Part 2

Andrew Testa / Panos Pictures

Part II reviews a new global dataset that provides a comprehensive

portrait of global political violence for the years 2002 and 2003. It also

surveys trends in human rights abuse, criminal violence and human

trafficking. It concludes with a discussion of the methodological

challenges facing researchers measuring human insecurity.

The Human Security Audit

P A R T I I

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

The Human Security Audit

Introduction

A new global dataset

The new Uppsala/Human Security Centre dataset includes two categories of political violence

not counted by other conflict datasets, as well as estimates of death tolls from state-based and

non-state armed conflicts, and one-sided violence.

Measuring human rights abuse

Measuring trends in human rights violations is not easy. There is evidence to suggest that respect

for human rights has improved modestly in five out of six regions of the developing world since

the mid-1990s.

Tracking criminal violence

Although criminal violence is a major threat to human security, attempts to map global and

regional trends in homicide and rape are frustrated by a lack of reliable data.

Human trafficking

Described as the ‘dark side of globalisation’, human trafficking has become a multi-billion-dol-

lar industry. Statistics are scarce, but some evidence suggests that trafficking might be on the

decline—at least in parts of Europe.

Creating a human security index.

The idea of creating a human security index that ranks states according to how secure their

citizens are has generated a lot of interest. Creating such a measure may not be possible—or

even desirable.

P A R T I I

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Introduction

Research findings published here for the fi rst

time help illuminate both the scope and the

depth of human insecurity around the world.

But obtaining accurate data remains a

major challenge.

Mapping trends in political violence is important both

for researchers seeking to understand their causes, and for

governments and international organisations attempting

to evaluate the impact of security policies.

Part II begins with a detailed review of the new dataset

on global political violence that the Human Security Centre

commissioned from Uppsala’s Conflict Data Program spe-

cifi cally for this report.

The Uppsala/PRIO dataset that was featured in

Part I only provides data on the number of ‘state-based’

conflicts—that is, those waged between a state and an-

other state, or a state and an armed rebel group.

The Uppsala/PRIO dataset does not count:

‘Non-state’ conflicts—intercommunal and other armed

conflicts in which a government is not one of the war-

ring parties.

‘One-sided’ violence—cases of unopposed deadly

assaults on civilians, like that in Rwanda in 1994.

Death tolls from state-based and non-state conflict, or

from one-sided violence.

The new Uppsala/Human Security Centre dataset pro-

vides data on all of the above.

Because the new dataset thus far covers just two

years—2002 and 2003—it does not yet provide trend data.

Even so, it has produced some arresting findings. It re-

veals, for example, that there have been far more conflicts

taking place around the world than are counted by the

Uppsala/PRIO dataset. Indeed, in both 2002 and 2003 there

were more non-state than state-based armed conflicts.

This raises an obvious question. How can we be sure

that all armed conflicts have declined since the end of the

Cold War when we have no record of non-state conflicts

before 2002. This question is examined—and answered—

in the section that follows.

In the absence of official statistics, Uppsala and the

compilers of other conflict datasets rely on reports from

the media, NGOs, academics and governments to count

battle-related death tolls. The fact that many battle-

related deaths are simply not reported, plus Uppsala’s strict

Leo Erken / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

data coding rules, mean that a degree of under-counting

is inevitable.

Other methods of estimating war deaths, including

epidemiological surveys, are in principle more accurate.

But they cannot be used to create global datasets.

Another issue that complicates our understanding of

trends in political violence is the attachment that the me-

dia, and some policymakers, have to many of the myths of

contemporary global security, some of which were briefly

discussed in the Overview. As one UN report on the dif-

fi culties of data collection noted:

When it comes to statistics .. . numbers take on a life

of their own, gaining acceptance through repetition,

often with little inquiry into their derivations. Journalists,

bowing to the pressures of editors, demand numbers,

any number. Organizations feel compelled to supply

them, lending false precision and spurious authority to

many reports.

A classic case of ‘numbers taking on a life of their own’

is the endlessly reiterated claim that 90% of those killed

in today’s wars are civilians. In fact, the 90% figure is

completely without foundation. (See ‘The myth of civilian

war deaths’ by Kristine Eck of Uppsala University’s Conflict

Data Program.)

