be
fewer instances of flagrant persecution. Rather,
former and official political authorities, military
personnel, and business elites may deploy violent intimidation against those challenging their
position. In many cases, such actions may be
reported erroneously as common or petty crime.
Even more problematic, in some post-conflict
settings experiencing fragmentation and division, armed violence can take on more anarchic
characteristics. Following the United States-led
armed intervention in Afghanistan in 2001, for
example, the vacuum created by the factionalization
of the security sector contributed to an escalation
in warlord-inspired violence.7
Photo ! A five-year-old
Hutu refugee boy stands
next to a Rwandan Army
soldier in Gisenyi, 1996.
© Jerome Delay/AP Photo
demobilization of commanders and rank and file,
or the introduction of democratic elections, different winners and losers emerge in the post-conflict
period. In addition, political elites may rely on
political armed violence to shore up their negotiating positions and lay out their agendas. The shape
and direction of such violence will be informed
by the dynamics of a given peace settlement or
internationally supported recovery strategy.
Many post-conflict environments are equally characterized by more routine state-led armed violence
perpetrated by its security apparatus. In certain
countries such as Guatemala, Mozambique, or
Angola, the military, police, and paramilitary
forces may be more inclined to pursue violent
strategies than to deliver public security after
the warfare has come to an end. The progressive
militarization of these security institutions may
be implicitly sanctioned, even if not explicitly
authorized, by politicians and public authorities.
Routine state armed violence can include what
Chaudhary and Suhrke (2008) label ‘encounter
violence’ (i.e. extrajudicial killing of suspected
criminals rather than arrest or prosecution) as
well as torture to obtain confessions. Security
agencies may also condone social cleansing
operations in slums and shanty towns as part of
law and order operations.
As noted by Chaudhary and Suhrke (2008), if one
party wins and controls a strong security apparatus this can lead to violent purges to eliminate
Another common feature of post-war societies is
economically motivated armed violence. Policymakers and researchers have focused on the way
An under-reported but nevertheless important
category of post-war armed violence relates to
community and informal justice and policing.
Because ‘modern’ law enforcement is often contested in post-conflict societies, informal policing
including vigilantism, lynching, gang patrols, and
customary forms of retributive justice can come
to the fore. As Chaudhary and Suhrke (2008)
observe, the lines between these various categories are fluid and shifting. For example, vigilante
groups are often formally structured and draw on
popular support (White, 1981).
Such violence may derive legitimacy through the
real and perceived protection of civilians from
daily insecurity, often with public support from
state authorities. In Liberia, for instance, the
Ministry of Justice (controversially) called for the
formation of vigilante groups to counter increasing
violent crime in the capital, Monrovia. Lynching
55
and mob justice also appear to enforce certain
forms of order and moral codes. 8 Community
policing can include elements of ‘gang’ violence,
just as neighbourhood gangs may also establish
elements of local control through the provision
of ‘protection services’. In post-war Nicaragua,
for example, urban youth gangs have evolved
from ‘providing micro-regimes of order as well as
communal forms of belonging’ in the mid-1990s,
to forming predatory organizations ‘concerned
with regulating an emergent drug economy in the
exclusive interest of the individual gang members
instead of protecting the local neighbourhood’
(Rodgers, 2006, p. 321).
A final category of post-conflict armed violence
relates to property disputes arising from competing claims registered by displaced populations.
Large-scale dislocation can generate renewed
armed violence if repatriated or returning families
find their house, land, and assets seized by some-
Photo " A mother with
A R M E D V I O L E N C E A F T E R WA R
illegal war economies, including their networks
of patronage, contribute to persistent armed violence at war’s end in countries such as Afghanistan,
Bosnia, Haiti, and elsewhere (Cooper, 2006; Spear,
2006; Goodhand, 2005; Pugh, 2005). Armed
groups that have not been effectively disarmed
and demobilized may morph into organized criminal networks. The entrenchment of economic
armed violence can persist due to the continued
presence of armed ex-combatants with experience
using violence and the absence of meaningful
employment and economic opportunities, as the
case of Iraq so painfully demonstrates. Government and state security forces may also seek to
continue to profit from illegal rents. As pointed
out by Chaudhary and Suhrke (2008), organized
crime of a certain scale cannot continue without
some degree of official complicity. Countries
such as Liberia, Northern Ireland, South Africa,
and others in Central America experienced violent
crime waves in the aftermath of war.
two children crouched
in the entrance of a
makeshift shelter at the
Kalma refugee camp,
Nyala, Sudan, 2007.
© Sven Torfinn/Panos
Pictures
1
2
3
4
5
6
7
56
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Box 3.3 Protecting the displaced from armed violence
Refugees and displaced persons are most often fleeing from conflicts, but dislocated populations can remain for long periods in protracted and ‘post-conflict’
situations. In these circumstances, violence may have subsided, but insecurity
is high and return impossible.
Overall refugee numbers are disputed. In 2007, The UN High Commissioner for
Refugees (UNHCR) recorded 11.4 million refugees under its mandate, of whom
about 2.3 million were in Africa alone (UNHCR, 2008, pp. 2, 7). Although there
are competing definitions of who counts as an ‘internally displaced person’
(IDP), the range of estimates is much higher in comparison to refugees. The
Internal Displacement Monitoring Center reported 26 million IDPs in December
2007, of whom 12.7 million were located in Africa (IDMC, 2008, p. 7). UNHCR
estimates that a total of 51 million IDPs have been displaced as a result of armed
conflict or natural disasters (UNHCR, 2008, p. 2).
Population dislocation is one of the world’s most urgent humanitarian and
development problems. A considerable proportion of the displaced population
resides in so-called protracted situations, often living in dilapidated settlements
over generations. Despite the emergence of new normative standards to promote protection from the UN High Commissioner for Refugees (among others),
insecurity remains widespread and poorly understood (Muggah, 2006).
Map 3.1 Distribution of IDP and refugee populations
in selected African countries
Q
Q
7)2)+%0
1990–2006
Q
7-)66%0)32)
1994–2001
79(%2
Q
Q
0-&)6-%
Q
2004–07
9+%2(%
Q
1989–97
1999–2003
1997 4IVMSHGSZIVIHF]GSYRXV]«WHEXE
&9692(1993–2006
0)+)2(
2YQFIVSJ-(4WERHVIJYKIIW
2YQFIVSJGEQTW
1996–2007
()13'6%8-'6)49&0-'
3*8,)'32+3
1997–2007
Source: Ford Institute for Human Security (2008)
A recent research assessment of what puts protracted
refugees and IDPs at risk of armed violence reviewed
more than 1,500 refugee and IDP camps in Burundi,
the DRC, Liberia, Senegal, Sierra Leone, Sudan, and
Uganda (Ford Institute for Human Security, 2008). It
identified more than 25 factors that intensified risks
and enhanced resilience. The assessment highlights
four strategies to enhance the safety and security of
displaced populations.
Robust protection of camps is much more effective
than small symbolic contribution of forces. Protected
camps were less likely to be attacked than unprotected camps. Of 1,180 documented attacks, fewer
than 20 per cent took place where there a protection
force was in place. Government forces, irrespective
of their size, are most likely to be attacked, though
they are also regularly accused of abusing the populations they are charged with protecting. International
peacekeeping forces are less likely to be attacked,
but a small symbolic force does not provide a robust
deterrent. A small force may in fact embolden wouldbe attackers.
Early protection of camps can save lives. There is an
important relationship between the duration of conflict and the number of attacks on camps. Attacks
tend to steadily increase in the early stages of war,
then decrease. Early protection can prevent belligerent forces from committing armed violence. In Sierra
Leone between 1997 and 2001, for example, in the
aftermath of a coup, more than three-quarters of all
camps were attacked at least once per year. These
rates dropped dramatically after 2001.
Improved access to water can potentially reduce armed
violence against displaced people. There appears to
be a relationship between water points, camps, and
the incidence of armed violence. Specifically, water
availability appears to motivate both the migratory
movements of refugees and IDPs and attacks by
belligerents. In Sudan, for example, a high percentage of attacks occur near water points.
Locating camps at some distance from international
borders does not necessarily increase the safety
of displaced residents. Current international
##
57
Larger camps tend to be more susceptible to
attacks than smaller ones. There is growing
evidence that the larger the refugee or IDP settlement, the more likely it is to be exposed to
armed violence. In Sudan, for example, according
to available data, more than two-thirds of the
101 camps with populations over 10,000 were
attacked. Approximately one-third of the 188
camps with populations of fewer than 10,000
were attacked over the same period.
Photo Georgian soldiers run near a blazing
building after a Russian bombardment in
Gori, 2008. © Gleb Garanich/Reuters
A R M E D V I O L E N C E A F T E R WA R
standards issued by UNHCR emphasize the importance of locating refugee and IDP camps at
least 50 km from neighbouring country borders.
But the 50 km buffer between camps and borders or conflict zones does not necessarily protect the camps.
Source: Ford Institute for Human Security (2008)
one else (see Box 3.3). In certain cases, entire
villages and population groups may have been
coercively evicted, as was the case with certain
Tamil and Sinhalese populations in Sri Lanka
between 2002 and 2008. Liberian Mandingos
who fled during the war found their land occupied
by other ethnic groups when they returned, and
attempts to reclaim it led to rioting and new forms
of communal violence (Chaudhary and Suhrke,
2008). Likewise, in post-war Kosovo, for example,
the Serb minority was particularly exposed to
Kosovo Albanians seeking to establish an ethnically homogeneous territory. Revenge or retribution killings over the death or maiming of family
and community members are also common in
many post-war societies. Such killings tend to
reflect the interests of narrow groups, which subtly distinguishes them from the community and
informal justice just described. In certain instances,
such killings can escalate and intensify smouldering tensions (Mac Ginty 2006).
1
2
3
4
5
6
7
58
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Box 3.4 When do countries relapse into civil war?
It is often said that countries coming out of civil war have a nearly 50 per
cent risk of sliding back into war within the first five post-conflict years.
The figure has circulated in the academic world, the United Nations system,
and the international donor community, and was used as a justification for
the establishment of the Peacebuilding Commission.
However, the broad acceptance of this figure stands in contrast to its general validity. The 50 per cent figure was established as part of an inquiry at
the World Bank into the economic aspects of armed conflict that was led
by Paul Collier and associates (Collier et al., 2003). Various authors have
suggested that this figure is misleading and probably too high. Revised
figures point to a 20–25 per cent risk of conflicts recurring, based on the
use of alternative datasets and independent retesting of the original data
(Walter, 2004; Suhrke and Samset, 2007). Even the authors of the World
Bank study revised their earlier figure downward to 40 per cent (Collier,
Hoeffler, and Söderboom, 2006, p. 14).
These differences matter. On the policy level, a high figure will bolster the
arguments for ‘robust’ international interventions in war-torn countries
and post-conflict situations. Since the figure is based on statistical averages,
Collier recommends that, as a rule, international peacekeeping missions
should last at least ten years to counter the high risk of conflict recurrence.
The lower-end estimate of 20–25 per cent, by contrast, would justify a more
modest and less intrusive engagement.
The different outcomes partly reflect the use of different time periods for
analysis (does war recur within five or ten years?), and different methods.
But this should be a strong warning about the complexities and uncertainties
of using a single estimate as an evidence base for policy. This is particularly the case in research on armed conflict, where the raw data often is
incomplete and uncertain. In this context, statistical analysis can provide
false certainty to policy-makers and support tendencies to fit the data to
the preferred policy position. While still resonating in policy circles, much
statistical research on civil war has been discredited on methodological
grounds (Nathan, 2005; Cramer, 2002).
The responsibility for preventing misuse of research lies with both scholars
and policy-makers. There is nothing unusual about figures changing as
methodologies and data evolve. Researchers need to acknowledge and
discuss openly the limitations of their data and, where appropriate, the
changing results over time—even if it means less support from policymakers who ask for certainty and general formulas. This is particularly so
where statistical methods seem to convey a high degree of certainty. Policymakers should acknowledge that most social scientific knowledge evolves,
and temper their expectations on certainty and general formulas as the
basis for developing policy.
Source: Suhrke and Samset (2007)
Risk factors facing postconflict societies
International concern with post-conflict armed
violence is motivated by its potential to reignite
war and contribute to persistent suffering and
insecurity. At the macro level, research suggests
that post-conflict societies are vulnerable—at
least to the risk of conflict recurrence, if not also
to high levels of armed violence. The oft-cited
statistic that countries emerging from war have a
50 per cent risk of sliding back within the next
five years is probably too pessimistic, but the
risk still is likely to be in the order of 20–25 per
cent—which remains significant from a policy
perspective. Box 3.4 provides an overview of this
debate. Similarly, although the data is poor, Paul
Collier and his colleagues find that ‘during the
first five years following a civil war [homicide] is
around 25 per cent higher than normal’ (Collier
and Hoeffler, 2004, p. 12).
Better evidence is needed on these macro risks,
since these differences matter for policy and programming. For example, the higher the estimated
risk of war recurrence, the more likely policymakers are to undertake robust interventions. The
less certainty that exists, the more cautious and
sensitive will be the likely external intervention.
At the social and individual levels, a host of risk
factors for armed violence affect both non-conflict
and post-conflict societies (Small Arms Survey,
2008). Understanding why violence occurs, who
commits violent acts, and who is at risk of victimization is at the core of strategies for violence
reduction. At the centre of these interventions
are risk factors, which paint a picture of perpetrators, victims, means, and types of violence in
a community. These in turn enable policy-makers
to design interventions to target those perpetrating
armed violence and protect the most vulnerable.
Table 3.4 Risk factors for youth violence
Individual
Family
Peer
School
!"Attention deficit
!"History of early aggression
!"Substance abuse
!"Low cognitive skills
!"Exposure to violence in the family
!"History of victimization
!"Poor parenting
!"Severe or erratic punishment
!"Poor family functioning
!"Parental substance abuse
!"Poor supervision
!"Associating with delinquent peers
!"Peer substance abuse
!"Involvement in gangs
!"Social rejection by peers
!"Lack of involvement in school
extra-curricular activities
!"Poor academic performance
!"Low commitment to school
!"Poor school environment
!"School bullying
Source: Small Arms Survey (2008, p. 262)
General risk factors for violence include substance
abuse, a history of victimization, violence in the
home, attitudes that support the use of violence,
and high levels of economic inequality. While the
presence of these general risk factors increases
the likelihood of violence, different types of violence appear to exhibit some unique risk factors,
as Table 3.4 shows for youth violence. Important
predictors for violence are the presence of gangs
in the neighbourhood, having an older sibling who
is in a gang, feeling unsafe at school or in the neighbourhood, and lack of economic opportunities.
Substance abuse, associating with delinquent peers,
and school bullying contribute to youth violence.
In addition, general conditions such as social
and economic exclusion, rapid urbanization and
social dislocation, unequal access to basic public services, unemployment, and living in poorer
and socially marginalized areas appear to be
correlated with the onset of criminal violence
(UNODC, 2005; Small Arms Survey, 2007). In some
cases, as in West and Central Africa, youth are
rapidly recruited (voluntarily and forcibly) from
urban slums into more structured political institutions such as militia or even rebel groups (Small
Arms Survey, 2006). Given that many of these
factors are associated with rapid urbanization,
greater attention to the dynamics of post-conflict
urban armed violence is needed. Cities are mag-
nets for the young, and youth are the most likely
to perpetrate and be victimized by armed violence
(WHO, 2008b).
Meanwhile, other structural risk factors are being
linked to the recurrence of conflict armed violence.
Sharp economic shocks, rising levels of income
inequality (Picciotto and Fukuda-Parr, 2008), the
expansion of unemployed youth populations
(Collier et al., 2003), horizontal inequalities, and
emerging grievances have all been offered as
explanations for the onset of armed conflict as
well as its contagion across borders. Although
debates persist over the influence of these risks,
the fact that many countries afflicted by war slip
back into conflict means that conflict-prevention
and peace-building interventions should focus
attention on reducing conflict-related violence
(OECD, 2008).
Despite increasing knowledge about risk factors
for violence, a number of important issues remain
unresolved. Little research has yet been undertaken to identify the specific risk factors that
might condition the onset and nature of postconflict armed violence, whether or not it erupts
into outright war. More attention also needs to
be paid to the factors that contribute to the resilience of individuals and societies in the face of
the extreme adversity that often characterizes
post-conflict settings.
A R M E D V I O L E N C E A F T E R WA R
59
1
2
3
4
5
6
7
60
Box 3.5 The demographics of discord
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
From the alleyways of Nairobi’s Kibera slum to the cocaine-processing
enclaves of Colombia’s highlands and militia encampments in
Darfur, the age of violence entrepreneurs is strikingly similar. The
overwhelming majority of those wielding arms are male and less
than 30 years old. This isn’t altogether surprising. Even in developed
countries males are responsible for four out of every five violent
crimes, and the proportion of young adults in a society is a fair (but
incomplete) predictor of homicide rates (see Figure 3.1). Likewise,
the proportion of young adults in a society gives a reasonable
indication of a country’s risk of stumbling into mass violence.
What is the youth bulge?
The youth bulge represents the relatively large proportion of young
adults (15 to 29 years of age) in a given society. More than 80 per
cent of all armed civil conflicts since the 1970s began in countries
where more than 60 per cent of the population was younger than
30. Most other conflicts involved both insurrections and the violent suppression of young populations. While the age structure of
a given population may not necessarily figure in the political and
strategic calculations that pave the way to war, their mobilization
is one ingredient that, together with capital availability, arms
supplies, grievances, and state weaknesses, completes the recipe.
When plotted graphically, the profile of the youthful population is
easily identified and distinguished from more mature ones. It
appears broadly pyramidal, providing a hint of the magnitude of
the challenges that developing states face in providing adequate
public services. Typically, countries with pyramidal age structures
experience growth rates in working age populations of three to
four per cent (compared to about 1 per cent in the United States).
An abundance of adolescents and young adults tends to promote
a vibrant and experimental youth culture. When this large group
matures into its working years, it tends to saturate the job market,
depressing wages and exacerbating unemployment. As a society’s
agricultural sector declines and urbanization intensifies, inequalities rapidly emerge.
Declines in women’s fertility dramatically alter this profile.9 As a
rule, youth bulges appear in countries that have experienced high
fertility rates 20 years previously. Because a bulge dissipates only
after about two decades of fertility decline, today—despite the
spread of modern contraception—15- to 29-year-olds still comprise
more than 40 per cent of the working-age population (15 to 64) in
over half the world’s countries. Most are in sub-Saharan Africa,
the Andes in South America, Central and South Asia, and the
Pacific Islands.
Youthful risks
A youthful society constitutes a potential risk, rather than a cause,
of the onset of collective armed violence. Since the 1960s, there
has been growing awareness that those countries with a large
proportion of young adults have an elevated risk of experiencing
the emergence of a new civil conflict, political violence, and domestic terrorism.10 Comparative studies indicate that the risk of conflict
associated with a large youth bulge is roughly comparable to risks
associated with low levels of per capita income or high levels of infant
mortality—around 2.3 times that of other intervening variables. 11
Political demographers hypothesize that a large youth bulge facilitates youth political mobilization and more formal recruitment
into state and non-state forces and criminal networks.
##
Figure 3.1 Youth population growth rates and murder rates in the United States, 1950–2005
Percentage of young adults (15–29 years)
in the working-age population (15–64 years)
44%
42%
40%
Legend:
Young adults
Murders per 100,000 population
11.0
Murder rate
10.0
Sources: Cohen and Land (1987);
9.0
US Department of Justice Statistics
38%
8.0
36%
7.0
34%
6.0
32%
5.0
30%
4.0
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
Map 3.2 The youth bulge in 2005
A R M E D V I O L E N C E A F T E R WA R
61
0)+)2(
=SYRKEHYPXW¦
EWETIVGIRXEKISJ
XLI[SVOMRKEKI
TSTYPEXMSR ¦
>48%
42–48%
36–42%
<36%
No data
1
AIDS-influenced
age structures
2
Youthful immigrantinfluenced age structures
3
4
Source: Cincotta (2008)
5
Declines in youth bulges are not immediately associated with rapid
reductions in civil conflict. During Northern Ireland’s ‘Troubles’
(1968–96) and Sri Lanka’s conflict (1983–present), collective
armed violence persisted after the population age structure had
experienced considerable maturation. There are some indications
that increasing age maturity together with economic development
can make recruitment into organized armed violence more expensive (ECONOMIC COSTS OF ARMED VIOLENCE).12 Even so, mediumand long-term strategies can reduce the demographic risks of high
levels of criminal and political violence.
Boosting job supply while decreasing job demand
In the medium term, development donors and development
banks can speed up the global migration of light industry to
youth-bulge countries by focusing on incentives and risk protection for private investors—particularly those who encourage
export-oriented industries, job growth, and apprenticeships for
young people, and are willing to work in post-conflict conditions.
Governments and NGOs could promote interventions that reduce
young males’ vulnerability by expanding their skill sets, promoting self-esteem, and developing entrepreneurial motivation and
opportunities to encounter peers. More job opportunities for
youth in high-risk countries as well as investments in girls’ education, maternal and child health, and family planning could also
help in the long term to ease demographic pressures while simultaneously reducing the risks associated with surging unemployed
populations.
Source: Cincotta (2008)
6
7
62
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Box 3.6 The mobilization of inequalities
The vast majority of multiethnic and multi-religious societies are
not excessively violent (Fearon and Laitin, 1996). Nevertheless,
policy-makers would do well to better understand the circumstances
under which violent ethnic and communal conflicts do break out.
A recent project by the Centre for Research on Inequality, Human
Security and Ethnicity (CRISE) at Oxford University focuses on the
role of ‘horizontal inequalities’ as a causal factor. The study focuses
on Latin America, South-east Asia, and West Africa, and finds that
leaders are instrumental in mobilizing latent horizontal inequalities into conflicts and occasionally armed violence. They play a
critical role in fomenting social cleavages along particular group
identities and in exacerbating tensions between communities for
instrumental gain.
Horizontal inequalities refer to the economic, social, and political
inequalities between culturally defined groups (Stewart, 2008).13
Most people have multiple social identities, including gender,
ethnicity, religion, language, profession, and geographic location.
The importance attached to some of these identities varies. In some
contexts where one’s group affiliation assumes more prominence,
however, they can lead individuals to fight, kill, and die in the
name of identity (Stewart, 2008). This is particularly likely to be
the case where groups have suffered vis-à-vis other groups in
terms of their economic advancement, educational and social
welfare, access to the state in terms of exercising political voice
or using services, or rights to express their cultural identity (Langer
and Brown, 2008; Diprose and Ukiwo, 2008; Stewart, 2008). Group
identities and the real and perceived relationships between groups,
are frequently a central feature of contemporary armed violence.
In Côte d’Ivoire, for example, differences in socio-economic status
between northerners and southerners were mobilized by political
leaders and the media (Langer, 2008). Likewise, in both Nigeria
and Indonesia localized identity differences were critical in mobilizing votes and gaining access to local government institutions
(Diprose and Ukiwo, 2008). Group affiliations can be mobilized
according to religious affiliation (e.g. Northern Ireland, India, and
the Philippines), ethnicity (e.g. Rwanda and Sri Lanka), class and
caste (e.g. Nepal), or some combination of these. Ultimately,
however, large-scale group mobilization is not likely to occur in
the absence of serious grievances experienced by both elites and
citizens.14
Both leaders and followers may become strongly motivated where
there are severe and persistent economic, social, and political
differences between culturally defined groups. Østby (2008) also
shows a significant rise in the probability of the onset of conflict
across countries with severe social and economic horizontal
inequalities, for 1986–2004. Mancini (2008) also finds that horizontal inequality in child mortality rates and its change over time
are positively (and significantly) associated with the patterns of
ethno-communal violence in Indonesia.
There are also connections between different types of horizontal
inequality. Inequalities in political power often lead to social and
economic inequalities. Lack of access to education leads to decreased economic opportunities, while low incomes tend to result
in poor educational access and achievement in a vicious cycle of
deprivation. There are also reinforcing cycles of privilege and
deprivation because of the way that one type of capital requires
others to be productive (Stewart, Brown, and Langer, 2008).
The nature of the state and its reaction to conflicts are important
elements determining the severity and persistence of conflict
over time. In Guatemala’s civil war (1960–96) the extremely violent and repressive state reaction to rebellion has been described
as ‘a campaign of state terror’ (Caumartin, 2005, p. 22) with massive
killings, particularly focused on the indigenous population. In
areas where the state is absent (whether by design or by default),
local institutions and local leaders’ reactions to emerging conflicts can determine the likelihood and persistence of violence.
There are ways to minimize the risk that such horizontal inequalities will be mobilized into violent conflicts. For example, in both
Nigeria and Indonesia the presence of formal and informal institutions in peace-building can prevent armed violence from breaking
out. Where the state gives equal treatment to competing sides
(e.g. accountability and incentives to resolve tensions), suspicion
can be reduced and social capital fostered.
There is also empirical evidence that power sharing (through state
structures) can reduce political horizontal inequalities. Likewise,
taxation, affirmative action, employment and education quotas,
and other factors are shown to have a significant impact on reducing socio-economic horizontal inequalities. Successful examples
include Malaysia, where systematic policies introduced in the
1970s have improved the position of the Bumiputera, and Northern
Ireland, where effective employment and education policies (among
others) have sharply narrowed the difference between Catholics
and Protestants and are one major factor behind the progress to
peace (Stewart, Brown, and Langer, 2008).
Source: Diprose and Steward (2008)
Conclusion: promoting security
after conflict
Armed violence and its aftershocks tend to
persist well after the formal fighting stops.15
Anticipating the many forms armed violence
can take in the post-conflict period is essential
63
to promoting sustainable security and development. Yet many contemporary post-conflict
security-promoting activities are simply illequipped to deal with the diverse and complex
faces of armed violence.
A R M E D V I O L E N C E A F T E R WA R
Given the potential importance of ‘youth bulges’
and ‘horizontal inequality’ as general factors
conditioning conflict and violence, a better understanding of these specific risk factors—whether for
criminal or inter-personal violence—is warranted.
Boxes 3.5 and 3.6 explore the impact of demographic factors and of horizontal inequality on
the incidence of armed violence and conflict.
Multilateral peace and security operations have
expanded to deal with irregular forms of war, up
to and including peace enforcement operations,
and to engage in the longer-term process of postconflict peace- and state-building and democracy
promotion. The vast majority of DDR and arms
control operations are also launched in post-war
and post-conflict settings, and (as Figure 3.2
indicates) they have expanded in scale since the
1990s. The development community has also
come to treat underdevelopment as ‘dangerous’
and to invest in interventions to bolster govern-
Figure 3.2 Number of DDR operations around the world, 1989–2008
1
2
Legend:
1989
1990
Sub-Saharan Africa
1991
South Asia
Middle East and
1992
North Africa
1993
Latin America and
1994
3
4
5
Caribbean
1995
Europe and Central Asia
1996
East Asia and Pacific
1997
Source: Muggah (2008)
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
0
5
10
15
20
25
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
64
Yet most contemporary forms of security promoBox 3.7 Transitional justice and DDR in Africa
tion in post-conflict environments tend to be
Conventional transitional justice measures include, inter alia, criminal
prosecutions, truth commissions, reparations for victims, and vetting or
other forms of institutional reform. Post-conflict countries in Africa have
witnessed some of the most well-known efforts in the emerging field of
transitional justice. Examples include the South African Truth and Reconciliation Commission, the International Criminal Tribunal for Rwanda, the
gacaca process in Rwanda, the Special Court for Sierra Leone, the Sierra
Leone Truth and Reconciliation Commission, and the first arrest warrants
issued by the International Criminal Court (ICC) against political leaders
and leaders of armed groups in the DRC, Sudan, Central African Republic,
and Uganda.16
adopted in response to war. As a result, these
Africa has also been the site of the greatest number of DDR operations.
Since the late 1980s there have been at least 11 UN peacekeeping operations in Africa in which the DDR of combatants has been included in the
mandate. In six of these there has also been some form of an internationally assisted transitional justice process.17 But in the remaining two dozen
DDR programmes undertaken in Africa since the early 1990s, connections
with transitional justice did not feature at all.18
as part of a UN Security Council Resolution or
There are good reasons to expect a rise in post-conflict situations where DDR
processes and transitional justice initiatives will coexist. While transitional
justice is focused on promoting justice and accountability, DDR is more
focused on stability and security promotion. Though supporters of the two
processes often compete, both are nevertheless intent on contributing to
longer-term peace and structural stability.
stability) will ensue in the anticipated post-conflict
interventions typically adopt a narrow conception of armed violence and specific categories of
armed actors and struggle to contend with the
more dynamic temporal, spatial, and demographic dimensions of armed violence before,
during, and after wars come to a close. Part of
the reason for this is political and bureaucratic—
programmes such as DDR, international policing,
and small arms control are routinely introduced
pursuant to a peace agreement with direct prescriptions on how such interventions should be
executed.
As such, they assume that conflict has passed its
‘peak’ and that some form of normalization (or
period. Only rarely are interventions developed
on the basis of robust evidence on the ground, to
deal with the combined forms of armed violence
identified above, or to anticipate the mediumand long-term importance of risk reduction.
Source: Muggah (2008)
Beyond a focus on the former warring parties,
and on instrumental policies (such as DDR) to
remove weapons and combatants from conflict
ance and growth opportunities in so-called fragile
dynamics, a number of other approaches can
or weak states.
be explored. One involves linking transitional
As this chapter shows, investment in armed violence prevention and reduction will have to account
justice to issues such as DDR, and is explored in
Box 3.7.
for the many dimensions of post-conflict violence,
Other approaches to containing arms and spoilers
investing in reducing known risk factors, and pro-
in post-conflict contexts could draw upon emerg-
moting violence-sensitive development. A failure
ing experiences of armed violence prevention and
to address effectively and comprehensively the
reduction in seriously violence-affected societies.
immediate and underlying causes of armed vio-
These approaches tend to focus on identifying
lence means that the embers can smoulder, waiting
and responding to risk factors, enhancing resil-
for the next spark to reignite into war.
ience at the municipal level, and constructing
Endnotes
Such programmes also, however, rely on comparatively robust and decentralized local authorities
and civil society—institutions that may be weakened by prolonged periods of warfare and comparatively underdeveloped. More positively, they
also encourage public and private actors to define
and design targeted programmes. Mirroring the
logic of participatory development, the initiative,
control, and responsibility of overseeing such
violence reduction activities rests at least as
much with local partners as with external actors.
Although such interventions are nascent, and
evidence of their effectiveness is patchy, they offer
a promising approach to dealing with some of the
complexities of post-conflict violence.19
1
For a review of the epidemiological literature on postconflict armed violence, consult Small Arms Survey (2005).
2
Reporting biases are common in post-conflict environments. In some cases, post-war killing may be classified
as common crime rather than banditry. In other cases,
the sudden and rapid expansion of reporting may give a
false impression that criminal violence is on the increase.
See, for example, Collier et al. (2003).
3
Reporters and human rights agencies may also underreport the scale of violence owing to repression and selfcensorship. In an era dominated by the ‘war on terror’,
governments may also describe simmering violence as
‘terrorism’.
4
See, for example, CRED surveys in its Complex Emergency
Database (CE-DAT) <http://www.cedat.be/database>.
5
This typology draws explicitly from Chaudhary and
Suhrke (2008) and is based on a project on Violence in
the Post-conflict State at the Chr. Michelsen Institute
(CMI) in Norway.
6
The Rwandan Patriotic Front (RPF), which took control
of the state after the 1994 genocide, used military means
to pursue the genocidaires and the wider ethnic group
associated with them as they fled into neighbouring
DRC, reportedly killing approximately 200,000 people
(Chaudhary and Suhrke, 2008).
7
For instance, militia leaders and rivals Abdul Rashid
Dostum and Atta Mohammed have repeatedly clashed in
their attempts to control the country’s northern provinces.
8
This is not new. Lynchings of African-Americans in the
post-civil war United States were sometimes announced
in newspapers beforehand.
9
Abbreviations
CMR
Crude mortality rate
As youthful populations progress through the demographic transition—descending from high to low birth
and death rates—their age structure matures gradually,
accumulating larger proportions in the middle and upper
parts of their profile while the proportion in younger
age groups shrinks. This transition, which began slowly
during the 18th century in western Europe, has picked
up dramatically: since the mid-1960s it generated an
unprecedented diversity of country-level age structures.
DDR
Disarmament, demobilization, and reintegration
10
See, for example, Staveteig (2005) and Urdal (2006).
DRC
Democratic Republic of the Congo
11
See, for example, Urdal (2006).
ICC
International Criminal Court
12
IDP
Internally displaced person
For example, as Northern Ireland’s youth bulge dissipated
during the early 1980s, the Irish Republican Army shifted
to its ‘long war’ strategy that disengaged from personnelintensive armed incursions. By the mid-1990s both nationalist and unionist militia were reduced to relatively small,
though ruthless and savvy, criminalized units. That effect
IRC
International Rescue Committee
SSR
Security sector reform
UNHCR
UN High Commissioner for Refugees
65
A R M E D V I O L E N C E A F T E R WA R
interventions based on identified needs. A variety
of armed violence prevention and reduction programmes were launched in municipal centres in
Colombia, El Salvador, Nicaragua, Haiti, and Brazil
during the 1990s and the early part of the next
decade. These included voluntary weapons collection, limits on weapon-carrying, alcohol restrictions, and targeted environmental design. These
and other interventions explicitly targeted the
diverse dimensions of arms availability, including the preferences of actors using them and the
real and perceived factors contributing to armed
violence.
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
66
seems also to be influencing Sri Lanka’s ongoing civil war
and the changing parameters of Colombia’s insurgency.
13
These are distinct from ‘vertical inequalities’, which are
typically described as inequalities between individuals.
14
One study in Indonesia that compared two areas in
Central Sulawesi Province with similar concentrations
of Muslims and Christians and inequalities in household
asset wealth demonstrated that only one experienced a
serious outbreak of armed violence. A major difference
between the two was that the difference in household
wealth at the elite level was much sharper in one community than in the other (Diprose and Stewart, 2008).
15
See, for example, Monthly Deaths by Collective Violence
from News Reports for a review of the way armed violence
can persist after outbreaks of collective violence. <http://
www.columbia.edu/~cds81/docs/violence_graphs.pdf>
16
The ICC is currently prosecuting political leaders and
leaders of armed groups in the DRC, Sudan, and
Uganda.
17
These six UN missions include: United Nations Assistance Mission for Rwanda (UNAMIR) in 1993, United
Nations Assistance Mission in Sierra Leone (UNAMSIL)
in 1999, the United Nations Mission in the Democratic
Republic of Congo (MONUC) in 1999, the United Nations
Mission in Liberia (UNMIL) in 2003, the United Nations
Operation in Burundi (UNOB) in 2004, and the United
Nations Mission in the Sudan (UNMIS) in 2005. See UN
Department of Peacekeeping Web site at <http://www.
unddr.org/partners.php?id=5>.
18
See, for example, Muggah (2008).
19
See Muggah (2008) for a review of such interventions
in Africa.
B
y far the largest aspect of the global
burden of armed violence is the deaths
and injuries that occur in non-conflict or
non-war settings. Countries such as South Africa,
Jamaica, and El Salvador suffer from extremely
high recorded levels of homicide, with more
deaths each year than in many contemporary
wars. This fact alone underlines the importance
of adopting a more comprehensive approach to
armed violence, since a narrow focus on conflictrelated deaths by development donors and practitioners excludes the significant burden of armed
violence that occurs in non-conflict settings.
This chapter provides a regional and subregional
breakdown of the global distribution of nonconflict violent deaths, both in absolute terms
and as rates per 100,000 population. It also examines the limited available trend data and provides
information on the burden of violence in cities,
firearm homicides, the gendered dimension of
violent deaths, and the issue of the effectiveness
of criminal justice systems.
The main findings of this chapter are as follows:
Approximately 490,000 deaths from homicide
are estimated to have occurred in 2004.1 The
world average homicide rate in 2004 was 7.6
per 100,000 population.
Southern Africa, Central America, and South
America are the three subregions with the
highest homicide rates. West and Central
Europe, East Asia, and South-east Europe are
the three subregions showing the lowest rates
of homicide.2
Approximately 60 per cent of all violent deaths
are committed with firearms, with variation
from a low of 19 per cent in West and Central
Europe to a high of 77 per cent in Central
America, based on data from 45 countries.
In countries with high homicide rates, women
make up around ten per cent of the victims.
As homicide rates drop, women make up a
greater percentage of victims, up to around 30
per cent in European countries. Available data
is seldom, however, disaggregated by sex.
Trend data shows few increases in homicide
rates over the past decade. The majority of
subregions examined show flat or slightly
increasing or decreasing trends. There is little
evidence that armed violence has, at least at
the subregional level, increased overall in
the Americas, Europe, and Central Asia and
Transcaucasia in recent years.
Arriving at these findings is a complex and delicate exercise, and the chapter also explains some
of the difficulties involved in measuring armed
violence. Existing statistics and data-gathering
mechanisms are underdeveloped, and greater
investment in effective measurement of the burden of armed violence will be needed in order to
develop a more accurate picture of its overall
scope and impact.3
67
N O N - CO N F L I C T A R M E D V I O L E N C E
Lethal Encounters:
Non-conflict Armed Violence
Chapter Four
1
2
3
4
5
6
7
Defining and measuring
violent deaths
A R M E D V I O L E N CE
68
G LO B A L B U R D E N
of
‘Homicide’ is a legal label used to gather information about a specific way in which people die. Most
generally, homicide can be defined as unlawful
death inflicted on a person by another person.
Such a broad definition encompasses a wide
range of acts that may result in death and a whole
spectrum of states of mind of the perpetrator.
Photo " Supporters of
the opposition armed
with machetes, clubs,
and axes run from teargas and bullets in Kibera
slum, Nairobi, Kenya.
© Jon Hrusa/EPA
The focus of this chapter is intentional homicide,
or murder. Intentional homicide requires that the
perpetrator purposefully intends to cause the
death or serious injury of a victim. Situations
where the perpetrator is reckless or grossly negligent, or where the perpetrator kills in self-defence,
are therefore usually excluded from the category
of intentional homicide. The fact that a person is
intentionally killed by another does not necessarily mean that the act is a homicide in law. The
killing of a person by a police officer acting legitimately in the line of duty is an obvious exclusion,
as is the killing of an enemy combatant during a
war or armed conflict.
Despite varying definitions, ‘homicide’ is the most
widely collected data source on non-conflictrelated armed violence across and within countries. The killing of a person is one of the most
serious crimes and therefore tends to be recorded
more effectively than other crimes. The fact of a
dead body is usually processed by the medical
or public health system, in addition to the police
and criminal justice system, creating two potential
sources of administrative statistics. In addition
to counting direct and indirect deaths from armed
conflict, numbers and rates of homicides are useful indicators to capture the non-conflict-related
burden of armed violence.
Armed violence also results in many tens of thousands more victims than the 490,000 homicide
victims in 2004. There are, however, no reliable
estimates for the number of people who are injured
(with either minor injuries or permanent disabilities), or who become victims of armed crimes
such as robbery, carjacking, or armed assault.
The legal label ‘homicide’ captures a wide range
of acts, including domestic disputes that end in
a killing; interpersonal violence; violent conflicts
over land, resources, grazing, or water rights;
inter-gang clashes over turf or control; and predatory violence and killing by armed groups. For
example, most of the deaths in Kenya in the aftermath of the disputed 2007 election would be
considered intentional homicide, as would the
more than 2,500 persons killed in drug-related
69
The difference between deaths arising from armed
conflict and non-conflict deaths is often described
by the organization of the killing. Homicide is
usually committed by individuals or small groups,
whereas the killing in armed conflict is committed
by more or less cohesive groups of up to several
hundred members (Collier and Hoeffler, 2004, p. 3).
But there is often little difference in intensity
between large-scale criminal violence and lowlevel armed conflict, and the line between the two
is often blurred.