Political violence is a ter m that embraces more

than simply war, genocide and terrorism. It also en-

compasses state repression: torture; extrajudicial, ar-

bitrary and summary executions; the ‘disappearance’

of dissidents; the use of death squads; and incarcera-

tion without trial. All of these are as much part of the

human security agenda as they are of the human

rights agenda.

But mapping global and regional trends in human

right abuse is extremely difficult. The UN’s Human Rights

Commission is far too politicised a body to even contem-

plate such an exercise, while the major human rights or-

ganisations resist as a matter of principle any attempt to

quantify human rights abuse.

Researchers running one little-publicised project—the

Political Terror Scale (PTS) now located at the University

of North Carolina, Asheville—have been collating and

coding data on human rights violations around the world

for more than 20 years. PTS data are coded in such a

way that making comparisons between countries is rela-

tively simple.

P TS researchers Linda Cornett and Mark Gibney used

PTS data to track regional trends in human rights viola-

tions from 1980 to 2003 for the Human Security Report.

Taken as a whole the data reveal little significant varia-

tion—certainly nothing like the dramatic post–Cold War

decline in armed conflicts.

However, the regional trend data indicate that the hu-

man rights situation worsened somewhat in four out of six

regions around the world from 1980 to 2003, but that after

1994 it improved modestly in five of the six regions.

In most states, most of the time, far more people are

killed or injured by criminal violence than by warfare.

But attempts to map the incidence of violent crime

around the world confront major data problems. The fact

that only a small percentage of countries report violent

crime statistics in a regular and timely manner, plus per-

vasive problems of under-reporting and under-recording,

mean that efforts to produce reliable global or regional

trend data on violent crime confront nearly insurmount-

able difficulties.

The discussion of global homicide and rape trends in

the section on criminal violence draws on a paper com-

Torture and other gross human rights abuses are often

perpetrated in secret, making the collection of reliable

trend data very difficult.

Andrew Testa / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

missioned for this report by Graeme Newman of the State

University of New

York in Albany. This section also includes

a discussion by Luke Dowdney of Viva Rio on children and

violent crime in Rio de Janeiro.

Human trafficking is another crime that of ten involves

violence, and has been described by UNICEF as the ‘larg-

est slave trade in history.’

There are no reliable data on

the numbers trafficked each year, but there is widespread

agreement that trafficking has become a significant cause

of human insecurity.

The final section in Part II asks whether or not it is

possible to create a human security index modelled on the

UN’s Human Development Index. It draws on data from

the Uppsala Conflict Data Program, the Political Terror

Scale and a relatively new World Bank dataset on politi-

cal violence and stability, to present a comparison of the

world’s least secure states. The fact that so many of the

same countries appear in all three ‘least secure’ lists speaks

to the interrelated nature of war, human rights violations

and political instability.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

A new global dataset

Commissioned by the Human Security Centre,

and published here for the fi rst time, this data-

set provides the most comprehensive picture yet

of the incidence, scope and intensity of political

violence around the world.

Counting wars is a complex and often contested busi-

ness. Most datasets that measure the incidence of armed

conflict, including the 1946 to 2003 Uppsala/PRIO data-

set that provided much of the data for Part I of this report,

only count the number of ‘state-based’ conflicts—those in

which a government is one of the warring parties.

One-sided violence is distinguished

from armed conflict because it in-

volves the slaughter of defenceless

civilians rather than combat.

But relying solely on counts of state-based conflicts

means ignoring the very large number of conflicts in which

a government is not involved—intercommunal conflicts be-

tween ethnic and religious groups, or fights between rival

warlords, for example. The Uppsala Conflict Data Program

describes these as ‘non-state’ conflicts.

Similarly, most armed conflict datasets don’t count

cases of what Uppsala calls ‘one-sided’ violence—the un-

opposed killing of 25 or more civilians during a calendar

year. This category is distinguished from armed conflict

because it involves the slaughter of defenceless civilians

rather than combat.

To gain a more comprehensive picture of the inci-

dence and intensity of political violence around the world,

the Human Security Centre commissioned the Uppsala

Conflict Data Program to collect data on the two previously

uncounted categories, as well as on the number of deaths

associated with the three types of political violence.

Although the new dataset thus far only covers two

years, it has already produced some surprising findings.

Some examples:

All categories of political violence declined between

2002 and 2003.

There were more non-state conflicts than state-based

conflicts in both 2002 and 2003.

Non-state conflicts killed two to five times fewer

people on average than did state-based conflicts.

The reported deaths from one-sided violence were

lower still.

Sven Torfi nn / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

In 2003 less than 5% of all armed conflicts were fought

between states.

In 2003 sub-Saharan Africa experienced more cases of

political violence of all types than any other region.

Figure 2.1 records a modest worldwide decline (10%)

in cases of political violence of all types (armed conflicts

and one-sided violence) between 2002 and 2003.

Regionally, Africa and then Asia experienced the great-

est number of cases of political violence. But while Asia

showed a small increase (4%) from 2002 to 2003, there was

a significant decrease (21%) in Africa.

In 2002 only one of the 66 armed conflicts (that be-

tween India and Pakistan) was coded as an interstate

conflict.

In 2003 two of the 59 ongoing armed conflicts

were coded as interstate conflicts (the fighting between

India and Pakistan and the US-led invasion of Iraq).

Figure 2.2 tells us about the number of countries in

each region that suffered from political violence in 2002

and 2003. Since some countries experience several conflicts

WHAT’S NEW ABOUT THE UPPSALA/HUMAN SECURITY CENTRE DATASET.

The Uppsala/Human Security Centre dataset covers the three main categories of political violence.

This report publishes the new data for 2002 and 2003.

The Human Security Report 2006 will publish

the data for 2004 and 2005.

Category of political violence

State-based armed co n. icts

Conflicts between states or between a state and a non-state actor, with at

least 25 battle-related deaths per year. Includes all inters tate wars and those

civil wa rs where the state is a warring party. Updated data on the number

of state-based armed conflicts compiled by the Uppsala Data Program are

published each year in the Journal of Peace Research and the SIPRI Yearbook.

What’s new. Data on the number of reported battle-related deaths for each

conflict and the death rate ( fatalities per 100, 000 population) for each

country experiencing conflict.

Non-state armed conflicts

Conflicts in which none of the warring parties is a government and which

incur at least 2 5 battle-related deaths per year.

What’s new. Data on the number and location of non-state conflicts and

the numbers killed have never before been systematically collected and

published annually.

One-sided violence

The deliberate unopposed slaughter of at least 25 civilians in one year by

a government or political group. Includes genocides, politicides and other

violent as saults on civilians.

What’s new. Barbara Harff’s dataset ( see Part I ) counts genocides and

politicides and the death tolls associated with them. However, the one-

sided violence considered in the new dataset is a broader category that goes

beyond genocide and politicide.

What’s counted.

The number of cases

of political violence

(armed conflicts plus

cases of one-sided

violence).

The number of countries

experiencing political

violence.

The number of

reported deaths from

political violence.

The number of

reported deaths per

100,000 population.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

in a single year, the figures are significantly lower than in

Figure 2.1.

Africa was the only region to show a marked year-on-

year change. Between 2002 and 2003 Africa became sig-

nificantly more secure, with 28% fewer countries being af-

fected by political violence.

As Figure 2.3 indicates, the seven countries that

had the highest number of conflicts and cases of one-

sided violence in 2002 were India (10), the Democratic

Republic of the Congo (9), Somalia (8), Niger ia (7),

Ethiopia (6) and Sudan and Burma (Myanmar) (with

5 each).

In 2003 India again suffered the highest number of

armed conflicts and cases of one-sided violence (15), fol-

lowed by Uganda (7), the Democratic Republic of the

Congo and Ethiopia (with 6 each), and Nigeria, Somalia

and Sudan (with 5 each).

Has the number of conflicts really declined.

The armed conflict statistics that Uppsala/PRIO update each

year, and that are published in the Stockholm International

Peace Research Institute’s SIPRI Yearbook and the Journal of

Peace Research, have become a valued and trusted source

of information on the trends in armed conflict around

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Figure 2.3 Cases of armed conflict and one-sided violence by country, 2002–2003

Source: Uppsala/Human Security Centre dataset, 2005

Individual country counts of state-based conflicts, non-state conflicts and cases of one-sided violence

provide a detailed picture of the location of political violence around the world.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

the world. Yet few, if any, non-specialists will have been

aware that a whole category of conflict was excluded from

these publications.

The single most important finding from the new

Uppsala/Human Security Centre dataset is that in 2002

and 2003 there were more non-state conflicts than there

were state-based conflicts.

Of the 66 armed conflicts in 2002, 34 (52%) were non-

state conflicts, and 32 (49%) were state-based.