A comparative analysis of homicide statistics
must be conducted cautiously. Legal definitions
of homicide vary among countries, and may or
may not include crimes such as assault leading
to death, euthanasia, infanticide, or assistance
with suicide. Societies define those killings that
are perceived as acceptable and others that are
not in their legal codes. Comparing intentional
homicide among countries and regions is, therefore, a comparison not only of the level of intended
killing of persons, but also of the extent to which
countries and regions deem that a killing should
be classified as such.
Official statistics rarely capture the number of
actual criminal events that have occurred. Figures
N O N - CO N F L I C T A R M E D V I O L E N C E
violence in Mexico in 2007–08 (BBC, 2008; Los
Angeles Times, 2008; Reuters, 2008). By contrast,
the 79 suspected gang members killed in clashes
with police in Sao Paulo in May 2006 may not be
counted as homicides (BBC, 2006). Similarly,
neither the nearly 3,000 persons killed in the
attacks on the United States on 11 September
2001, nor the nearly 200 persons killed in terrorist
attacks on 11 March 2004 in Madrid, Spain were
recorded as homicides. These examples highlight
that while ‘homicide’ is a broad category that goes
beyond interpersonal violence, it does not capture
all intentional killing.
and rates should therefore be assumed to be
conservative estimates. Homicide can be reported
by relatives and witnesses, but obviously cannot
be measured through reports by victims. The
quality of homicide figures is also affected by
different criteria and approaches to case recording,
and the capacity of national institutions to gather
data and accurately record events (Aebi, 2004).
The capacity gap between developed and developing countries particularly affects the crossnational comparison of police-recorded crime
statistics (UN, 2007a), with the result that administrative statistics are not a particularly strong
basis for the study of cross-national differences
in criminal activity (Aebi, 2004, p. 163). Some
analysts (Soares, 2004a, p. 851) have demonstrated that variations in crime reporting rates
are ‘strongly related to measures of institutional
stability, to police presence, and . . . to a subjective index of corruption’ (see also Soares, 2004b).
Cross-national differences in reported crime must
therefore take into account both state capacity
and crime victim reporting rates.
Photo ! Police officers
patrol near the house
where two Chinese
students were found
murdered in Newcastle,
UK, August 2008.
© Paul Ellis/AFP/
Getty Images
1
2
3
4
5
6
7
70
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
There are also important differences between data
obtained from public health, police, or criminal
justice institutions. All measure subtly different
phenomena and are therefore unlikely to provide
identical numbers. 4 The differences between
health and police statistics are especially marked
in developing countries, with some analysts noting
that health statistics may be up to 45 per cent
higher than police-recorded figures. In higher
income countries, such as those in West and Central
Europe, significant differences remain for some
countries between police and health statistics
(Shaw, Van Dijk, and Romberg, 2003, pp. 46–47).
Such differences may be linked to limitations in
Map 4.1 Homicide rates per 100,000 population, by subregion, 2004
0)+)2(
Per 100,000
population
>30
25–30
20–25
10–20
5–10
3–5
0–3
Note: The boundaries and designations used on this map do not imply endorsement or acceptance.
Source: UN Office on Drugs and Crime (UNODC) estimates
the capacity of police and law enforcement agencies to identify and record homicide events, and
other factors such as the lethality of assaults.
Despite the proliferation of increasingly dangerous
weapons and an increase in the number of serious
criminal assaults in developing countries since
1960, the lethality of such assaults has dropped
dramatically due to developments in medical
technology and medical support services, in both
North America and Western Europe (Harris et al.,
2002; Aebi, 2004). As a consequence, not only is
it difficult to explain long-term homicide trends
in one region without taking into account improvements in health care, but it is also difficult to draw
Estimating global homicide levels
This section disaggregates the estimated 490,000
non-conflict violent deaths using results from
analysis of available national-level data. 5 Data is
presented in this section as subregional aggregates
due to the difficulties in comparing homicide data
directly at the country level. The resulting homicide
estimates are expressed as the number of homicides per 100,000 people in one year.
Map 4.1 shows the global distribution of homicide captured as population-weighted homicide
levels for 16 subregions for 2004. These subregional figures are calculated from 201 individual
The world average for 2004—the most recent year
for which comprehensive data is available—is 7.6
homicides per 100,000 population. The highest
homicide rates are concentrated in Africa (with
the exception of North Africa) and Central and
South America, and fall within the higher homicide
rate ranges of from 20 to more than 30 homicides
per 100,000 population. By contrast, East and
South-east Asia and West and Central Europe
show the lowest homicide levels, with rates lower
than 3 homicides per 100,000 population. The
Caribbean and East Europe are affected by relatively high homicide rates that are in the range of
10–20 homicides per 100,000 population. North
Africa, North America, and Central Asia follow with
Figure 4.1 Homicide rates per 100,000 population by region and subregion, 2004
71
N O N - CO N F L I C T A R M E D V I O L E N C E
country or territory homicide level estimates, each
derived from available national-level administrative data. 6
comparisons between regions of the world that
have different healthcare systems.
1
2
3
Southern Africa
Central America
4
South America
West and Central Africa
5
East Africa
Africa
6
Caribbean
Americas
Note: Regional and subregional estimates are derived both
East Europe
from public health and police or criminal justice data sources
North Africa
at the national level. The full methodology is described in the
World
on-line appendix at <http://www.genevadeclaration.org>. Data
North America
for Africa derives primarily from public health sources, while
Central Asia and Transcaucasian countries
data for Europe and Asia uses police data as the preferred
Europe
source. Data for the Americas represents both public health
Near and Middle East/South-west Asia
and police data. As set out in this chapter, police and health
Oceania
statistics measure subtly different phenomena, with the result
South Asia
that data sets may not be directly comparable. Where possible,
Asia
such differences have been taken into account at the national
South-east Europe
level, prior to the calculation of subregional figures.
East and South-east Asia
Source: UNODC estimates
West and Central Europe
0
5
10
15
20
25
30
35
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
72
rates between 5 and 10, while Oceania, the Near
and Middle East/South-west Asia, South Asia,
and South-east Europe show homicide rates in
the range 3–5 per 100,000 population.
Figure 4.1 provides in graphic form details of the
regional and subregional distribution of homicide rates. In Africa, high homicide rates may be
associated with a series of social and economic
indicators also linked to crime. These include, for
example, a low overall Human Development Index
(HDI),7 low economic performance,8 high levels of
income inequality,9 a youthful population,10 rapid
rates of urbanization, poorly resourced criminal
justice systems, and a proliferation of firearms,
related in part to the recurrence of conflict in all
regions of the continent (UNODC, 2005, p. ix).
Systematic analysis of the nature of these linkages,
however, remains to be done (see Box 4.1).
Box 4.1 Homicide and human development
Analysis of homicide rates by level of human development reveals the concentration of violent deaths in countries marked by a lack of resources and
poverty. Figure 4.2 shows the population-averaged homicide rate for 176
countries, grouped by low, medium, and high levels of human development
as assigned by the United Nations Development Programme (UNDP) HDI. The
HDI combines measures of life expectancy, literacy, education, and gross
domestic product (GDP) per capita as means of measuring and comparing
levels of human development.
The homicide rate in countries with low levels of human development is
more than three times higher than the average rate in countries with high
or medium levels of human development. This should come as no surprise:
crime rarely occurs in isolation and is one of a range of co-factors associated
with underdevelopment. High levels of income inequality, rapid urbanization,
a high share of unemployed youth in the population, poorly resourced criminal justice systems, and the proliferation of firearms are all associated with
both crime and low levels of development. However, while Figure 4.2 suggests
broad links between development and homicide levels, a strong correlation
does not exist between the two at the level of individual countries.11 Rather,
the HDI captures development indicators that are both affected by and partly
symptomatic of the level of violence in a given society.
In Africa, some conflict-related deaths may appear
in homicide statistics, but overall the number of
direct conflict-related deaths in Africa (approximately 17,700 conflict deaths were recorded via
incident reporting in 2004 12) pales compared to
an estimated 180,000 non-conflict violent deaths
in 2004. There is nevertheless a link between
conflict and non-conflict violence. Armed conflict
has the potential to influence violent crime both
during and after the end of hostilities (ARMED
VIOLENCE AFTER WAR). Contemporary conflicts
often also overlap with organized criminal activity and other forms of looting and predation. The
psychological impact of war, destruction of social
fabric, loss of livelihoods, social displacement,
and increased availability of weapons may also
all contribute to high post-conflict levels of crime
and insecurity that are reflected in homicide levels
(UNODC, 2005, p. x).
The Americas, with the exception of North America,
show the second-highest regional homicide levels.
Central and South American rates are higher than
the global average, representing the second- and
third-highest subregional rates globally: 29.3 and
25.9 homicides per 100,000 population, respecFigure 4.2 Homicide and HDI: homicide rate per
100,000 population, 2004*
22 countries, low HDI
69 countries, high HDI
85 countries, medium HDI
0
5
10
15
20
25
* The classification of high, medium, and low human development
is used in the UNDP Human Development Report to describe countries that have a HDI value of 0.800 or above (high), 0.500–0.799
(medium), or less than 0.500 (low). See UNDP (2008, <http://hdr.
undp.org/en/media/hdr_2007 2008_readers_guide.pdf>, p. 222).
Source: UNODC estimates
73
N O N - CO N F L I C T A R M E D V I O L E N C E
tively. The Caribbean rate of 18.1 is more than twice
as high as the global average (7.6 per 100,000
population).
However, the socioeconomic situation of the
Americas is qualitatively different to that of Africa.
GDP per capita for the Caribbean and South and
Central America is about double that of Africa
and the average HDI is 0.78, as compared with
0.53 for Africa.13 Out of a total of 41 main armed
conflicts globally, 16 occurred in Africa, while
there were only 2 in the Americas (DIRECT
CONFLICT DEATH).
This suggests a different set of factors associated
with a high homicide rate. While the Americas
region does have some history of armed conflict
(especially in Central and South America), it is
drug trafficking, criminal activity, and youth gangs
that play a more significant role in driving homicide levels, particularly in Central America and the
Caribbean (OTHER FORMS OF ARMED VIOLENCE).
The drug trade fuels crime in numerous ways:
through violence linked to trafficking; by normalizing illegal behaviour; by diverting criminal justice
resources from other activities; and, importantly
with respect to homicide, by contributing to the
widespread availability of firearms (UNODC, 2007,
p. 15; UNODC and World Bank, 2007, pp. i–ii).
By comparison, as a region, Asia has the lowest
average intentional homicide rate. However, its
subregions show considerable variability, from
6.6 per 100,000 population for Central Asia and
Transcaucasia to 2.8 for East and South-east Asia.
South Asia is slightly higher at 3.4 per 100,000,
as is the Near and Middle East/South-west Asia
at 4.4 per 100,000 population. It is worth noting
that average homicide levels in South Asia are
almost six times lower than for Africa, even
though average GDP per capita in South Asia is
approximately equal to that for Africa. There is
1
2
3
4
no clear explanation for this, but it does call for a
nuanced perspective on the association between
economic performance (as measured by GDP) and
levels of armed violence.
Oceania shows a homicide rate slightly higher than
for Asia, at 4.0 per 100,000 population. Factors
particularly affecting this comparatively low rate
may include the unique geographic and demographic features of Oceania, with some 23 out of
26 countries or territories having a population
under 1,000,000 persons. Fifteen of these do not
reach 100,000 inhabitants. While the regional
average is low, countries within Oceania show
considerable variability, ranging from 15.2 to less
than 1 per 100,000 population.
Photo ! An armed gang
member in a Rio favela.
© Q. Sakamaki/Redux
5
6
7
Map 4.2 Absolute homicide counts by subregion, 2004
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
74
0)+)2(
,SQMGMHIGSYRXW
F]WYFVIKMSR
50,000–100,000
10,000–50,000
1,000–10,000
0–1,000
100,000
50,000
10,000
1,000
Note: The boundaries and designations used on this map do not imply endorsement or acceptance.
Source: UNODC estimates
The South-east and West/Central European subregions have among the lowest rates of homicide
worldwide, at 3.2 and 1.5 homicides per 100,000
population, respectively. The overall average for
Europe, 5.4 homicides per 100,000, is influenced
by the high value for East Europe of 15.7 homicides
per 100,000 population. West and Central Europe,
taken as a whole, has detailed homicide statistics
available from police and criminal justice sources,
which implies comparatively efficient police forces
capable of crime prevention, detection, and investigation functions. This may be a significant factor
in the low figure for West and Central Europe and
may partly explain the consistently decreasing
trend of homicide levels. Figures from EUROSTAT,
for example, suggest that homicides recorded by
the police fell by about three per cent annually in
European Union member states where consistent
figures could be provided for the period 1995–
2005 (Tavares and Thomas, 2007, p. 2). This pattern
is most noticeable in South-east Europe, where
absolute numbers of homicides declined by around
50 per cent between 1998 and 2006 (UNODC,
2008, p. 39).
The global burden of homicide can also be expressed in absolute counts. These figures are
not representative of homicide levels, because
they are unrelated to the population from which
Of around 490,000 people who were killed in
homicides in 2004, the largest number died in
the subregion of South America: some 95,000,
representing 19 per cent of the total. West and
Central Africa followed with an estimated total
of 78,000 deaths. Homicides in Africa and the
Americas together represent 66 per cent of the
overall figure; 37 per cent and 29 per cent, respectively. Asia follows with 25 per cent of global homicides. Europe accounts for around 9 per cent of
homicide deaths and Oceania for 0.3 per cent of
the total.
Behind the numbers: trends and
distribution of violent deaths
A global analysis of homicide trends over the past
fifty years points to no clear trends. Twelve out of
thirty-four countries for which World Health Organization (WHO) mortality statistics were available
showed significant increases—also described as
‘crime booms’—in homicide levels between 1956
and 1998 (LaFree and Drass, 2002). However, there
75
Box 4.2 Guns and homicide
Firearms are not the only weapons used in armed violence, and death is not
the only outcome. Death by firearm is nonetheless a crucial aspect of the
global burden of armed violence. Using figures from the Ninth UN Survey
on Crime Trends and Operations of Criminal Justice Systems (UN, 2006),
Figure 4.3 presents the percentage of homicides committed by firearm for
countries from eight subregions for which data was available.
N O N - CO N F L I C T A R M E D V I O L E N C E
the homicide count is derived. Nonetheless, Map
4.2 presents a representation of absolute numbers
of homicides by subregion and provides a broad
idea of the global distribution of non-conflict
violent deaths.
The percentage of homicides committed by firearm varies from 19 per cent
in West and Central Europe to 77 per cent in Central America. On a global
scale, percentages may be divided into subregions with more than 50 per
cent of homicides committed by firearm—Central America, South America,
the Caribbean, the Near and Middle East, South-west Asia, and North
America—and those under 50 per cent—Central Asia and Transcaucasia,
South-east Europe, and West and Central Europe.
Although a number of interpretations may be given to the data, such as the
effect of gun control laws and differing availability of small arms and light
weapons between subregions, the results must be interpreted with caution.
Countries operate different recording systems and may inaccurately record
the number of homicides committed by firearms. This may be a result of
limited criminal justice statistics-gathering capacity, factual difficulties in
identifying the cause of death, or simply a lack of follow-through from operational case notes to official police statistics. Some homicide by firearm
statistics reported to the UN Survey on Crime Trends (UN, 2006; 2008) (and
not included in the above analysis) reveal inconsistencies either with data
from previous years or as compared to the total homicide figure provided.
1
2
Despite these difficulties, the available data suggests that approximately
60 per cent of total homicides in the eight subregions were carried out
with a firearm. This figure excludes all of Africa, Oceania, East and Southeast Asia, and South Asia, for which no reliable figures were available. It is,
however, worth noting that if the 60 per cent figure is applied to the global
total of 490,000 estimated total homicides in 2004, the result (approximately
245,000 firearms deaths) is somewhat higher than previously estimated
(Richmond, Cheney, and Schwab, 2005; Small Arms Survey, 2004).14
3
4
5
6
7
Figure 4.3 Percentage of homicides committed with a firearm for countries in eight subregions, 2004 or closest available year
5 countries in Central America
7 countries in South America
5 countries in the Caribbean
3 countries in Near and Middle East/South-west Asia
3 countries in North America
3 countries in Central Asia and Transcaucasia
Source: UNODC elaboration of Crime
6 countries in South-east Europe
Trends Survey Data (UN, 2006)
18 countries in West and Central Europe
0
10
20
30
40
50
60
70
80
76
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
is no conclusive evidence to support the argument
that crime booms have been universal since the
Second World War.15 More recently, analysis of
homicide and homicide attempts in the 1990s in
Europe shows an increase between 1990 and 1992,
followed by a gradual but consistent decrease in
homicide levels between 1992 and 2000 (Aebi,
2004). According to data from EUROSTAT, this
decline has continued to the year 2006 (Tavares
and Thomas, 2007).
Photo " An Italian
soldier guards a train
station in suburban Rome.
© Tony Gentile/Reuters
In a longer historical perspective, however, all
analysts agree that homicide rates in Western
Europe have dropped more or less steadily—and
dramatically—over the past several centuries.
Homicide rates dropped roughly by half from the
medieval to the early modern period (late 16th and
early 17 th centuries), and by the 19th century had
dropped five to ten times further. This holds from
England and Scandinavia to Germany, Switzerland,
the Netherlands, and Italy. The homicide rate in
England dropped from about 23 per 100,000
population in the 13th and 14th centuries to 4.3
per 100,000 by the end of the 17 th century, to 0.8
per 100,000 by the first half of the 20th century.
In the Netherlands and Belgium, equivalent figures were 47, 9.2, and 1.7 per 100,000; while in
Germany and Switzerland, the figures fell from
37 per 100,000 to below 2.0 for the 20th century
(Eisner, 2001; Gurr, 1981; Monkkonen, 2001).
Although the exact timing and scope of the decline
varies from place to place, there is no doubt about
the historical decline in lethal violence within
European states.
Various explanations have been advanced for
this decline, including increases in state capacity
(policing, criminal justice), increased urbanization
and levels of education, and changing norms
towards interpersonal violence. Whatever the
causes, the long-term decline in lethal violence
should provide some insight into contemporary
global trends analysed over a short time period.
The analysis presented below looks at homicide
trends in selected countries based on results
from multiple data sources. It captures the best
available data for the period 1998–2006 in order
to provide a temporal context to the subregional
estimates presented above for 2004.16 This trend
analysis refutes the existence of ‘crime booms’
in the Americas, Europe, and Central Asia and
Transcaucasia in recent years. It shows that there
were very few sustained increases of greater
than ten per cent in homicide levels. The majority
Table 4.1 National-level homicide trend analysis by subregion, 1998–2006
Central
America
North
America
South
America
Central
Asia and
Transcaucasia
East
Europe
Southeast
Europe
West and
Central
Europe
Total
Increasing trend
1
2
–
7
1
–
–
–
11
Decreasing trend
3
1
2
1
6
3
6
11
33
Flat trend
1
2
1
1
–
–
1
11
17
Single dominant change
–
–
–
–
1
1
–
5
7
Total
5
5
3
9
8
4
7
27
68
16
7
3
13
8
4
9
34
94
Number of countries/
territories in subregion
Source: UNODC estimates
of subregions examined show flat or slightly
increasing or decreasing trends.
The examination of homicide trends over time
can be undertaken, provided that reporting and
recording practices, as well as legal definitions
of the offence, do not change during the period
considered. Trend analysis further requires a
rigorous approach to data completeness: it is
important that data from the same set of countries
is compared year to year and that, where subregional or regional trends are examined, data is
collected from as many representative countries
as possible. Reliable trend analysis also usually
requires that countries with fewer than one million inhabitants be excluded, as small numbers
may contribute to a lack of statistical reliability
(Aebi, 2004).
National-level time series data was examined for
the existence of possible trends, and countries
(or territories) classified as ‘increasing’, ‘decreasing’, ‘flat’, or ‘single dominant change’. The category ‘single dominant change’ describes the
situation where homicide levels show a ‘!’- or
‘"’-shaped trend. Countries exhibiting short-term
fluctuations or cyclic changes with multiple peaks
N O N - CO N F L I C T A R M E D V I O L E N C E
Caribbean
77
and troughs, but no overall trend, were classified
as flat. The full methodology used to produce
trend data and to classify it according to these
four categories is described in the methodological annex available on the Geneva Declaration
Web site.
Table 4.1 shows the results of homicide time
series data for 68 countries in eight subregions
for which sufficient data was available. In 33 out of
68 countries, the trend is declining. The majority
1
2
3
4
5
of countries with an increasing trend are in Central
6
and South America. A large number of countries
7
in West and Central Europe show no overall upward or downward trend, although only a few of
these exhibited a completely flat trend, with the
rest showing significant year-on-year variation.17
A number of countries in West and Central Europe,
East Europe, and Central Asia and Transcaucasia
showed a ‘!’- or ‘"’-shaped trend over the
period, suggesting some short- to medium-term
change in homicide trends. Figure 4.4 shows
overall trend graphs by subregion for the period
1998–2006.
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
78
Figure 4.4 Trends in intentional homicide in the Americas, Europe, and
Central Asia and Transcaucasia, 1998–2006
120
100
80
60
40
20
0
1998
1999
2000
2001
2002
2003
2004
2005
2006
Legend:
8 countries in Central Asia and Transcaucasia
4 countries in East Europe
7 countries in South-east Europe
27 countries in West and Central Europe
120
100
80
60
40
20
0
1998
1999
2000
2001
2002
2003
2004
2005
2006
Legend:
5 countries in the Caribbean
5 countries in Central America
9 countries in North America
3 countries in South America
Note: The figure provides a visual representation of the overall subregional trend classifications.
As with the 16 subregional homicide estimates for the year 2004, the subregional trends in the
figure represent population weighted averages for each year. The trend is made up from data
corresponding to the same sub-set of countries within the subregion in each year. The practice of
excluding data from countries with a population of less than one million persons has been followed.
The graphs have all been set to a nominal starting value of 100 in order to allow direct comparability of trends, irrespective of actual homicide levels, i.e. real homicide rates are not shown. The
full methodology used for trend calculation is provided in the on-line appendix at <http://www.
genevadeclaration.org>.
Source: UNODC estimates
Between 1998 and 2006, subregional homicide
levels appear relatively stable. Rates change
reasonably slowly and consistently and do not
generally exhibit unpredictable large increases
or decreases from year to year. In the Americas,
for example, only four data points show a five
per cent change or greater as compared with the
previous year.
In Europe and Central Asia, rates are slightly less
stable. Only South-east Europe and Central Asia
and Transcaucasia, however, show a significant
number of changes of greater than five per cent
between individual years. During the whole period,
a change of greater than ten per cent between
individual years occurs only three times, each time
in South-east Europe. A change greater than 10
per cent occurs as an increase from 1999 to 2000
(20 per cent), and a decrease from 2000 to 2001
(12 per cent) and from 2004 to 2005 (17 per cent).
East Europe shows a particular turning point in
2001. Homicide rates were gradually increasing
prior to this date and began a consistent decline
thereafter. It is possible that this change is due,
in part, to increased rule of law initiatives and
reform within the subregion introduced around
this time.18
In other European subregions, homicide trends
are generally decreasing. In South-east Europe,
homicide rates declined between 2001 and 2006
by over 40 per cent after a peak in 2000: an annual
average decline of 5.1 per cent. This pattern is
matched, although less dramatically, in Central
Asia and Transcaucasia, with an annual average
decrease in the same period of 4.2 per cent. West
and Central Europe shows a decreasing trend
throughout the period 1998–2006, with an average decrease of 2.8 per cent. As a subregional
average, however, this masks the fact that, as
shown in Table 4.1, some countries showed con-
By contrast, South America shows the greatest
rate of consistent increase between 1998 and
2002 (four per cent). The Central America rate
fell between 1998 and 1999, but increased consistently thereafter. North America decreased
between 1998 and 2002, with an average annual
decrease of 2.4 per cent. The Caribbean shows
Trend analysis for the Americas, Europe, and
Central Asia and Transcaucasia provides a con-
Figure 4.5 Homicide country rate per 100,000 population plotted against
average % change in country homicide levels
Legend:
Homicide rate, 2004
Countries in:
70
Caribbean
79
N O N - CO N F L I C T A R M E D V I O L E N C E
no clear linear increase, but presented a homicide
rate six per cent higher in 2002 than 1998. The
increasing trend in the Caribbean links with previous findings of rising crime in the subregion
and a vulnerability to narcotics trafficking and
the violence associated with it (UNODC and World
Bank, 2007, p. ii).
sistent increases during the time period, while
others demonstrated ‘!’- or ‘"’-shaped trends.
65
Central America
North America
60
South America
55
Central Asia and Transcaucasia
50
East Europe
South-east Europe
1
45
West and Central Europe
2
40
Square with outline: statistically non-significant
35
Source: UNODC estimates
3
30
4
25
20
5
15
6
10
7
15
-25
-20
-15
-10
-5
0
5
10
15
20
25
% change
Notes:
Each square represents data for one country, coloured by subregion. Data points in the top right of the chart indicate a high and increasing
homicide rate. Data points in the bottom left indicate a low and decreasing homicide rate.
The plot in this figure represents the superimposition of national homicide levels per 100,000 population at the end of the trend period
measured, with the corresponding average percentage change in homicide levels for that country over the time period. It should be noted
that the period over which the average percentage change is measured is not identical among subregions. Homicide trend analysis was
only possible for the years 1998–2002 in the Americas and for 1998–2005 in Europe, and Central Asia and Transcaucasia. Only countries
showing a decreasing, increasing, or flat trend are plotted. It is not possible to calculate an average percentage change figure for those
countries showing a single dominant change or where significant year-on-year variation occurred. These countries are excluded from the
figure, which, as a result, is provided for visual comparison only.
80
text to the global subregional estimates for 2004
low homicide rates in 2004—West and Central
Europe, South-east Europe, and Central Asia and
Transcaucasia—have achieved such values through
consistent and, in some cases, marked decreases
since 1998.
homicide value for South America (25.9 per
100,000 population), for example, is a result of a
consistent increase in homicide levels between
1998 and 2002. At the lower end of the scale, it
Figure 4.5 summarizes the homicide trends. It
provides a visual indication of homicide levels
can be seen that subregions with comparatively
Table 4.2 Female homicides for selected countries, 2005
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
presented above. The high subregional 2004
Country
Belarus
Total homicides
Female homicides
Female homicides as % of
total homicides
Rate per 100,000
population
1,135
393
34.6
7.6
48,600
4,520
9.3
4.4
Bulgaria
372
100
26.9
2.5
Canada
663
180
27.1
1.1
18,111
1,493
8.2
6.4
167
58
34.7
1.1
El Salvador
3,778
390
10.3
11.5
Germany
2,723
974
35.8
2.3
Guatemala
5,338
518
9.7
8.0
Honduras
2,417
171
7
5.0
Hungary
165
77
46.7
1.5
Ireland
62
9
14.5
0.4
1,471
141
9.6
10.6
Kyrgyzstan
491
106
21.6
4.1
Netherlands
198
67
33.8
0.8
Nicaragua
729
60
8.2
2.2
18,528
2,409
13
10.1
Turkey
6,573
1,266
19.3
3.5
Ukraine
3,529
961
27.2
3.8
Brazil
Colombia
Czech Republic
Jamaica*
South Africa**
* 2004 figure.
** South African Police Service statistics are given from April 2005 to March 2006.
Source: World Bank Group (2008); SIM Datasus (n.d.); Colombian National Police;19 Observatorio Centroamericano sobre Violencia (2007); Campana de prevencion de
violencia de género en El Salvador (2006); IIDH (2006); Jamaica Police Constabulary;20 South African Police Service;21 UNECE (2008)
81
N O N - CO N F L I C T A R M E D V I O L E N C E
Box 4.3 Sex, age, and armed violence
Figure 4.6 Female homicide rates per
100,000 population, 2005
Sex disaggregated data on homicide shows that male homicides vastly outnumber female homicides. There are no comprehensive and reliable statistics
El Salvador
disaggregated by sex, but data from various sources—which are not directly
Jamaica (2004)
comparable with the dataset used in this chapter—indicates that male homicide
rates are usually four or five times greater than female homicide rates.
Table 4.2 presents female homicide data for a selection of states. Aside from
South Africa
Guatemala
Belarus
Colombia
Honduras
the great variation in rates—between 0.4 and 11.5 per 100,000 population—
Brazil
one potential relationship stands out: as a country’s rate of female homicide
Kyrgyzstan
decreases, the percentage of its total homicide victims that are women increases.
In countries that have relatively high overall homicide levels, female homicides
Ukraine
Turkey
Bulgaria
represent between 7 and 13 per cent of total homicides. Colombia, El Salvador,
Germany
Jamaica, and South Africa have particularly high female homicide rates. For
Nicaragua
countries with lower overall rates of homicide (Germany, the Netherlands, and
Canada, for example), the proportion of female homicides is higher, falling
between 27 and 46 per cent.
This suggests that as homicide levels rise, the deaths are concentrated among
Hungary
Canada
Czech Republic
Netherlands
Source: see Table 4.2
Ireland
young men, perhaps linked to larger patterns of criminal activity (e.g. drugs,
0
gangs, etc.). It also suggests that intimate partner violence may not necessarily
2
4
6
8
10
12
decline along the same path that other forms of lethal violence follow. This might
be linked to the persistence of traditional gender roles and violent masculinities across time and place. Data and analysis for many more countries would
2
Figure 4.7 World estimates for homicide
rates per 100,000 population by age, 2004
3
be needed, however, in order to test this observation.
Data from WHO also confirms the general notion that men between the ages of
20 and 29, or 30 and 44, are the most vulnerable to being victims of lethal
armed violence compared to other age categories (see Figure 4.7). Women, by
4
Under 1
5–9
and to have a roughly equal level of vulnerability from age 20 onwards. This
10–14
girl children in many societies, based on the greater value accorded to male
20–29
While these figures provide an overall picture of the distribution of homicide
30–44
among different sex and age groups, much remains to be done to improve data
45–59
but these are often of limited utility due to their incompleteness. 22 In order to
Male
6
7
60–69
develop a better understanding of the distribution among different sexes and
70–79
age groups, data gathering on the national level should include these catego-
80+
5
Female
15–19
children (ARMED VIOLENCE AGAINST WOMEN).
collection. So far, most sex- and age-disaggregated datasets are from WHO,
Legend:
1–4
contrast, are shown to be most vulnerable to homicide in their first year of life,
might be linked to practices of female infanticide and the general neglect of
1
Source: WHO
(forthcoming)
ries in standard reporting mechanisms on homicide.
0
5
10
15
20
25
30
35
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
82
Box 4.4 Up close and personal:
arms availability and female homicide
The simple existence of a gun in a household increases the risk for women
becoming a homicide victim. In the United States, between 40 and 50 per
cent of all female homicides are intimate partner homicides. Of these homicides, 67–80 per cent involve physical abuse of the female by the male
partner before the homicide. Access to a gun and previous threats with a
gun have been found to increase the risk of homicide by about three times
(Kellermann et al., 1993; Campbell et al., 2003, p. 1089).
In 2005 in the United States, 1,858 females were murdered by men. More
than 50 per cent (52 per cent) of these female homicides were committed
with firearms, and more than 90 per cent (92 per cent) of the victims were
murdered by someone they knew (VPC, 2007). In South Africa, 43 per cent
of female homicides were committed with a gun in 2000, making it a major
external cause of death for women. The majority of these homicides are
committed by legally possessed firearms. Thus, rather than contributing
to higher levels of protection, gun ownership at home can increase the risk
of homicide by a family member or intimate partner (Campbell et al., 2003,
p. 1084; NIMSS, 2001, p. 21).
and rates of change across eight subregions for
47 countries (but see the note beneath the figure).
The countries of South and Central America predominantly fall towards the top right of the graph,
indicating high and increasing homicide levels
between 1998 and 2002. The countries of West
and Central Europe, South-east Europe, and
North America fall to the bottom left, indicating
low and decreasing homicide levels (between
1998 and 2005). Countries of the Caribbean and
Central Asia are more widely distributed. They
generally fall higher in the graph and to the left.
A number of outliers, however, show strongly
increasing trends with resultant effects on overall
subregional trends. Overall, this figure suggests
that homicide rates are highly sensitive to local
factors, including, as discussed above, crossnational differences in healthcare systems.
Armed violence and the criminal
justice system
An effective criminal justice response to armed
violence is an important element of prevention
and reduction policies—both for its deterrent
effect and for the prevention of reoffending. An
effective and successful criminal justice system
boosts public confidence and perceptions of
security. As might be expected, however, the
‘success’ of a criminal justice system in detecting
crime and bringing perpetrators to justice depends
on many factors. These range from the efficiency
and level of resources and training of police and
justice personnel to the level of sophistication of
criminal activity in a particular country and the
degree to which corruption and bribery allow
criminals to operate with impunity.
The measurement of ‘success’ is a complex task,
and a number of possible tools have been proposed. The justice attrition rate compares the
number of recorded cases of armed violence, the
number of persons arrested for this crime, the
number of persons prosecuted, the number of
persons convicted, and the number of persons
sentenced to deprivation of liberty. The utility of
the method suffers, however, from the fact that
police, prosecution, court, and penal systems
frequently use different methods of case recording and different definitions, and from the problem
that cases may take a significant amount of time
to be processed by the police and justice system.
As a result, comparison of such figures as published in official statistics is rarely appropriate.
Another tool is the police detection rate. The detection rate is frequently defined simply as the number
of cases solved divided by the number of cases
recorded (Smit, Meijer, and Groen, 2004, p. 229).
The Tenth UN Survey of Crime Trends and Operations
of Criminal Justice Systems (UN, 2006) defined a
case ‘solved’ if it conforms to the following criteria:
The offender was caught in the act (even if he
denies all guilt) or;
The person who committed the offence has
been identified (regardless of whether he is in
custody, on provisional release, still at large,
or dead) or;
Police investigations reveal that no penal
offence was in fact committed (UN, 2007b,
p. 39).
and 10 European) the percentage of homicide
cases solved was greater than 90 per cent, while
in 7 countries (4 Asian and 3 European) the value
was less than 80 per cent. The differences between
the subregional medians are relatively modest,
and, as noted above, a range of factors may affect
police performance in resolving cases. In particular, as the data relates to recorded cases in one
particular year (2005), cases solved in the next
year involving crimes committed in 2005 may not
be taken into account.
Figure 4.8 highlights preliminary results for the
number of recorded homicide cases that are
solved, based on state responses to a question in
the UN Survey on Crime Trends. It must be noted
that only a limited number of responses were
received, from countries predominantly in Europe
and Asia, and that these countries have very different criminal justice systems. In general, however,
responding countries indicated a very high percentage of homicide cases solved.
The overall median value for all 24 countries responding to the question of the number of solved
cases was 90 per cent. For 16 countries in Europe,
the median was 92 per cent, while for 8 countries
in East Asia, Central Asia, and Transcaucasia the
median was 76 per cent. In 13 countries (3 Asian
Photo ! A member of the
18th Street Gang lies
dead, gunned down by a
rival gang member in
Guatemala City, 2001.
© Donna DeCesare
1
2
3
4
5
Commentators note that the majority of solved
cases are solved at the moment of registration or
shortly thereafter (Smit, Meijer, and Groen, 2004,
p. 229). Moreover, the standard as to what constitutes ‘satisfied of a suspect’s guilt’ or ‘the person
6
7
Figure 4.8 Median percentage of recorded homicide cases solved in 24 countries by subregion, 2005
5 countries in East Asia
3 countries in East Europe
10 countries in West and Central Europe
3 countries in South-east Europe
3 countries in Central Asia and Transcaucasia
Source: UNODC elaboration from
Crime Trends Survey Data (UN, 2008)
0
10
20
30
40
50
60
70
N O N - CO N F L I C T A R M E D V I O L E N C E
83
The police are satisfied of a suspect’s guilt
because there is a corroborated confession
and/or because of the weight of the evidence
against him or;
80
90
100
84
A R M E D V I O L E N CE
Map 4.3 Homicide clearance rates in Central America
G LO B A L B U R D E N
of
+9%8)1%0%
)07%0:%(36
2-'%6%+9%
0)+)2(
,SQMGMHIGPIEVERGI
1YVHIVW[MXLWYWTIGXW
1YVHIVW[MXLSYXWYWTIGXW
'378%6-'%
Source: UNODC elaboration from Crime Trends Survey Data (UN, 2006)
who committed the offence has been identified’
may vary between countries. The suspect may
or may not have to be formally charged before this
Figure 4.9 Respondents saying it is possible to bribe a judge to get a
reduced sentence in Central American countries (%), 2004
Source:
Latinobarómetro (2004)
El Salvador
Costa Rica
Panama
Nicaragua
Guatemala
Honduras
0
5
10
15
20
25
30
35
40
criterion is satisfied. Overall, while little may be
said about the differences between subregions,
the results indicate a generally high level of success of the various criminal justice systems.
Nevertheless, these figures should not underestimate the significance of the problem of ineffective
justice and correctional services for violence prevention and reduction. In Guatemala, for example,
in the year 2000 there were 2,707 murders with a
suspect and only 197 without suspects (UNODC,
2007, p. 32). In addition, 37 per cent of respondents
in a survey for Latinobarómetro (2004) indicated
that it is possible to bribe a judge to receive a
reduced sentence (see Figure 4.9). Other Central
The inability to prosecute offenders, corruption,
and the absence of adequate prison facilities
foster a perception of impunity for homicide. The
experience of justice reform in Jamaica and the
Dominican Republic highlights that better cooperation among the police, justice, and correctional services (supported by integrated information systems)
and embedding justice reform in a broader multisector strategy of violence and crime prevention
can help in dealing with impunity and increase the
effectiveness of institutional responses to crime
(UNODC and World Bank, 2007, pp. 126–27).
Conclusion: knowledge gaps
and policy implications
The use of international homicide data as an indirect means to assess the global burden of armed
violence is in its infancy. This chapter has made
use of extensive and rigorous data gathering and
analysis in order to provide a comprehensive
snapshot of the scale and magnitude of lethal
non-conflict armed violence. It has also attempted
to provide some indication of recent trends, and
of the possible spatial, demographic, and socioeconomic factors that might affect levels of
armed violence.
Some cross-national comparisons of homicide levels
have recently begun to appear in developmentrelated publications, including the Human Develop-
85
Box 4.5 Violent death in the city
Received wisdom claims that victimization by more serious crimes is correlated with increases in the proportion of the population of a country living
in larger cities. Criminologists frequently argue that urban density is thought
to be associated with crime, since greater concentrations of people lead to
competition for limited resources, greater stress, and increased conflict
(Glaeser and Sacerdote, 1996; Van Dijk, 1998, p. 69; Naudé, Prinsloo, and
Ladikos, 2006).
When it comes to urban armed violence, however, its frequency and effect
is strikingly heterogeneous and it results from multiple causes. It is linked
to factors such as the drug trade, the availability of weapons, and forms of
social organization such as street gangs and militia or quasi-militia groups
(Small Arms Survey, 2007). While not all urban violence ends in homicide,
homicide rates are related to more general violent acts.
The complexity of urban armed violence is highlighted by the fact that there
is no clear correlation between city population and levels of urban homicide
(Small Arms Survey, 2007). In order to provide further insight into differences between urban and rural homicide rates, research was undertaken
to identify homicide rates per 100,000 population in major cities. Data for
the largest available city in 67 countries was located. The cities ranged
from a population of just over 6,000 persons to more than 14 million persons. The median city population was slightly over 1.2 million. The results
of this comparison are presented in Figure 4.10 and are summarized by
subregion. Insufficient data prevented meaningful comparisons being
carried out for Africa, Oceania, and Asia, with the exception of East Asia.
The methodology for major city/rest of country comparisons is described
in full in the on-line appendix (www.genevadeclaration.org).
A common theme in the literature is that crime levels are higher in urban
areas than rural areas (UN-HABITAT, 2006). While this may be true for
North America, Central Asia and Transcaucasia, West and Central Europe,
and South-east Europe, the reverse appears to be true for East Europe,
Central America, and East Asia. South America shows only a small difference between urban and rural homicide rates.