Of the 59 armed conflicts in 2003, 30 (51%) were non-

state conflicts, and 29 (49%) were state-based.

This raises an important question about one of the

central claims made in Part I of this report—namely, that

the number of armed conflicts has declined quite dra-

matically in the past dozen years. That claim was based

on trends revealed in the Uppsala/PRIO dataset on state-

based conflicts.

How can we be sure that there has been a major decline

in all armed conflicts since the end of the Cold War if the

Uppsala/PRIO dataset counts only state-based conflicts.

In 2002 and 2003 there were more

non-state conflicts than there were

state-based conflicts.

It is at least theoretically possible that non-state con-

flicts increased more than state-based conflicts decreased

during this period. If this were true, there would have been

a net increase, not decrease, in armed conflicts of all types

over the past decade, and the central thesis of this report

could not be supported.

But we can be confident that non-state conflicts

declined in the post–Cold War era. Indeed, the de-

cline may well be steeper than the drop in state-based

conflicts. Why.

First, there are the findings of the Minorities at Risk

Project at the University of Maryland on violence between

communal groups from 1990 to 1998. Project director Ted

Robert Gurr concluded that during this period ‘serious

intercommunal conflict [i.e., non-state conflict] followed

the same rise-and-fall path of violent ethnic challenges to

states [i.e., state-based conflict]’. The intercommunal con-

flicts examined in Gurr’s study declined by more than 50%

between 1993 and 1998.

Second, Monty G. Marshall of the Center for Int-

ernational Development and Conflict Management

has created a dataset that counts the number of coun-

tries experiencing all forms of warfare—including non-

state conflict—each year from 1946 to 2004.

Marshall’s

data reveal an even steeper decline since the end

of the Cold War than the Uppsala/PRIO state-based

conflict dataset.

Together, the studies by Gurr and Marshall indicate

that non-state conflicts have followed the same down-

ward trend in the post–Cold War years as state-based

conflicts—and that the decline in non-state conflict has

likely been greater than the decline in state-based con-

flict. It follows that the number of armed conflicts of all

types has declined.

Can we trust the death toll data.

In addition to tracking the number of armed conflicts and

cases of one-sided violence around the world, the Uppsala/

Human Security Centre dataset provides a count of report-

ed, verifiable and codable deaths from political violence in

every country each year.

The dataset also records whether the deaths were

caused by governments or non-state armed groups, and

whether the conflict was about the struggle for control of

the government or over territory.

Estimating the number of armed conflicts or cases

of one-sided violence is relatively easy. Determining the

numbers killed by political violence is both more difficult

and requires more resources.

Once it was possible to count bodies on the battle-

field when an engagement was over. Not any more. The

typical conflict today is spread over huge areas and many

months—sometimes years. Fighting, particularly between

non-state actors, often takes place in remote areas.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Many of the estimates of war deaths that get publicised

in the media and used by governments, NGOs and even

UN agencies are simply guesses—and are often greatly ex-

aggerated.

In 1995, for example, there were widely publicised

claims that some 200,000 people had been killed in the

fighting in Bosnia and Herzegovina. This figure was not

based on any serious assessment of the evidence and was

subsequently found to be hugely in.ated.

Many of the estimates of war

deaths that get publicised in the

media and used by governments,

NGOs and even UN agencies are

simply guesses.

Uppsala’s approach to data collection is cautious,

conservative and subject to stringent coding rules that

inevitably lead to a degree of under-counting. However,

Uppsala’s methodolog y is currently the only one that can

produce annual national, regional and global trend data

for all three categories of political violence and publish

them in a timely manner.

Three approaches to estimating war deaths

A number of different methodologies are currently used to

measure war deaths in a systematic manner. Each of the

following approaches has advantages and disadvantages—

and each serves different analytic and policy purposes.

Report-based methodologies

The Uppsala/Human Security Centre and Correlates of

War datasets and the International Institute of Strategic

Studies’ Armed Conflict Database

all rely on reports of

deaths from political violence.

In Uppsala’s case, the relevant information is culled

electronically from the huge Factiva news database using

purpose-built automated software.

The selected data are

then reviewed and coded. This approach, as noted above,

has a systematic bias toward under-counting. There are

two main reasons for this.

First, some deaths simply never get reported. This is

particularly true of conflicts where the media are excluded,

such as in Chechnya. Deaths that are not reported cannot

be recorded.