A first possible explanation for differences may relate to differing degrees
of urbanization in the rest of the country. The ‘major city’ rate may, for
instance, be compared with a ‘rest of country’ rate that itself contains many
large urban centres. A look at urbanization rates only partially explains the
differences, however. While a low urbanization rate in Central Asia and
Transcaucasia (50 per cent) corresponds with a homicide rate one and a half
times as high in the largest city as for rest of country, the pattern is more
complicated in other subregions. The four countries examined in East
Europe and the five countries in East Asia, for example, show average urbanization rates (both around 60 per cent) lower than those for the countries
examined in North America (80 per cent) and West and Central Europe (73
per cent).23 This is despite the fact that East Europe and East Asia show
higher homicide rates for ‘rest of country’ than major cities.
##
N O N - CO N F L I C T A R M E D V I O L E N C E
American countries have similar trends, even if
they are not as acute as Guatemala (Map 4.3). In
Africa, the chances of a murder resulting in a conviction are only around 11 per cent. This figure
increases to 18 per cent in South Africa and stands
in comparison to 56 per cent in the United States
and 61 per cent in the United Kingdom (UNODC,
2005, p. 13).
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
86
ment Report 2007/8, The Economist’s Global
Peace Index, and the Ibrahim Index of African
Governance. These analyses all stress the negative impact that high levels of lethal violence can
have on states and societies, and the utility of
homicide as a proxy to capture overall levels of
armed violence and insecurity.
Another reason may be that homicide rates are not dictated by simple urban–
rural distinctions, but by the nature of urban settings themselves. Small
towns may have levels of violent crime as high as in large cities because
people are more likely to remain in contact, leading to pressure to solve
ongoing conflict (Garrido, Stangeland, and Redondo, 2001). Rapid urbanization in subregions such as Central America may lead to the growth of
many small towns and a subsequent higher homicide rate in the rest of the
country as compared to the largest city.
Other studies conducting cross-national comparison and interpretation of data using police and
public health statistics on homicide (Neapolitan
and Schmalleger, 1997; LaFree, 1999; 2005) have
attempted to describe the phenomenon with reference to time series data and correlations with
other variables. Van Wilsem (2004) notes, for
example, a statistical connection between homicide
and other forms of violent crime. Other research
has detected correlations between homicide levels
and political, economic, and social variables, in
an attempt to identify co-determinates of homicide
(Collier and Hoeffler, 2004; Fajnzylber, Lederman,
and Loayza, 1998; Bye, 2008). A few studies have
also attempted to make cross-national or historical comparisons of homicide levels (Gartner and
Parker, 1990; Stamatel, 2008). But the overall
results are still inconclusive, in part because
analysts are working with poor or insufficiently
detailed data.
The results suggest that a number of factors may be at work in different social,
cultural, and national contexts. Patterns of violence may differ between
urban and rural areas according to whether the perpetrator is an individual,
a gang, or an organized criminal group, and whether the crime is driven by
factors such as drugs, personal vendettas, or simple opportunism. Police
presence and effective state control are also likely to differ between urban
and rural areas, particularly in developing countries.
In the more developed countries of North America and West and Central
Europe, higher homicide rates in major cities may actually indicate a concentration of violent offences in urban areas, because police and medical
systems usually provide effective country-wide coverage. In East Asia and
East Europe, it is difficult to conclude whether violent crime is indeed higher
outside of the major cities or whether other factors, such as differences in
the urban–rural availability of medical care, are responsible for the apparent
difference. In South and Central America, it is possible that a range of forms
of violent crime operate across the countries examined. These can include
organized crime and drug trafficking or opportunism and banditry, giving
rise to similar homicide rates for major cities and the rest of the country.
Finally, the blurring of traditional classifications of urban and rural through
the widespread growth of shantytowns and super-conurbations dictates
that comparisons should be interpreted with caution. Such effects make
accurate definition of the population of a ‘major city’ an extremely difficult
task. In turn, when population figures do not correspond with the area
covered by police administrative statistics, a significant degree of error
may be introduced into the urban–rural comparison.
These studies represent the first steps towards
providing solid policy- and programme-relevant
Figure 4.10 Ratio of homicide rates in major cities and rest of country, 2005
3 cities in 3 countries in North America
28 cities in 28 countries in West and Central Europe
4 cities in 4 countries in Central Asia and Transcaucasia
8 cities in 8 countries in South-east Europe
Note: Bars to the right of 1.0 indicate a higher
homicide rate in the major city than in the rest of the
country. Bars to the left of 1.0 indicate a lower homicide
rate in the major city than in the rest of the country.
5 cities in 5 countries in East Asia
9 cities in 9 countries in South America
6 cities in 6 countries in Central America
Source: UNODC estimates
4 cities in 4 countries in East Europe
2.0
1.5
1.0
1.5
2.0
2.5
Abbreviations
GDP
gross domestic product
HDI
Human Development Index
UNDP
United Nations Development Programme
UNODC
United Nations Office on Drugs and Crime
WHO
World Health Organization
system cannot determine the legal existence of an intentional homicide, merely the fact that a person has been
killed by an act of violence that appears to have been
carried out intentionally. Sometimes, doctors may even
be reluctant to classify a death as a homicide for social
reasons or as a result of pressure from the victim’s family.
5
Described in the on-line appendix at
<http://www.genevadeclaration.org>.
6
The results represented in Map 4.1, and also in Figure 4.1,
correspond to population weighted averages. As a result,
they are sensitive to the distorting effect of countries with
particularly high or low homicide rates (outliers). An alternative method of calculation of subregional figures is the
use of median values. These are available for comparison in
the on-line appendix at <http://www.genevadeclaration.org>.
7
Countries in Africa have an average HDI of 0.53. See UNDP
(2008, <http://hdrstats.undp.org/indicators/1.html>).
8
Countries in Africa have an average gross domestic product
per capita just over one-third that of countries in the Americas and around one-sixth that of countries in Europe. See
UNDP (2008, <http://hdrstats.undp.org/indicators/5.html>).
9
On average, the richest 10 per cent in Africa earn 28 times
more than the poorest 10 per cent. See UNDP (2008,
<http:// hdrstats.undp.org/indicators/145.html>).
10
Forty-three per cent of the population are under the age of
15. See UNDP (2008, <http://hdrstats.undp.org/indicators/
44.html>).
11
R 2 = 0.2, for 176 countries.
12
This figure accounts for 21 armed conflicts in Africa in
2004 (see Chapter 1).
13
See UNDP (2008, <http://hdrstats.undp.org/indicators/1.
html>; <http://hdrstats.undp.org/indicators/5.html>).
14
Richmond, Cheney, and Schwab (2005) estimate total nonconflict-related firearm mortality at between 196,000 and
229,000; Small Arms Survey (2004, p. 200) estimates it
to be between 180,000 and 250,000.
Endnotes
1
2
3
4
The most recent date for which comprehensive global
data is available is 2004.
The world’s regions are subdivided as follows: Africa:
East Africa, North Africa, Southern Africa, West and Central
Africa; Americas: Caribbean, Central America, North America,
South America; Asia: Central Asia and Transcaucasia, East
and South-east Asia, Near and Middle East/South-west
Asia, South Asia; Europe: East Europe, South-east Europe,
West and Central Europe. Oceania is not subdivided.
An on-line appendix (<http://www.genevadeclaration.org>)
also provides a comprehensive account of the methodology
used to arrive at the figures given in this chapter, including
an explanation of data sources and the calculations of
subregional estimates, homicide trends, major city/rest
of country homicide ratios, and the percentage of homicides committed with firearms.
In official public health statistics, important differences
may arise among cause-of-death recording systems. The
individual responsible for determining the cause of death
and the manner in which such decisions enter official
statistics may also vary. In one country, doctors may enter
a cause of death on a death certificate; however, in another
country, a medico-legal coroner may be required to certify
the cause of death. Most importantly, the public health
15
In the 12 ‘boom’ countries identified, rates were reported
to have increased from around 2 homicides per 100,000
population in 1956 to nearly 3 per 100,000 population in
1998, and from just below 4 per 100,000 population in
1956 to 7 per 100,000 by 1995 in developing countries.
Over all 34 countries, while 30 were reported to show an
upward trend direction, this was characterized as ‘sustained’ in only 15 countries, including the 12 considered
to show a homicide boom (LaFree and Drass, 2002).
16
Insufficient data was available to enable reliable trend
analysis in Africa, Oceania, and Asia, with the exception
of Central Asia and Transcaucasia. In the remaining eight
subregions, however, sufficient national-level data was
available for trend analysis between 1998 and 2002 in
87
N O N - CO N F L I C T A R M E D V I O L E N C E
evidence on the patterns and distribution of nonconflict violence. Greater information on the effectiveness of criminal justice systems, and on who
is at risk, from what kind of violence, from what
source, and where and when they are vulnerable
are all important keys to improving the ability of
the international community to design practical
policies to reduce the global incidence of armed
violence.
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
88
17
18
the Americas, and between 1998 and 2005 in Europe, and
Central Asia and Transcaucasia.
19
Figures on Colombia based on data provided by the Colombian National Police.
In West and Central Europe, some 8 countries demonstrated significant fluctuations of up to 50 per cent from
year to year, with no overall upward or downward trend in
homicide levels.
20
Figures on Jamaica based on data provided by the Jamaica
Police Constabulary.
21
South African Police Service statistics received through
written correspondence with Angelica Pino from the Centre
for Study of Violence and Reconciliation, <http://www.csvr.
org.za/>.
22
These datasets include the WHO mortality database (WHO,
n.d.); the WHO World Report on Violence and Health (WHO,
2002); the PAHO mortality database (PAHO, n.d.); the
PAHO age-standardized mortality rate (PAHO, n.d.); WHO
(forthcoming); and projected deaths by WHO region, age,
sex, and cause (WHO, 2006).
23
See UNDP (2008, <http://hdrstats.undp.org/indicators/
41.html>).
During the early to mid-1990s, for example, both the
Russian Federation and Ukraine adopted significant legislative acts aimed at providing a modern framework for
policing. This was followed in the mid-1990s by the adoption of ‘Concept of Development’ Programmes for the reform
of police in the Russian Federation and Ukraine, which
included short-, medium-, and long-term plans relating to
police activity, resulting in changes to police legal status,
organizational structure, operational police forces, work
patterns, and supervision and control (Robertson, 2004).
A
rmed violence imposes costs across
multiple levels of society and especially
on the poorest and most vulnerable.
Although armed violence can benefit a small
minority of the population—some gain from (new)
employment opportunities and (often illicit) wealth
transfers—there is overwhelming evidence of the
ways it diminishes development prospects for the
majority and hinders achievement of the Millennium
Development Goals.1
The annual global economic costs of armed violence run into the hundreds of billions of dollars.
These financial, fixed, and human capital costs
need to be considered in any estimate of the global
burden of such violence. The precise dimensions
of this economic burden depend, however, on
how ‘costs’ are measured. The ‘costs’ of armed
violence are here defined as the short- and longterm measurable effects that are convertible to
welfare losses. Although there are many ways to
calculate the economic costs, their true extent is
ultimately shaped by the duration, severity, and
spatial distribution of armed violence.
Specifically, this chapter finds that:
Non-conflict armed violence produces direct
and indirect economic effects that can exceed
the costs of armed conflict. The economic
costs of non-conflict armed violence in just
90 countries—measured in terms of lost
productivity—is USD 95 billion and may reach
as high as USD 163.3 billion, or 0.14 per cent
of the annual global gross domestic product
(GDP) in 2004.
The overall costs of armed violence escalate
higher still when the consequences of armed
conflict are taken into consideration. Violent civil
conflict decreases the GDP growth of an average
economy by at least two per cent per year.2
The subjective experience of armed violence
generates tremendous economic costs. Using
contingent valuation approaches, the global
cost of ‘insecurity’ generated by conflict
amounts to up to USD 70 per person, or a
global annual burden of USD 400 billion.
This chapter adopts a broad approach to measuring the economic costs of armed violence.
Looking beyond the narrow financial costs, it
finds that the negative economic impacts of
armed violence are more extensive than often
assumed and include:
fiscal effects (macroeconomic instability,
increases in inefficient military and policing
expenditures, and decreases in welfare
spending);
losses in productive capital (destruction of
infrastructure, land, houses, and assets);
depleted financial capital (capital flight, soaring inflation and depreciating investments,
and rising transaction costs);
eroded human capital (due to communicable
disease, reduced nutrition, diminished edu-
89
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
What’s in a Number? Estimating
the Economic Costs of Armed Violence
Chapter Five
1
2
3
4
5
6
7
cation, displacement, and out-migration
measured as a function of years of life lost);
A R M E D V I O L E N CE
rising transaction costs (lowered consumer
and investor confidence, particularly in cities);
and
G LO B A L B U R D E N
of
90
Photo " An Iraqi man
the reallocation of development assistance (to
less risky environments).
mourns the death of his
family and destruction of
his home in fighting in
Baghdad, 2003.
© Q. Sakamaki/Redux
It is only in accounting for all of these costs that
one can achieve a genuinely global picture of the
economic consequences of armed violence.
Developing a better estimate of the economic
burden of armed violence is essential for prioritizing, designing, financing, and implementing
effective interventions. The Geneva Declaration
on Armed Violence and Development emphasises how measurement of the economic costs
of armed violence can encourage investment in
preventive action and hedging against future
losses.3 Demonstrating who loses what, where,
when, and under what circumstances, can
assist policy-makers, activists, and researchers
in identifying constituencies to support armed
violence reduction in the public and private
sectors.
Unfortunately, most efforts to calculate the costs
of armed violence at the global or even national
level have been frustrated by the absence of a
unified conceptual framework, complementary
methodologies, or the availability of data over
time. As discussed below, there are consequently
extreme variations in estimates. 4
The importance of documenting the economic
dimensions of armed violence is today widely
appreciated. But research is often narrowly focused
on two core manifestations of armed violence—war
and crime—with issues such as intimate partner
and sexual violence often left hidden from view.
In the case of war, researchers frequently adopt a
case study or cross-country comparative approach,
measuring economic effects as a function of GDP
losses in absolute or relative terms. 5 Such costs
imply a loss of income (and purchasing power)
that, in most developing countries, would otherwise be devoted to the acquisition of basic needs,
such as food, shelter, and clothing.
These studies reveal that the annual burden of
war-related violence ranges from 2 to 20 per cent
of a country’s GDP.6 While the studies have limitations and contradictions, they overwhelmingly
This chapter examines different approaches to
measuring the economic costs of armed violence,
in order to increase our awareness of its broad
implications for development. The first section of
the chapter introduces a three-fold approach to
costing armed violence by drawing on accounting,
modelling, and contingent valuation methodologies. The second section considers the economic
costs of armed violence in a selection of countries for which there is adequate data and draws
explicitly on all three approaches. The third section provides a short overview of the distributional
effects of armed violence. The fourth section
discusses the possible positive effects of armed
violence. The chapter closes with some brief conclusions and considers next steps for development
policy-makers and practitioners.
Approaches to costing
armed violence
The earliest assessments of the costs of war were
undertaken by Werner Sombart (1913) and John
Maynard Keynes, particularly in the latter’s seminal The Economic Consequences of Peace in 1919
(Keynes, 2005). As mainstream economists became
interested in the issue, they sought to demonstrate
whether investment and destruction arising from
armed conflict had the potential to generate new
efficiencies and release productive energies. 8
Following the Second World War, however,
emerging research highlighting the negative
consequences of collective armed violence
gained more credence, particularly in the context
of civil wars. By the end of the 20th century, contemporary analysis of the negative economic
consequences of civil wars and criminal violence
began to grow in breadth and sophistication.9
Evidence began to mount of the way upward
shifts in military and policing spending constituted unproductive expenditure and detracted
from welfare spending. The primary metrics of
these negative costs consisted of macroeconomic
functions such as GDP growth or simply government revenue.10
This chapter considers three ways of measuring
the economic burden of armed violence: accounting, modelling, and contingent valuation. Each of
these approaches adopts different assumptions,
methods, and data sources, and they are not
necessarily comparable. But each offers important insights into the scale and magnitude of the
economic burden of armed violence. A first step
to generating a realistic estimate of the economic
costs of armed violence, then, is to recognize the
differences among the approaches.11
The accounting approach is essentially a balance
sheet of the accumulated costs of specific factors
to the economy. Whether determined from a macro
or micro perspective, it requires reliable data
and the ability to identify appropriate cost factors
associated with fatal and non-fatal injury rates.
This is the principle methodology applied by
public health economists associated with the
91
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
observe that war has negative effects on economic
growth and well-being.7 As for criminal violence,
researchers frequently account for public spending on law enforcement and the judicial sector,
together with foregone investment and nonproductive expenditures. A review of these studies suggests that, in developing countries, public
expenditures on law enforcement consume 10–15
per cent of GDP, as compared to 5 per cent in
developed states (IADB, 2006; Londoño and
Guererro, 1999). There is ample evidence that
criminal violence also undermines human welfare and, ultimately, social development (UNODC,
2007a; 2007b).
1
2
3
4
5
6
7
92
applied to generate estimates of the costs of civil
wars (Collier, 1999).
A R M E D V I O L E N CE
The contingent valuation approach seeks to capture the amount theoretically assigned by the
‘market’, or individuals’ willingness to pay to
improve their security and reduce the incidence
of armed violence.
G LO B A L B U R D E N
of
It is useful to recall that armed violence is not
necessarily bad for everyone. There are winners
and losers in situations of war and crime-related
violence. But, in economic terms, those who profit
from violence are gaining from inefficient and
unproductive activities. Put another way, those
doing well out of armed violence simply reallocate wealth and do not increase the productive
capacity of an economy (on the contrary, they
often destroy value). While the development of warrelated technologies can theoretically increase
overall economic productivity, there is comparatively less evidence that this is the case in practice.
In order to shed light on the externalities generated
by armed violence, the chapter considers the
‘distributional costs’ that shape the transfer of
assets and income arising from armed violence.
Accounting for armed violence
Photo ! An ex-combatant
receives carpentry skills
training at a UNICEFfunded rehabilitation
centre, DRC. © Giacomo
Pirozzi/Panos Pictures
World Health Organization (WHO) and other
agencies (Butchart et al., 2008; Muggah, 2008).
The modelling approach requires establishing a
credible counter-factual situation and then determining the difference between expected and actual
economic growth. It is most often measured as a
function of GDP losses. This is the principal method
The accounting approach first identifies different
categories of armed violence and then tabulates
an overall burden. Categories include ‘direct
costs’ arising from medical and rehabilitation
services, policing, criminal justice, and private
security; ‘indirect costs’, including lost earnings,
reduced savings, and losses in investment and
human capital; and ‘social multipliers’ relating to
loss of social capital and reduced political participation. One way of accounting for multiple categories is by assessing the bottom-up distribution
of external mortality from national surveillance
systems (see Box 5.1).
Table 5.1 A typology for costing armed violence
Health economists often distinguish between
the direct and indirect costs of armed violence.
Direct costs arise directly from acts of intentional
violence and require payments by individuals or
institutions. They can be further subdivided into
medical and non-medical costs. Indirect costs refer
to lost resources and opportunities resulting
from armed violence. Studies tend to emphasize
the tangible costs (e.g. reduced productivity of
survivors, lost investment in social capital, and
reduced productivity of perpetrators), together
with reduced quality of life. While these costs
likely only reveal the tip of the iceberg, they
can be accounted for and are reproduced in the
typology given in Table 5.1.
Cost category
Type of cost
Components
Direct
Medical
!"Hospital inpatient
!"Hospital outpatient
!"Transport
!"Physician
!"Drugs/tests
!"Counselling
Non-medical
imprisonment
!"Legal services
!"Foster care
!"Private security
Indirect
Tangible
!"Lost investment in
social capital
!"Life insurance
These guidelines were subsequently tested to
assess the costs of armed violence in ‘non-conflict’
contexts: Brazil, Jamaica, and Thailand. The
preliminary assessment drew primarily from
national surveillance data for the most recent
years available.
In Jamaica, the direct medical costs of interpersonal
violence in 2006 totalled some USD 29.5 million
(JMD 2.1 billion)—the vast majority of which was
concentrated among young males. Indirect medical costs were ten times higher, exceeding USD
385 million (JMD 27.5 billion). Direct medical
##
costs accounted for approximately 12 per
!"Loss of productivity
(earnings and time)
WHO, the Centers for Disease Control, and the
Small Arms Survey recently elaborated economic
costing guidelines to assess the direct and indirect burden of violence (Butchart et al., 2008).
In Brazil, the direct medical costs of interpersonal violence in 2004 totalled USD 235 million
(BRL 382 million— more than three-quarters of
which were attributed to injuries among men).
Indirect costs exceeded more than USD 9.2 billion
(BRL 15.4 billion).12 Taken together, the direct
medical costs of injuries amounted to 0.4 per
cent of the total health budget, while indirect
costs amounted to 12 per cent of all health
expenditures, or 1.2 per cent of GDP.13
!"Policing and
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
Box 5.1 Accounting for the costs of
violence: a typology and examples
93
Intangible
!"Indirect protection
1
!"Health-related quality
2
of life (pain and
suffering)
!"Other quality of life
(reduced job oppor
tunities, access to
public services, and
participation in public
life)
Source: Butchart et al. (2008)
cent of Jamaica’s total health expenditure, while the combined direct and
indirect impacts were equivalent to four per cent of GDP.
In Thailand, the direct medical costs of interpersonal violence in 2005
amounted to approximately USD 40.3 million (THB 1.3 billion).14 Indirect
medical costs were an order of magnitude higher at USD 432.7 (THB 14.4
billion). During 2005, the direct medical costs of both interpersonal and
self-directed violence accounted for four per cent of the country’s health
budget, while the indirect costs accounted for approximately 0.4 per cent
of GDP.
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
94
The annual direct costs of firearm injuries in Brazil
and Colombia, for example, are an estimated USD
88 million and USD 38 million per year, respectively.15 When extrapolated to account for indirect
costs arising from morbidity, foregone earnings,
and policing, however, the costs skyrocket: they
reach as high as USD 10 billion in Brazil and USD
4 billion in Colombia—or 0.5 and 1 per cent of their
respective GDPs for a single year (Small Arms
Survey, 2006, p. 207). In Guatemala, the direct
and indirect costs of armed violence-related injuries in 2005 amounted to some USD 2.4 billion, or
7.3 per cent of GDP. In El Salvador, the accumulated direct and indirect costs of armed violence
rise to some 14 per cent of GDP (UNDP, 2006).16
Box 5.2 Modelling the economic costs of civil war
A recent study by Chauvet, Collier, and Hegre (2008) estimates the range
of ‘core’ costs of a ‘typical’ civil war as USD 60–250 billion. These authors
claim that the economic costs averaged about USD 123 billion per year
over the past four decades. Likewise, coups were associated with core and
total costs of USD 4–16 billion per annum.
Although these are only estimates based on a simple model, they do offer
important insights into how economists determine the economic costs of
war. Such estimates assume that the (economic) ‘feasibility’ of war is,
ultimately, a key determinate of its onset. Countries predisposed to war
therefore face low income, sluggish growth, high dependence on commodity
exports, and (geographically) rough terrain.
Likewise, social characteristics—small populations, large share of youth,
and social fractionalization—are also connected to war onset. Finally, and
much more controversially, ‘democracy’ appears to have a benign effect—
but, in some cases, even increases rather than decreases the risk of war
onset (Chauvet, Collier, and Hegre, 2008). This latter finding is challenged
by Elbadawi (2008) and Bodea and Elbadawi (2007), who see robust democracies as mitigating conflict recurrence.
More importantly, efforts by Chauvet, Collier, and Hegre to model the economic
costs of civil war offer a number of straightforward and compelling policy
prescriptions. They suggest that aid packages provided to a ‘typical’ postconflict country of about USD 4 billion could potentially generate an overall
benefit of USD 10.7–13.8 billion over a ten-year period, depending on the discount rate and cost of conflict (Chauvet, Collier, and Hegre, 2008, pp. 61–63).
Modelling armed violence
Economists studying war commonly adopt a
modelling approach to measuring the economic
costs of collective armed violence. They estimate
the costs of armed conflict by undertaking growth
simulations in countries affected by civil wars.
Such estimates should take account of the social
and geographic concentration of the effects of war
(particularly among the poor), the opportunity
costs for development, the persistence of the
economic costs of war over time, and spillover
effects, such as crime, disease, and terrorism.
A variety of researchers have shown that a civil
war of five years can reduce the annual average
growth rate of a country by approximately 2–2.2
per cent (Collier, 1999; Hoeffler and Reynal-Querol,
2003).17 Drawing on more recent data and estimation techniques, researchers show that civil
conflict likely decreases the growth of GDP for
an average economy by 2.17 per cent (Restrepo
et al., 2008).18
Put in straightforward country terms, a ‘typical’
civil war is estimated to cost a country at least
USD 64 billion.19 This includes an estimated USD
49 billion in military expenditure and economic
losses, USD 10 billion in post-conflict effects,
and USD 5 billion in national healthcare costs
above what might have been expected had war
not taken place (Collier and Hoeffler, 2004b).
Recent modelling of armed conflicts in Africa
between 1990 and 2005 estimated economic
losses at approximately USD 284 billion (Oxfam-GB,
2007). By focusing on GDP losses, the modelling
approach accounts for reduced growth and finds
that these effects can persist long after wars
come to an end (Bates, 2008).
Modelling can provide an especially robust account
of the costs of armed violence at the country or
sub-national level. Estimates of the economic costs
The modelling approach also suffers from limitations in comparability since methodologies and
datasets often differ significantly among studies.
In Nicaragua, for example, models estimating lost
GDP range from 0.8 to 90 per cent (Stewart, Huang,
and Wang, 2001; Lopez, 2001). In countries not
affected by armed conflict, modelling armed violence reveals a tremendous array of hidden costs.
For example, the estimated costs of interpersonal
violence in the United States range from USD 1.8
billion to USD 507 billion depending on how violence and its consequences are measured (WHO,
2004, pp. 13–14). Likewise, in the United Kingdom
and Wales, the costs of criminal violence were
estimated at between USD 40.2 billion and USD
63.8 billion per year (Brand and Price, 2000).
Contingent valuation of armed violence
Contingent valuation or ‘willingness-to-pay’
approaches are also commonly employed to estimate the costs of armed violence. Such techniques
measure what individuals and households are
prepared to pay in order to improve their safety
from, or live free of the threat of, armed violence.
As with the two other methods, the contingent
valuation approach requires a number of basic
assumptions. It assumes that people (or individuals, households, and firms) seek to avoid
uncertainty and are prepared to give up some
degree of their consumption permanently in order
to live in a less uncertain world. 20 Contingent
valuation does not necessarily address all the
possible economic costs of armed violence. For
example, material impacts associated with lost
assets and inefficiencies generated by changes
in behaviour (induced by criminal violence, for
example) are not easily captured by this method
(Merlo, 2004). Even so, there is evidence that
suggests that non-monetary costs of armed
violence and crime are at least as important as
material ones (Soares, 2006).
Recent analysis indicates that individuals living
in conflict-affected countries would be, on average, prepared to contribute the equivalent of
eight per cent of their annual consumption (per
annum) to live in a more peaceful environment. If
extrapolated on the basis of international datasets,
the average global cost of ‘insecurity’ generated
by armed violence amounts to roughly USD 70
Photo ! A women’s
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
95
of conflict and collective armed violence reveal
significant GDP losses. During a protracted crossborder conflict with Iraq in 1979–81, for example,
Iran experienced a cumulative loss of some 48
per cent of GDP. Iraq was also significantly affected,
having lost an estimated 11 per cent of GDP over
two conflicts (1977–93). Internal or civil wars also
generate significant losses. For example, Ethiopia
lost approximately four per cent of expected GDP
(1977–93), Liberia nearly two per cent (1984–95),
and Sri Lanka 2–16 per cent, depending on the
periods under review (1983–87 and 1983–94)
(Stewart, Huang, and Wang, 2001, p. 96).
work brigade paid by the
Russian federal authorities
to sweep the streets of
war-ravaged Grozny,
Chechnya. © Martin
Adler/Panos Pictures
1
2
3
4
5
6
7
Figure 5.1 Social value of violence reduction, % of GDP, selected countries, 1995
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
96
Box 5.3 Measuring the health
dimensions of violence using
contingent valuation
The economic costs of armed violence extend
beyond material losses to longer-term welfare
losses arising from higher mortality. Mortality
rates and their distribution across age groups
can determine reductions in life expectancy
that can then be valued using a marginal
willingness-to-pay approach.
A seminal work by Soares (2006) examines
the health dimensions of the welfare costs
of violence in 73 countries. It is the first
comprehensive cross-country estimate of
the non-monetary costs of violence and a
first attempt at using the ‘value of life’
methodology to estimate the social value of
violence reduction.21 Together with the age
distribution of a population, the willingnessto-pay approach can be used to estimate the
social value attached to violence reduction—
or the welfare costs of violence. 22
The study finds that the reduction in life
expectancy due to homicidal violence represents a substantial welfare loss—in the same
order of magnitude of the direct material costs
of crime.23 On average, one year of life expectancy lost due to violence is associated with
a yearly social cost of 3.8 per cent of GDP.
Taking account of all related health dimensions increases the estimated social costs
of violence by 40 per cent in the United
States and 57 per cent in Latin America
(Londoño and Guerrero, 1999).
Figure 5.1 reveals the social value attached
to violence reduction as a share of GDP for
all countries sampled by Soares (2006),
ordered from highest to lowest. Of the top ten
countries, eight are found in Latin America.24
The 11 remaining countries that complete
the top 20 are all in Latin America and the
Caribbean or are former Communist regions.
At the other extreme of the distribution, the
ten lowest values are for Western European
countries and Japan (Soares, 2006).
Colombia*
Philippines*
Venezuela
Chile
El Salvador
Belize
Suriname
Mexico
Brazil
Russian Federation
Puerto Rico
Kazakhstan
Bahamas
Croatia
Latvia
Argentina
Saint Kitts & Nevis
Estonia
Ukraine
Kyrgyzstan
Azerbaijan
Tajikistan
United States
Belarus
Ecuador
Israel
Costa Rica
Trinidad & Tobago
Republic of Moldova
Lithuania
Turkmenistan
Albania
Barbados
Portugal
Cuba
Kuwait
Grenada
Singapore
Uzbekistan
Armenia
Georgia
Finland
Sweden
Poland
France
Uruguay
United Kingdom
Luxembourg
Czech Republic
Belgium
Republic of Korea
Slovak Republic
Bulgaria
Australia
New Zealand
Hong Kong
Canada
Hungary
Germany
Iceland
Romania
Mauritius
Malta
Japan
Slovenia
Italy
FYR of Macedonia
Norway
Netherlands
Austria
Ireland
Greece
Spain
percentage of gdp
$
$
* The values ascribed to
Colombia and the Philippines
are roughly 280 per cent.
Source: Soares (2006)
0
10
20
30
40
50
60
70
80
90
100
Table 5.2 The social value of violence reduction in selected WHO regions, 1990s*
Life expectancy
(years)
Homicide rate (per
100,000)
GDP per capita
(USD)
Expected years of
life lost
Social value, future
generation (USD
billions)
Social value as %
of GDP
Latin America and
the Caribbean
71.4
21.8
7,708
0.6
49.8
57
North America
76.1
6.5
25,672
0.2
456.14
15
Western Europe
76.2
4
11,383
0.1
7.23
7
Former Communist
68.9
17.2
6,009
0.4
6.59
20
76
7.8
17,839
0.2
82.3
46
Western Pacific
* Regional statistics are unweighted averages. Due to data availability, the only African country included in the sample is Mauritius, and the only eastern Mediterranean
country is Kuwait (these regions are not included in the table).
Source: Soares (2006)
per person, or a global annual ‘cost’ of USD 400
billion (Hess, 2003). Such general estimates must,
however, be treated with caution.
More common, however, are studies that focus on
the willingness of people to live free of certain
forms of criminal violence. One recent contingent
valuation assessment examines the value of permanent reductions in homicide for individuals in
more than 70 countries (see Box 5.3).25 In examining the health dimensions of the costs of violence,
the study reveals that homicide alone contributed
to a reduction of approximately 9.7 per cent of
Colombian GDP in 1995 and 0.9 per cent of US GDP
in the same year.26
Costing armed violence in a
sample of countries
There are comparatively few cases in which the
economic burden of armed violence has been
carefully measured using various types of methods.
Part of the reason for this relates to the relative
novelty of the study of the economic costs of
armed violence. Investment in such research can,
however, illustrate the huge economic costs of
conflict and non-conflict armed violence.
This section considers four countries—Uganda,
Sri Lanka, Nicaragua, and Guatemala—for which
there is data that offers important insights into
the applicability of a multi-method approach to
estimating the economic burden of armed violence.
In all four countries, a combination of accounting, modelling (counterfactual), and contingent
valuation approaches were attempted by various
researchers to generate comprehensive estimations of the costs of armed violence for society. It
finds that the economic burden depends in large
part on the duration, severity, and geographic
spread of armed violence, as well as the types of
indicators used and quality of available data. The
section also considers the costs of armed violence
that are often obscured from view, including violence against women (see Box 5.4).
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
WHO region
97
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
98
Box 5.4 The economic costs of intimate partner and
sexual violence during war
Though often hidden from view, the economic costs of intimate
partner and sexual violence against women act as a development
disabler. Although there are no comparative assessments of the
economic burden of intimate partner violence, a number of case
studies exist in high- and medium-income country contexts.
According to UNIFEM (2007) and CDCP (2003), the direct medical
costs of intimate partner assault, rape, and related victimization
amounts to at least USD 5.8 billion per annum, while the indirect
costs total some USD 1.8 billion.
In less-developed countries, particularly those affected by war,
sexual violence directed against women undermines formal and
informal economic productivity. Female single-headed households are often confronted with the pain and suffering related to
missing relatives, as well as economic uncertainties. In many
cases, missing or killed male relatives served as the primary
breadwinners and/or the household property owner. In Chechnya,
(northern) Kenya, Liberia, Nepal, Somalia, and Sri Lanka, a noticeable rise in female-headed households was observed in the wake
of armed violence (CICS, 2005).
In both war and peace, female-headed households face an increased
workload. They also regularly find themselves excluded from formal
economic activities, thus leading to reduced earning options.
Such households make up a disproportionate share of the poor.
In southern Sudan, for example, where women outnumber men,
widow-headed households represent up to 50 per cent of the poor
and poorest quintiles (Burns-Mackenzie and Buchanan-Smith,
2005). In Bosnia and Herzegovina, women-headed households
(16 per cent of all households in 1998) often live in precarious
conditions, with some members resorting to prostitution to make
ends meet (Bastick, Grimm, and Kunz, 2007).
Women also face a range of additional challenges tied to discrimination and social exclusion. The economic condition of displaced
women may further deteriorate due to a decline in access to formal and informal credit from their social networks (Brück and
Vothknecht, 2007).
During bouts of intense collective violence, gender roles can change
and adapt. The protracted absences of male family members and
the destruction of productive assets can force women and girls
into the labour market in new ways. In Nepal and Kenya, for example, large numbers of women are involved in farm management
and labour migration, work traditionally reserved for men. Others
have observed an increase of the share of women in the formal
and informal labour force in Bosnia and Herzegovina, Cambodia,
El Salvador, Georgia, Guatemala, and Rwanda in the aftermath of
war (Brück and Vothknecht, 2007).
Photo A victim of a gang rape by five men hides her
face at the Ndosho centre in central Goma, DRC, 2006.
© Per-Anders Pettersson/Getty Images
More than two decades of armed conflict in Sri
Lanka has stimulated a range of analyses of its
economic consequences.27 Data generated by the
state, the national bank, and international agencies
such as the World Bank allow for robust accounting and modelling. Depending on the period under
consideration, the independent variables assessed,
and the regions of the country that are considered,
the economic costs of collective armed violence
in Sri Lanka range from USD 333 million to USD
1.93 billion per year. These costs are attributed
primarily to lost earnings arising from foregone
foreign investment (42 per cent), military expenditures (27 per cent), lost tourism revenue (10 per
cent), depleted infrastructure (8 per cent), and
other factors (Bozzoli et al., 2008, p. 19).
In the wake of an insurgency in Nicaragua
launched in 1980, several econometric studies
were under taken to examine the extent of the
costs of mass armed violence. 28 Drawing on data
from the Nicaraguan government and international
agencies such as the UN and the World Bank, it
is possible to examine the economic implications
of external embargoes, military expenditures,
and even changes in the behaviour of economic
actors (shifts in propensity to import and consume).
The overall estimated costs of civil war range
from USD 80 million to more than USD 1.1 billion
(FitzGerald, 1987; DiAddario, 1997; Stewart,
Humphreys, and Lea, 1997). The primary impacts
were reported in relation to export revenues, fiscal deficits, and inflation rates, and easily rivalled
official development flows to the country.
More than a decade after a protracted internal
conflict, Guatemala continues to suffer from one
of the highest rates of armed violence in the world.
The UN Development Programme has estimated
that the costs of armed violence amounted to
almost USD 2.4 billion in 2005, or 7.3 per cent of
GDP (UNDP, 2006). The estimate incorporates
health sector costs, institutional costs, private
security expenditures, impacts on the investment
climate, and material losses.
99
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
In Uganda, a protracted armed conflict between
the Lord’s Resistance Army (in the north) and the
Ugandan People’s Defence Forces reveals the
heavy economic costs of mass violence. In examining a range of variables from the mid-1980s to
2002 and data from national and sub-national
sources, it appears that the economic costs of
more than two decades of war in northern Uganda
accounted for at least USD 1.3 billion. These effects
are primarily related to direct military expenditure
(28 per cent), loss of income from cash crops (27
per cent), and reductions in tourism revenue (14
per cent) (Dorsey and Opeitum, 2002). Drawing
on additional variables and country data, the
estimated total rises threefold to more than USD
3.5 billion (Bozzoli et al., 2008).
1
Estimating the global economic
costs of non-conflict armed violence
There are comparatively few attempts to estimate
the global costs of homicidal violence. The Conflict
Analysis Resource Center (CERAC) in Colombia
recently generated a global estimate on the basis
of the lost product due to violent deaths (LPVD) in
more than 90 countries. The use of the LPVD method
highlights the cost of lethal intentional violence
above ‘normal’ or ‘expected’ rates. ‘Normal’ is
defined levels observed in countries with a low
or very low incidence of homicides. 29
The approach first considers the potential gains
in life expectancy that would be achieved by
reducing the risk of violent death. This is represented by the added years of life expectancy the
population would gain if deaths from armed violence were reduced or eliminated. Estimating
2
3
4
5
6
7
100
A R M E D V I O L E N CE
potential gains in life expectancy for a country
requires mid-year population estimates by age
and gender, data on all causes of mortality by
age and gender, an estimate of total homicides,
purchasing power parity (PPP) indices, GDP per
capita, and GDP per capita growth rates.
G LO B A L B U R D E N
of
The assessment generated a range of insights into
the global economic burden of homicidal violence.
Of the more than 400,000 reported homicides in
2004 (from 90 countries), the lost product due to
violent deaths per homicide amounted to USD
85,000–363,000 (2007 US dollars), depending on
the rate at which future earnings are discounted
(ten and three per cent, respectively).30
Box 5.5 What is a discount rate?
A discount rate is the deduction that is applied to
a future value when brought back to the present
in order to make it comparable to current values.
It is equivalent to what a given investment would
yield if put to productive uses.
Overall, the annual lost productivity from lethal
non-conflict armed violence is roughly USD 95
billion per year. These losses could range from
as high as USD 163.3 billion (at a three per cent
discount rate) to as low as USD 38.3 billion (if a
ten per cent discount rate is used). This amounts
Map 5.1 Potential gains in life expectancy in years in the absence of non-conflict armed violence, by country, 2004
0)+)2(
4SXIRXMEPKEMRWMRPMJI
I\TIGXERG]MR]IEVW
1.00–1.81
0.66–1.00
0.42–0.66
0.26–0.42
0.00–0.26
0.00
Male
Female
Not included
Source: CERAC
101
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
Map 5.2 Global lost product due to violent deaths, 2004
0)+)2(
Lost product as a percentage
of 2004 GDP in PPP 2007
(discount rate of 5%)
1.1921–2.1420
0.6331–1.1920
0.2921–0.6330
0.0931–0.2920
0.0000–0.0930
1
2
Not included
3
4
Source: Restrepo et al. (2008)
5
to between 0.03 and 0.14 per cent of global GDP.
Put another way, the annual global LPVD is equivalent to the GDP of Chile, Hungary, or Romania,
and 47 times that of Burundi, in 2004 (Restrepo
et al., 2008).
There are considerable regional variations in the
economic costs of non-conflict armed violence.
North America features the highest loss of life
expectancy and lost economic productivity in
the world. In this region, homicide reduces male
life expectancy by 0.44 years and females by 0.26
years. Likewise, North America experiences the
highest loss of GDP when compared with other
regions, though Latin America and Africa feature
a larger share of GDP lost to homicidal violence.
Indeed, Latin America and the Caribbean region
feature the highest rates of homicide per 100,000
population, and the lost product due to violent
deaths is USD 79,000–304,000 per homicide,
again depending on the discount rate (ten or three
per cent) (Restrepo et al., 2008).
Reviewing homicidal violence in 15 countries also
highlights the national variations in lost productivity. For example, Jamaica, Colombia, Angola,
South Africa, and Bolivia experienced among the
highest homicide rates in 2004.31 Not surprisingly,
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
102
Table 5.3 Lost product due to violent deaths, USD million (2007)
and % of GDP (2004)
Region
Lost product due to violent
deaths, USD million (2007)
Lost product due to violent
deaths as % of GDP (2004)
Discount rate
3%
5%
Figure 5.2 Potential gains in life expectancy (years)
in the absence of violent deaths by region, 2004
World
Americas
Discount rate
10%
3%
5%
Africa
10%
Europe
South-east
Asia
24,540
14,513
5,765
0.03
0.02
0.01
Western
Pacific
35,068
18,510
6,179
0.27
0.14
0.05
North America
46,760
26,756
10,417
0.37
0.21
0.08
Eastern
Mediterranean
1,870
1,236
590
0.12
0.08
0.04
Europe
9,946
5,963
2,513
0.10
0.06
0.03
Africa
6,404
4,771
2,750
0.41
0.30
0.17
38,762
23,694
10,133
1.21
0.74
0.32
Source:
Small Arms Survey and
CERAC calculations
Eastern Mediterranean
Western Pacific
0
0.1
United States
Jamaica experienced the highest potential gains
in life expectancy if such violence had not occurred,
at 1.81 years for males and 1.0 for females, followed
by Colombia, El Salvador, Guatemala, Venezuela,
Bolivia, and Honduras—all of which are in the Latin
American and Caribbean region. The United States
experienced the highest total lost product due to
violent deaths with a value of USD 45.1 billion (at
a discount rate of three per cent), followed by China,
Colombia, Indonesia, India, and the Republic of
Korea (see Figure 5.3 and Table 5.4).
Positive effects of armed violence?
Armed violence generates effects in all directions.
It can result in the loss of capital and opportunity
costs, but can also redistribute wealth and build
0.3
0.4
Discount rate, USD million (2004)
3%
Sources: Small Arms Survey and CERAC calculations
0.2
Table 5.4 Aggregate lost product due to violent
deaths, selected countries, 2004
Country
Latin America
and the
Caribbean
Legend:
Female
Male
South-east Asia
5%
10%
45,112
25,846
10,080
China
24,620
12,992
4,232
Brazil
23,140
13,815
5,656
Indonesia
6,376
3,495
1,092
Colombia
6,231
3,935
1,766
India
6,179
3,803
1,277
Korea
5,586
2,499
607
Thailand
5,503
2,931
964
Nepal
4,723
3,161
1,525
South Africa
4,435
3,289
1,878
Mexico
4,110
2,575
1,155
Philippines
2,247
1,417
629
Canada
1,648
910
338
United Kingdom
1,477
824
308
Russian
Federation
1,358
942
492
Source: Small Arms Survey and CERAC calculations
Jamaica
Colombia
El Salvador
Guatemala
Venezuela
Bolivia
Honduras
Somalia
South Africa
Sudan
Angola
Dominican Republic
Brazil
Liberia
Russian Federation
Legend:
Male Female
Source:
Restrepo et al. (2008)
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
state capacity.32 Certain benefits arising from organized armed violence, including official recruitment
into armies or armed groups and the reshuffling
of opportunities within an unequal society, are
more obvious than others.33 In both cross-border
and civil wars, for example, armed violence can
lead to political and economic transformations that
result in new monopolies, cartels, and other forms
of informal resource accumulation (Cramer, 2006).
Actors who successfully control economic niches
and opportunities often become powerful and may
perpetuate violence to extend their economic
reach. Groups ranging from gangsters and mafia
in the Balkans to warlords in West Africa, Colombia, and Afghanistan may also seek to transform
themselves into legitimate political actors and
find ways of laundering their newly acquired
resources into the formal economy (Keen, 1998;
Reno, 1999). But these gains are offset by losses
in other areas: every resource spent or destroyed
in armed conflict is a resource that would be more
efficiently used for purely productive purposes.
Nevertheless, the motivations, interests, and
outcomes among those who ‘profit’ from armed
violence are still critical to explain the onset and
perpetuation of armed violence.
The dynamics of the informal and illicit economies
are often not captured in formal statistics. The
boundaries between the criminal and the informal
economies are, in some cases, blurred. This blurring is especially significant in lower- and middleincome countries. For example, armed violence
may generate opportunities for involvement in
informal activity, such as narcotics production,
‘conflict diamonds’, or trade in contraband. While
such activities may undermine the legitimacy of
the state, they can also contribute to household
incomes and local markets and spill over into the
formal economy.34
Armed violence may also result in the transfer of
assets from one set of actors to another. The extent to which such redistribution is ‘progressive’
must be carefully scrutinized. When armed groups
with no clear political agendas redistribute the
spoils of conflict, it is likely that armed violence
negatively affects the most vulnerable. It is the elite
who most often benefit from such redistribution
(Collier and Hoeffler, 2004a; 2004b). Though the
costs frequently outweigh the benefits, there may
nevertheless be benign transfers, particularly when
armed violence yields more ‘neutral’ transfers of
goods from wealthy to wealthy or from poor to poor.35
While debate over the potentially positive economic effects of armed violence persists, there
is evidence that intense bursts of violence are
103
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
Figure 5.3 Potential gains in life expectancy (years) in
the absence of violent deaths, top 15 countries, 2004
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
104
bad for an economy. Even the nominal growth
accompanying a ‘post-conflict’ transition is unlikely
to help a society catch up quickly to levels experienced prior to the conflict. The ‘peace dividend’
is more akin to an ‘efficiency dividend,’ as previously inefficiently mobilized resources are redirected to productive ends. After a long guerrilla
war in Uganda culminated in victory in 1986, for
example, Uganda’s National Resistance Movement
presided over the country’s longest period of economic development since independence (Gutierrez,
2008; Mutebi, 2008a). Similarly, the Rwandan
Patriotic Front also presided over sustained economic development after it came to power following the genocide in 1994. More controversially, in
Somaliland, protracted armed violence led to the
formation of a (largely unrecognized) state that
appears to ensure a degree of economic security
for its residents (Gutierrez, 2008; Ahmed and
Box 5.6 Armed violence and investment
Does armed violence universally deter foreign direct investment (FDI)
(Mihalache, 2008)? While armed violence may increase transaction and
transport costs, disrupt labour and commodity markets, and put a company’s
assets and personnel in danger, the relationship is not as straightforward
as it may appear. For example, in Algeria, Eritrea, the Democratic Republic
of the Congo, and Sri Lanka, FDI actually peaked during periods of intense
collective violence (Mihalache, 2008). One of the reasons for this is that
investors are not homogeneous or equally sensitive to risk, including the
risk of armed violence.
The real and relative effects of armed violence on investors will depend
in part on the characteristics of the investor and the nature of the risk.
Characteristics include the scale of physical assets, the expected costs of
an exit strategy, and whether outputs are directed to foreign or domestic
consumption. The energy and natural resource sectors tend to be more
vulnerable to targeted armed violence than the finance, service, telecommunications, or construction sectors. The nature of the risk is tied to the
geographic distribution of armed violence: if a company is based primarily
in a capital, but violence occurs in remote areas of the country, than there
will be comparatively fewer effects on routine operations (Fielding, 2003).
Green, 1999). Ultimately, determining who prevails
after bouts of collective armed violence may inform a country’s possible ‘trajectories’. Some
countries may rise from the ashes, while others
may remain in limbo.
While often neglected, some actors in the private
sector can do well in situations of armed violence.
The assumption that foreign direct and local investment always tumbles in contexts of acute armed
violence does not hold (see Box 5.6) (Mihalache,
2008). It is often the case, however, that the private sector profits due to monopolies or inefficiencies. But it is important to recall that the private
sector is both heterogeneous and expanding
rapidly in lower- and middle-income countries.
In some instances, extractive and smaller-scale
companies can rapidly develop specific niches in
societies severely compromised by armed violence
(Ballentine and Sherman, 2003). For example, in
Guatemala, high profits were accumulated in the
post-war period by a modest number of business
elites who effectively secured rents through tight
connections with the government (Joras, 2007).
In some other countries, however, due in large
part to the opportunity costs and uncertainties
generated by armed violence, companies can
also help broker peace, such as in South Africa
or Northern Ireland.36
The effects of armed violence on the business
climate—whether due to homicide, kidnapping,
extortion and the destruction of physical infrastructure or in relation to the prospects of international
sanctions and heavily conditioned loans and
credits—are severe. In some cases, these economic
costs may render active or tacit complicity in armed
violence unbearable. In such environments, private
sector actors may support preventive initiatives
and mobilize networks to reach out to national
stakeholders. In El Salvador, Brazil, Colombia,
and Nicaragua, the private sector can play or is
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
105
1
playing a role in promoting an end to armed conflict and criminal violence (UNODC, 2007a).37
Conclusion
Agreement on the meaning and use of different
approaches to measuring the economic costs of
such violence is a core priority. Drawing on accounting, modelling, and contingent valuation to understand the economic burden of armed violence are
important steps forward. Designing and investing
in reliable data collection and analysis tools to
monitor and measure these costs is another critical pillar to generate consensus on priorities, entry
points for action, and benchmarks of success.
Developing comprehensive longitudinal assessments of the economic burden of armed violence
requires the generation of reliable and continuous
data. Current datasets are highly dispersed, piecemeal, and poorly funded. Investment in data
generation must be commensurate with the real
challenges on the ground. The development of a
network of standardized information gathering
mechanisms on armed conflict and criminal violence and the pooling of data for public use are
of clear value.
Quantifying the costs of armed violence is critical
to draw attention to the way such violence impedes
development. While this will not by itself improve
the livelihoods of those affected, a better understanding of the factors contributing to armed violence onset and severity; the temporal, demographic
and spatial relationships between armed violence
and human development; the role and motivations
of armed violence entrepreneurs; and the ways in
which armed violence affects growth can potentially
enhance preventive and reduction initiatives.
Photo ! Police officials
patrol the impoverished
neighbourhood of El
Milagro, Guatemala City,
2004. © Rodrigo Abd/
AP Photo
2
3
4
5
6
7
Abbreviations
BRL
Brazilian real
CERAC
Conflict Analysis Resource Center
foreign direct investment
gross domestic product
JMD
Jamaican dollar
LPVD
lost product due to violent deaths
PPP
purchasing power parity
THB
Thai baht
of
FDI
GPD
USD
United States dollar
G LO B A L B U R D E N
A R M E D V I O L E N CE
106
WHO
World Health Organization
Endnotes
1
In situations of war or organized crime, the poor often
have the opportunity to join the ranks of fighters or private security agents. Enlistment may be an opportunity
for upward social mobility and the acquisition of status
(Small Arms Survey, 2006, pp. 189–213).
2
In a random effects model, a 2.53 per cent decrease per
year was detected (Restrepo et al., 2008).
3
For more information on the Geneva Declaration, see
<http://www.genevadeclaration.org>.
4
The wide discrepancies in estimates are often due to noncomparable cost factors, different time periods of analysis,
distinct ‘contexts’ shaping collective or interpersonal
violence, and different levels of analysis (from the international to the local level) (Sköns, 2006, pp. 172–73).
5
In some cases, studies also account for international
spillover effects, as well as long-lasting consequences
(Murdoch and Sandler, 2004; Arunatilake, Jayasuriya,
and Kelegama, 2001). More advanced assessments also
seek to account for the role of the informal economies of
the countries concerned, which are often left out of official
GDP estimates (Bozzoli et al., 2008).
6
In Africa alone, the cost of conflict is estimated at USD
284 billion (1990–2005) and approximately 15 per cent of
continental GDP (Oxfam-GB, 2007).
7
Single case studies tend to lack comparable and consistent frameworks and contain inconsistencies caused by
double counting and latent biases (Bozzoli et al., 2008).
Cross-country studies tend to draw on conventional
econometrics and do not sufficiently account for different
types of conflicts.
8
This ‘creative destruction’ or ‘phoenix factor’ resulted from,
it was argued, enhanced state control over key industries,
replacement of obsolescent capacities with more efficient
infrastructure, technological innovation, and other factors
(Sombart, 1913).
9
FitzGerald’s (1987) analysis on the US-backed destabilization of Nicaragua by the right-wing Contras is considered
by experts to be the first contemporary analysis of the
economic costs of mass violence. The assessment by
Stewart and FitzGerald (2001) is also the first comprehensive
account of the relationships between mass violence and
economic development in non-Organisation for Economic
Co-operation and Development countries. Considering the
economic costs at the macro-, meso-, and micro-levels, they
assess the impacts of armed violence in relation to average
rates of income, share in the agricultural subsistence sector,
foreign exchange effects, flexibility of the economic system,
monetary aspects of poverty, education and literacy, health
and nutrition, coping strategies, and other factors.
10
Although some of these studies extend beyond GDP and
government revenues, key assessments include Grobar
and Gnanaselvam (1993), Stewart, Humphreys, and Lea
(1997), Stewart, Huang, and Wang (2001), Hess (2003),
Hoeffler and Reynal-Querol (2003), and Chen, Loayza, and
Reynal-Querol (2007).
11
The various approaches are potentially connected. The
modelling approach should provide a statistical estimate
of what a society has lost economically as a result of armed
violence. Contingent valuation should, in turn, inform the
accounting approach in identifying potential imbalances.
Where there may be major differences in the outcomes of
the two approaches, the accounting specialists may need to
refine their core variables. Applying all of these approaches
together helps to elaborate a more sophisticated assessment by emphasizing the ways in which armed violence
affects different sectors of society.
12
A recent study by the Institute for Applied Research found
that the estimated cost of violence in Brazil amounted to
more than USD 56.5 billion (BRL 92.2 billion), of which
roughly one-third was linked to public sector expenditures and the remainder tied to tangible and intangible
costs paid by the private sector (Cerquiera et al., 2007).
13
By way of comparison, the annual costs of road accidents
in Latin America and the Caribbean (including Brazil) are
estimated to be about one per cent of GNP (Butchart et al.,
2008).
14
A further USD 17.1 million was attributed to self-directed
violence.
15
These medical costs appear relatively consistent with those
of other developing countries, e.g. El Salvador and South
Africa (Small Arms Survey, 2006, p. 196).
Specifically, the health and years lost in Guatemala
amounted to an estimated three per cent of GDP. Institutional costs relating to police and justice provision were
between one and two per cent of GDP. Costs associated
with private security amounted to between two and three
per cent of GDP, while foregone tourism ranged from 0.2
to one per cent (UNDP, 2006, p. 11; 2007).
17
Collier (1999) found that the annual growth rate is reduced
by 2.2 per cent using a sample of 92 countries for the
period 1960–89. Hoeffler and Reynal-Querol (2003) draw
on data for more than 200 countries (for the period 1960–90)
and note a reduction of 2 per cent.
18
This rises to as high as 2.53 per cent, if random effects are
taken into consideration (Restrepo et al., 2008).
19
This figure assumes certain temporal parameters relating
to conflict and post-conflict duration. Collier and Hoeffler
(2004a; 2004b) define these parameters as 7 ‘war years’
and 14 ‘post-conflict years’.
20
Soares (2006), for example, assumes that violence affects
life expectancy. The extension of one’s expected lifetime
by a small amount yields a marginal utility benefit that
can be measured. The equivalent consumption value to
achieve this benefit can be calculated.
21
22
23
Data on the number and cause of deaths is derived from
WHO statistics in order to determine the age-specific reduction in survival probabilities. Contingent valuation is than
applied in order to estimate the monetary value of the reductions in survival probabilities for individuals at a given age.
The value of violence reductions is described by Soares
(2006) as ‘the marginal willingness to pay of an 18-yearold individual, as the lifetime aggregate social value for
the current population and for the future generations, and
as the sum of these aggregate values as percentages of
the 1995 GDP’ (Soares, 2006, p. 830).
Crime and justice expenditures are expected to amount
to 2.1 per cent of GDP per annum in the United States and
3.6 per cent in Latin America (Londoño and Guerrero, 1999).
24
These include Colombia (281 per cent), followed by the
Philippines (280 per cent), Venezuela (95 per cent), Chile
(86 per cent), El Salvador (73 per cent), Belize (71 per cent),
Suriname (67 per cent), Mexico (67 per cent) and Brazil
(65 per cent) (Soares 2006).
25
Homicidal violence generates extraordinary welfare costs
across countries: from Colombia, where it contributes to
the reduction of 2.2 years in life expectancy at birth, to
the United States and Western Europe, where violence
reduces life expectancy at birth by 0.3 and 0.1 years,
respectively (Soares, 2006).
26
In Western Europe, on the other hand, the average social
value of violence eradication measured in terms of yearly
income corresponds to only 0.24 per cent of the 1995 GDP.
These findings do not necessarily imply that additional
expenditures on armed violence reduction should be
pursued. As Soares (2006, p. 829) notes, ‘the desirability
of increased investments in public safety depends on
whether further reductions in violence can be achieved at
a cost lower than the social willingness to pay’.
27
See, for example, Grobar and Gnanaselvam (1993); Harris
(1997); Richardson and Samarsinghe (1991); Kelegama
(1999); Arunatilake Jayasuriya, and Kelegama (2001).
28
The first study was undertaken by FitzGerald (1987) as
evidence for a case brought to the International Court of
Justice. A second study was done by DiAddario (1997)
and was later supplemented by others.
29
‘Normal’ levels are defined as the average homicide rate
of two groups of countries (classified as having low-level
and very low-level rates of homicide) that report mortality
statistics to the WHO. The average homicide rate for these
27 countries was 1.24 per 100,000 population in 2004.
30
The total number of homicides used to calculate this—
449,865—is about ten per cent lower than the figure
presented in the chapter on non-conflict armed violence
(NON-CONFLICT ARMED VIOLENCE), but is drawn from the only
country-level data that is available (Restrepo et al., 2008).
31
Measured as per 100,000 population, homicides rates
were as follows: Jamaica (53), Colombia (48), Angola (47),
South Africa (36), and Bolivia (43) (Restrepo et al., 2008).
32
Tilly (1992; 2003) describes the pro-growth and inherently
developmental functions of collective armed violence in
the growth of modern European states.
33
As Keynes (1978) notes, wars do not only produce destruction, but also change the baseline. Keynesian policies
emphasize military production, infrastructure and transport
construction, technological innovation, the transformation
of women’s roles in the labour market, and the overall
strengthening of the state.
34
Natural resources played an important function in financing a number of African conflicts, as evidenced in recent
reports of the UN Security Council Sanctions Monitoring
Mechanism in Angola, the Democratic Republic of the
Congo, and Sierra Leone (Smillie, Gberie, and Hazleton,
2000). But such resources were only one of many financing options for armed violence (Jean and Rufin, 2006). As
such, interventions focusing exclusively on single commodities through sanctions or even multistakeholder initiatives,
such as the Kimberley Process and anti-terrorist financing,
may only address one part of the problem.
35
Gutierrez (2008) observes that a distributional analysis
should take into account (1) programming effects (i.e. the
take-over of assets, income, or political rights of adversaries
107
E CO N O M I C CO S T S O F A R M E D V I O L E N C E
16
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
108
or third parties); (2) incentive systems (i.e. the distribution
of prizes and punishments to mobilize different sectors of
the population); (3) patterns of action (i.e. the possibilities
for massive redistribution); (4) organizational structures
(i.e. the type of organization can have long-term implications for patterns of redistribution); and (5) baselines (i.e.
the ability to recruit and promote high-risk collective action
depends on disaffected and/or risk-prone critical mass)
(Gutierrez, 2008, p. 16).
36
See, for example, Portland Trust (2007); Ben-Porat (2005);
Charney (1999); Wennmann (2007).
37
See, for example, UNODC (2007a), which describes how,
throughout Central America, crime and corruption are
considered leading problems for business leaders. More
than 80 per cent of 455 Guatemalan businesses polled
said they saw crime as a major problem, as compared to
the global weighted average of 23 per cent (UNODC,
2007a, p. 18).
Armed Violence Against Women
A
rmed violence affects women, men,
girls, and boys in different ways—as both
perpetrators and targets of armed violence.
Across cultures, most acts of violence are committed by men, and men and boys also account
for the majority of firearm-related deaths and
injuries. In Rio de Janeiro, for example, young
men are 24 times more likely than women to be
killed by armed violence, while men between the
ages of 15 and 29 are twice as likely to die from
armed violence as the rest of the male population
(Dreyfus et al., 2003; CICS, 2005, p. 14).
The present report has focused on the main indicators used to capture and quantify the burden
of armed violence and its impact on development,
including homicide, direct conflict deaths, indirect conflict deaths, and economic costs. While
these indicators provide valuable information on
the burden of armed violence at the population
level, they are limited when we turn our attention
to women’s experience of armed violence.
Women and girls are affected by armed violence
in different ways, including by direct and indirect
conflict violence, and by lethal and non-lethal
non-conflict violence. They are also more likely
than men and boys to die through intimate partner
violence (IPV). The World Health Organization
(WHO) suggests that 40–70 per cent of all female
homicides are committed by an intimate partner
(Krug et al., 2002, p. 93).
A number of forms of gender-based violence specifically target women and girls because of their
sex; this chapter refers to them as ‘violence
against women’. Such violence, including rape,
domestic violence, murder, and sexual abuse, is
a significant cause of female mortality and a leading cause of injury for women aged 15 to 44 years
(UNIFEM, 2007). The severe impact of violence
against women has prompted the United Nations
Development Fund for Women (UNIFEM) to describe
it as ‘a universal problem of epidemic proportions’
(UNIFEM, n.d.). And violence against women, in
its many forms, is responsible for ‘more than 100
million missing women’ due to ill-treatment, lack
of access to food and health care, and genderbased violence such as female infanticide and
sex-selective abortion (Sen, 1990).
This chapter examines some of the specific gender
dimensions of the global burden of armed violence
and provides an overview of forms of violence
specifically directed at women. It complements
the examination of gender issues in previous
chapters. Its main findings are:
The majority of victims of IPV are women, and
IPV is the most common form of violence
against women.
A number of forms of violence specifically
target women and have significant physical,
psychological, social, and economic costs.
Data collected on violence against women is
sparse and unsystematic; significant investments in improved data collection and analysis
should be made.
ARMED VIOLENCE AGAINST WOMEN
109
Chapter Six
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
110
A gender approach broadens conventional
understandings of ‘arms’ and ‘armed violence’
because conventional definitions provide only
a partial picture of how women experience
armed violence.
Photo " This 35-year-old
woman was shot in the
face. The bullet entered
her eye and exited behind
her ear. Abunduruk,
Sudan, near the Chadian
border. © Lynsey Addario
To assess the burden of armed violence on women,
it is more useful to focus on the broader question
of violence against women rather than armed
violence in order to understand broader patterns
of violence ranging from the abuse of women in
intimate partner violence to the impact of armed
conflict on women.
The gender dimensions of
armed violence
Different experiences of armed violence are determined by gender roles. Gender (as opposed to
sex) refers to the construction of social roles that
operate through various mechanisms, such as
institutions or stereotypes. Gender constructions
reflect deeply rooted relations of power and determine the roles, behaviour, values, and relationships associated with masculinity and femininity.
These are the roles and behaviour that a man or
a woman is expected to adopt in a given setting.
These roles vary between and within different cultures and are learnt behaviour acquired through
socialization (Connell, 1995, p. 44).
A gender approach is useful to account for the
different ways in which armed violence affects
women and men. Focusing on gender rather than
women allows one to include gender-based violence
against men and boys as well as gay, lesbian, transgender, and transsexual people. This is important
because violence is not only used by men to claim
and assert power over women, but it is also instrumental in enforcing the gender hierarchy of power
among men. A gender-sensitive approach highlights the power relations inherent in much armed
violence. Finally, such an approach does not limit
women to the role of victims and men to the role
of perpetrators, since it recognizes that women
can also be the perpetrators of armed violence,
while men are also among the victims.
Gendered power relations and forms of violent
masculinities are key underlying factors shaping
the dynamics of armed violence. In many societies, violence and weapons use by boys and
men are socially expected or accepted (Widmer,
Barker, and Buchanan, 2006). Boys are socialized
into violent behaviour through weapon-related
rites of passage from boyhood to manhood
There is, however, not just one form of masculinity
and femininity in any given society: different types
of masculinities exist and are interlinked by relationships of power, hierarchy, and exclusion. The
hierarchy of different forms of masculinity is an
important source of conflict and violence among
men, as challenges to one’s masculinity are common sources of disputes and injuries or even
murder (Connell, 2003, pp. 1–2). Gang turf wars,
for example, are often linked to honour-related
issues and challenges to ‘status’, ‘toughness’,
and ‘manhood’.
Perceived threats to one’s masculinity can also
arise from economic dislocation, unemployment,
or social transformations. In societies where the
male gender role is intricately tied to being the
main ‘breadwinner’, unemployment can leave
men feeling ‘emasculated’ and powerless, and
wanting to demonstrate that they are ‘real men’
(Widmer, Barker, and Buchanan, 2006, p. 3). The
resort to armed violence is often linked to a crisis
of masculinity and a ‘fear of loss of power and
privilege’ (Messner, 1990) through social transformations. Weapons can also be used as status
symbols, as tools to achieve economic and social
gain, or to acquire power over unarmed persons
in order to reassert one’s masculinity (Myrttinen,
2003, p. 37).
Although men are the main perpetrators of acts
of armed violence, women and children also use
armed violence (Bennett, Bexley, and Warnock,
1995). During the armed conflict in El Salvador,
for example, women held 40 per cent of leadership
and 30 per cent of combatant roles (Schroeder,
2005, p. 1), and women and girls are involved in
gangs in Haiti (OTHER FORMS OF ARMED VIOLENCE).
In the armed conflict in Liberia, child soldiers as
young as nine years old reportedly committed
killings and atrocities often under the influence
of drugs and alcohol used to induce aggression
and suppress fear (HRW, 2004, pp. 2–3).
Gender roles influence not only who perpetrates
armed violence but also who becomes the victim.
This is especially so with gender-based violence,
‘an umbrella term for any harm that is perpetrated
against a person’s will, and that results from power
inequities that are based on gender roles’ (RHRC,
2003, p. 9). Gender-based violence may be physical, sexual, psychological, economic, or sociocultural, such as intimate partner violence, sexual
assault, honour killings, dowry-related violence,
or trafficking. Categories of perpetrators include
intimate partners, family members, community
members, and those acting on behalf of cultural,
religious, or state actors.
The distinction between victim and perpetrator
of gender-based violence does not necessarily
follow gender fault lines: while men are the main
perpetrators, women also commit acts of genderbased violence, and even though women are the
main victims, men, boys, and transgender/transsexual people are also among the victims. Forms
of gender-based violence specifically directed
against men include sex-selective killings, forced
conscription, and sexual violence (Carpenter, 2006).
For example, in the armed conflicts of the Central
African Republic and the Democratic Republic of the
Congo (DRC), numerous cases of sexual violence
against men and boys were reported (INDIRECT
CONFLICT DEATHS). Among non-combatants in
the former Yugoslavia, adult civilian men were
the most likely to be massacred by enemy forces
(Carpenter, 2003). Such sex-selective killings of
111
ARMED VIOLENCE AGAINST WOMEN
(Myrttinen, 2003, p. 38). The media and popular
culture often link violence, arms, and masculinity,
reinforcing images of conventional gender roles.
In some cases violence becomes an expression
of masculinity.
1
2
3
4
5
6
7
‘He was very angry and he took his Kalashnikov
112
A R M E D V I O L E N CE
. . . The neighbours said: “Leave her alone” . . .
But then he didn’t stop, he shot my legs, I
could not feel them, they were numb, the sun
was setting, I was looking at the sky, I said to
the men: “I don’t want to die.” They took me
to the hospital.’
of
— Fatima, 19 years old, shot by her husband in Iraq in 2003
G LO B A L B U R D E N
(AI, 2004)
Photo ! This woman was
only 13 years old when
she was raped during the
1992–95 war in Bosnia.
© Robin Hammond/
Panos Pictures
men are rooted in the assumptions of male wartime roles, reproducing gendered hierarchies
(Carpenter, 2006, pp. 88–89).
The experience of armed violence is influenced not
only by gender but also by other factors, such as
age, race, ethnicity, class, or religion. During the
civil war in Guatemala, for example, women and
children of ethnic Mayan origin were specifically
targeted (Commission for Historical Clarification,
1999, §85–88, §91). In the Rwandan genocide,
sex-selective killings targeted specifically Tutsi
men, whereas Tutsi women frequently became
the victims of sexual violence (Carpenter, 2006,
p. 89; Ward, 2002; HRW, 1996).
Acts of gender-based violence do not necessarily
involve the use of weapons. However, arms are
often directly or indirectly linked to violence,
either through their presence or as the indirect
consequences of armed violence. Surveys have
shown, for example, that the presence of a gun in
the household generally increases threefold the
risk of becoming a homicide victim (Kellermann
et al., 1993).
Attitudes and roles shaped by armed violence,
for example through army training or the experiences of war, also contribute to gender-based
violence. A study on domestic violence in Bosnia
and Herzegovina shows that men returning from
war face a ‘masculinity crisis’, which increases
the likelihood of male violence and the abuse of
women (CARE and ICRW, 2007, p. 8).
Violence against women in
conflict settings
Women also die on the battlefield as combatants
or members of fighting forces. Women have actively
participated in armed conflicts in at least 57 countries since 1990 (Williams, 2005), including in
Chechnya, El Salvador, Eritrea, Ethiopia, Liberia,
Mozambique, Nepal, Nicaragua, Sierra Leone,
South Africa, Sri Lanka, and Uganda (Barth, 2002;
Peimani, 2004; McKay and Mazurana, 2004, pp.
21–23). In most cases, little data is available on
the proportion of female combatants in armed
forces or armed groups.
The percentage of female soldiers in NATO countries’ armed forces varied between 0.5 and 20 per
cent in 2005 and 2006 (Office on Women in the
Fatality figures for armed groups with a high
representation of women are not often available.
However, during Eritrea’s war of independence,
historians estimated that one-third of the 65,000
combat fatalities were women (Clodfelter, 2002,
p. 612). The Iraq Coalition Casualty Count reports,
as of 2 August 2008, 108 female fatalities among
the Coalition Forces (including US forces) in Iraq,
representing 2.4 per cent of a total 4,452 causalities. With increased gender equality in many
armed forces, more women will be deployed to
war theatres and the share of female combatant
battle deaths may be expected to increase.
While more men get killed on the battlefield,
women and children are often disproportionately
targets of other forms of potentially lethal violence during and after conflict. These include
sexual violence, secondary violence against survivors of sexual violence (such as honour killings),
and death from pregnancies or sexually transmitted infections resulting from rape. The scope
and nature of the violence vary tremendously
between conflicts (Wood, 2006).
Women and girls are also likely to disproportionately suffer from the indirect consequences of
armed conflict, such as reduced access to food,
clean water, and health care (Plümper and Neumayer,
2006; Ghoborah, Huth, and Russett, 2003, p. 189).
This leads both to indirect conflict deaths of women
113
and girls, and to women and girls bearing the
burden of others’ deaths and injuries, destroyed
infrastructure, and the breakdown of law and
order.
This gendered burden is often neglected in assessments of the impact of armed conflict. It is not
reflected either in conflict or battle death figures
or in narrow calculations of costs of armed conflict to the economy Little quantitative evidence
is available. This section therefore looks at various health-related and socio-economic aspects
of the gendered burden of armed conflict on
women to highlight some areas relevant to the
global burden of armed violence.
Photo " FARC soldiers
march in a military
parade at the main
square of San Vicente del
Caguan, Colombia, 2001.
© Ricardo Mazalan/
AP Photo
ARMED VIOLENCE AGAINST WOMEN
NATO Forces and The Women’s Research &
Education Institute, n.d.). In non-state armed
groups, however, the proportions can be much
higher. More than 30 per cent of the fighters in
the following non-state armed groups were
observed to be women: the Liberation Tigers of
Tamil Eelam (LTTE), the Communist Party of Nepal–
Maoists, the Fuerzas Armadas Revolucionarias
de Colombia (FARC), and the Sandinista National
Liberation Front.1
1
2
3
4
5
6
7
114
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
The impact of armed conflict on women’s health
is difficult to ascertain as relevant and reliable
data is sparse in developing countries and even
less available during armed conflict. Mortality
studies tend to focus on age groups rather than male
and female mortality. For example, an analysis of
37 datasets on conflict-related mortality (GuhaSapir and von Panhuis, 2004) compares the risk
of dying for children younger than five years old
and persons older than five years during armed
conflict with the pre-conflict risk. Findings show
very high vulnerability for children under five, and
increased mortality due to diarrhoeal diseases,
severe malnutrition, respiratory infections, and
measles.
Maternal mortality, defined as the annual number
of deaths from pregnancy-related causes per
100,000 live births, is a good indicator of women’s
health condition, and can be used to assess the
indirect impact of armed conflict on women.2 The
findings in Table 6.1, from a study assessing the
impact of armed conflict on maternal mortality and
under-five mortality, showed increased maternal
and under-five mortality rates in countries that
had recently experienced armed conflict (O’Hare
and Southall, 2007). The maternal and under-five
mortality rates are both 44 per cent higher than
the baseline rates.
Table 6.1 Comparison of maternal mortality and under-five mortality in
42 sub-Saharan countries
Mortality rates
Countries with recent
armed conflict
Countries without recent
armed conflict
Maternal mortality rate
(median)
1,000/100,000 births
690/100,000 births
Under-five mortality rate
(median)
197/1,000 live births
137/1,000 live births
Source: O’Hare and Southall (2007). The study covered 42 sub-Saharan countries, of which 21 have
experienced armed conflict since 1990.
Box 6.1 Armed conflict and HIV/AIDS
A common assumption is that armed conflict
increases HIV infections, and that refugees and
internally displaced people are particularly at
risk and likely to experience a higher incidence
of HIV infections. This assumption has been
fuelled by increased reporting on widespread
rape of women and girls during armed conflict
and high levels of HIV/AIDS in some armed
groups. However, the findings of a recent study
by UNHCR and the University of Copenhagen on
the incidence of HIV infections among conflictaffected and displaced people in seven subSaharan African countries3 could not confirm
these assumptions at the population level due
to insufficient data (Spiegel et al., 2007). Further
research is thus needed.
News and NGO reports on the armed conflicts in
Darfur or the DRC frequently refer to horrifying
stories of sexual violence, especially rape, against
women and girls. Data on the scope and magnitude
of sexual violence is, however, scarce, making it
impossible to estimate the overall extent of sexual
violence in armed conflicts (INDIRECT CONFLICT
DEATHS, Box 2.2). Evidence from a WHO survey
on women’s experience of violence during and
after the conflict in Liberia found that 81.6 per cent
of 1,216 randomly selected women and girls had
been subjected to one or multiple violent acts
during and after the conflict. The most commonly
reported violent acts were detention against a
woman’s will, being threatened by a weapon,
beating, kicking, and rape (of which more than
70 per cent were gang rapes) (Omanyondo, 2005).
Beyond battle, armed conflict has many disruptive
consequences for women’s lives. Women carry
the burden of family displacement and of becoming
the sole breadwinner when male relatives join
fighting forces, are detained, are taken hostage,
115
ARMED VIOLENCE AGAINST WOMEN
go missing, or are killed. In such situations, women
take on additional responsibilities of income
generation and of caring for their children and
wounded, disabled, sick, and elderly people.
Women face these challenges in environments
that are not only war-torn but contain social and
legal obstacles that may seriously hamper women’s
livelihood and other opportunities.
Discrimination against women and gender inequality are the main reasons why ‘indirect negative
consequences on health and mortality are likely
to affect men and women differently’ (Plümper
and Neumayer, 2006). In situations of scarce
resources and deteriorated health services, the
lower socio-economic status of women and girls
exacerbates the negative consequences of armed
conflict for women’s health.
1
Non-conflict violence
against women
The UN General Assembly’s Declaration on the
Elimination of Violence against Women (1993)
defines violence against women as:
any act of gender-based violence that results in,
or is likely to result in, physical, sexual or psychological harm or suffering to women, including
threats of such acts, coercion or arbitrary deprivation of liberty, whether occurring in public or
in private life.
Article 2 of the Declaration makes clear that violence against women takes many forms:
2
3
genital mutilation and other traditional practices
Photo ! A Tamil woman
harmful to women, non-spousal violence and
stays at her war-torn
violence related to exploitation;
physical, sexual and psychological violence
occurring within the general community, including rape, sexual abuse, sexual harassment and
intimidation at work, in educational institutions
and elsewhere, trafficking in women and forced
prostitution;
physical, sexual and psychological violence
perpetrated or condoned by the State, wherever
it occurs. (UNGA, 1993)
physical, sexual and psychological violence
occurring in the family, including battering, sexual abuse of female children in the household,
dowry-related violence, marital rape, female
While men are the main perpetrators of violence
against women, women also commit such violence: female infanticide, for example, is often
house in Jaffna, Sri Lanka.
© Q. Sakamaki/Redux
4
5
6
7
116
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Box 6.2 The costs of violence against women
Violence against women, like all forms of violence, creates a wide range of
economic and development costs, some direct and some indirect. Yet the
true cost of this violence remains unknown.
Most attention has been paid to the costs of intimate partner or domestic
violence in developed countries, where attention to the issue of violence
against women is greatest. Table 6.2 summarizes the findings of several
different studies that used various definitions and methods. It cannot be
used to make comparisons, but it does highlight the potential scope of the
socio-economic costs that violence against women imposes on communities and societies.
Table 6.2 Selected studies on costs of intimate partner violence and/or
domestic violence
Country
Year
Area of
study
Categories
analysed
Costs (USD)
Australia
2002–03
National
Health, production,
consumption,
administration,
second-generation
costs
6.1 billion
(excluding pain
and suffering)
Canada
2002
National
Direct medical
1.1 billion
Chile
1999
310 women
in Santiago
Lost productivity
1.7 billion
Colombia
2003
National
Prevent, detect,
and offer services
to survivors of
family violence
73.7 million
Netherlands
1997
National
Direct medical,
costs of legal
services, costs of
incarceration, other
monetary costs,
costs of policing
142.2 million
USA
2002
National
Legal and medical
services, judicial
system costs and
lost productivity
12.6 billion
Sources: Australia: Access Economics (2004); Colombia: Sánchez et al. (2004); all others: Waters
et al. (2004)
practised by women. Despite its variety, violence
against women is a manifestation of unequal
power relations between men and women—an
asymmetric relationship that is also reflected in
the lower social and economic status of women
in many cultures and societies across all regions.