Second, Uppsala’s stringent coding rules require:

That there be a minimum of 25 deaths per year.

That the cause of death be identified as political rather

than criminal violence.

That the group responsible for the deaths be reli-

ably identified.

The case of Iraq clearly shows the effect this last re-

quirement can have on battle-related death counts.

Most of the killings that have taken place during the

post-war insurgency in Iraq have been carried out by un-

identified perpetrators and cannot, therefore, be coded.

This means that thousands of instances of what are very

likely acts of political violence have gone unrecorded.

The coding difficulties in the Iraq case are unusual,

but they highlight the need to record a new category of

deaths—those that are likely due to political violence but

are not codable for lack of sufficient information. This cat-

egory will be included in future Human Security Reports.

Epidemiological surveys

Epidemiological surveys in wartorn countries are most-

ly undertaken to provide information for humanitarian

agencies whose primary interest is the health consequenc-

es of war, not its causes. But such surveys are increas-

ingly used to estimate death tolls from combat-related

violence, as well as deaths from war-exacerbated disease

and malnutrition.

Standard population health survey methodologies

based on a randomly chosen sample of the population are

used to establish death rates from various causes. Death

rates during or after the conflict are then compared with

pre-conflict death rates to determine the war-induced ‘ex-

cess’ death rate. (‘Excess’ deaths are those that would not

have occurred had there been no conflict.)

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

When the sample size is large enough and appropriate-

ly selected, researchers can have considerable confidence

in the accuracy of the extrapolated death toll estimates.

In-depth historical investigations

Another approach to counting war deaths, one developed

by Patrick Ball and colleagues at the Human Rights Data

Analysis Group,

relies on exhaustive historical investi-

gations of particular conflicts using painstakingly cross-

checked reports from human rights organisations, data

from exhumations, extensive interviews and other relevant

sources. In-depth historical investigations have been car-

ried out in Guatemala, Peru, Kosovo, East Timor and Haiti,

and their findings are typically used to provide information

on gross human rights abuses for truth and reconciliation

commissions

or as evidence for war crimes tribunals.

These studies tend to uncover large numbers of deaths

that have not been previously reported.

The strengths of report-based methodologies

In June 2005 the Small Arms Survey 2005 (SAS)

published

a comparison of estimates of death totals compiled by vari-

ous report-based datasets (including Uppsala’s) with those

of epidemiological surveys and in-depth historical inves-

tigations. It found that the death estimates of the report-

based methodologies were two to four times lower than

death estimates produced by the other methodologies.

Neither epidemiological surveys nor

in-depth historical studies can be

used to produce timely global and

regional death toll data.

While the SAS analysis of Uppsala’s data was prob-

lematic for a number of reasons,

its claim that report-

based methodologies under-count battle-related deaths

was clearly correct.

So if epidemiological surveys and in-depth historical

analyses like those undertaken by the Human Rights Data

Analysis Group produce a more complete picture of the

numbers of people killed by political violence, why doesn’t

the Human Security Report rely on them to provide death

toll data.

The short answer is that neither epidemiological sur-

veys nor in-depth historical studies can be used to pro-

duce timely global and regional death toll data. There are a

number of reasons for this:

Global coverage. Only methodolog ies like Uppsala’s

record deaths for all conflicts in all countries

each year.

N ational coverage. Many epidemiological surveys

carried out in post-conflict societies, or those still at

war, focus only on the areas that are of greatest con-

cern to humanitarian agencies. But the findings of

surveys of a particular region of a country cannot be

extrapolated to generate national death toll estimates.

Comparability. Different epidemiological surveys

have different coding rules, which makes compari-

sons problematic. For example, some surveys lump

murders and combat-related deaths together under a

sing le descriptive category of ‘violent deaths’. Others

distinguish between deaths caused by political vio-

lence and those caused by criminal violence.

Timeliness. In order to establish comparable annual

death toll estimates, data for each country in conflict

must be collected and published in a timely manner.

Only report-based methodologies currently do this.

Cost. Nationwide epidemiological surveys and in-

depth historical studies are expensive (relative to re-

port-based methodologies). Currently, there is simply

no funding available to conduct them annually for ev-

ery country experiencing political violence.

Feasibility. Even when funding is available, epide-

miological surveys and in-depth historical analyses

normally require the permission, if not the coopera-

tion, of governments. Sometimes that permission will

not be granted. Report-based approaches are not

similarly constrained.