Gender-based violence does not necessarily
involve physical strength or armed violence but
can nevertheless be lethal. It does not always
involve ‘arms’ as conventionally defined, but can
involve tools that are turned into arms for the
purpose of violence against women. This includes
such things as the use of acid in attacks, the practice of sex-selective abortion, or female infanticide. Examining these different forms of violence
against women forces us to broaden our understanding of ‘arms’ and ‘armed violence’, since
conventional definitions often only partially account
for women’s experience of violence.
Even when it is not lethal, violence against women—
especially such forms as sexual violence in conflict (INDIRECT CONFLICT DEATHS)—can have severe
and long-lasting health (physical and psychological) and socio-economic consequences for the
victims. Beyond the impacts on the individual
survivor, violence against women also has serious
consequences for the family and the community
of the victim, and for society as a whole. Victims
are often unable to care for their families, which
has serious implications in societies with weak
social and support services. In addition, violence
against women affects the productivity of women
and represents a considerable burden on the health
system. There are as yet, however, no good crossnational studies that demonstrate systematically
the scope and scale of these consequences.
Despite its grave consequences, violence against
women often goes unreported and remains hidden
from view. Comprehensive and comparative sex-
117
Often tolerated as part of cultural or historical
tradition, sexual violence tends to be improperly
reflected in victimization surveys and datasets,
and such datasets often do not contain sexdisaggregated data. For instance, reliable data
on homicide of women is still rare. While accurate
data is available for certain subregions, for many
regions—especially Africa—data is either nonexistent or incomplete. International efforts to
improve our understanding of violence against
women are, however, under way (Johnson, Ollus,
and Nevala, 2008).
Intimate partner violence
Intimate partner violence is the most common
form of violence against women, and the majority
of its victims are women (Krug et al., 2002, p. 89).
IPV, also known as ‘domestic violence’, is perpetrated by a current or former intimate partner or
spouse. It can take many forms, both lethal and
non-lethal, including acts of physical aggression—
such as slapping, battering, hitting, kicking, and
beating—or psychological abuse—such as intimidation and humiliation. Intimidation can be such
that the victim does not search for help or report
domestic violence, but rather endures an ongoing
abusive relationship. It has been estimated that
ARMED VIOLENCE AGAINST WOMEN
disaggregated data is still not available for most
forms of violence against women. For example,
studies of violent and coerced sex by intimate or
non-intimate partners are rare. Intimidation and
the taboo and stigma attached to violence against
women prevent victims from reporting such crimes,
which leads to a high rate of under-reporting,
including in official crime statistics. In many
countries incidents remain unreported because
victims fear the consequences of the perceived
‘soiling’ of the family honour (UNIFEM, 2007).
it takes as many as 35–37 repeated incidents over
an average period of seven years before women
report IPV to an agency (Hall and Wright, 2003).
Photo ! Three years
While gender-based violence committed by strangers is considered a crime in many countries,
intimate partner violence is often regarded as a
‘private matter’ and therefore not adequately
reported and penalized. Crimes of IPV against
men are even less reported, as the stigma for
men is even higher than for women. Studies on
the relationship between small arms availability
and intimate partner violence show that, even
without the direct use of armed violence, intimate
partner violence can be linked to the presence
of arms).3
apartment with her ex-
Studies on IPV have been conducted in 71 countries, according to the UN General Assembly’s
study on all forms of violence against women.
For each year, between 13 and 61 per cent of the
women interviewed reported being physically
assaulted by an intimate male partner at some
after her divorce, this
1
victim of domestic
violence still shares an
husband. Yekaterinburg,
Russia. © Olivia Arthur/
Magnum Photos
2
3
4
5
6
7
118
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
point in their lives (UNGA, 2006, §114). Often,
IPV is not an isolated event, but includes multiple
acts of aggression over a long period of time. There
appears to be no difference in the prevalence of
IPV between high, middle, and low-income countries: women in developed countries are as much
exposed to IPV as are women in less developed
countries (García-Moreno et al., 2005, pp. 27–41,
83–84).
IPV often involves sexual violence (Krug et al.,
2002, p. 151). Figure 6.1 shows the percentage of
Figure 6.1 Percentage of surveyed women reporting on IPV, selected cases
Bangladesh city (2001)
Bangladesh province (2001)
women experiencing any form of sexual violence
by a current or former spouse or partner in selected
countries. Although reporting rates vary widely,
the incidence of any form of sexual violence ranges
from less than 5 per cent to more than 50 per cent.5
The International Violence against Women Survey
conducted in a number of countries found varied
experience of intimate partners using a gun or a
knife among the women interviewed. In Hong Kong,
the Philippines, and Switzerland one per cent of
the women interviewed reported such an experience, two per cent in Denmark and Mozambique,
three per cent in the Czech Republic and Poland,
five per cent in Australia, and up to eight per cent
in Costa Rica (Johnson, Ollus, and Nevala, 2008,
pp. 44–45).
Botswana (2000)
Brazil city (2000–01)
Brazil province (2000–01)
Sexual violence
Canada (1993)
Ethiopia province (2002)
Sexual violence is a form of gender-based violence that occurs in many different settings, with
a variety of motives, perpetrators, and victims.
Sexual violence is commonly defined as:
Japan city (2000–01)
Lesotho (2000)
Mozambique (2002)
Namibia (2002)
Nigeria (1998)
Peru city (2000)
any sexual act, attempt to obtain a sexual act,
Peru province (2000)
unwanted sexual comments or advances, or acts
Samoa (2000)
Serbia and Montenegro city (2000)
to traffic a person’s sexuality, using coercion,
South Africa (2000)
threats of harm or physical force, by any person
Swaziland (2000)
regardless of relationship to the victim, in any
Tanzania city (2001)
setting, including but not limited to home and
Tanzania province (2001)
work. (IASC, 2005, p. 8)
Thailand city (2000)
Thailand province (2000)
Uganda (2000)
Zambia (2000)
Zimbabwe (1996)
Percentage
0
10
20
30
40
50
Sources: García-Moreno et al. (2005, p. 28); Naudé, Prinsloo, and Ladikos (2006, p. 107); FederalProvincial-Territorial Ministers Responsible for the Status of Women (2002)
Sexual violence takes many forms, including sexual harassment, sexual abuse and exploitation,
rape, gang-rape or attempted rape, sexual slavery, forced pregnancy, abortion, sterilization or
contraception, and trafficking for the purpose of
sexual exploitation (IASC, 2005, p. 8; RHRC, 2003,
pp. 8–11).
Sexual violence is not about sex but about power
relations: ‘rape is not an aggressive expression of
sexuality, but a sexual expression of aggression
. . . a manifestation of anger, violence and domination . . .’ (Seifert, 1992). The specific motives
for such acts vary according to the context. In
intimate partner violence, acts of sexual violence
are common as a form of domination. During armed
conflict, sexual violence may be used as an explicit strategy to achieve military objectives, to
punish and humiliate an enemy group, or even to
destroy a particular social or ethnic group, such
as in the Rwandan conflict. Within armed forces
and groups, sexual violence may serve to affirm
aggression and brutality, and it may be used as
a ‘morale booster’ or a ‘reward.’
Sexual violence often has grave health implications,
both physical (such as direct injuries, infections
or infertility, and sexually transmitted diseases
including HIV/AIDS) and psychological (such as
severe trauma and depression, sometimes leading to suicide). In some cases, victims may be
re-victimized, or even murdered through honour
killings. Sexual violence can also have severe
socio-economic implications, whereby survivors
are rejected by their partners, stigmatized and
sometimes excluded from the family or the community, and unable to find work or to care for their
families (Bastick, Grimm, and Kunz, 2007, p. 15).
However, acts of sexual violence often remain
unreported and hidden due to the victims’ shame
and the stigma attached to such forms of violence.
The WHO Multi-country Study on Women’s Health
and Domestic Violence against Women reports
the prevalence of women having experienced
attempted or completed forced sex by an intimate
partner in their lifetime as ranging from 6.3 per
cent in Serbia and Montenegro up to 49.7 per cent
of women in Bangladesh (García-Moreno et al.,
2005, p. 167). A UN Interregional Crime and Justice
Research Institute study comparing ‘sexual incidents’ (rape, attempted rape, indecent assault,
or offensive behaviour) across regions finds that
10 per cent of women in Asia, 15 per cent of women
in Latin America, and 33 per cent of women in
Africa are victimized in this manner (Zvekic and
Alvazzi del Frate, 1994). For any such analysis,
however, one must acknowledge that in many
societies sexual violence perpetrated by known
or unknown individuals remains unreported.
119
ARMED VIOLENCE AGAINST WOMEN
Acts of sexual violence occur in many different
contexts, including at home or in the workplace,
during armed conflict, or in refugee or post-conflict
settings. Sexual violence is not limited to women
and girls; men, boys, and transsexual/transgender
people may also be victims of sexual violence, as
has been reported in the armed conflicts in the
Central African Republic, the DRC, and in Liberia
(Bastick, Grimm, and Kunz, 2007, pp. 35, 42, 49).
While women and girls are the majority of victims
of acts of sexual violence, the main perpetrators
are men and boys. However, women and girls have
also been reported to incite and commit sexual
violence, for example in the Rwandan genocide
(Bastick, Grimm, and Kunz, 2007, p. 55).
1
2
3
4
5
6
7
Box 6.3 Gang rapes
Gang rapes of women—an extreme form of sexual violence—are commonly
reported in countries including South Africa, Papua New Guinea, and the
United States (Watts and Zimmerman, 2002). A rape is classified as ‘gang
rape’ when it involves at least two perpetrators (Krug et al., 2002, p. 153).
A South African surveillance study for the inner-city of Johannesburg found
that one third of all rapes are gang rapes (Vetten and Haffejee, 2005, p. 33).
In the United States about one out of ten acts of sexual assault is committed
by multiple perpetrators (Greenfeld, 1997, p. 4). Gang rapes are mostly committed by people unknown to the victim (Krug et al., 2002, p. 153).
Honour killings
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
120
Photo " A Turkish
woman at the grave of a
woman who was stoned
to death in 2003 in an
honour-related crime.
© Lynsey Addario/Corbis
A so-called ‘honour killing’ is a murder committed
by (male) relatives in reaction to a perceived violation of the community, family, or individual honour (Vlachová and Biason, 2005, p. 27; UNIFEM,
2007). Most honour killings are perpetrated against
women and girls, based on cultural perceptions
of women as bearers of the family honour. In some
cultures, women are subjected to strict social norms
of behaviour; perceived ‘immoral’ behaviour in
breach of such norms is blamed on women and
can lead to honour killings. The most common
reasons for honour killings are perceived ‘provocative’ behaviour, the refusal of an arranged
marriage, extra-marital affairs, demanding a
divorce, or being a victim of sexual violence.
Honour killings are a global phenomenon but
have mainly been reported in Egypt, Iran, Jordan,
Lebanon, Morocco, Pakistan, Syria, Turkey, Yemen,
and other Mediterranean and Gulf states (UNIFEM,
2007). Through migration, incidents of honour
killings have been exported to western European
countries and North America. The United Nations
Population Fund estimates that worldwide 5,000
women fall victim to honour killings every year
(UNFPA, 2000).
In Pakistan, 4,000 women and men were reportedly killed between 1998 and 2003 ‘in the name of
honour’, with women representing more than half
of the victims. After the fall of Saddam Hussein’s
government in Iraq, 400 women and girls were
reportedly raped between April and October 2004,
of which more than half were later killed for ‘honourrelated’ reasons (MADRE, 2007, p. 16). The perpetrators of honour killings are often male family
members. In Jordan and Lebanon, 70–75 per cent
of all perpetrators of honour killings are the girls’
or women’s brothers (UNIFEM, 2007).
In addition to gendered notions of honour, discriminatory laws contribute to the persistence of such
crimes by granting impunity to perpetrators, thus
allowing honour killings to go unpunished. In
Haiti, for example, the penal code states that the
murder by a husband of his wife and/or her partner
immediately upon discovering them in flagrante
delicto in the conjugal residence is pardonable. A
wife who kills her husband upon discovering him
in the act of adultery is not excused. The Syrian
penal code grants immunity or a significantly
reduced sentence to a man who murders a female
relative. Human Rights Watch reports that in
Guatemala and elsewhere in Latin America police
rarely investigate hundreds of murders of women
each year because they are assumed to be ‘crimes
of passion’ (GCSKSW, n.d.).
A dowry is the money, goods, or estates that are
given by the bride’s family to her husband at
marriage. The practice of dowry payment is particularly common in some South Asian countries
but also occurs in other countries. Dowry disputes,
which may arise due to an unsatisfactory dowry or
the husband’s wish to pursue another marriage in
order to receive an additional dowry, can lead to
gender-based violence including the killing of the
woman. Some women also commit suicide after
continuous harassment by their husbands or in-laws.
In certain societies, the future husband instead
pays a ‘bride wealth’ to the bride’s family, often
leading to the belief that the spouse becomes
his ‘property’. Families sometimes refuse to ‘take
back’ their daughter even in cases where she is
being maltreated, out of inability, or fear of being
obliged, to pay back the bride wealth.
According to UNIFEM, 6,822 women were victims
of dowry-related killings in 2006 (UNIFEM, 2007).
The same year 2,276 Indian women were reported
to have committed suicide as a result of dowry
Figure 6.2 Reported incidents of acid attacks
in Bangladesh, May 1999–July 2008
Number of incidents
400
Source: ASF (2008)
350
300
disputes with their husbands. The figures were
even higher in 2005 and 2004: 2,305 and 2,585
suicides, respectively (Niazi, 2008). It cannot be
ruled out that a certain percentage of these suicides may actually have been homicides committed
by the husband or in-laws.
Acid attacks
Acid attacks are a form of gender violence occurring mainly in Bangladesh, Pakistan, India, and
other Asian countries. While men and boys may
be victims of acid attacks, girls and women represent the majority of victims. The Acid Survivors
Foundation estimates that about 68 per cent of
acid attacks in Bangladesh are directed against
girls and women (ASF, 2006, p. 7).
In this form of gender violence, acid is thrown at
the victim’s body, especially at the face and genitalia of women. Acid attacks are usually motivated
by conflicts over land, property and money, by
refusal of love, marriage, or sexual services, or by
family or dowry-related disputes (ASF, 2006, p. 8).
150
100
50
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
1
2
3
4
5
Box 6.4 Acid attacks in Bangladesh
Bangladesh reports a relatively high level of acid attacks—up to one incident
every two days (ASF, 2006, p. 3). Such attacks have grown in prominence
since the early 1990s, coinciding with a trend of women’s growing financial
power and increased social standing, notably through micro-credit development strategies (Woolf, n.d.). Acid attacks peaked in 2002, when 490
people were injured, and have since declined (ASF, 2008).
250
200
121
ARMED VIOLENCE AGAINST WOMEN
Dowry-related violence
The Acid Survivors Foundation has launched public awareness-raising campaigns to encourage victims to report incidents. It also provides guaranteed
legal assistance, promulgates the existence of laws against acid crimes,
offers free medical care—such as burn treatment, nursing, plastic surgery,
physical therapy, and psychotherapy—and ensures access to counselling
and rehabilitation for victims. These efforts also help to reintegrate victims
into their families and communities, avoiding their isolation (Scholte, 2006).
6
7
122
A R M E D V I O L E N CE
Consequences include permanent marks on the
body, disfiguration, potential blindness, loss of
hearing, and sometimes death. Social isolation is
a further indirect effect. Victims of acid attacks
rarely marry, thus remaining a burden to their
families.
G LO B A L B U R D E N
of
Female infanticide and sexselective abortion
Photo " This 45-year-old
woman was left blind
after an acid attack in
1998. Dhaka, Bangladesh,
2006. © Olivier Hanigan/
WPN
Female infanticide has likely accounted for millions
of sex-selective deaths throughout history. The
UN Children’s Fund defines female infanticide as
the killing of a girl child within the first few weeks
of her birth. Infanticide is practised as a method
of family planning in societies where boys are
valued, economically and socially, above girls.
Methods of ending a baby girl’s life can be cruel,
including poisoning, smothering, or feeding her
unhulled rice to puncture the infant’s windpipes.
While infanticide of newborn girls still takes place,
ultrasound technology has given female infanticide
a modern face in the form of sex-selective abortion.
Substantial disparities between the numbers of
girls and boys born suggest the extent of sexselective abortion. The ratio of girls to boys born
in Europe and North America is approximately
95:100, but in countries such as China, Taiwan,
South Korea, India, and Pakistan, as well as some
sub-Saharan African countries, the ratio is lower.
China and India show the most extreme disparities
(Watts and Zimmerman, 2002). 6
In China, approximately 84 girls are born for every
100 boys (UNFPA, 2007, p. 5), and in some regions
female birth is even lower. The practice of killing
or abandoning female infants markedly increased
in China during the 1980s and is generally attributed to China’s strict ‘one couple, one child’ policy.
It is estimated that by 2020 China could be ‘missing’
around 30 million women. China’s State Population and Family Planning Commission recently predicted that within 15 years one in every ten men
aged between 20 and 45 will be unable to find a
wife (Macartney, 2007; UNFPA, 2007, pp. 5–7).
Already, a shortage of brides is seen as the cause
of increased kidnapping and slave trade of women,
wife selling, and prostitution (Manthorpe, 1999).
In 1996 India’s census showed there to be only
929 females to every 1,000 males. In India’s 1901
census figures, there were 972 females to every
1,000 males. The selective killing of female foetuses is suggested by research that shows that
fewer females are born as second or third children
to families that have yet to have a boy. The ‘most
plausible explanation for the low female-to-male
Female infanticide and sex-selective abortion are
driven by both economic and cultural forces. In a
traditional South Asian family, a son is expected
to earn an income, inherit property, and care for
his parents, while a daughter requires a dowry to
be paid, often incurring substantial debt. However,
the practice cannot be explained by income level
alone. Cultural factors also seem to play an important role. In India, for example, it has been observed
that abortion of female foetuses is most prevalent
in some of the poorest and in some of the richest
states (Sen, 2003).
Conclusion
Using a gender approach reveals the full extent of
the direct and indirect impact of the global burden
of armed violence. This is crucial to understanding
the gender-specific impacts of armed violence,
and the forms of violence specifically targeted
against women. Such an analysis is also important
in terms of policy-making and programme development, allowing for the development of policies
that take into account the specific needs of different groups.
Analysing the gender dimensions of the global
burden of armed violence demonstrates the great
variety of forms of violence and their multiple physical, psychological, social, and economic impacts.
It becomes clear that the picture is highly complex,
defying simplistic notions of women as victims and
men as perpetrators. Finally, a gender approach
broadens understandings of ‘arms’ and ‘armed
violence’ since conventional definitions often
only partially account for women’s experience
of violence.
In times of conflict and social upheaval, women
suffer from lethal, non-lethal, direct, or indirect
armed violence. However, paradoxically, such
situations have sometimes offered a space for
women’s emancipation, be it through women’s
participation in armed groups, or through women
taking on new responsibilities and asserting their
rights. This contradictory relationship is worthy
of further research.
Violence against women is one of the most common
but least punished categories of crime in societies
around the world. Inadequate data, discriminatory
laws or ineffective implementation, widespread
immunity for perpetrators, and a lack of political
will to condemn such crimes all contribute to this
situation. There is a need to review existing data
collection methods and indicators in order to
present a more balanced picture of the gendered
experiences of violence, which make up an important part of the global burden of armed violence.
123
ARMED VIOLENCE AGAINST WOMEN
sex ratios reported at birth is prenatal sex determination followed by selective abortion’ (Jha and
Oster, 2006).
1
2
Abbreviations
DRC
Democratic Republic of the Congo
IPV
Intimate partner violence
UNIFEM
United Nations Development Fund for Women
3
4
5
6
7
Endnotes
1
Bouta, Frerks, and Bannon (2005, p. 11); Gyawali and
Shrestha (2006, p. 147); Marón (2003); and Karame
(1999).
2
The definition of maternal death (by WHO): ‘The death of
a woman while pregnant or within 42 days of termination
of pregnancy, irrespective of the duration and site of the
pregnancy, from any cause related to or aggravated by
the pregnancy or its management but not from accidental
or incidental causes.’
3
These countries include the DRC, (southern) Sudan, Rwanda,
Uganda, Sierra Leone, Somalia, and Burundi.
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
124
4
See AI, IANSA, and Oxfam (2005); Jackson et al. (2005);
Greenfeld (1997); Kellermann et al. (1993); WomenWar
Peace.org (n.d.).
5
The definition of sexual violence by an intimate partner
includes the following elements: the woman was physically forced to have sexual intercourse when she did not
want to; she had sexual intercourse when she did not
want to because she was afraid of what the partner might
do; she was forced to do something sexual that she
found degrading or humiliating (WHO, 2005, pp. 13–16).
6
It should be noted, however, that different studies produce
different data.
T
he global burden of armed violence
extends well beyond acute death and
injury rates arising during war or as a
consequence of crime. Other forms of social and
predatory violence are routinely committed through
such acts as intimidation and assaults, extortion
and kidnapping, or gang violence. Similarly, political violence is often deployed against citizens in
the form of extrajudicial killings and disappearances. The effects of armed violence are routinely
experienced by women afraid to walk in certain
neighbourhoods at night; by partners in abusive
relationships; and by children in slums that lack
adequate lighting, safe schools, and public security.
Throughout the world’s rapidly urbanizing cities
and shanty towns, many citizens are beginning
to fill these ‘protection gaps’ with alternative
means of security provision at the community
level. From the Americas and Africa to the South
Pacific, gangs and vigilante groups are a major, if
poorly understood, security concern. In the absence
of legitimate military and policing authorities,
civilians are increasingly investing in private
security companies, barbed wire and higher
protective walls, neighbourhood watch associations, and even gun-free zones.
This chapter considers ‘other forms’ of armed
violence that are not easily classified under the
rubric of war or crime. It finds that while largely
hidden from view and rarely discussed, such
violence can be present in ostensibly ‘peaceful’
contexts. Likewise, these other forms of armed
violence may contribute to or result in direct
conflict deaths or homicide. While such violence
defies easy description or categorization, the
chapter nevertheless finds the following:
Armed violence perpetrated by armed groups
and gangs is under-studied and contributes
to insecurity in urban settings, with 70,000–
200,000 gang members in Central America
alone.
A high proportion of armed violence by agents
of the state is concentrated in just over 30
countries (in 2006). Disappearances are difficult to calculate but appear to be common
in a similar number of countries.
More than 50 extrajudicial killings were
registered in 2006 for at least 12 countries
unaffected by war, with most not being captured in typical surveillance systems.
Recorded enforced disappearances declined
from an annual average of 1,442 cases between
1964 and 1999 to the annual average of 187
cases between 2000 and 2003, and 140
between 2004 and 2007.
There was an annual global average of 1,350
reported kidnapping for ransom cases from
1998 to 2002. These appeared to have increased
to 1,425 in 2007.
The five countries registering the most kidnapping cases in 2007 included Mexico, Venezuela,
Nigeria, Pakistan, and Colombia.
125
OTHER FORMS OF ARMED VIOLENCE
Other Forms of Armed Violence:
Making the Invisible Visible
Chapter Seven
1
2
3
4
5
6
7
126
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
The violent death rate for aid workers is 60 per
100,000 workers per year, and a considerable
proportion of these killings are carried out
using arms.
Photo " Children in one
of the slum areas in the
centre of Kathmandu,
Nepal.
© Espen Rasmussen/
Panos Pictures
The chapter seeks to enhance the understanding
of other forms of armed violence around the world.
It offers a general overview of their different manifestations and considers various risk factors that
contribute to their onset and duration. The first
section considers specific agents responsible for
armed violence in urban and peri-urban areas. It
focuses on the role of gangs, especially maras
and pandillas in Central America, as symptoms of
larger political, economic, and social processes.
The second section focuses on extrajudicial vio-
lence and enforced disappearances, categories
of illegitimate state-led violence that frequently
are poorly recorded or ignored. It also considers
kidnapping—a tactic adopted by armed groups,
gangs, and common criminals alike—and another
source and outcome of armed violence. The third
section considers the incidence of armed violence
against aid workers.
Armed groups and gangs
Armed groups—including rebels and organized
gangs—do not emerge in a political vacuum. They
reflect a complex combination of economic and
ideological interests. While certain groups reportedly mobilize out of greed or profit, researchers
are discovering that motivations for recruitment
are much more multifaceted than narrow monetary
interest.1 Although prospects for loot clearly provide a motivation for some, in many situations
there are multiple factors that shape the resort
to violence. For example, political elites may have
long-established systems of personal rule and
patronage, and may draw on armed groups to
shore up their authority. Similarly, members of
armed groups may join out of the more routine
and pragmatic desire to protect their neighbourhoods or communities from violence.
Armed groups are highly heterogeneous and exhibit tremendous dynamism and enterprise. In
some cases, members may be popularly described
as ‘thugs’ or ‘bandits’, while in others they may
be seen as heroes in their communities. Gang
members may be viewed with apprehension, particularly if recruits were forcibly removed from
their families and social milieu. In situations
where political institutions and public security
providers suffer from weak governance, alternative
forms of political authority and security delivery
Armed groups are also frequently connected
through power, patronage, and political affiliation.
Groups include both formal and informal actors
such as soldiers, police, paramilitaries, rebel
groups, and ex-combatants and their dependents,
together with mercenaries, militia groups, criminal and predatory gangs. Figure 7.1 presents a
stylized typology of different types of organized
armed groups and the ways in which they are
potentially interlinked.
In many cases, armed groups emerge in the context
of a wider social crisis or malaise, itself potentially shaped by macroeconomic distortions and
political disorder. Armed group members and
their backers may coalesce as a reaction to social
and economic exclusion rather than as a direct
political project, as was the case of the Bakassi
Boys or O’odua People’s Congress of Nigeria.
Likewise, the Mai Mai militia of the Democratic
Republic of the Congo (DRC) and former diamond
miners in Sierra Leone initially banded together
to defend their communities from predation, though
their motives changed over time (Weinstein, 2007).
The domestic (and in some cases international)
legitimacy bestowed on such groups is linked in
large measure to their capacity to provide public
goods such as security and services otherwise
lacking to ordinary civilians (Muggah and Jütersonke,
forthcoming).
While not a new phenomenon, gangs are emerging
as a major concern for policy-makers and practitioners around the world. Gangs are found in all
societies, with the vast majority constituting little
more than ephemeral groups of youth engaged
in behaviour labelled ‘anti-social’ or ‘delinquent’.
Gangs in the more formal sense are defined social
organizations that display institutional continuity
independent of their membership. They exhibit
fixed conventions and rules that may include, for
example, initiation rites, ranking systems, induction
127
OTHER FORMS OF ARMED VIOLENCE
will likely emerge. Armed groups may therefore
be widely regarded as more legitimate than state
institutions in the eyes of the community (Moser
and McIlwaine, 2006; Moser, 2004).
1
2
3
4
Figure 7.1 A typology of armed groups and related actors
5
6
O RG ANIZED
Armed forces
Private security companies
Military police and special forces
Rebels/gangs/vigilante groups
Police and gendarmerie
Organized crime/mafia
STATE
S E CU RITY
NON-STATE
Paramilitaries/militia
Political elites
Paid informants
Commercial elites
Religious leaders/faith-based groups
S PO N TA NEOUS
Source: Muggah and Jütersonke (forthcoming)
7
128
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Box 7.1 Gender and gang violence
The gender dimensions of gang violence are complex and contradictory.
Armed gang violence is mostly a male phenomenon, and victim rates are
highest for young men. However, women are also affected by gang activities in a number of ways. Gang violence can have oppressive or protective
implications for women. In some cases, women are exposed to homicide,
robbery, and sexual assault; in others, they are protected from attacks by
other gangs (UNODC, 2007).
Women also participate in gang activities and have multiple roles. They often
act in support roles such as cooking and washing for male gang members,
or providing logistical assistance like hiding guns, or transporting drugs or
weapons from one point to another. However, women and girls also sell
stolen goods, drugs, and weapons and use armed violence themselves in
some circumstances.
In general, female gang members are responsible for less serious, sporadic
delinquencies than male gang members (Miller and Decker, 2001). While
gang membership may present an opportunity for some women to break
out of traditional gender roles, these roles are often reproduced within
gangs. Independent of their role, participation in gangs makes women a
target for violent acts between gangs.
A study on the participation of women and girls in gang violence in Haiti
reveals the complex gender dimensions of gang activities (Loutis, 2006).
An overview of the variety of female roles within the different gangs and
armed groups in the townships of Cayes, Port-au-Prince, and Gonaïves—
the main urban centres of violence—highlights that women and girls are
perpetrators, dependents, supporters, and victims of gang violence.
Activities differ from one group to another, but mostly they entail support
functions such as cooking and washing, and transmitting messages,
news, and warnings of incursions of rival gangs. Women gang members
may be forced to deliver sexual services or be used as human shields during gang disputes.
Women also commit acts of violence. One female gang in Haiti, composed
of young women, reportedly participates in the kidnapping and raping of
girls, sometimes in concert with male gangs. In some cases, they were
also reported to surrender the kidnapped girls to other groups to be raped
again. It is not clear whether members of this female gang are armed, but
there is evidence that the male groups they act with are armed. Other groups,
such as the ‘Brigades de Vigilance’ in Gonaïves, unified to protect their
neighbourhood against raids and attacks by criminals and gangs. The
members of this group are mainly women; they do not possess firearms
but fight with stones and machetes (Loutis, 2006).
ceremonies, rules of conduct, or specific behaviour patterns. Gangs are often associated with
a particular territory, and relationships with
local communities can be either oppressive or
protective.
Current estimates of the proportion of all regional
violence in Central America committed by gangs
vary from 10 to 60 per cent, suggesting that the
range may be more a question of inclusion and
exclusion criteria than violence itself (UNODC,
2007, p. 64). Gangs are regularly accused of committing (and more rarely prosecuted for) crimes
ranging from delinquency, mugging, theft, and
harassment to rape, assault, and drug dealing.
In other cases, gangs are linked to insurrections
and global terrorism. They are described as a
kind of ‘new urban insurgency’ with the objectives
of deposing or controlling the governments of
certain countries through ‘coups de street’. For
example, the US government recently announced
that gangs constituted the greatest problem for
national security in Central America and Mexico
(Rodgers, 2007; Bruneau, 2005).
State responses to armed groups tend to reproduce violent behaviour rather than contain it
(Small Arms Survey, 2007). Such interventions
generate localized conditions of insecurity and
symbolically demonstrate the power of the state.
The most visible manifestation of this is the ‘war
on gangs’ launched by Central American governments (and others) in the past decade (Muggah
and Stevenson, forthcoming). While anticipating
a major deterrent effect, the war on gangs—or
mano dura—has led instead to the fracturing of
gangs and their adopting more violent tactics.
While this is interpreted by some officials as the
state ‘winning’ the war, it also seems likely that
gangs have adapted and become less conspicuous
in their activities.
Gangs in Central America
129
Central America is a region in which gangs constitute a real contemporary concern from the regional
to the community level. However, their interests
and activities remain relatively poorly understood.
Reliable data and analysis of gangs are limited,
and official statistics are especially problematic
owing to chronic under-reporting, deficient data
collection, and issues of political interference.
Although official figures suggest there are some
70,000 gang members operating in Central
America, the estimates of NGOs and certain scholars suggest that the number could be as high as
200,000 (UNODC, 2007, p. 60). Even using the
low estimate suggests that there are at least as
many gang members as there are military personnel in Central America (World Bank, 2008).
OTHER FORMS OF ARMED VIOLENCE
There is a great diversity of gangs among countries in Central America. For example, El Salvador,
Guatemala, and Honduras are experiencing considerably higher rates of gang violence than Costa
Rica and Nicaragua. The distribution of armed
violence attributed to gangs therefore varies
considerably, although the overwhelming majority
of such activity is urban, including in capital cities.
This is not entirely surprising: gangs are an urban
phenomenon, partly because they require a critical mass of youth to allow them to emerge and
be sustained over time. Recent studies suggest
that as many as 15 per cent of all youth within
gang-affected communities can end up joining a
gang (Rodgers, 2004; 2007). They remain tightknit and small-scale, with between 15 and 100
members (although average sizes tend to be
20–25 members).
One of the strongest predictors of gang membership and related violence relates to demographic
factors, including so-called youth bulges (ARMED
VIOLENCE AFTER WAR). The vast majority of gang
1
2
3
4
members are young urban males, often unemployed
and from lower-income segments of a given community. Although female gang members exist (allfemale gangs are operating in Nicaragua and
Guatemala), perpetrators and victims are most
frequently boys and young men (see Box 7.1). While
the age of gang members ranges from 7 to 30 years,
the average entry into gangs is approximately 15
years of age (Muggah and Stevenson, forthcoming).
Although there is a tendency to treat Central
American gangs generically, a distinction can be
made between maras and pandillas (see Box 7.2).
Specifically, maras are a phenomenon with transnational roots, while pandillas are more localized
and home-grown. In contrast to the many sensa-
Photo ! Honduran
police officers stand
above three Mara
Salvatrucha gang
members after an
anti-gang operation in
Tegucigalpa, Honduras,
2005. © Ginnette
Riquelme/AP Photo
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
130
tionalist claims linking Central American gangs to
migrant trafficking, kidnapping, and international
organized crime, it appears that most maras and
pandillas are involved in small-scale localized
crime and delinquency, such as theft and muggings (Rodgers, 2006).
While there is some evidence that certain mara
groups in El Salvador, Guatemala, and Honduras
are involved in extortion and racketeering, these
often extend no further than the territories they
physically control. There are, however, growing
risks of their assuming a more prominent role in
the drug trade in the coming decade, owing to
the way in which Central America is assuming an
important transit function in the trafficking of
narcotics from South America to North America.
Box 7.2 Maras and pandillas in Central America
Maras are organizations that can be directly linked to specific migratory
patterns. Formerly, there were just two mara groups—the Dieciocho and
the Salvatrucha, which today operate in El Salvador, Guatemala, Honduras,
and southern Mexico. The gangs find their origins in the gangs of Mexican,
Salvadoran, and Guatemalan refugees and migrants in the United States
during the 1960s, 1970s, and 1980s. Following the imposition of strict antigang laws and immigration reform in the United States, however, many gang
members were repatriated back to Central America. Between 1998 and 2005,
the United States deported almost 46,000 convicts to Central America—
with El Salvador, Guatemala, and Honduras receiving more than 90 per cent
of the total. These gang members reproduced many of the structures and
functions they exhibited in the United States (Rodgers, 2006).
Pandillas have their origins in the Central American peace processes of the
1990s. Demobilized former combatant youths in Nicaragua, El Salvador,
and Guatemala returned home to situations of heightened insecurity and
socio-economic uncertainty. Many eventually formed localized vigilantestyle self-defence groups in an effort to provide safety for themselves and
their families. From relatively organic beginnings, however, they rapidly
expanded and developed semi-ritualized patterns of behaviour, including
gang warfare. Some acquired new names—Los Dragones, Los Rampleros,
and Los Comemeuertos in Nicaragua—and assumed strict hierarchies.
Source: Muggah and Stevenson (forthcoming)
Extrajudicial killings,
disappearances, and kidnapping
A number of forms of armed violence perpetrated
by individuals acting in the name of the state can
be classified as illegitimate. Two described here
include extrajudicial killings and enforced disappearances. Owing to their political nature, they
often remain purposefully hidden from view. It
is, of course, important to recognize that not all
uses of force are illegitimate. International norms
and, in most cases, domestic laws recognize a
state’s legal monopoly of the legitimate use of
armed force to protect and safeguard citizens,
institutions, and core values.
But states are also bound by international law and
human rights principles, together with national
laws, to exercise only legitimate force. Excessive
or inappropriate uses of force can contravene
international and domestic laws, and can thus be
declared illegitimate. In certain cases, the illegitimate use of force by public actors against the
population—ranging from extortion and harassment to extrajudicial killings and disappearances—
can undermine the legitimacy of the state and its
institutions and generate negative socioeconomic
impacts. As a result, many multilateral and bilateral assistance programmes are seeking to build
effective and accountable security institutions.
Extrajudicial killings
Human rights groups, such as Amnesty International and Human Rights Watch or trade unionists
and local NGOs, have long decried the use of
extrajudicial armed violence. Until recently, little
data existed to compare the severity of such violence among countries and over time. Activists and
others are frequently unable to disclose precise
information on extrajudicial violence for legal and
ethical reasons. Similarly, owing to international
One crucial source is the Cingranelli–Richards (CIRI)
Human Rights Data Project, which features crosscountry data on extrajudicial killings. Such killings
are broadly defined as the illegitimate use of fatal
armed violence by agents of the state against its
citizens. They may result from the deliberate,
illegal, and excessive use of force by the police,
security forces, or other state actors against
criminal suspects, detainees, prisoners, or other
individuals or groups, and can also include murders committed by private groups, if instigated
by the government (Cingranelli and Richards,
2008b, p. 7).
The CIRI Human Rights Data Project collects data
on extrajudicial killings along with other human
rights variables. Data is drawn from reports of the
US State Department and Amnesty International.
They are coded in three categories that capture
whether extrajudicial killings occur not at all (0
deaths), occasionally (1–49 deaths), or frequently
(more than 50 deaths). When available, numerical
Map 7.1 Global extrajudicial killings, 2006
131
OTHER FORMS OF ARMED VIOLENCE
norms condemning such violence, states are seldom prepared to volunteer such information for
public consumption. 2
1
2
3
4
5
6
7
0)+)2(
)\XVENYHMGMEPOMPPMRKW
*VIUYIRX"
3GGEWMSREP¦
2SRI
2SHEXE
Note: Data for Somalia is not available in this map as there was no central political authority in the country in 2006.
Source: Cingranelli and Richards (2008a)
132
A R M E D V I O L E N CE
counts of extrajudicial killings are used to classify
countries into the three categories (Cingranelli
and Richards, 2008b, pp. 7–10).3
G LO B A L B U R D E N
of
In 2006 there were at least 31 countries4 in which
extrajudicial killings occurred frequently (more
than 50 deaths) and 73 countries5 in which they
occurred occasionally (between 1 and 49 deaths)
(see Map 7.1). These figures serve as a reminder
that the burden of other forms of armed violence
requires more investigation and attention.
Photo " A man holds up
a picture of his son, who
has been on death row
for more than a decade as
a result of a confession
he made under torture.
© Ian The/Panos Pictures
Comparison of the distribution of extrajudicial
killings to maps generated by Uppsala’s Conflict
Database displaying the distribution of direct
conflict deaths for 2006 shows that at least 12
countries register more than 50 extrajudicial killings but are not considered to be in ‘conflict’.
These countries include Cambodia, China, Côte
d’Ivoire, the Democratic People’s Republic of Korea,
the Dominican Republic, the DRC, Jamaica, Kenya,
Mexico, Nigeria, South Africa, and Venezuela.
Although certain national human rights agencies
can provide confidential information to international organizations, there are few monitoring
mechanisms to track trends and investigations
in this area over time. In 1982 the UN established
a special rapporteur on extrajudicial, summary, or
arbitrary executions, with a mandate to perform
country visits. In a recent report on Brazil, for
example, the special rapporteur found that many
killings registered by on-duty police were classified as ‘acts of resistance’ or cases of ‘resistance
followed by death’, suggesting that such events
were under-diagnosed. Indeed, in 2007 in Rio de
Janeiro, the police recorded 1,330 resistance killings, a figure that accounts for 18 per cent of the
total number of killings in the city (HRC, 2008a,
para. 10).
There are a number of reasons why comprehensive statistics on extrajudicial killings have not
been tabulated and publicized. Existing human
rights practice tends to focus on individual cases
rather then cross-country comparisons. As a
result, few comprehensive databases exist within
the human rights community. Similarly, the UN
special rapporteur was issued a mandate to investigate ‘situations’ rather than establish global or
even national datasets on extrajudicial killings.
It is thus extremely difficult to verify and validate
extrajudicial killings. In many cases, human rights
agencies render assessments on the basis of
information transferred to them by local people
or local NGOs. Allegations frequently contradict
official accounts, and legal cases can take years,
even decades, to build. Equally challenging is
the fact that instances of such killing frequently
go unreported, for the simple reason that there
is nobody to report them or a lack of awareness
about reporting practices and a fear of the legitimacy of relevant institutions.
Disappearances
Typically described as ‘enforced disappearances’,
such acts constitute yet another facet of illegitimate
armed violence. In certain cases, disappearance
may include the eventual killing of the person who
is abducted. In many cases, the victim’s family
does not know whether the disappeared person
is alive, contributing to their pain and suffering.
Disappearances are also frequently linked to
criminal violence, including social cleansing; executions; displacement; and, in certain circumstances,
rape, sexual violence, and forced recruitment.
The category of ‘enforced disappearances’ is
invoked by human rights specialists to describe
violence by state officials. While exceptions exist,
the term does not usually refer to disappearances
committed by non-state actors. 6 The illegitimacy
of such actions is enshrined in a number of legal
instruments, including the 2006 International
Convention for the Protection of all Persons
from Enforced Disappearances. Included in the
definition are those who suffer:
arrest, detention, abduction or any other form of
deprivation of liberty by agents of the State or
by persons or groups of persons acting with the
authorization, support or acquiescence of the
133
Box 7.3 Gender-based violence by state officials
State security and justice agencies bear the responsibility to protect citizens
and to ensure the protection of human rights and the maintenance of public
order. However, all too often, these agencies are involved in gender-based
violence (GBV) in the execution of their tasks. GBV occurs either through
discriminatory laws and policies, their inadequate implementation, or
granting impunity for acts of GBV committed by state officials, including
police officers, prison guards, and soldiers.
The absence of an impartial and effective criminal justice system often
protects delinquent state officials. GBV includes, for example, the abuse
of persons in custody by supervising authorities or other detainees. As
required by Rule 8 of the United Nations Standard Minimum Rules for the
Treatment of Prisoners, men and women should be detained separately.
But due to a lack of facilities or inadequate policies, women and girls share
facilities with men in many countries. This exposes women to a high risk of
sexual violence by other detainees.
In the United States, at least two-thirds of imprisoned women have experienced violence, sexual harassment, and abuse by male guards, and at
least one out of four women has been sexually assaulted while in state
custody (Vlachová and Biason, 2005, pp. 96–97; HRW, 1996). The police
can also perpetrate acts of GBV through the mistreatment and revictimization of survivors, or their unwillingness to investigate such crimes. In Haiti
and Zimbabwe, for example, ‘political rapes’ against women were committed
by government officials in retaliation for supporting political opposition
OTHER FORMS OF ARMED VIOLENCE
The absence of data makes it difficult to compare
extrajudicial killings across time and space. One
element of violence reduction policies, however,
could include improving reporting rates, enhancing access to legitimate justice mechanisms, and
providing meaningful protection. At the most basic
level, the pooling of information on extrajudicial
killings by the special rapporteur and human rights
organizations could be one step forward in generating awareness of the frequency and magnitude
of this form of armed violence.
1
2
3
groups (Bastick, Grimm, and Kunz, 2007, pp. 67, 79).
4
5
State, followed by a refusal to acknowledge the
deprivation of liberty or by concealment of the
fate or whereabouts of the disappeared person,
which place such a person outside the protection
of the law (UNGA, 2006, art. 2).
The distribution of enforced disappearances
around the world suggests that they are highly
concentrated (see Map 7.2). Although probably
an undercount, there appear to be at least 12
countries where such disappearances are frequent
(defined as 50 or more cases annually) and another
22 countries where such actions are more occasional (defined as fewer than 49 cases annually).7
6
7
Map 7.2 Enforced and involuntary disappearances, 2006
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
134
0)+)2(
(MWETTIEVERGIW
*VIUYIRX"
3GGEWMSREP¦
2SRI
2SHEXE
Note: Data for Somalia is not available in this map as there was no central political authority in the country in 2006. No information was available for Iran.
Source: Cingranelli and Richards (2008a)
The global magnitude of enforced disappearances
remains poorly understood. The Working Group
on Enforced or Involuntary Disappearances of the
UN Office of the High Commissioner for Human
Rights recorded a total of 51,763 cases between
1964 and 2007. At least 41,257 of these cases—
some 80 per cent—remain unresolved in 2008
(HRC, 2008b, pp. 104–6). It should be noted,
however, that most of these reported incidents
occurred before 2000. In Iraq, for example, 15,853
out of 16,517 cases occurred before 1989. In Sri
Lanka, 9,443 out of 12,085 cases were recorded
in 1989 and 1990. 8
Indeed, between 2000 and 2007, the Working
Group recorded just 1,307 cases, which represent
approximately 2.5 per cent of all recorded cases.9
While this represents only a small proportion of
global enforced disappearances, it provides insight into recent figures of recorded enforced
disappearances, and suggests that these have
declined from an annual average of about 187
between 2000 and 2003 to 140 between 2004
and 2007 (see Table 7.1). Comparing these to the
annual average of 1,442 for the period 1964 to
199910 further highlights the dramatic decrease
of recorded enforced disappearances.
Table 7.1 Recorded cases of enforced or involuntary disappearances,
selected countries, 2000–03 and 2004–07
2000–03
Total cases
Algeria
2004–07
Annual average
Total cases
Annual average
15
3.75
8
2
9
2.25
0
0
China
15
3.75
6
1.5
Colombia
80
20
34
8.5
Ethiopia
0
0
4
1
Guatemala
1
0.25
0
0
Honduras
0
0
2
0.5
India
54
13.5
10
2.5
Indonesia
43
10.75
1
0.25
Iran
1
0.25
2
0.5
Iraq
1
0.25
0
0
Lebanon
3
0.75
0
0
16
4
2
0.5
1
0.25
0
0
307
76.75
153
38.25
Pakistan
6
1.5
31
7.75
Peru
2
0.5
0
0
13
3.25
38
9.5
105
26.25
23
5.75
Sri Lanka
17
4.25
185
46.25
Sudan
54
13.5
61
15.25
Turkey
4
1
0
0
747
186.75
560
140
Argentina
Mexico
Morocco
Nepal
Philippines
Russian Federation
Total/average
Source: Calculations based on HRC (2008b, pp. 107–20)
OTHER FORMS OF ARMED VIOLENCE
Country
135
1
2
3
4
5
6
7
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
136
Developing a more robust capacity to monitor
and track enforced disappearances is a priority,
since many cases still go unreported due to factors such as illiteracy, fatalism, fear of reprisal,
weaknesses in the policing and judicial system,
ineffective reporting channels, or a culture of
impunity. These factors do not encourage a victim’s
kin or family to file a case with local prosecutors,
human rights bodies, or ombudspersons, much
less with the UN Working Group or other international mechanisms. As in the case of all reported
indicators of armed violence, high reporting
rates may reveal a higher awareness of reporting
practices or a robust surveillance system rather
than actual incidence.
Kidnapping
Unlike disappearances, which are ostensibly
‘political’, kidnapping is primarily criminally motivated. Kidnapping is frequently undertaken by
Box 7.4 The burden of kidnapping in Venezuela
Between 1996 and 2006 approximately 1,732 kidnapping events were
recorded in Venezuela. Kidnapping progressively shifted from an isolated
activity to a well-planned and -organized industry. Gangs devoted to kidnap
and ransom usually include 10–20 people who are specialized in activities
such as identifying victims, researching their movements, valuing their
possessions, carrying out the kidnapping, guarding the victim, and negotiating the ransom.
Kidnapping targets include wealthy male executives but also middle-class
businesspeople and children. Middle-class victims tend to be viewed as
easier targets, since they usually feel less at risk of kidnapping and do not
adopt preventive measures. In the first six months of 2007, 147 kidnappings
were registered, of which 20 per cent were foreign nationals. In 36 per cent
of these cases, victims were released without ransom, while 20 per cent
were rescued by the police. Just 19 per cent were released after payment,
and three per cent were ultimately murdered. Only three per cent escaped
from their captors, while the remainder are still in captivity.
Source: Armour Group (2007)
armed groups or individuals and involves a high
degree of coercive force. Although most kidnapping victims are ultimately freed, the physical
and psychological consequences are serious and
persist long after the event. Pain and suffering
extend to the victim’s family, who suffer considerable emotional duress during the period of captivity. In certain cases, the relationships between
the victims and their families may also alter permanently, depending on the trauma experienced.
Similarly, from Colombia and Haiti to the United
States and Western Europe, the financial expenditures associated with freeing victims from kidnappers are frequently substantial. These include
ransom payments that deplete household savings, lost income due to the protracted detention
of income earners, and protection costs of other
family members.
Kidnapping rates—like those of extrajudicial
killings and disappearances—are notoriously
difficult to monitor. While there are no multilateral
agencies devoted exclusively to the task, the
firm Control Risks has collated a unique global
database on kidnapping extending back to 1975
that includes records for more than 35,600
unique kidnapping cases.11 ‘Kidnap for ransom’
cases are defined by Control Risks as ‘the abduction of a person or persons with the intent of their
detention in an unknown location until a demand
is met’. Further, Control Risks determines that
‘cases include political and criminal perpetrators
and political or financial demands must be met
prior to release of the victim’.12
There were at least 1,350 reported cases of kidnapping per year between 1998 and 2002, or
some 6,753 cases reported over the entire period.
While undoubtedly an undercount, this figure
offers insight into the changing patterns and
dynamics of kidnapping worldwide. Three-quarters
of all kidnap for ransom incidents (74 per cent)
Table 7.2 Kidnap for ransom cases by region, 1998–2002
1998–2002
Latin America
137
Annual average
4,997
999
Asia
945
189
Europe and the FSU
473
95
Africa and the Middle East
203
41
US, Canada, and the Caribbean
135
27
6,753
1,350
Total
Source: Control Risks estimates
Legend:
Latin America (74%)
Source:
Control Risks estimates
Table 7.3 Kidnap for ransom cases by region, 2007
Asia (14%)
Europe and the FSU (7%)
2007
Africa and the Middle East (3%)
US, Canada, and the Caribbean (2%)
Figure 7.3 Kidnap for ransom cases worldwide,
by region, 2007
Latin America
684
Asia
356
Africa and the Middle East
228
US, Canada, and the Caribbean
114
Europe and the FSU
Total
Legend:
Latin America (48%)
Source:
Control Risks estimates
Asia (25%)
Africa and the Middle East (16%)
US, Canada, and the Caribbean (8%)
Europe and the FSU (3%)
took place in Latin America, another 14 per cent
of all reported kidnap for ransom events occurred
in Asia, and 7 per cent in Europe and the Former
Soviet Union (FSU) (see Figures 7.2 and 7.3 and
Tables 7.2 and 7.3).13
OTHER FORMS OF ARMED VIOLENCE
Figure 7.2 Kidnap for ransom cases worldwide,
by region, 1998–2002
1
43
2
1,425
3
Source: Control Risks estimates
4
Patterns of kidnapping also appear to be dynamic.
In 2007 the global total of recorded kidnap for
ransom cases increased slightly to 1,425. Although
half of all reported cases occurred in Latin America, there appears to be a growing tendency for
kidnapping in Asia and Africa (see Figure 7.3). It
appears that the overall decline in Latin America
can also be attributed to a general decline in kidnappings in Colombia, despite moderate increases
in Mexico and Venezuela. The surge of kidnapping
in Afghanistan, Pakistan, India, Iraq, and Nigeria
accounts for the growth in other regions. The top
ten countries for kidnap for ransom cases in 2007
were Mexico, Venezuela, Nigeria, Pakistan, Colombia, India, Haiti, Afghanistan, Brazil, and Iraq.
5
6
7
Armed violence and aid workers
138
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
Aid workers provide humanitarian assistance to
millions of people around the world. They are a
group specifically exposed to armed violence,
because most of their work occurs in conflict or
post-conflict environments. In this context, the
organizations involved in humanitarian assistance find themselves weighing difficult choices
between interrupting life-saving relief activities and
safeguarding the security of their staff. Violence
against aid workers has captured the attention of
the media, and various researchers have set out
to develop a better understanding of the dynamics of this type of armed violence.
Legend:
Figure 7.4 Types of fatal attacks against aid
workers, 1997–2003
Ambush (44%)
Murder (not in vehicle)
(25%)
Car/truck bombing
(9%)
Landmine (9%)
Anti-aircraft attack
(8%)
Aerial bombardment
(5%)
Source:
Fast and Rowley (2008)
Figure 7.5 Location of intentional violence cases
against aid workers, 2002–05
Legend:
En route to field
activities (61%)
NGO central office
(11%)
NGO sub-office (6%)
Personal residence
(8%)
Other/don’t know (14%)
Source:
Fast and Rowley (2008)
Intentional armed violence is one of the leading
causes of death for aid workers around the world.
A recent estimate by researchers at Johns Hopkins
University estimated the violent death rate of relief
personnel at 60 per 100,000 aid workers per year
(Fast and Rowley, 2008). This figure—higher than
the intentional homicide rate for almost all countries—shows that aid workers face a high risk of
victimization. Although international and local
personnel regularly face various threats to their
health and well-being, research points to the role
of arms availability and misuse as a critical risk
factor (Buchanan and Muggah, 2005; Beasley,
Buchanan, and Muggah, 2003) (see Figure 7.4).
It is difficult to predict with certainty regional or
country-level risks. Nevertheless, it appears that
Africa remains the site of most relief worker deaths
and injuries. While intentional violence is a threat
to aid workers, criminality and other manifestations of routine interpersonal violence also impact
on morbidity, stress, and mental health. The most
dangerous activity for aid workers is travelling
between sites (home and office), while road
ambushes—often involving the use weapons—
are the most frequently reported type of armed
violence (see Figure 7.5). Finally, national (and
not international) staff bear the largest brunt of
intentional violence, particularly drivers, guards,
and those working directly in the field (Fast and
Rowley, 2008).
The present evidence base does not necessarily
suggest that the overall incidence of intentional
violence is increasing, but rather that it has
kept pace with the expansion in the number of
humanitarian personnel working on the ground.
Nevertheless, there is a need to enhance monitoring of these trends in order to develop a better
understanding of the risks aid workers face in
specific countries. Aid worker deaths have not
been fully incorporated into the global cost of
ties suffered by aid workers, interruptions to the
Abbreviations
CIRI
Cingranelli–Richards (Human Rights Data
Project)
for conflict-affected populations in terms of their
DRC
Democratic Republic of the Congo
access to food, water, shelter, and other forms of
FSU
Former Soviet Union
life support. The costs of armed violence against
GBV
gender-based violence
delivery of assistance have major consequences
aid workers are therefore high both for those who
need the assistance and for those who provide it.
Endnotes
Conclusions
The various forms of armed violence reviewed in
1
See, for example, Marchal (2006); Esser (2004); Rodgers
(2004); Hillier, Greene, and Gesyllas (2000).
2
The legal doctrine on extrajudicial killings is based on the
‘right to life’ as enshrined in the 1948 Universal Declaration
of Human Rights and the 1966 International Covenant on
Civil and Political Rights. In 1982 the UNOHCHR underlined
that states are required to prevent and punish deprivation
of life by criminal acts, as well as by killings committed by
their own security forces (UNOHCHR, 1982, para. 3). In
1982 the UN Commission on Human Rights established
a special rapporteur on extrajudicial, arbitrary and summary executions with a mandate to investigate situations
of extrajudicial killings around the world by holding governments to account when state agents were responsible
for killings, or when the state has not done everything in
its power to prevent or respond to killings committed by
others. Article 4 of the 1989 Principles on the Effective
Prevention and Investigation of Extra-legal, Arbitrary, and
Summary Executions further enshrines the protection of
the right to life.
this chapter warrant special attention, if only
because they are often hidden from view. The violence from armed groups and gangs, extrajudicial
killings or forced disappearances, kidnappings,
and the victimization of aid workers are part of
the global burden of armed violence and need to
be recognized as such. However, much remains
to be done to develop a better understanding of
the magnitude and distribution of these types of
armed violence.
The forms of armed violence discussed in this
chapter do not lend themselves to simple policy
interventions. Gang violence, for example, may be
met with robust force, or with policies designed
3
to stem recruitment into gangs and erode their
economic foundations. Few policies, either forceful or preventative, have been systematically
tested. Similarly, responding to extrajudicial killings is often complicated by competing accounts
of the circumstances that led to the killing of an
individual or group. But by broadening the optic
beyond a simple count of fatalities, the chapter
signals how different forms of armed violence
generate effects that extend out from victims, to
families, households, communities, and society
at large.
4
Thresholds were determined from the CIRI database. In
cases without numerical estimates, categorization relies
on the wording within the reports. In cases where extrajudicial killings occur frequently, language describing the
violations includes adjectives such as ‘gross’, ‘widespread’, ‘systematic’, ‘epidemic’, ‘extensive’, ‘wholesale’,
‘routine’, or ‘regularly’. In cases in which extrajudicial
killings occur occasionally, adjectives include ‘numerous’,
‘many’, or ‘various’ (Cingranelli and Richards, 2008b,
pp. 7–10).
Countries in which extrajudicial killings occur frequently
include Algeria, Bangladesh, Brazil, Burundi, Cambodia,
Central African Republic, Chad, China, Colombia, Democratic
People’s Republic of Korea, DRC, Côte d’Ivoire, Dominican
Republic, India, Iran, Iraq, Israel, Jamaica, Kenya, Mexico,
Myanmar, Nepal, Nigeria, Pakistan, Philippines, South Africa,
Sri Lanka, Sudan, Thailand, Uganda, and Venezuela.
139
OTHER FORMS OF ARMED VIOLENCE
armed violence, since, beyond the direct casual-
1
2
3
4
5
6
7
5
G LO B A L B U R D E N
of
A R M E D V I O L E N CE
140
Countries in which extrajudicial killings occur occasionally include Afghanistan, Angola, Argentina, Azerbaijan,
Bahamas, Barbados, Bolivia, Botswana, Burkina Faso,
Cameroon, Canada, Ecuador, Egypt, El Salvador, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, Georgia,
Ghana, Guatemala, Guinea, Guyana, Haiti, Honduras,
Indonesia, Jordan, Kazakhstan, Kyrgyz Republic, Laos,
Lesotho, Liberia, Libya, Madagascar, Malawi, Malaysia,
Mauritius, Mongolia, Morocco, Mozambique, Namibia,
Nicaragua, Niger, Papua New Guinea, Paraguay, Peru,
Portugal, Republic of the Congo, Romania, Russian Federation, Rwanda, Saint Lucia, São Tomé and Príncipe, Saudi
Arabia, Senegal, Sierra Leone, Suriname, Swaziland,
Syria, Tajikistan, Tanzania, Timor-Leste, Togo, Trinidad
and Tobago, Tunisia, Turkey, Turkmenistan, United States,
Vietnam, Yemen, Zambia, and Zimbabwe.
6
The Rome Statute of the International Criminal Court is an
exception, since it makes reference to political groups as
potential perpetrators and to ‘the intention of removing
[the victim] from the protection of the law for a prolonged
period of time’.
7
This information is based on the CIRI Human Rights database, which codes annual reports from the US State
Department and Amnesty International (Cingranelli and
Richards, 2008b, pp. 13–17). Countries in which disappearances occur frequently include Bangladesh, Colombia,
Democratic People’s Republic of Korea, Ethiopia, India,
Iraq, Myanmar, Pakistan, Philippines, Russian Federation,
Sri Lanka, and Thailand. Countries in which disappearances
occur occasionally include Algeria, Argentina, Azerbaijan,
Brazil, Burundi, Chad, China, DRC, Côte d’Ivoire, Eritrea,
Gambia, Honduras, Mexico, Nepal, Paraguay, Sudan, Syria,
Togo, Uganda, Uzbekistan, Venezuela, and Zimbabwe.
8
A statistical analysis is presented in the online methodological appendix at <http://www.genevadeclaration.org>.
9
This figure is based on counting together all cases from
22 countries presented in Table 7.1 amounting to 1,307
cases and relating them to the total of 51,763 cases as
reported in HRC (2008b, pp. 104–6).
10
This figure is based on 50,456 recorded enforced disappearances for the 35 years between 1964 and 1999, resulting
in an annual average of 1,441.6.
11
The kidnapping database is maintained by a team of four
analysts who carry out daily searches for such cases worldwide using a variety of sources and who update the database accordingly.
12
This is Control Risks’ working definition of ‘kidnap for
ransom’ cases (correspondence, 10 June 2008).
13
The distribution of this relative weight was stable over the
five years. The Former Soviet Union includes the Russian
Federation as well as Armenia, Azerbaijan, Belarus, Estonia,
Georgia, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Moldova,
Tajikistan, Turkmenistan, Ukraine, and Uzbekistan.
Bibliography
Executive Summary
CNP (Colombian National Police). n.d. Crime Report 2005.
Bogota: CNP.
Peden, Margie, Kara McGee, and Etienne Krug, eds. 2002.
Injury: A Leading Cause of the Global Burden of Disease
2000. Geneva: World Health Organization.
World Health Organization. 2008a. Preventing Violence and
Reducing its Impact: How Development Agencies Can Help.
Geneva: World Health Organization.
——. 2008b. ‘Violence Prevention Highlights’. <http://www.
emro.who.int/VIP/violence_prevention.htm>
——. 2008b. Conflict Data, 1975–2001. Accessed 11 June 2008.
<http://www.cred.be/cedat/conflict/conflict.htm>
BIBLIOGR APHY
141
Eck, Kristine. 2005. A Beginner’s Guide to Conflict Data: Finding
and Using the Right Dataset. UCDP Paper, No. 1. Uppsala:
Uppsala Conflict Data Programme.
Eriksson, Mikael, et al. 2002. ‘Armed Conflict 1946–2001:
A New Dataset.’ Journal of Peace Research, Vol. 39, No. 5,
pp. 615–37.
Garfield, Richard. 2008. ‘Editorial: Measuring Deaths from Conflicts.’ British Medical Journal. Online First. <http://www.
bmj.com/ cgi/content/extract/336/7659/1446?maxtosho
w=&amp;HITS=10&amp;hits=10&amp;RESULTFORMAT=
&amp;fulltext=obermeyer&amp;searchid=1&amp;FIRST
INDEX=0&amp;resourcetype=HWCIT>
1
Chapter One
Geneva Declaration on Armed Violence and Development. 2008.
Web site. <http://www.genevadeclaration.org>
Ball, Patrick, et al. 2003. How Many Peruvians Have Died? An Estimate of the Total Number of Victims Killed or Disappeared
in the Armed Internal Conflict between 1980 and 2000.
Washington, D.C.: American Association for the Advancement of Science. 28 August.
Gleditsch, Kristian. 2007. Expanded War Data, v.1.52. 23 May.
Accessed 2 May 2008. <http://privatewww.essex.ac.uk/
~ksg/expwar.html>
3
Gleditsch, Nils and Bethany Lacina. 2005. ‘Monitoring Trends in
Global Combat: A New Dataset of Battle Deaths.’ European
Journal of Population, Vol. 21, Nos. 2–3, pp. 145–66.
5
Center for International Development and Conflict Management.
2005. Minorities at Risk Dataset. Accessed 10 February
2008. <http://www.cidcm.umd.edu/mar/>
Center for Systemic Peace. 2004. Major Episodes of Political
Violence 2004 Database. Accessed 5 February 2008.
<http://www.systemicpeace.org/inscr/MEPV2004.xls>
——. 2007. Major Episodes of Political Violence, 1946–2006.
Accessed 5 February 2008. <http://members.aol.com/
CSPmgm/warlist.htm>
Coghlan, Benjamin, et al. 2008. Mortality in the Democratic Republic of Congo: An Ongoing Crisis. New York and Melbourne:
International Rescue Committee and Burnet Institute.
COW (Correlates of War). 2007. Militarized Interstate Disputes,
v.3.10. Accessed 10 February 2008. <http://www.correlates
ofwar.org/COW2%20Data/MIDs/MID310.html>
CRED (Centre for Research on the Epidemiology of Disasters).
2008a. The Complex Emergency Database (CE-DAT). Accessed
11 June 2008. <http://www.cedat.be/>
Human Security Centre. 2005. Human Security Report 2005:
War and Peace in the 21st Century. New York: Oxford University Press.
——. 2006. Human Security Brief 2006. Vancouver: University of
British Columbia.
Human Security Report Project. 2008. Human Security Brief 2007:
Dying to Lose: Explaining the Decline in Global Terrorism.
Vancouver: Simon Fraser University.
Huth, Paul and Benjamin Valentino. 2008. ‘Mass Killings of
Civilians in Times of War, 1945–2000.’ In J. Joseph Hewitt,
Jonathan Wilkenfeld, and Ted R. Gurr, eds. Peace and
Conflict 2008. Center for International Development and
Conflict Management. Boulder: University of Maryland
and Paradigm, pp. 79–93.
ICG (International Crisis Group). 2008. Crisis Watch, No. 55. 1 March.
<http://www.crisisgroup.org/home/index.cfm?id=5325>
2
4
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
142
IISS (International Institute for Strategic Studies). 2008. Armed
Conflict Database. Accessed 23 April 2008. <http://acd.
iiss.org/armedconflict>
Iraq Body Count. 2008. Web site. Accessed 6 August 2008.
<http://www.iraqbodycount.org>
Obermeyer, Ziad, et al. 2008. ‘Fifty Years of Violent War Deaths
from Vietnam to Bosnia: Analysis of Data from the World
Health Survey Programme.’ British Medical Journal. Online
First. <http://bmj.org>
PITF (Political Instability Task Force). 2008. ‘State Failure Problem
Set: Internal Wars and Failures of Governance, 1955–2006.’
Web site. Accessed February 2008. <http://globalpolicy.
gmu.edu/pitf/pitfpset.htm>
PRIO (International Peace Research Institute, Oslo). Centre for the
Study of Civil War. 2008. Battle Deaths Data, v.2.0, 1945–
2005. Accessed 5 February 2008. <http://www.prio.no/
CSCW/Datasets/Armed-Conflict/Battle-Deaths/>
Project Ploughshares. 2007. Armed Conflicts Report. Accessed
5 May 2008. <http://www.ploughshares.ca/libraries/
ACRText/ACR-TitlePageRev.htm>
PTS (Political Terror Scale). 2008. Accessed 2 May 2008.
<http://www.politicalterrorscale.org/>
Richardson, Lewis F. and Quincey Wright. 1960. Statistics of
Deadly Quarrels. London: Stevens and Sons.
Sarkees, Meredith Reid. 2000. ‘The Correlates of War Data on
War: An Update to 1997.’ Conflict Management and Peace
Science, Vol. 18, No. 1, pp. 123–44.
Singer, J. David, ed. 1979. The Correlates of War, Volume 1:
Research Origins and Rationale. New York: Free Press.
SIPRI (Stockholm International Peace Research Institute). 2007.
SIPRI Yearbook 2007: Armaments, Disarmament and
International Security. Stockholm: Oxford University Press.
Small Arms Survey. 2005. Small Arms Survey 2005: Weapons at
War. Oxford: Oxford University Press.
UCDP (Uppsala Conflict Data Programme). Department of Peace
and Conflict Research, Uppsala University. 2006a. UCDP
Non-State Conflict Dataset, v.1.1, 2002–05. Accessed
2 May 2008. <http://www.pcr.uu.se/publications/UCDP_
pub/UCDP%20Non-state%20Conflict%20dataset%20
version%201.1.xls>
——. 2006b. UCDP Battle-Deaths Dataset, v.4.1, 2002–05. Accessed
2 May 2008. <http://www.pcr.uu.se/publications/UCDP_
pub/UCDP%20Battle-deaths%20dataset%20version%20
4.1.xls>
——. 2006c. UCDP One-Sided Violence Dataset, v.1.2, 1989–2005.
Accessed 2 May 2008. <http://www.pcr.uu.se/publications/
UCDP_pub/UCDP%20One-sided%20violence%20data
set%20version%201.2.xls>
——. 2008. Armed Conflict Database. Accessed 5 May 2008.
<http://www.ucdp.uu.se/database>
—— and Centre for the Study of Civil War. 2007. UCDP/PRIO
Armed Conflict Dataset. <http://www.prio.no/misc/Down
load.aspx?file=%2fprojects%2f dataset-website-workspace
%2fArmed%2520Conflicts%2520Version%252042007%2f
Main_Conflict_Table.xls>
Vasquez, John. 1987. ‘The Steps to War: Towards a Scientific
Explanation of Correlates of War Findings.’ World Politics,
Vol. 40, No. 1, pp. 108–45.
WHO (World Health Organization). Forthcoming. The Global Burden
of Disease. Geneva: WHO.
——. 2004. Mortality and Burden of Disease Estimates for WHO
Member States in 2002. <http://www.who.int/entity/
healthinfo/statistics/bodgbddeathdalyestimates.xls>
Chapter Two
Ahuja, Anjana. 2007. ‘Could 650,000 Iraqis Really Have Died
because of the Invasion?’ The Times. 5 March. <http://
www.timesonline.co.uk/tol/news/world/iraq/article
1469636.ece>
AI (Amnesty International). 2004. Liberia: The Promises of Peace
for 21,000 Child Soldiers. London: AI. 17 May.
Alkhuzai, Amir, et al. (Iraq Family Health Survey Study Group).
2008. ‘Violence-related Mortality in Iraq from 2002 to 2006.’
New England Journal of Medicine, Vol. 358, No. 5, pp. 484–93,
31 January.
Asher, Jana, David Banks, and Fritz J. Scheuren, eds. 2008.
Statistical Methods for Human Rights. New York: Springer.
BBC (British Broadcasting Corporation). 2008. Women on the
Front Line – Democratic Republic of Congo. BBC World.
Bijleveld, Catrien, Olivier Degomme, and Shanna Mehlbaum.
2008. A Very Dark Number: Estimating Direct and Indirect
Mortality in Southern Sudan 1983–2005. Working paper.
Amsterdam: Vrije Universiteit Amsterdam.
Bijleveld, Catrien and Lotte Hoex. 2008. Direct and Indirect Mortality During the Conflict in Sierra Leone 1991–2002.
Working paper. Amsterdam: Vrije Universiteit Amsterdam.
Burnham, Gilbert, Riyadh Lafta, Shannon Doocy, and Les Roberts.
2006. ‘Mortality after the 2003 Invasion of Iraq: A Crosssectional Cluster Sample Survey.’ The Lancet, Vol. 368,
pp. 1421–28, 21 October.
Brookings Institution. 2008. ‘Iraq Index: Tracking Reconstruction
and Security in Post-Saddam Iraq.’ 21 April. <http://www.
brookings.edu/saban/iraq-index.aspx>
——. 1998. Working Document II: Quantifying Genocide in the
Southern Sudan and the Nuba Mountains 1983–1998.
Washington, DC: US Committees for Refugees.
Checchi, Francesco and Les Roberts. 2005. Interpreting and
Using Mortality Data in Humanitarian Emergencies: a
Primer for Non-Epidemiologists. Network Paper No. 52.
London: Overseas Development Institute. September.
Coghlan, Benjamin et al. 2006. ‘Mortality in the Democratic
Republic of Congo: A Nationwide Survey.’ The Lancet.
Vol. 367, No. 9504. January, pp. 44–51.
——. 2008. Mortality in the Democratic Republic of Congo: An
Ongoing Crisis. New York: International Rescue Committee.
Conibere, Richard et al. 2004. Appendix 1: Statistical Appendix to
the Report of the Truth and Reconciliation Commission of
Sierra Leone. Palo Alto, California: Human Rights Data
Analysis Group, Beneficent Technology Inc.
CRED (Centre for Research on the Epidemiology of Disasters).
2006. Reinforcing Data Reporting From Refugee Settings.
Brussels: CRED.
——. 2008. ‘Database on the Human Impact of Complex Emergencies (CE-DAT).’ Brussels: CRED. Accessed June 2008.
<http://www.cedat.be/?q=database>
Daponte, Beth O. 2008. ‘Why Estimate Direct and Indirect Causalities From War? The Rule of Proportionality and Casualty
Estimates.’ In Jana Asher et al., pp. 51–63. New York:
Springer.
Dobbs, Michael. 2007. ‘Counting Civilian Deaths in Iraq.’ The Factchecker. Washington Post Web site. 1 October. <http://blog.
washingtonpost.com/fact-checker/2007/10/counting_
civilian_deaths_in_ir.html>
Fischer, Hannah. 2008. ‘Iraqi Civilian Deaths Estimates.’ CRS
Report for Congress, RS22537. 16 May. <http://ftp.fas.org/
sgp/crs/mideast/RS22537.pdf>
Fottrell, Edward. F. 2008. Dying to Count: Mortality Surveillance
Methods in Resource-Poor Settings. Doctoral dissertation,
Faculty of Medicine, Umeå University.
Garfield, Richard M. and Alfred I. Neugut. 1991. ‘Epidemiologic
Analysis of Warfare: A Historical Review.’ Journal of the
American Medical Association (JAMA). Vol. 266, No. 5.
August, pp. 688–92.
Gayer, Michelle, Dominique Legros, Pierre Formenty, and Maire
A. Connolly. 2007. ‘Conflict and Emerging Infectious Diseases.’ Emerging Infectious Diseases. Vol. 13, No. 11.
November, pp. 1625–31.
Ghobarah, Hazem Adam, Paul Huth and Bruce Russett. 2003.
‘Civil Wars Kill and Maim People—Long After the Shooting
Stops.’ American Political Science Review. Vol. 97, No. 2.
May, pp. 189–202.
Guha-Sapir, Debarati and Olivier Degomme with Mark Phelan.
2005a. Darfur: Counting the Deaths: Mortality Estimates
from Multiple Survey Data. Brussels: CRED. May.
Guha-Sapir, Debarati and Olivier Degomme. 2005b. Darfur: Counting the Deaths (2): What Are the Trends? Brussels: CRED.
Guha-Sapir, Debarati, Olivier Degomme, and Chiara Altare. 2007.
Crude Mortality Rate and Global Acute Malnutrition. Brussels:
CRED.
143
BIBLIOGR APHY
Burr, M. 1993. A Working Document: Quantifying Genocide in the
Southern Sudan 1983–1993. Washington DC: US Committees
for Refugees.
Guha-Sapir, Debarati and Wilbert Gijsbert van Panhuis. 2002.
Armed Conflict and Public Health: A Report on Knowledge
and Knowledge Gaps. Brussels: CRED.
——. 2004. Conflict Mortality: Comparing the Data. Brussels: CRED.
January.
Guha-Sapir, Debarati and Olivier Le Polain de Waroux. n/d. Marburg
and Ebola Hemorrhagic Fever Outbreaks: Piecing the
Epidemiological Puzzle. Unpublished working paper.
Brussels: CRED.
Guha-Sapir, Debarati, Willem Gijsbert van Panhuis, Olivier
Degomme, and Vicente Teran. 2005. ‘Civil Conflicts in
Four African Countries: a Five-Year Review of Trends in
Nutrition and Mortality.’ Epidemiologic Reviews. Vol. 27,
pp. 67–77.
Hagan, John and Alberto Palloni. 2006. ‘Death in Darfur.’ Science.
Vol. 313, No. 5793. 15 September, pp. 1578–79.
Harbom, Lotta, ed. 2007. States in Armed Conflict 2006. Research
Report No. 79. Uppsala: Department of Peace and Conflict
Research, Uppsala University.
HRC (Human Rights Council). 2008. Report of the Special Rapporteur on Violence against Women, its Causes and Consequences. Yakin Ertürk: Mission to the Democratic Republic
of the Congo. A/HRC/7/6/Add.4 of 28 February.
HSC (Human Security Centre). 2004. Deadly Connections: The
War/Disease Nexus. Workshop Report. Vancouver: HSC.
——. 2005. The Human Security Report 2005. Wars and Political
Violence. New York: Oxford University Press.
Hynes, Michelle and Barbara Lopes Cardozo. 2000. ‘Sexual Violence against Refugee Women.’ Journal of Women’s Health
and Gender-Based Medicine. Vol. 9, No. 8, pp. 819–23.
Iraq Body Count. 2008. Iraq Body Count. Accessed June 2008.
<http://www.iraqbodycount.org/>
Iraq Coalition Casualty Count. 2008. Web site. Accessed 21 April.
<http://icasualties.org/>
IRC (International Rescue Committee). 2000. ‘Mortality in Eastern
DRC: Results from Five Mortality Surveys.’ New York: IRC.
Web edition, June 2008. <http://www.theirc.org/media/www/
mortality_study_eastern_dr_congo_aprilmay_2000.html >
1
2
3
4
5
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
144
——. 2001a. ‘Mortality in Eastern Democratic Republic of Congo:
Results from Eleven Mortality Surveys.’ New York: IRC.
Web edition, June 2008. <http://hei.unige.ch/sas/files/
portal/issueareas/victims/Victims_pdf/2001_IRC_DRC.pdf>
——. 2002a. ‘Disability Prevalence Survey. Bujumbura Rural
Province, Burundi.’
——. 2002b. ‘Disability Prevalence Survey. Muyinga Province,
Burundi.’
——. 2002c. ‘Disability Prevalence Survey. Makamba Province,
Burundi.’
——. 2003a. ‘Disability Prevalence Survey. Bujumbura Rural
Province, Burundi.’
——. 2003b. Mortality in the Democratic Republic of Congo:
Results from a Nationwide Survey. New York: IRC.
——. 2004a. Mortality in the Democratic Republic of Congo:
Results from a Nationwide Survey. New York: IRC.
——. 2004b. Mortality in the Pool Region: Results of a Mortality,
Immunization, and Nutrition Survey in Kinkala-Boko Health
District, Pool Region, Republic of Congo. New York: IRC.
——. n.d. ‘The IRC in Burundi’. New York: IRC. Web edition, June 2008.
<http://www.theirc.org/where/the_irc_in_burundi.html>
IRIN (Integrated Regional Information Networks). 2008. ‘Burundi:
Fighting displaces 20,000’. Nairobi: IRINNEWS. 9 May.
Web edition, June 2008. <http://www.irinnews.org/Print
Report.aspx?ReportId=78129>
Lacina, Bethany and Nils Petter Gleditsch. 2005. ’Monitoring
Trends in Global Combat: A New Dataset of Battle Deaths.’
European Journal of Population. Vol. 21, pp. 145–66.
Li, Quan and Ming Wen. 2005. ‘The Immediate and Lingering
Effects of Armed Conflict on Adult Mortality: A Time-Series
Cross-National Analysis.’ Journal of Peace Research. Vol. 42,
No. 4, pp. 471–92.
Mills, Edward and Frederick Burkle. (2008). ‘Correspondence.’
New England Journal of Medicine, Vol. 359, No. 4, p. 432,
25 July.
Murray, Christopher J.L., Gary King, Alan D. Lopez, Niels Tomijima,
and Etienne Krug. 2002. ‘Armed Conflict as a Public Health
Problem.’ British Medical Journal. Vol. 324, pp. 346–49.
Ndayimirije, Nestor and Mary Kay Kindhauser. 2005 ‘Marburg
Hemorrhagic Fever in Angola: Fighting Fear and a Lethal
Panthogen.’ New England Journal of Medicine. Vol. 352,
No. 21, pp. 2155–57.
Nielsen, Jens, Henrik Jensen, Per Kragh Andersen, and Peter Aaby.
2006. ‘Mortality Patterns During a War in Guinea-Bissau
1998–99: Changes in Risk Factors?’ International Journal
of Epidemiology. Vol. 35, No. 2, pp. 438–46.
Obermeyer, Ziad et al. 2008. ‘Fifty Years of Violent War Deaths
from Vietnam to Bosnia: Analysis of Data from the World
Health Survey Programme.’ British Medical Journal. Online
First. <http://bmj.org>
OCHA/IRIN (United Nations Office for the Coordination of Humanitarian Affairs/Integrated Regional Information Networks).
2005. Broken Bodies – Broken Dreams: Violence against
Women Exposed. Kenya: OCHA/IRIN.
Opinion Business Research. 2007. ‘More than 1,000,000 Iraqis
Murdered.’ September. Web edition, June 2008. <http://
www.opinion.co.uk/Newsroom_details.aspx?NewsId=78>
PHR (Physicians for Human Rights). 2002. War-Related Sexual
Violence in Sierra Leone: A Population-Based Assessment.
Boston: PHR. Web edition, June 2008. <http://physicians
forhumanrights.org/library/documents/reports/sexualviolence-sierra-leone.pdf>
Ratnayake, Ruwan Olivier Degomme, Chiara Altare, and Debarati
Guha-Sapir. 2008. Methods and Tools to Evaluate Mortality
in Conflicts: Critical Review, Case-Studies and Applications.
Occasional Paper No. 237. Brussels: CRED.
Refugees International. 2004. Liberia: Major Effort Needed to
Address Gender-Based Violence. 19 January. Web edition,
May 2008. <http://interaction.org/newswire/detail.php?
id=2567>
Roberts, Les, Riyadh Lafta, Richard Garfield, Jamal Khudhairi, and
Gilbert Burnham. 2004. ‘Mortality before and after the
2003 Invasion of Iraq: Cluster Sample Survey.’ The Lancet,
Vol. 364, pp. 1857–64.
Setel, Philip W. et al. 2006. ‘Core Verbal Autopsy Procedures
with Comparative Validation Results from Two Countries.’
PLoS Medicine. Vol. 3, No. 8. August, pp. 1282–91.
Silva, Romesh and Patrick Ball. 2006. The Profile of Human
Rights Violations in Timor-Leste, 1974–1999. A Report by
the Benetech Human Rights Data Analysis Group to the
Commission on Reception, Truth and Reconciliation of
Timor-Leste. Palo Alto: The Benetech Initiative.
Small Arms Survey. 2005. Small Arms Survey 2005: Weapons at
War. Oxford: Oxford University Press.
SMART (Standardized Monitoring and Assessment of Relief and
Transitions). 2005. Measuring Mortality, Nutritional Status,
and Food Security in Crisis Situations: the Smart Protocol.
Smart Methodology. New York and Washington, DC:
UNICEDF and USAID.
Sphere Project. 2004. Humanitarian Charter and Minimum Standards in Disaster Response. Geneva: Sphere Project.
Spiegel, Paul and Peter Salama. 2000. ‘War and Mortality in
Kosovo, 1998–1999: An Epidemiological Testimony’. The
Lancet. Vol. 355. 24 June, pp. 2204–09.
Tapp, Christine, et al. 2008. ‘Iraq War Mortality Estimates: A
Systematic Review.’ Conflict and Health, Vol. 2, No. 1.
tarian Emergencies.’ Bulletin of the World Health Organization. Vol. 83, No. 8. August, pp. 584–89.
Thoms, Oskar N. and James Ron. 2007. ‘Public Health, Conflict
and Human Rights: Toward a Collaborative Research
Agenda.’ Conflict and Health. Vol. 1, No. 11. November.
doi:10.1186/1752-1505-1-11.
Toole, Michael J. and Ronald, J. Waldman. 1997. ‘The Public
Health Aspects of Complex Emergencies and Refugee
Situations.’ Annual Review of Public Health. Vol. 18,
pp. 283–312.
TRC (Truth and Reconciliation Commission). 2005. Witness to
Truth: Report of the Sierra Leone Truth and Reconciliation
Commission. Volume 3b. Freetown: Government of Sierra
Leone.
UBOS (Uganda Bureau of Statistics). 2006. 2002 Uganda Population and Housing Census. Kampala: UBOS.
UNECOSOC (United Nations Economic and Social Council),
Commission on Human Rights. 1996. Report on the
Situation of Human Rights in Rwanda submitted by Mr.
René Degni-Segui, Special Rapporteur of the Commission
on Human Rights under Paragraph 20 of the Resolution
S-3/1 of 25 May 1994. UN doc. E/CN.4/1996/68 of 29
January.
UNHCR (United Nations High Commissioner for Refugees). 2007.
Statistical Yearbook 2006: Trends in Displacement, Protection and Solutions. Geneva: UNHCR.
UNICEF. 2005. The Impact of Conflict on Women and Girls in West
and Central Africa and the UNICEF Response. New York:
UNICEF. February.
UNSC (United Nations Security Council). 2008. Resolution 1820
(2008). S/RES/1820 (2008) of 19 June.
Utzinger, Jürg and Mitchell G. Weiss. 2007. ‘Editorial: Armed
Conflict, War and Public Health.’ Tropical Medicine and
International Health. Vol. 12, No. 8, pp. 903–06.
Wakabi, Wairagala. 2008. ‘Sexual Violence Increasing in Democratic Republic of Congo.’ The Lancet. Vol. 371, No. 9606.
January, pp. 15–16.
WHO (World Health Organization). 2005. Health and Mortality
Survey Among Internally Displaced Persons in Gulu,
Kitgum and Pader Districts, Northern Uganda. Geneva:
WHO.
Working Group for Mortality Estimation in Emergencies. 2007.
‘Wanted: Studies on Mortality Estimation Methods for
Humanitarian Emergencies: Suggestions for Future Research.’
Emerging Themes in Epidemiology. Vol. 4, No. 9. June.
doi:10.1186/1742-7622-4-9.
Chapter Three
Bastick, Megan, Karin Grimm, and Rahel Kunz. 2007. Sexual
Violence in Armed Conflict: Global Overview and Implications for the Security Sector. Geneva: Democratic
Control of Armed Forces.
Caumartin, Corinne. 2005. Racism, Violence, and Inequality:
An Overview of the Guatemalan Case. Working Paper
No. 11. Oxford: CRISE, University of Oxford.
Chaudhary, Torunn and Astri Suhrke. 2008. Postwar Violence.
Unpublished background paper. Geneva: Small Arms
Survey.
145
BIBLIOGR APHY
Thieren, Michael. 2005. ‘Health Information Systems in Humani-
Cincotta, Richard. 2008. Young Guns: The Demographics of
Discord. Unpublished background paper. Geneva:
Small Arms Survey.
Collier, Paul and Anke Hoeffler. 2004. Murder by Numbers:
Socio-economic Determinants of Homicide and Civil
War. Working Paper No. 2004-10. Oxford: Centre for the
Study of African Economies, University of Oxford.
—— and Måns Söderboom. 2006. Post-Conflict Risks. Working
Paper No. 256. Oxford: Centre for the Study of African
Economies, University of Oxford. <http://www.bepress.
com/csae/paper256>.
Collier, Paul, Lani Elliott, Håvard Hegre, Anke Hoeffler, Marta
Reynal-Querol, and Nicholas Sambanis. 2003. Breaking
the Conflict Trap: Civil War and Development Policy.
Washington, DC: World Bank.
Cohen, Lawrence E., and Kenneth C. Land. 1987. ‘Age Structure
and Crime: Symmetry Versus Asymmetry and the Projection of Crime Rates through the 1990s.’ American
Sociological Review, Vol. 52, No. 2. April, pp. 170–83.
Cooper, Neil. 2006. ‘Peaceful Warriors and Warring Peacemakers.’ Economics of Peace and Security Journal,
Vol. 1, No. 1. January, pp. 20–24.
Cramer, Christopher. 2002. ‘Homo Economicus Goes to War:
Methodological Individualism, Rational Choice and the
Political Economy of War.’ World Development, Vol. 30,
No. 11. November, pp. 1845–64.
Daponte, Beth. 1993. ‘A Case Study in Estimating Casualties
from War and its Aftermath: The 1991 Persian Gulf War.’
PSR Quarterly, Vol. 3, No. 2, pp. 57–66.
Darby, John. 2001. The Effects of Violence on Peace Processes.
Washington, D.C.: United States Institute of Peace.
——, ed. 2006. Violence and Reconstruction. Notre Dame:
University of Notre Dame.
Diprose, Rachel and Frances Stewart. 2008. Horizontal Inequalities and Violent Conflict. Unpublished background paper.
Geneva: Small Arms Survey.
1
2
3
4
5
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
146
Diprose, Rachel and Ukiwa Ukiwo. 2008. The Complexities of
Managing Ethno-religious Diversity: A Comparison of
Decentralization and Conflict Dynamics in Indonesia and
Nigeria. Working Paper No. 49. Oxford: CRISE, University
of Oxford.
Mancini, Luca. 2008. ‘Horizontal Inequality and Communal
Violence: Evidence from Indonesian Districts.’ In Frances
Stewart, ed. Horizontal Inequalities and Conflict: Understanding Group Violence in Multiethnic Societies. London:
Palgrave Macmillan, pp. 106–135.
Fearon, James, and David Laitin. 1996. ‘Explaining Interethnic
Cooperation.’ American Political Science Review, Vol. 90,
No. 4, pp. 715–35.
Mac Ginty, Roger. 2006. ‘Post Accord Crime’, in John Darby,
ed., pp. 101–19.
Ford Institute for Human Security. 2008. What Makes a Camp
Safe? The Protection of Children from Abduction in Internally Displaced Persons and Refugee Camps. Pittsburgh:
University of Pittsburgh. Accessed July 2008. <http://
www.fordinstitute.pitt.edu/docs/23182ReportPR11.pdf>
Ghobarah, Hazem, Paul Huth, and Bruce Russett. 2003. ‘Civil
Wars Kill and Maim People—Long After the Shooting
Stops.’ American Political Science Review, Vol. 97, No. 2.
May, pp. 189–202.
Goodhand, Jonathan. 2005. ‘Frontiers and Wars: The Opium
Economy in Afghanistan.’ Journal of Agrarian Change,
Vol. 5, No. 2, pp. 191–216.
Hoeffler, Anke and Marta Reynal-Querol. 2003. Measuring the
Costs of Conflict. Oxford: Center for the Study of African
Economies, University of Oxford.
IDMC (Internal Displacement Monitoring Center). 2008. Internal
Displacement: Global Overview of Trends and Developments in 2007. Geneva: IDMC and Norwegian Refugee
Council. <http://www.unhcr.org/refworld/docid/48074
b842.html>
IRC (International Rescue Committee). 2007. Mortality in the
Democratic Republic of Congo: An Ongoing Crisis. Atlanta:
IRC and Burnet Institute.
IRIN. 2008. ‘Sierra Leone: Sex Crimes Continue in Peacetime.’
Freetown: IRINNEWS. 20 June. <http://www.irinnews.org/
report.aspx?ReportID=78853>
Kellah, Amie-Tejan. 2007. ‘Establishing Services in Postconflict Sierra Leone.’ Forced Migration Review. No. 27.
January.
Kreutz, Joakim, Nicholas Marsh, and Manuela Torre. 2007.
Regaining State Control: Arms and Violence in PostConflict Countries. Unpublished background document.
Oslo: PRIO.
Langer, A. 2008. ‘When Do Horizontal Inequalities Lead to
Conflict? Lessons from a Comparative Study of Ghana
and Côte d’Ivoire’. In Frances Stewart, ed. Horizontal
Inequalities and Conflict: Understanding Group Violence
in Multiethnic Societies. London: Palgrave Macmillan,
pp. 163–89.
Licklider, Roy. 1995. ‘The Consequences of Negotiated Settlements in Civil Wars, 1945–1993.’ American Political
Science Review, Vol. 89, No. 3, pp. 681–90.
Muggah, Robert, ed. 2006. No Refuge: The Crisis of Refugee
Militarization in Africa. London: Zed Books.
——, ed. 2008. Security and Post-Conflict Reconstruction:
Dealing with Fighters in the Aftermath of War. New York:
Routledge.
Nathan, Laurie. 2005. The Frightful Inadequacy of Most of the
Statistics: A Critique of Collier and Hoeffler on Causes of
Civil War. Discussion Paper No. 11. London: Crisis
States Research Centre, London School of Economics.
OECD (Organisation for Economic Co-operation and Development). 2008. Armed Violence Reduction Guidance:
Draft. Paris: Small Arms Survey/SecDev Associates.
Østby, Gudrun. 2008. ‘Polarization, Horizontal Inequalities
and Violent Civil Conflict.’ Journal of Peace Research,
Vol. 45, No. 2, pp. 143–62.
PHR (Physicians for Human Rights). 2002. War-related Sexual
Violence in Sierra Leone: A Population-based Assessment.
Boston: PHR. Web edition. <http://physiciansforhuman
rights.org/library/documents/reports/sexual-violencesierra-leone.pdf>
Picciotto, Robert and Sakiko Fukuda-Parr. 2008 (forthcoming).
Conflict Prevention and Development Cooperation in
Africa. A Joint project of JICA and UNDP.
Pugh, Michael. 2005. ‘The Political Economy of Peacebuilding:
A Critical Theory Perspective.’ International Journal of
Peace Studies, Vol. 10, No.2, pp. 23–42.
Reynal-Querol, Marta. 2002. ‘Ethnicity, Political Systems, and
Civil wars’. Journal of Conflict Resolution, Vol. 46, No. 1,
pp. 29–54.
Rodgers, Dennis. 2006. ‘The State as a Gang: Conceptualizing
the Governmentality of Violence in Contemporary Nicaragua.’ Critique of Anthropology, Vol. 26, No. 3, pp. 315–30.
Spear, Joanna. 2006. ‘From Political Economies of War to Political Economies of Peace: The Contribution of DDR after
Wars of Predation.’ Contemporary Security Policy,
Vol. 27, No. 1, pp. 168–89.
Small Arms Survey. 2005 Small Arms Survey 2005: Weapons
at War. Oxford: Oxford University Press.
——. 2006. Small Arms Survey 2006: Unfinished Business.
Oxford: Oxford University Press.
——. 2007. Small Arms Survey 2007: Guns and the City.
Cambridge: Cambridge University Press.
Staveteig, Sarah. 2005. ‘The Young and the Restless: Population Age Structure and Civil War.’ In Environmental
Change and Security Project Report, Issue 11, pp. 12–19.
Stewart, Frances. 2008. ‘Horizontal Inequalities and Conflict:
An Introduction and Some Hypotheses’. In Frances
Stewart, ed. Horizontal Inequalities and Conflict: Understanding Group Violence in Multiethnic Societies. London:
Palgrave Macmillan, pp. 3–24.
——, Graham Brown, and Arnim Langer. 2008. ‘Policies towards
Horizontal Inequalities’. In Frances Stewart, ed. Horizontal Inequalities and Conflict: Understanding Group
Violence in Multiethnic Societies. London: Palgrave
Macmillan, pp. 301–25.
Suhrke, Astri, and Ingrid Samset. 2007. ‘What’s in a Figure?
Estimating Recurrence of Civil War’. International
Peacekeeping, Vol. 14, No. 2, pp. 195–203.
UCDP (Uppsala Conflict Data Program). Conflict Termination
Dataset v.2.0, 1946–2006. Data retrieved 18 June 2008.
Uppsala: Department of Peace and Conflict Research.
UNHCR (United Nations High Commissioner for Refugees). 2008.
2007 Global Trends: Refugees, Asylum-seekers, Returnees,
Internally Displaced and Stateless Persons. Geneva: UNHCR.
United Nations Population Division. 2007. Population Prospects,
the 2006 Revision. United Nations: New York. <http://
www.un.org/esa/population/publications/wpp2006/
wpp2006.htm>
UNODC (United Nations Office on Drugs and Crime). 2005.
Crime and Development in Africa. Vienna: UNODC.
UNOWA (United Nations Office for West Africa). 2007. Urbanisation and Insecurity in West Africa: Population Movements,
Mega Cities and Regional Stability. Dakar: UNOWA.
Urdal, Henrik. 2006. ‘A Clash of Generations? Youth Bulges
and Political Violence.’ International Studies Quarterly,
Vol. 50, No. 3, pp. 607–30.
US Department of Justice. Bureau of Justice Statistics—Data
Online. Accessed 1 May 2008. <http://www.ojp.usdoj.
gov/bjs>
Walter, Barbara F. 2004. ‘Does Conflict Beget Conflict? Explaining Recurring Civil War.’ Journal of Peace Research, Vol. 41,
No. 3, pp. 371–88.
White, Richard. 1981. ‘Outlaw Gangs of the Middle Border:
American Social Bandits.’ Western Historical Quarterly,
Vol. 12, No. 4, pp. 387–408.
Chapter Four
Aebi, Marcelo F. 2004. ‘Crime Trends in Western Europe from
1990 to 2000.’ European Journal on Criminal Policy and
Research, Vol. 10, Nos. 2–3, pp. 163–86.
BBC (British Broadcasting Corporation). 2006. ‘Police Revise Sao
Paulo Toll Down.’ Web edition, 24 May. Accessed June 2008.
<http://news.bbc.co.uk/2/hi/americas/5010962.stm>
——. 2008. ‘Around Kenya: After the Violence.’ Web edition,
12 February. Accessed June 2008. <http://news.bbc.co.uk/
2/hi/africa/7239554.stm>
147
BIBLIOGR APHY
——. 2008. Small Arms Survey 2008: Risk and Resilience.
Cambridge: Cambridge University Press.
Bye, Elin K. 2008. ‘Alcohol and Homicide in Eastern Europe: A
Time Series Analysis of Six Countries.’ Homicide Studies,
Vol. 12, No. 1, pp. 7–27.
Campaña de prevención de violencia de género en El Salvador.
2006. ‘Cada 18 segundos podemos construir une vida
diferente.’ Campaign organized by Una vida diferente.
Accessed June 2008. <http://www.unavidadiferente.org/
IE01OCT2006.html>
Campbell, Jacquelyn C., et al. 2003. ‘Risk Factors for Femicide in
Abusive Relationships: Results From a Multisite Case
Control Study.’ American Journal of Public Health, Vol. 93,
No. 7, pp. 1089–97. July.
Collier, Paul and Anke Hoeffler. 2004. Murder by Numbers: SocioEconomic Determinants of Homicide and Civil War. Centre
for the Study of African Economies Working Paper No. 210.
Connecticut: EconWPA. Accessed June 2008. <http://www.
bepress.com/csae/paper210>
Eisner, Manuel. 2001. ‘Modernization, Self-Control and Lethal
Violence: The Long-term Dynamics of European Homicide
Rates in Theoretical Perspective.’ British Journal of Criminology, Vol. 41, pp. 618–38.
Fajnzylber, Pablo, Daniel Lederman, and Norman Loayza. 1998.
Determinants of Crime Rates in Latin America and the
World. Washington, DC: World Bank.
Garrido, Vicente, Per Stangeland, and Santiago Redondo. 2001.
Principios de criminologia, 2nd edn. Valencia: Tirant Lo
Blanch.
Gartner, Rosemary and Robert Nash Parker. 1990. ‘Cross-National
Evidence on Homicide and Age Structure of the Population.’
Social Forces, Vol. 69, No. 2, pp. 351–71. Chapel Hill:
University of North Carolina Press.
Glaeser, Edward and Bruce Sacerdote. 1996. Why Is There More
Crime in Cities? NBER Working Paper No. W5430. Cambridge,
Mass.: National Bureau of Economic Research. January.
WHO (World Health Organization). 2008a. WHO Report on
Violence in Africa. Geneva: WHO.
Gurr, Ted Robert. 1981. ‘Historical Trends in Violent Crime: A
Critical Review of the Evidence.’ Crime and Justice, Vol. 3,
pp. 295–353.
——. 2008b. Preventing Violence and Reducing its Impact: How
Development Agencies and Governments Can Help.
Geneva: WHO.
Harris, Anthony R., et al. 2002. ‘Murder and Medicine: The Lethality
of Criminal Assault 1960–1999.’ Homicide Studies, Vol. 6,
No. 2, pp. 128–66.
1
2
3
4
5
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
148
IIDH (Instituto Interamericano de Derechos Humanos Secretaria
Técnica). 2006. Situación y análisis del feticidio en la Región
Centroamericana. Web edition. Accessed May 2008. <http://
www.mujereshoy.com/media/Femicidio2006IIDHDocAdj.pdf>
Jamaica Police Constabulary. 2005. ‘Homicide by Gender.’
Unpublished data.
Kellermann, Arthur L., et al. 1993. ‘Gun Ownership as a Risk Factor
for Homicide in the Home.’ New England Journal of Medicine, Vol. 329, No. 15, pp. 1084–91. Accessed June 2008.
<http://content.nejm.org/cgi/content/full/329/15/1084>
LaFree, Gary. 1999. ‘A Summary and Review of Cross National
Comparative Studies of Homicide.’ In D. Smith and M. A.
Zahn, eds. Homicide Studies: A Sourcebook of Social
Research. Thousand Oaks: Sage, pp. 125–48.
——. 2005. ‘Evidence for Elite Convergence in Cross-National
Homicide Victimization Trends, 1956 to 2000’. Sociological
Quarterly, Vol. 46, No. 1, pp. 191–211.
—— and Kenneth A. Drass. 2002. ‘Counting Crime Booms among
Nations: Evidence for Homicide Victimization Rates, 1956
to 1998.’ Criminology, Vol. 40, No. 4, pp. 769–800. Rockville: NCJRS.
Latinobarómetro. 2004. Summary Report Latinobarómetro 2004:
A Decade of Measurements. Santiago: Corporación
Latinobarómetro. Web edition. Accessed June 2008.
<http://www.latinobarometro.org/>
Los Angeles Times. 2008. ‘Toll Mounts in Mexico’s Drug War.’
Web edition, 3 June. Accessed June 2008. <http://www.
latimes.com/news/nationworld/world/la-fg-border3-2008
jun03,0,1109808,print.story>
Monkkonen, Eric. 2001. ‘New Standards for Historical Homicide
Research.’ Crime, histoire et sociétés, Vol. 5, No. 2, pp. 5–26.
Guyancourt: International Association for the History of
Crime and Criminal Justice.
Naudé, Beaty, Johan Prinsloo, and Anastasios Ladikos. 2006.
Experience of Crime in Thirteen African Countries: Results
from the International Crime Victim Survey. Turin: UNICRI–
UNODC. Accessed June 2008. <http://rechten.uvt.nl/icvs/
pdffiles/ICVS%2013%20African%20countries.pdf>
Neapolitan, Jerome L. and Frank Schmalleger. 1997. Cross National
Crime: A Research Review and Sourcebook. Westport:
Greenwood Press.
PAHO (Pan American Health Organization). n.d. PAHO Basic
Indicator Data Base. Accessed June 2008. <http://www.
paho.org/English/DD/AIS/coredata.htm>
Reuters. 2008. ‘Suspected Drug Gangs Kill Mayor in Western Mexico.’
Web edition, 2 June. Accessed June 2008. <http://www.
reuters.com/article/worldNews/idUSN0230481520080602>
Richmond, Therese S., Rose Cheney, and C. William Schwab.
2005. ‘The Global Burden of Non-conflict Related Firearm
Mortality.’ Injury Prevention, Vol. 11, pp. 348–52. London:
BMJ Publishing Group. Accessed June 2008. <http://injury
prevention.bmj.com/cgi/ content/full/11/6/348>
Robertson, Annette. 2004. ‘Police Reform in Russia.’ In Marina
Caparini and Otwin Marenin, eds. Transforming Police in
Central and Western Europe: Process and Progress.
Muenster: Lit Verlag, pp. 289–304.
Shaw, Mark, Jan van Dijk, and Wolfgang Rhomberg. 2003. ‘Determining Global Trends in Crime and Justice: An Overview of
Results from the United Nations Surveys of Crime Trends
and Operations of Criminal Justice Systems.’ Forum on
Crime and Society, Vol. 3, Nos. 1–2, pp. 35–63.
SIM Datasus. n.d. Database. Accessed June 2008. <http://www.
datasus.gov.br/>. Data processed by Viva Rio.
Small Arms Survey. 2004. Small Arms Survey 2004: Rights at
Risk. Oxford: Oxford University Press.
——. Small Arms Survey 2007: Guns and the City. Cambridge:
Cambridge University Press.
Smit, Paul R., Ronald F. Meijer, and Peter-Paul J. Groen. 2004.
‘Detection Rates, an International Comparison.’ European
Journal on Criminal Policy and Research, Vol. 10, Nos. 2–3,
pp. 225–53.
Soares, Rodrigo R. 2004a. ‘Crime Reporting as a Measure of
Institutional Development.’ Economic Development and
Cultural Change, Vol. 52, pp. 851–71.
——. 2004b. ‘Development, Crime and Punishment: Accounting
for the International Differences in Crime Rates.’ Journal
of Development Economics, Vol. 73, pp. 155–84.
Stamatel, Janet P. 2008. ‘Using Mortality Data to Refine Our
Understanding of Homicide Patterns in Selected Postcommunist Countries.’ Homicide Studies, Vol. 12, No. 1,
pp. 117–35.
NIMSS (National Injury Mortality Surveillance System). 2001. A
Profile of Fatal Injuries in South Africa, 2000. Second
Annual Report. NIMSS. October.
Tavares, Cynthia and Geoffrey Thomas. 2007. ‘Crime and Criminal
Justice.’ Statistics in Focus: Population and Social Conditions. Brussels: European Community. Web edition. Accessed
January 2008. <http://epp.eurostat.ec.europa.eu/cache/
ITY_OFFPUB/KS-SF-07-015/EN/KS-SF-07-015-EN.PDF>
Observatorio Centroamericano sobre Violencia. 2007. ‘Victimas
de Homicidios Dolosos en Centroamérica y Republica
Dominicana (1999–2007).’ Accessed June 2008. <http://
www.ocavi.com/docs_files/file_379.pdf>
UN (United Nations). 2006. The Ninth United Nations Survey on
Crime Trends and the Operations of Criminal Justice Systems (2003–2004). Accessed July 2008. <http://www.
unodc.org/unodc/en/data-and-analysis/Ninth-United-
——. 2007a. World Crime Trends and Responses: Integration and
Coordination of Efforts by the United Nations Office on
Drugs and Crime and by Member States in the Field of
Crime Prevention and Criminal Justice. E/CN.15/2007/2.
Commission on Crime Prevention and Criminal Justice,
Sixteenth Session. Economic and Social Council, Vienna,
23–27 April.
——. Division of Policy Analysis and Public Affairs. 2007b. Questionnaire for the Tenth United Nations Survey of Crime
Trends and Operations of Criminal Justice Systems, covering
the period 2005–2006. Vienna: United Nations Office on
Drugs and Crime. Accessed July 2008. <http://www.unodc.
org/pdf/crime/tenthsurvey/10th_CTS_form_English.pdf>
VPC (Violence Policy Center). 2007. When Men Murder Women:
An Analysis of 2005 Homicide Data. Washington, DC: VPC.
September.
WHO (World Health Organization). n.d. Database. Accessed July
2008. <http://www.who.int/healthinfo/morttables/en/
index.html>
——. 2002. World Report on Violence and Health. Geneva: WHO.
——. 2006. WHO Projected Deaths by WHO Region, Age, Sex and
Cause. Geneva: WHO.
——. Forthcoming. The Global Burden of Disease. Geneva: WHO.
149
BIBLIOGR APHY
Nations-Survey-on-Crime-Trends-and-the-Operations-ofCriminal-Justice-Systems.html>
World Bank Group. 2008. World Development Indicators. WDI
online. Web edition. Accessed May 2008. <http://ddpext.
worldbank.org/ext/DDPQQ/member.do?method=get
Members&userid=1&queryId=6>
——. 2008. The Tenth United Nations Survey on Crime Trends and
the Operations of Criminal Justice Systems (2005–2006).
Accessed June 2008. <http://www.unodc.org>
UNDP (United Nations Development Programme). 2008. Human
Development Report Statistics 2007/2008. <http://hdrstats.
undp.org/indicators/>
UNECE (United Nations Economic Commission for Europe). 2008.
Gender Statistics: Crime and Violence. Online database.
Accessed June 2008. <http://w3.unece.org/pxweb/DATA
BASE/STAT/30-GE/07-CV/07-CV.asp>
UN-HABITAT (United Nations Human Settlement Programme). 2006.
State of the World’s Cities 2006/7. London: Earthscan.
Chapter Five
Ahmed, Ismail and Reginald Green. 1999. ‘The Heritage of War
and State Collapse in Somalia and Somaliland: Local-level
Effects, External Interventions and Reconstruction.’ Third
World Quarterly, Vol. 20, No. 1, pp. 113–27.
Arunatilake, Nisha, Sisisra Jayasuriya, and Saman Kelegama.
2001. ‘The Economic Cost of the War in Sri Lanka.’ World
Development, Vol. 29, No. 9, pp. 1483–500.
UNODC (United Nations Office on Drugs and Crime). 2005. Crime
and Development in Africa. Vienna: UNODC. Accessed June
2008. <http://www.unodc.org/pdf/African_report.pdf>
Ballentine, Karen and Jake Sherman. 2003. The Political Economy
of Armed Conflict: Beyond Greed and Grievance. Boulder:
Lynne Rienner.
——. 2007. Crime and Development in Central America: Caught in
the Crossfire. Vienna: UNODC. Accessed June 2008. <http://
www.unodc.org/pdf/research/Cr_and_Vio_Car_E.pdf>
Bastick, Megan, Karing Grimm, and Rahel Kunz. 2007. Sexual
Violence in Armed Conflict: Global Overview and Implications for the Security Sector. Geneva: DCAF.
——. 2008. Crime and Its Impact on the Balkans and Affected
Countries. Vienna: UNODC. Accessed June 2008. <http://
www.unodc.org/pdf/research/Cr_and_Vio_Car_E.pdf>
Bates, Robert. 2008. When Things Fell Apart: State Failure in
Late Century Africa. Cambridge: Cambridge University
Press.
—— and World Bank. 2007. Crime, Violence and Development:
Trends, Costs and Policy Options in the Caribbean. Vienna:
UNODC. Accessed June 2008. <http://www.unodc.org/
pdf/research/Cr_and_Vio_Car_E.pdf>
Ben-Porat, Guy. 2005. ‘Between Power and Hegemony: Business
Communities in Peace Processes.’ Review of International
Studies, Vol. 31, No. 2, pp. 325–48.
Van Dijk, Jan J. M. 1998. ‘Criminal Victimisation: A Global View.’
In Surveying Crime: A Global Perspective. Proceedings of the
International Conference, 13–21 November. Rome: UNICRI.
Van Wilsem, Johan. 2004. ‘Criminal Victimization in Cross-National
Perspective: An Analysis of Rates of Theft, Violence and
Vandalism across 27 countries.’ European Journal of
Criminology, Vol. 1, No. 1, pp. 89–109.
Vision of Humanity. 2007. The Global Peace Index. Accessed
January 2008. <http://www.visionofhumanity.org/gpi/
home.php>
Bilmes, Linda and Joseph Stiglitz. 2006. The Economic Costs of
the Iraq War: An Appraisal Three Years after the Beginning
of the Conflict. NBER Working Papers, No. 12054. Cambridge:
National Bureau for Economic Research.
Bodea, Christina and Ibrahim A. Elbadawi. 2007. Riots, Coups
and Civil War: Revisiting the Greed and Grievance Debate.
Post-Conflict Transition-Policy Research Working Paper,
No. 34397. Development Economic Research Group.
Washington, DC: World Bank.
Bozzoli, Carlos, Tilman Brück, Thorsten Drauzenburg, and Simon
Sottas. 2008. Economic Costs of Mass Violent Conflict:
1
2
3
4
5
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
150
Final Report for the Small Arms Survey. Politikberatung
Kompakt, No. 42. Berlin: Deutsches Institut für
Wirtschaftsforschung.
Brand, Sam and Richard Price. 2000. The Economic and Social
Cost of Crime. Home Office Research Study, No. 217. London:
Home Office.
Brück, Tilman. 2002. ‘Ludwig Erhard in Africa: War Finance and
Post-war Reconstruction in Germany and Mozambique.’
In J. Bruaer and P. Dunne, eds. The Economics of Military
Expenditure and Arms Production and Trade in Developing
Countries. London: Macmillan, pp. 235–50.
CICS (Centre for International Cooperation and Security). 2005.
The Impact of Armed Violence on Poverty and Development.
Full Report of the Armed Violence and Poverty Initiative.
Bradford: University of Bradford.
Coale, Ansley and Guang Guo. 1989. ‘Revised Regional Model
Life-tables at Very Low Levels of Mortality.’ Population
Index, Vol. 55, No. 4, pp. 613–43.
Collier, Paul. 1999. ‘On the Economic Consequences of Civil War.’
Oxford Economic Papers, Vol. 51, pp. 168–83.
—— and Anke Hoeffler. 2004a. ‘Aid, Policy and Growth in Postconflict Societies.’ European Economic Review, Vol. 48,
pp. 1125–45.
—— and Marc Vothknecht. 2007. Women and Post-Conflict Economic Recovery. Background Paper, Project on Fostering
Post-Conflict Economic Recovery, Second Advisory Panel
Meeting, UNDP Bureau for Crisis Prevention and Recovery.
New York: Manhasset.
——. 2004b. ‘Conflicts.’ In Bjorn Lamborg, ed. Global Crises,
Global Solution. Cambridge: Cambridge University Press,
pp. 129–74.
Burns-Mackenzie, Regina and Margie Buchanan-Smith. 2005.
Armed Violence and Poverty in Southern Sudan: A Case
Study for the Armed Violence and Poverty Initiative. Centre
for International Cooperation and Security. Bradford:
University of Bradford.
Development Initiatives. 2008. Global Humanitarian Assistance.
London: Development Initiatives.
Butchart, Alexander, et al. 2008. Manual for Estimating the
Economic Costs of Injuries due to Interpersonal and Selfdirected Violence. Geneva: WHO.
CBO (Congressional Budget Office). 2007. Estimated Cost of U.S.
Operations in Iraq and Afghanistan and of Other Activities
Related to the War on Terrorism. Washington, DC: Congress
of the United States. 24 October.
CDCP (Centers for Disease Control and Prevention). 2003. Costs
of Intimate Partner Violence against Women in the United
States. Atlanta: CDCP.
Cerquiera, Daniel, et al. 2007. Analise dos custos e consequencias da violencia no Brazil. Brasilia: Insituto de Pesquisa
Economica Aplicada.
Charney, Craig. 1999. ‘Civil Society, Political Violence and Democratic Transitions: Business and the Peace Process in South
Africa: 1990–1994.’ Comparative Studies in Society and
History, Vol. 41, No. 1, pp. 192–206.
Cramer, Christopher. 2006. Civil War Is not a Stupid Thing: Accounting for Violence in Developing Countries. London: Hurst.
DiAddario, Sabrina. 1997. ‘Estimating the Economic Costs of
Conflict: An Examination of the Two-Gap Estimation Model
for the Case of Nicaragua.’ Oxford Development Studies,
Vol. 25, No. 1, pp. 123–42.
Dorsey, Jeff and Steven Opeitum. 2002. The Net Economic Cost
of the Conflict in the Acholiland Sub-region of Uganda.
Background Paper for Civil Society Organisation for Peace
in Northern Uganda. <http://www.scribd.com/doc/
322330/ economic-cost-of-conflict-final-edited-version6-oct-20025b15d>
Elbadawi, Ibrahim A. 2008. The Security Challenge in Conflictprone Countries. Conflicts Challenge Paper. Copenhagen:
Copenhagen Consensus Centre. Web edition. Accessed
June 2008. <http://www.humansecuritygateway.info/
documents/CP_Mack_securitychallengesinconflictprone
countries.pdf>
Fielding, David. 2003. ‘Counting the Cost of the Intifada: Consumption, Saving and Political Instability in Israel.’ Public Choice,
Vol. 116, Nos. 3–4, pp. 297–312. September.
Chauvet, Lisa, Paul Collier, and Haavard Hegre. 2008. The Security
Challenge in Conflict-prone Countries. Conflicts Challenge
Paper. Copenhagen: Copenhagen Consensus Centre. Web
edition. Accessed June 2008. <http://www.humansecurity
gateway.info/documents/CP_Collier_securitychallengein
conflictpronecountries.pdf>
FitzGerald, Valpy. 1987. ‘An Evaluation of the Economic Costs to
Nicaragua of U.S. Aggression: 1980–1984.’ In Rose J.
Spalding, ed. The Political Economy of Revolutionary
Nicaragua. Boston: Allen and Unwin, pp. 195–213.
Chen, Siyan, Norman Loayza, and Marta Reynal-Querol. 2007. The
Aftermath of Civil War. Post-Conflict Transitions Working
Paper, No. 4. Washington, DC: World Bank.
Grobar, Lis and Shiranthi Gnanaselvam. 1993. ‘The Economic
Effects of the Sri Lankan Civil War.’ Economic Development
and Cultural Change, Vol. 41, No. 2, pp. 395–405.
Chiang, Chin. 1984. The Life Table and Its Applications. Malabar:
Robert E. Krieger.
Gutierrez, Francisco. 2008. Stupid and Expensive? A Critique of
the Costs of Violence Literature. Small Arms Survey Back-
Galtung, J. 1969. ‘Violence, Peace, and Peace Research.’ Journal
of Peace Research, Vol. 6, No. 3, pp. 167–91.
Harris, Geoff. 1997. ‘Estimates of the Economic Cost of Armed
Conflict: The Iran–Iraq war and the Sri Lankan Civil War.’
In Jurgen Brauer and William Gissy, eds. Economics of
Conflict and Peace. Aldershot: Avebury, pp. 269–91.
Hess, Gregory. 2003. ‘The Economic Welfare Cost of Conflict: An
Empirical Assessment.’ Working Papers in Economics.
Claremont: Claremont Colleges.
Hoeffler, Anke and Marta Reynal-Querol. 2003. Measuring the
Costs of Conflict. Washington, DC: World Bank.
IADB (Inter-American Development Bank). 2006. ‘Preventing
Violence.’ Technical Note, No. 5. Washington, DC: IADB.
Jean, François and Jean-Christophe Rufin, eds. 1996. Economie
des guerres civiles. Paris: Hachette.
Centre Working Paper, Vol. 28, No. 2. London: London
School of Economics.
Naylor, Thomas. 2002. Wages of Crime: Black Markets, Illegal
Finance, and the Underworld Economy. Ithaca: Cornell
University Press.
Oxfam-GB. 2007. ‘Africa’s Missing Billions: International Arms
Flows and the Costs of Conflict.’ Briefing Paper, No. 107.
Oxford: Oxfam-GB.
Portland Trust. 2007. Economics in Peacemaking: Lessons from
Northern Ireland. London: Portland Trust.
Preston, Samuel, Patrick Heuveline, and Michel Guillot. 2001.
Demography: Measuring and Modeling Population Processes. Malden: Blackwell.
Reno, William. 1999. Warlord Politics and African States. Boulder:
Lynne Rienner.
Keen, David. 1998. The Economic Functions of Violence in Civil
Wars. Adelphi Papers, No. 320. London: Routledge.
Restrepo, Jorge, Brodie Ferguson, Jukliana M. Zúñiga, and
Adriana Villamarin. 2008. Estimating Lost Product Due to
Violent Deaths in 2004. Unpublished background paper
for the Small Arms Survey. Geneva/Bogota: Small Arms
Survey/CERAC.
Kelegama, Saman. 1999. ‘Economic Costs of Conflict in Sri Lanka.’
In Robert Rotberg, ed. Creating Peace in Sri Lanka: Civil
War and Reconciliation. Washington, DC: Brookings Institution Press, pp. 71–87.
Richardson, John and S. W. R. Samarasinghe. 1991. ‘Measuring
the Economic Dimensions of Sri Lanka’s Ethnic Conflict.’ In
S. W. R. Samarasinghe and Reed Coughlan, eds. Economic
Dimensions of Ethnic Conflict. London: Pinter, pp. 194–223.
Keynes, John M. 1978. ‘Paying for War.’ In D. Moggridge, ed.
Collected Writings of J. M. Keynes. London: Macmillan.
SIPRI (Stockholm International Peace Research Institute). 2007.
SIPRI Yearbook 2007: Armaments, Disarmament and
International Security. Oxford: Oxford University Press.
Joras, Ulrike. 2007. Companies in Peace Processes: A Guatemalan
Case Study. Bielefeld: Transcript Verlag.
——. 2005 [1919]. The Economic Consequences of Peace. New
York: Cosimo.
Londoño, Juan L. and Rodrigo Guerrero. 1999. Violencia en America Latina: Epidemiologia y Costos. Washington, DC: IADB.
Lopez, Humberto. 2001. The Cost of Armed Conflict in Central
America. Washington, DC: World Bank.
Merlo, Antonio. 2004. ‘Introduction to Economic Models of Crime.’
International Economic Review, Vol. 45, No. 3, pp. 677–79.
Mihalache, Andreea. 2008. ‘Gambling on Conflict: Profiling Investments in Conflict Countries.’ Presentation to Midwest
Political Science Association 66th Convention. Chicago.
Muggah, Robert. 2008. ‘An Armed Violence Prevention and Reduction Agenda.’ Medicine, Conflict and Survival, Vol. 24, No. 1.
Murdoch, James and Todd Sandler. 2004. ‘Civil Wars and Economic
Growth: Spatial Dispersion.’ American Journal of Political
Science, Vol. 48, No. 1, pp. 138–51.
Mutebi, Frederick. 2008a. Collapse, War and Reconstruction in
Uganda: An Analytical Narrative on State-making. Crisis
States Research Centre Working Paper, Vol. 27, No. 2.
London: London School of Economics.
——. 2008b. ‘Collapse, War and Reconstruction in Rwanda: An
Analytical Narrative on State-making.’ Crisis States Research
151
BIBLIOGR APHY
ground Paper. Geneva/London: Small Arms Survey/London
School of Economics.
Sköns, Elisabeth. 2006. ‘The Cost of Armed Conflict.’ In International Task Force on Global Public Goods. Peace and Security.
Expert Paper, No. 5. Stockholm: International Task Force
on Global Public Goods.
Small Arms Survey. 2006. Small Arms Survey 2006: Unfinished
Business. Oxford: Oxford University Press.
Smillie, Ian, Lansane Gberie, and Ralph Hazleton. 2000. The
Heart of the Matter: Sierra Leone, Diamonds and Human
Security. Toronto: Partnership Africa–Canada.
Soares, R. 2006. ‘The Welfare Cost of Violence across Countries.’
Journal of Health Economics, Vol. 25, pp. 821–46.
Sombart, Werner. 1913. Krieg und Kapitalismus. Munich and
Leipzig: Duncker and Humblot.
Stewart, Frances and Valpy FitzGerald, eds. 2001. War and Underdevelopment, Volume 1: The Economic and Social Consequences of Conflict. Oxford: Oxford University Press.
Stewart, Frances, Cindy Huang, and Michael Wang. 2001. ‘Internal Wars: An Empirical Overview of Economic and Social
Consequences.’ In Frances Stewart and Valpy FitzGerald,
eds. War and Underdevelopment, Volume 1: The Economic
and Social Consequences of Conflict. Oxford: Oxford University Press, pp. 67–103.
1
2
3
4
5
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
152
Stewart, Frances, Frank Humphreys, and Nick Lea. 1997. ‘Civil
Conflict in Developing Countries over the Last Quarter of a
Century: An Empirical Overview of Economic and Social
Consequences.’ Oxford Development Studies, Vol. 25, No. 1,
pp. 11–42.
Tilly, Charles. 1992. Coercion, Capital and European States: AD
990–1992. Cambridge: Blackwell.
——. 2003. The Politics of Collective Violence. Cambridge:
Cambridge University Press.
UNDP (United Nations Development Programme). 2006. El costo
económico de la violencia en Guatemala. Guatemala: UNDP.
——. 2007. Informe estadistico de la violencia en Guatemala.
Guatemala: UNDP.
UNIFEM (United Nations Development Fund for Women). 2007.
Violence against Women: Facts and Figures. Web edition.
Accessed June 2008. <http://www.unifem.org/campaigns/
vaw/facts_figures.php>
University of Bradford. 2005. Armed Violence and Poverty Initiative.
Web edition. Accessed June 2008. <http://www.bradford.
ac.uk/acad/cics/publications/AVPI/briefing/>
UNODC (United Nations Office on Drugs and Crime). 2005. Crime
and Development in Africa. Vienna: UNODC.
——. 2007a. Crime and Development in Central America. Vienna:
UNODC.
——. 2007b. Crime, Violence, and Development: Trends, Costs,
and Policy Options in the Caribbean. Vienna/Washington,
DC: UNODC/World Bank.
Wennmann, Achim. 2007. Money Matters: Economic Dimensions
in Peace Mediation. PSIS Occasional Paper, No. 4. Geneva:
Graduate Institute of International Studies.
WHO (World Health Organization). 2004. The Economic Dimension
of Interpersonal Violence. Geneva: WHO.
——. 2005. Armed Violence Prevention Programme. Geneva: WHO.
Web edition. Accessed June 2008. <http://www.who.int/
violence_injury_prevention/violence/activities/avpp.pdf>
Chapter Six
Access Economics. 2004. The Cost of Domestic Violence to the
Australian Economy: Part I. Canberra: Commonwealth of
Australia.
AI (Amnesty International). 2004. Iraq: Stop Violence against
Women. ‘I Don’t Want to Die’. Domestic Violence in Iraq.
AI Index: MDE 14/001/2004. <http://www.amnesty.org/fr/
library/asset/MDE14/001/2004/en/dom-MDE14001200
4en.pdf>
––, IANSA, and Oxfam. 2005. The Impact of Guns on Women’s
Lives. London: AI.
ASF (Acid Survivors Foundation). 2006. Acid Survivors Foundation:
Annual Report 2006. Dhaka: ASF.
––. 2008. ‘Acid Attack Statistics.’ Accessed August 2008.
<http://www.acidsurvivors.org/statistics.html>
Barth, Elise Fredrikke. 2002. Peace as Disappointment: The
Reintegration of Female Soldiers in Post-Conflict Societies:
A Comparative Study from Africa. PRIO Report No. 3. <http://
www.prio.no/page/Publication_details//9429/40567.html>
Bastick, Megan, Karin Grimm, and Rahel Kunz. 2007. Sexual Violence in Armed Conflict: Global Overview and Implications
for the Security Sector. Geneva: Geneva Centre for the
Democratic Control of Armed Forces.
Bastick, Megan, Karin Grimm, and Jasna Lazarevic. 2008. ‘The
Global Burden of Armed Violence for Women.’ Background
paper. Geneva: Geneva Centre for the Democratic Control
of Armed Forces.
Bennet, Olivia, Jo Bexley, and Kitty Warnock, eds. 1995. Arms to
Fight, Arms to Protect: Women Speak Out About Conflict.
London: Panos.
Bouta, Tsjeard, Georg Frerks, and Ian Bannon. 2005. Gender,
Conflict, and Development. Washington, D.C.: World Bank.
CARE and ICRW (Cooperative for Assistance and Relief Everywhere
and International Centre for Research on Women). 2007.
Exploring Dimensions of Masculinity and Violence: WesternBalkan Gender-Based Violence Prevention Initiative. Accessed
July 2008. <http://www.wavenetwork.org/images/doku/
balkan_gender_based_violence_report.pdf>
Carpenter, Charli R. 2003. ‘Women and Children First: Gender,
Norms and Humanitarian Evacuation in the Balkans 1991–
95.’ International Organization, Vol. 57, No. 4, pp. 661–94.
––. 2006. ‘Recognizing Gender-Based Violence against Civilian
Men and Boys in Conflict Situations.’ Security Dialogue,
Vol. 37, No. 1, pp. 83–103.
CICS (Centre for International Cooperation and Security). 2005.
The Impact of Armed Violence on Poverty and Development.
Full report of the Armed Violence and Poverty Initiative.
Bradford: University of Bradford. March.
Clodfelter, Michael. 2002. Warfare and Armed Conflicts: A Statistical Reference to Casualty and Other Figures, 1500–2000.
Jefferson, North Carolina: McFarland.
Commission for Historical Clarification. 1999. Guatemala: Memory
of Silence. Report of the Commission for Historical Clarification; Conclusions and Recommendations. <http://shr.
aaas.org/guatemala/ceh/report/english/toc.html>
Connell, R.W. 1995. Masculinities. Berkeley: University of California
Press.
Dreyfus, Pablo, Carolina Iootty de Paiva Dias, Benjamin Lessing, and
William Godnick. 2003. Small Arms Control in MERCOSUR.
Latin America Series No. 3. Monitoring the implementation
of small arms controls (MISAC). Rio de Janeiro and London:
International Alert and Viva Rio. June.
Federal-Provincial-Territorial Ministers Responsible for the Status
of Women. 2002. Assessing Violence against Women:
A Statistical Profile. Ottawa: Federal-Provincial-Territorial
Ministers Responsible for Status of Women. <http://www.
swc-cfc.gc.ca/pubs/0662331664/200212_0662331664_
e.pdf>
García-Moreno, Claudia, Henrica A.F.M. Jansen, Mary Ellsberg, Lori
Heise, and Charlotte H. Watts 2005. WHO Multi-country
Study on Women’s Health and Domestic Violence against
Women. Geneva: WHO.
GCSKSW (Global Campaign to Stop Killing and Stoning Women).
n.d. ‘Frequently Asked Questions about “Honour Killing”.’
Accessed July 2008. <http://www.stop-stoning.org>
Ghobarab, Hazem A., Paul Huth, and Bruce Russett. 2003. ‘Civil
Wars Kill and Maim People—Long after the Shooting
Stops.’ American Political Science Review, Vol. 92, No. 2,
pp. 189–202.
Greenfeld, Lawrence. 1997. Sex Offences and Offenders: An
Analysis of Data on Rape and Sexual Assault. Washington,
D.C.: Office of Justice Programs, Bureau of Justice Statistics,
US Department of Justice. NCJ-163392.
Guha-Sapir, Debarati and Willem Gijsbert von Panhuis. 2004.
‘Conflict-related Mortality: An Analysis of 37 Datasets.’
Disasters, Vol. 28, No. 4, pp. 418–28.
Gyawali, Tirth Prasad and Shobha Shrestha. 2006. ‘Small Arms
Proliferation and Its Impact on Women.’ In Bishnu Raj
Upreti and Rohit Kumar Nepali, eds. Nepal at Barrel of Gun.
Kathmandu: South Asia Partnership Nepal.
Hall, Tina and Sam Wright. 2003. ‘Making It Count. A Practical
Guide to Collecting and Managing Domestic Violence Data.’
Community safety practice briefing. July. London: Nacro.
HRW (Human Rights Watch). 1996. Shattered Lives: Sexual Violence
during the Rwandan Genocide and Its Aftermath. New York:
HRW. <http://www.hrw.org/reports/1996/Rwanda.htm>
––. 2004. How to Fight, How to Kill: Child Soldiers in Liberia.
Vol. 16, No. 2. February. <http://www.hrw.org/reports/
2004/liberia0204/liberia0204.pdf>
IASC (Inter-Agency Standing Committee). 2005. Guidelines for
Gender-based Violence Interventions in Humanitarian
Settings: Focusing on Prevention of and Response to
Sexual Violence in Emergencies. Task Force on Gender
and Humanitarian Assistance. <http://www.humanitarian
info.org/iasc/content/subsidi/tf_gender/gbv.asp>
Iraq Coalition Casualty Count. 2008. Female Fatalities. Accessed
July 2008. <http://icasualties.org/oif/Female.aspx>
Jackson, Thomas, Nicholas Marsh, Taylor Owen, and Anne Thurin.
2005. Who Takes the Bullet? The Impact of Small Arms
Violence. Understanding the Issues No. 3. Oslo: PRIO and
Norwegian Church Aid.
Jha, Prabhat and Emily Oster. 2006. ‘Male-to-female Sex Ratio of
Children Born in India: National Survey of 1.1 Million Households.’ The Lancet, Vol. 367, No. 9506. February, pp. 211–18.
153
BIBLIOGR APHY
––. 2003. ‘Masculinities and Masculinity Politics in World Society.’
Sydney: University Institute for Research on Women,
University of Sydney. <http://irw.rutgers.edu/lectures/
connelllecture.pdf>
Johnson, Holly, Natalia Ollus, and Sami Nevala, eds. 2008. Violence against Women: An International Perspective. New
York: Springer.
Karame, K. 1999. ‘Women and War: A Highly Complex Interrelation.’
In NUPI. Women and Armed Conflicts: A Study for the
Norwegian Ministry. Oslo: NUPI, pp. 4–31.
Kellermann, Arthur L. et al. 1993. ‘Gun Ownership as a Risk Factor
for Homicide in the Home.’ The New England Journal of
Medicine, Vol. 329, No. 15, pp. 1084–91.
Krug, Etienne G., Linda L. Dahlberg, James A. Mercy, Anthony B. Zwi,
and Rafael Lozano, eds. 2002. World Report on Violence
and Health. Geneva: WHO.
Macartney, Jane. 2007. ‘China Faces Population Imbalance Crisis.’
Times Online. Web edition, 12 January. <http://www.times
online.co.uk/tol/news/world/asia/article1292295.ece>
MADRE. 2007. Promising Democracy, Imposing Theocracy: GenderBased Violence and the US War on Iraq. New York: MADRE.
Manthorpe, Jonathan. 1999. ‘China Battles Slave Trading in
Women: Female Infanticide Fuels a Brisk Trade in Wives.’
Vancouver Sun, 11 January.
Marón, Karen. 2003. ‘Mujeres guerrilleras (extractos).’ International Committee of the Red Cross. <http://www.icrc.org/
web/spa/sitespa0.nsf/htmlall/5TECQS?OpenDocument>
Messner, Michael A. 1990. ‘When Bodies are Weapons: Masculinity and Violence in Sport.’ International Review for the
Sociology of Sport. Vol. 25, No. 3, pp. 203–18.
McKay, Susan and Dyan Mazurana. 2004. Where are the Girls?
Girls in Fighting Forces in Northern Uganda, Sierra Leone
and Mozambique: Their Lives during and after War. Québec:
Rights & Democracy.
Morrison, Andrew and María Beatriz Orlando. 2004. The Cost and
Impacts of Gender-based Violence in Developing Countries:
Methodological Considerations and New Evidence. New
York: World Bank. <http://siteresources.worldbank.org/
INTGENDER/Resources/costsandimpactsofgbv.pdf>
Myrttinen, Henri. 2003. ‘Disarming Masculinities.’ Disarmament
Forum, Vol. 4, pp. 37–46. <http://www.unidir.org/pdf/
Gender/6%20myrttinen.pdf>
1
2
3
4
5
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
154
Naudé, Beaty, Johan Prinsloo, and Anastasios Ladikos. 2006.
Experience of Crime in Thirteen African Countries: Results
from the International Crime Victim Survey. Turin: UNODC/
UNICRI.
Niazi, Shuriah. 2008. ‘India: State of Dowry Deaths.’ Women’s
Feature Service. WFS Ref. INDH212. February.
Office on Women in the NATO Forces and The Women’s Research
& Education Institute. n.d. Percentages of Female Soldiers
in NATO Countries’ Armed Forces. <http://www.nato.int/
issues/women_nato/perc_fem_soldiers_2001_2006.pdf>
O’Hare, Bernadette and David P. Southall. 2007. ‘First Do No Harm:
The Impact of Recent Armed Conflict on Maternal and
Child Health in Sub-Saharan Africa.’ Journal of the Royal
Society of Medicine, Vol. 100, No. 12, pp. 564–70.
Omanyondo, Marie-Clarie O. 2005. Sexual Gender-Based Violence
and Health Facility Needs Assessment: Liberia. Geneva: WHO.
Peimani, Hooman. 2004. Armed Violence and Poverty in Chechnya:
A Mini Case Study for the Armed Violence and Poverty
Initiative. Bradford: Centre for International Cooperation
and Security, University of Bradford. November.
Plümper, Thomas and Eric Neumayer. 2006. ‘The Unequal Burden
of War: The Effect of Armed Conflict on the Gender Gap in
Life Expectancy.’ International Organizations, Vol. 60, No. 3,
pp. 723–54.
RHRC (Reproductive Health Response in Conflict). 2003. Genderbased Violence Tools Manual. New York: RHRC. <http://www.
rhrc.org/resources/gbv/gbv_tools/manual_toc.html>
Sánchez, Fabio et al. 2004. ‘Segunda Parte: Estudio de Mujeres.’
In: Los costos de la violencia intrafamiliar en Colombia.
Bogota: Centro de Estudios sobre Desarrollo Económico,
Universidad de los Andes. Cited in Andrew Morrison and
María Beatriz Orlando.
Scholte, Marianne. 2006. ‘Acid Attacks in Bangladesh: A Voice
for the Victims.’ Spiegel Online. 17 March. <http://www.
spiegel.de/international/0,1518,druck-406485,00.html>
Schroeder, Emily. 2005. Multi-Country Demobilization and Reintegration Program. MDRT Gender Desk Study. New York:
World Bank. July. <http://www.mdrp.org/PDFs/Study_
GenderDesk_En.pdf>
on Reproductive and Sexual Health and Rights.’ New
York: UNFPA.
Spiegel, Paul B. et al. 2007. ‘Prevalence of HIV Infection in ConflictAffected and Displaced People in Seven Sub-Saharan African
Countries: A Systematic Review.’ The Lancet, Vol. 369. June,
pp. 2187–95.
UNFPA (United Nations Populations Fund). 2000. The State of
World Population: Lives Together, World Apart: Men and
Women in a Time of Change. New York: UNFPA.
––. 2005. State of World Population 2005: The Promise of Equality:
Gender Equity, Reproductive Health and the Millennium
Development Goals. New York: UNFPA.
UNGA (United Nations General Assembly). 1993. Declaration on
the Elimination of Violence against Women. UN doc. A/RES/
48/104 of 20 December.
––. 2006. In-depth Study on All Forms of Violence Against Women:
Report of the Secretary General. UN doc. A/61/122/Add.1
of 6 July.
UNIFEM (United Nations Development Fund for Women). 2007.
Violence against Women—Facts and Figures. November.
<http://www.unifem.org/attachments/gender_issues/
violence_against_women/facts_figures_violence_against_
women_2007.pdf>
––. n.d. Violence against Women: Global Issue. <http://www.
cities-localgovernments.org/uclg/upload/docs/speech
byk.annanonviolenceagainstwomen.pdf>
Vetten, Lisa and Sadiyya Haffejee. 2005. ‘Gang Rape: A Study in
Inner-city Johannesburg.’ SA Crime Quarterly, No. 12. June,
pp. 31–36.
Vlachová, Marie and Lea Biason, eds. 2005. Women in an Insecure World. Violence against Women: Facts, Figures and
Analysis. Geneva: Geneva Centre for the Democratic Control
of Armed Forces.
Ward, Jeanne. 2002. If Not Now, When? Addressing Gender-based
Violence in Refugee, Internally Displaced, and Post-conflict
Settings: A Global Overview. New York: RHRC.
Waters, Hugh, et al. 2004. The Economic Dimensions of Interpersonal Violence. Geneva: Department of Injuries and
Violence Prevention, WHO.
Seifert, Ruth. 1992. War and Rape: Analytical Approaches. Geneva:
WILPF. <http://www.wilpf.int.ch/publications/1992ruth
seifert.htm>
Watts, Charlotte and Cathy Zimmerman. 2002. ‘Violence against
Women: Global Scope and Magnitude.’ The Lancet, Vol. 359.
April, pp. 1232–37.
Sen, Amartya. 1990. ‘More than 100 Million Women are Missing.’
New York Review of Books, Vol. 37, No. 20. 20 December.
<http://ucatlas.ucsc.edu/gender/Sen100M.html>
Widmer, Mireille, Gary Barker, and Cate Buchanan. 2006. Hitting
the Target: Men and Guns. RevCon Policy Brief. Geneva:
Centre for Humanitarian Dialogue. June. Web edition.
<http://www.hdcentre.org/files/MenandGuns.pdf>
––. 2003. ‘Missing Women—Revisited.’ British Medical Journal,
Vol. 327. December, pp. 1297–98.
Shuzhuo Li. 2007. ‘Imbalanced Sex Ratio at Birth and Comprehensive Intervention in China. 4th Asia Pacific Conference
Williams, Louise. 2005. ‘Assessing the Rights and Wrongs of
Women at War.’ Sydney Morning Herald. March. Web edition.
<http://www.smh.com.au/news/Opinion/Assessing-the-
WomenWarPeace.org. n.d. Issue Brief on Small arms and Light
Weapons. New York: WomenWarPeace.org. <http://www.
womenwarpeace.org/webfm_send/1129>
Wood, Elisabeth Jean. 2006. ‘Variation in Sexual Violence during
War.’ Politics and Society, Vol. 34, No. 3, pp. 307–41.
Woolf, Linda. n.d. The Atrocity of Acid Attacks. Webster University.
<http://www.webster.edu/~woolflm/acidattacks.html>
Zvekic, Ugljesa and Anna Alvazzi de Frate. 1994. Criminal
Victimization in the Developing World. Publication No. 55.
Turin: UNICRI.
Chapter Seven
AHRC (Asian Human Rights Commission). 2007. The State of
Human Rights in Eleven Asian Nations—2007. Hong Kong:
AHRC.
AI (Amnesty International). 2006. Philippines Political Killings,
Human Rights and the Peace Process. Document No. ASA
35/006/2006. London: AI.
APWLD (Asia Pacific Forum on Women, Law and Development).
2006. ‘APWLD/FA Statement on Extrajudicial Killing in Sri
Lanka, Philippines and Chechnya at the UN Human Rights
Council.’ 19 September. Web edition. Accessed April 2008.
<http://www.apwld.org/ejk.htm>
Armour Group. 2007. Kidnapping Special Report, January–June
2007. Caracas: Armour Group.
Bastick, Megan, Karin Grimm, and Rahel Kunz. 2007. Sexual Violence in Armed Conflict: Global Overview and Implications
for the Security Sector. Geneva: Geneva Centre for the
Democratic Control of Armed Forces.
Beasley, Ryan, Cate Buchanan, and Robert Muggah. 2003. In the
Line of Fire: Surveying the Perceptions of Humanitarian and
Development Personnel of the Impacts of Small Arms and
Light Weapons. Geneva: Small Arms Survey and Centre
for Humanitarian Dialogue.
Bruneau, Thomas. 2005. ‘The Maras and National Security in
Central America.’ Strategic Insights, Vol. 4, No. 5. <http://
www.ccc.nps.navy.mil/si/2005/May/bruneauMay05.asp>
Buchanan, Cate and Robert Muggah. 2005. No Relief: Surveying
the Effects of Gun Violence on Humanitarian Aid and
Development Personnel. Geneva: Small Arms Survey and
Centre for Humanitarian Dialogue.
Cingranelli, David L. and David L. Richards. 2008a. The Cingranelli–
Richards Human Rights Dataset v.2008.03.12. Web edition.
Accessed April 2008. <http://www.humanrightsdata.org>
——. 2008b. The Cingranelli–Richards (CIRI) Human Rights Data
Project Coding Manual v.2008.3.13. <http://ciri.binghamton.
edu/documentation/ciri_coding_guide.pdf>
155
Esser, Daniel. 2004. ‘The City as Arena, Hub and Prey: Patterns
of Violence in Kabul and Karachi.’ Environment and Urbanization, Vol. 16, No. 2, pp 31–38.
BIBLIOGR APHY
rights-and-wrongs-of-women-at-war/2005/03/07/1110160
750185.html>
Fast, Larissa and Elizabeth Rowley. 2008. Mortality and Victimization of Aid Workers. Unpublished background paper
prepared for the Global Burden of Armed Violence report.
Hillier, Bill, Margarita Greene, and Jake Desyllas. 2000. ‘Selfgenerated Neighbourhoods: The Role of Urban Form in
the Consolidation of Informal Settlements.’ Urban Design
International, Vol. 5, No. 2, pp. 61–96. <http://eprints.ucl.
ac.uk/archive/00001401/>
HRC (Human Rights Council). 2006. Civil and Political Rights,
Including the Questions of Disappearances and Summary
Executions: Report of the Special Rapporteur on Extrajudicial, Summary or Arbitrary Executions, Philip Alston:
Addendum: Mission to Guatemala. 21–25 August. A/HRC/
4/20/Add.2 of 19 February 2007. Geneva: United Nations.
——. 2008a. Promotion and Protection of All Human Rights, Civil,
Political Economic and Cultural Rights, Including the Right
to Development: Report by the Special Rapporteur on Extrajudicial, Summary or Arbitrary Executions: Addendum:
Preliminary Note on Mission to Brazil. November. A/HRC/
8/3/Add.4 of 14 May. Geneva: United Nations.
1
——. 2008b. Promotion and Protection of All Human Rights, Civil,
Political Economic and Cultural Rights, Including the Right
to Development: Report of the Working Group on Enforced
or Involuntary Disappearances. A/HRC/7/2 of 10 January.
Geneva: United Nations.
2
HRW (Human Rights Watch). 1996. All too Familiar: Sexual Abuse
of Women in U.S. State Prisons. December. Web edition.
<http://hrw.org/reports/1996/Us1. htm#_1_2>
5
——. 2008. Brazil Country Summary 2008. New York: HRW. Web
edition. Accessed April 2008. <http://hrw.org/wr2k8/pdfs/
brazil.pdf>
KNCHR (Kenya National Commission on Human Rights). 2007.
Preliminary Report on Alleged Executions and Disappearance of Persons between June and October 2007. Nairobi:
KNCHR.
Loutis, Wiza. 2006. Evaluation de la situation des femmes dans
le cadre de la violence armée en Haïti: Communes de Portau-Prince, des Cayes et des Gonaïves. Port-au-Prince: UNDP
Haiti and MINUSTAH. June.
Marchal, Roland. 2006. ‘Resilience of a City at War: Territoriality,
Civil Order and Economic Exchange in Mogadishu.’ In
Deborah Fahy Bryceson and Deborah Potts, eds. African
Urban Economies: Viability, Vitality or Vitiation? Basingstoke and New York: Palgrave Macmillan, pp. 207–29.
3
4
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
156
Miller, Jody and Scott H. Decker. 2001. ‘Young Women and Gang
Violence: Gender, Street Offending, and Violent Victimization in Gangs.’ Justice Quarterly, Vol. 18, No. 1. March,
pp. 115–40.
——. 2007. ‘When Vigilantes Turn Bad: Gangs, Violence, and
Social Change in Urban Nicaragua.’ In David Pratten and
Atryee Sen, eds. Global Vigilantes: Anthropological Perspectives on Justice and Violence. London: Hurst.
Moser, Caroline. 2004. ‘Urban Violence and Insecurity: An Introductory Roadmap.’ Environment and Urbanization, Vol. 16,
No. 2. October, pp. 3–16.
Small Arms Survey. 2007. Small Arms Survey 2007: Guns and the
City. Cambridge: Cambridge University Press.
—— and Cathy McIlwaine. 2006. ‘Latin American Urban Violence
as a Development Concern: Towards a Framework for
Violence Reduction.’ World Development, Vol. 34, No. 1,
pp. 89–112. <http://www.brookings.edu/~/media/Files/rc/
articles/2006/01globaleconomics_moser/200601moser.pdf>
Muggah, Robert and Oliver Jütersonke. Forthcoming. ‘Endemic
Urban Violence and Public Security.’ Cities and Urban
Violence. Ottawa: Ministry of Foreign Affairs.
Muggah, Robert and Chris Stevenson. Forthcoming. ‘On the Edge:
Armed Violence and Arms in Central America.’ World Bank
Development Report: Central America. Washington, D.C.:
World Bank.
Pyakurel, Subodh R. 2007. Human Rights Situation in Nepal after
Loktantra. Kathmandu: Informal Sector Service Centre. Web
edition. Accessed April 2008. <http://www.inseconline.org/
report/Human%20Rights%20Situation%20in%20Nepal
%20after%20Loktantra.pdf>
Rodgers, Dennis. 2004. ‘“Disembedding” the City: Crime, Insecurity, and Spatial Organisation in Managua, Nicaragua.’
Environment and Urbanization, Vol. 16, No. 2, pp. 113–24.
——. 2006. ‘Living in the Shadow of Death: Gangs, Violence, and
Social Order in Urban Nicaragua, 1996–2002.’ Journal of
Latin American Studies, Vol. 38, No. 2, pp. 267–92.
UNDP (United Nations Development Programme). 2007. Informe
estadístico de la violencia en Guatemala. Guatemala:
UNDP.
UNGA (United Nations General Assembly). 2006. International
Convention for the Protection of all Persons from Enforced
Disappearances. Paris, 20 December. Web edition. Accessed
April 2008. <http://untreaty.un.org/English/notpubl/IV_
16_english.pdf>
UNODC (United Nations Office on Drugs and Crime). 2007. Crime
and Development in Central America: Caught in the Crossfire. Vienna: UNODC. May. <http://www.unodc.org/pdf/
Central%20America%20Study.pdf>
UNOHCHR (Office of the High Commissioner for Human Rights).
1982. General Comment No. 6: The Right to Life. 30 April.
Web edition. <http://www.unhchr.ch/tbs/doc.nsf/(Symbol)/
84ab9690ccd81fc7c12563ed0046fae3?Opendocument>
Vlachová, Marie and Lea Biason, eds. 2005. Women in an Insecure
World: Violence against Women—Facts, Figures and Analysis.
Geneva: Geneva Centre for the Democratic Control of
Armed Forces.
Weinstein, Jeremy. 2007. Inside Rebellion: The Politics of Insurgent
Violence. Cambridge: Cambridge University Press.
World Bank. 2008. Urban Crime and Violence in LAC: Status Report
on Activities. Washington, D.C.: World Bank.
List of Illustrations
Boxes
Box 1.1 Methods to estimate direct conflict deaths
.............................................................................................................................
Box 1.2 Using population surveys to estimate direct conflict deaths
.................................................................................
11
13
L I S T O F I L L U S T R AT I O N S
157
Box 1.3 The dramatic impact of particularly violent conflicts .................................................................................................... 22
Box 1.4 Accounting for direct deaths in Sudan
.......................................................................................................................................
24
...............................................................................................................................................
26
..................................................................................................................................................................................
34
Box 1.5 International vs. intrastate conflict
Box 2.1 Crude mortality rates
Box 2.2 Sexual violence in armed conflict
..................................................................................................................................................
Box 2.3 A very dark number: direct and indirect mortality in southern Sudan, 1999–2005
Box 2.4 Direct and indirect mortality in Sierra Leone, 1991–2002
38
.......................
43
......................................................................................
44
Box 2.5 Armed violence in Iraq: what’s in a number? ....................................................................................................................... 46
1
2
3
4
5
Box 3.1 Post-conflict violence in the Democratic Republic of the Congo
Box 3.2 Sexual violence in the aftermath of war
........................................................................
52
...................................................................................................................................
53
Box 3.3 Protecting the displaced from armed violence
.................................................................................................................
56
..............................................................................................................................
58
.............................................................................................................................................................
60
Box 3.4 When do countries relapse into civil war?
Box 3.5 The demographics of discord
Box 3.6 The mobilization of inequalities
......................................................................................................................................................
Box 3.7 Transitional justice and DDR in Africa
.........................................................................................................................................
64
..............................................................................................................................................
72
......................................................................................................................................................................................
75
Box 4.1 Homicide and human development
Box 4.2 Guns and homicide
62
6
7
G LOBAL BURDEN
of
ARMED VI O L EN CE
158
Box 4.3 Sex, age, and armed violence
............................................................................................................................................................
Box 4.4 Up close and personal: arms availability and female homicide
Box 4.5 Violent death in the city
........................................................................
82
...........................................................................................................................................................................
85
Box 5.1 Accounting for the costs of violence: a typology and examples
Box 5.2 Modelling the economic costs of civil war
.........................................................................
93
.............................................................................................................................
94
Box 5.3 Measuring the health dimensions of violence using contingent valuation
Box 5.4 The economic costs of intimate partner and sexual violence during war
Box 5.5 What is a discount rate?
.............................................
96
.................................................
98
........................................................................................................................................................................
Box 5.6 Armed violence and investment
Box 6.1 Armed conflict and HIV/AIDS
100
...................................................................................................................................................
104
.............................................................................................................................................................
114
Box 6.2 The costs of violence against women
Box 6.3 Gang rapes
81
.......................................................................................................................................
116
.........................................................................................................................................................................................................
119
Box 6.4 Acid attacks in Bangladesh
121
...............................................................................................................................................................
Box 7.1 Gender and gang violence .................................................................................................................................................................... 128
Box 7.2 Maras and pandillas in Central America
................................................................................................................................
130
Box 7.3 Gender-based violence by state officials
...............................................................................................................................
133
.................................................................................................................................
136
Box 7.4 The burden of kidnapping in Venezuela
Maps
Map 1.1 The GBAV conflict-affected countries
..........................................................................................................................................
21
Map 1.2 Increases and decreases in direct conflict deaths in
selected armed conflicts, 2004 and 2007
....................................................................................................................................................
23
Map 1.3 The risk of dying violently in armed conflict
per 100,000 population per year, average, 2004–07
........................................................................................................................
Map 3.1 Distribution of IDP and refugee populations in selected African countries
Map 3.2 The youth bulge in 2005
28
............................................
56
.........................................................................................................................................................................
61
............................................................................
70
................................................................................................................
74
.....................................................................................................................
84
Map 4.2 Absolute homicide counts by subregion, 2004
Map 4.3 Homicide clearance rates in Central America
Map 5.1 Potential gains in life expectancy in years in the absence
of non-conflict armed violence, by country, 2004
..............................................................................................................................
Map 5.2 Global lost product due to violent deaths, 2004
Map 7.1 Global extrajudicial killings, 2006
100
...........................................................................................................
101
...............................................................................................................................................
131
Map 7.2 Enforced and involuntary disappearances, 2006
........................................................................................................
159
L I S T O F I L L U S T R AT I O N S
Map 4.1 Homicide rates per 100,000 population, by subregion, 2004
134
Tables
Table 1.1 Direct conflict deaths by region and subregion,
and as percentage of total direct conflict deaths, 2004–07
.........................................................................................................
17
1
Table 1.2 Estimates of the regional distribution of
direct conflict deaths, 2004–07
.............................................................................................................................................................................
18
3
Table 1.3 Top ten direct conflict death countries, relative and
cumulative total percentages, 2004–07
........................................................................................................................................................
Table 1.4 Estimated direct conflict deaths in South Sudan, 2002–07
Table 1.5 Estimated direct conflict deaths in Darfur, 2003–07
22
..............................................................................
25
................................................................................................
25
Table 1.6 Epidemiological surveys of direct conflict deaths in Sudan, 2003–05
Table 1.7 Direct conflict death rate by country, 2004–07
....................................................
25
..............................................................................................................
27
Table 2.1 Comparison of methods for measuring excess mortality
.....................................................................................
37
Table 2.2 Estimated incidents of rape in selected armed conflicts
.....................................................................................
39
Table 2.3 Direct vs. indirect deaths in several recent armed conflicts
Table 2.4 Violent deaths reported in Iraq, 2003–07
2
.............................................................................
40
.........................................................................................................................
46
Table 2.5 Violent death estimates from three mortality surveys
...........................................................................................
Table 2.6 Overview of indirect death estimates from three mortality surveys
.........................................................
47
47
4
5
6
7
Table 3.1 Selected post-conflict countries: 1995–2005 ................................................................................................................... 51
Table 3.2 Approximate crude mortality rates (CMRs) in east and west DRC, 1999–2007
..............................
52
............................................................................................................................
53
......................................................................................................................................................
59
Table 3.3 Typology of post-conflict armed violence
Table 3.4 Risk factors for youth violence
Table 4.1 National-level homicide trend analysis by subregion, 1998–2006 ............................................................. 77
Table 4.2 Female homicides for selected countries, 2005
..........................................................................................................
80
...................................................................................................................................
93
G LOBAL BURDEN
of
ARMED VI O L EN CE
160
Table 5.1 A typology for costing armed violence
Table 5.2 The social value of violence reduction in selected WHO regions, 1990s
...............................................
97
Table 5.3 Lost product due to violent deaths,
USD million (2007) and % of GDP (2004) .................................................................................................................................................... 102
Table 5.4 Aggregate lost product due to violent deaths,
selected countries, 2004 ........................................................................................................................................................................................... 102
Table 6.1 Comparison of maternal mortality and
under-five mortality in 42 sub-Saharan countries
..............................................................................................................................
114
Table 6.2 Selected studies on costs of intimate partner
violence and/or domestic violence ................................................................................................................................................................... 116
Table 7.1 Recorded cases of enforced or involuntary disappearances,
selected countries, 2000–03 and 2004–07 ............................................................................................................................................ 135
Table 7.2 Kidnap for ransom cases by region, 1998–2002
Table 7.3 Kidnap for ransom cases by region, 2007
.......................................................................................................
137
.........................................................................................................................
137
Figures
Figure 1 Categories of deaths
....................................................................................................................................................................................
3
Figure 1.1 Total direct conflict deaths by database, main armed conflicts, 2004–07 ......................................... 15
Figure 1.2 Estimates of the regional distribution of direct conflict deaths, 2004–07
Figure 1.3 Direct conflict deaths by armed conflict type, 2004–07
........................................
....................................................................................
17
26
......................................................................................................................................................
Figure 2.2 Distribution of killings in Sierra Leone, 1991–2001
33
...............................................................................................
44
..........................................................................................................................................................................
60
Figure 3.1 Youth population growth rates and murder rates
in the United States, 1950–2005
Figure 3.2 Number of DDR operations around the world, 1989–2008
.............................................................................
Figure 4.1 Homicide rates per 100,000 population by region and subregion, 2004
63
...........................................
71
............................................................
72
..........................................................................................
75
Figure 4.2 Homicide and HDI: homicide rate per 100,000 population, 2004
Figure 4.3 Percentage of homicides committed with a firearm
for countries in eight subregions, 2004 or closest available year
161
L I S T O F I L L U S T R AT I O N S
Figure 2.1 Typology of conflict mortality
Figure 4.4 Trends in intentional homicide in the Americas, Europe,
and Central Asia and Transcaucasia, 1998–2006 ................................................................................................................................. 78
Figure 4.5 Homicide country rate per 100,000 population plotted
against average % change in country homicide levels
.....................................................................................................................
Figure 4.6 Female homicide rates per 100,000 population, 2005
79
.......................................................................................
81
..............................................................................................................................................................
81
Figure 4.7 World estimates for homicide rates
per 100,000 population by age, 2004
2
Figure 4.8 Median percentage of recorded homicide cases
solved in 24 countries by subregion, 2005
.................................................................................................................................................
83
to get a reduced sentence in Central American countries (%), 2004
..................................................................................
Figure 4.10 Ratio of homicide rates in major cities and rest of country, 2005
4
.........................................................
84
6
86
7
Figure 5.1 Social value of violence reduction, % of GDP, selected countries, 1995 ............................................. 96
Figure 5.2 Potential gains in life expectancy (years) in the
...............................................................................................................................................
102
Figure 5.3 Potential gains in life expectancy (years) in the
absence of violent deaths, top 15 countries, 2004
3
5
Figure 4.9 Respondents saying it is possible to bribe a judge
absence of violent deaths by region, 2004
1
............................................................................................................................
Figure 6.1 Percentage of surveyed women reporting on IPV, selected cases
..........................................................
Figure 6.2 Reported incidents of acid attacks in Bangladesh, May 1999–July 2008
.......................................
103
118
121
G LOBAL BURDEN
of
ARMED VI O L EN CE
162
Figure 7.1 A typology of armed groups and related actors
.......................................................................................................
Figure 7.2 Kidnap for ransom cases worldwide, by region, 1998–2002
Figure 7.3 Kidnap for ransom cases worldwide, by region, 2007
127
......................................................................
137
.......................................................................................
137
Figure 7.4 Types of fatal attacks against aid workers, 1997–2003
...................................................................................
Figure 7.5 Location of intentional violence cases against aid workers, 2002–05
..............................................
138
138
Fly UP