Each of the three approaches to measuring death tolls

reviewed here serves a different purpose and each adds to

our understanding of political violence around the world.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

But while in-depth historical studies and epidemiolog-

ical surveys provide the most detailed picture of the human

costs of political violence in individual conflicts, they can-

not be used to create annually updated datasets that track

national, regional and global trends in deaths from political

violence. And it is the data on trends that are of greatest

importance to policymakers and to researchers investigat-

ing the causes of war and peace. A systematic bias in the

recorded death toll data toward under-counting does not

compromise their value in tracking trends.

Trend data are important not only because they help

policymakers dtermine whether or not their policies are

working, but because they also reveal long-term changes

that might otherwise be overlooked. For example, the rap-

id rise in the number of armed conflicts around the world

during the ‘Long Peace’ of the Cold War passed largely

unnoticed by a scholarly community that were focused

on relations between the major powers and the East-West

confrontation. This dramatic shift only became obvious

when reliable trend data were published. The same trend

data revealed the subsequent substantial drop in armed

conflicts in the post–Cold War era.

The 1946 to 2002 Lacina and Gleditsch battle-death

dataset reviewed in Part I provides another example of the

utility of trend data. While the accuracy of many of the indi-

vidual death tolls, for particular countries in particular years,

can certainly be challenged, the finding that there has been

a dramatic decrease in the deadliness of conflict over the

past 50-plus years is not in question. This surprising finding

helps us understand how changes in the modes of combat

during this period have made warfare much less deadly.

Trend data are important because

they help policymakers determine

whether or not their policies

are working.

Deaths from political violence: The new dataset

Globally, state-based conflicts killed more people (57%

of the total in 2002, and 75% in 2003) than either non-

state conflicts (26% in 2002, 14% in 2003) or one-sided

violence (18% and 10%). But as Figure 2.4 shows, there

was considerable regional variation.

Remarkably, in spite of the Iraq war in 2003, the re-

ported global death count from all forms of political vio-

lence held virtually steady from 2002 to 2003.

Figure 2.4 Numbers of reported deaths from political violence, 2002–2003*

Source: Uppsala/Human Security Centre dataset, 2005

*Fatality figures are ‘best estimates’

From 2002 to 2003 total reported deaths from all categories of political violence decreased in all

regions except the Middle East, where the Iraq war drove deaths from state-based conflict dram-

atically upward.

1 9

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Figure 2.5 Numbers of reported deaths from political violence by country, 2002–2003*

*Fatality figures are ‘best estimates’.

Source: Uppsala/Human Security Centre dataset, 2005

Number of fatalities per 100,000 of population, rounded to the nearest decimal. Population data come from the World Bank’s World

Development Indicators database and are for 2002.

Reported death counts and death rate data for individual countries reveal a more detailed picture of

the costs of political violence.

2 2

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

As it was, the Iraq war meant that the body count

for the Middle East increased at least sixfold, and likely

much more. Elsewhere, deaths from political violence

fell substantially from 2002 to 2003: in the Americas by a

massive 62%, in Europe by 32%, in Asia by 35% and even

in wartorn Africa by 24% .

Most of the increase in the battle-related death toll in

the Middle East is attributable to state-based conflict in

Iraq. In all other regions, deaths from state-based conflict

dropped significantly—down 11% in Africa and 58% in

the Americas.

Non-state conflict death tolls were down everywhere,

from a 10% fall in the Middle East to a massive 92% drop

in Asia.

And from one-sided violence, reported deaths fell in

all regions except Europe. The decrease ranged from 19%

in the Middle East to 51% in Africa.

Figure 2.5 shows that in 2002 the five countries with

the highest number of reported deaths from all three

forms of political violence were the DRC, India, Nepal,

Sudan and Uganda.

In 2003 the picture had changed significantly. The

five countries with the highest number of reported deaths

from political violence were Iraq,

Sudan, India, the DRC

and Liberia.

However, these reported death-count figures over-

state the significance of high absolute numbers of deaths

in more populous countries. A different picture emerges

Similar claims are regularly made by UN agencies

(including UNDP

2 3

and UNICEF

) and are quoted in the

European Union’s security strategy.

2 5

Many journalists,

NGOs, academics and policymakers accept the 90% figure

as an uncontested truth.

And yet it has no basis in fact.

The claim can be traced back to two sources. In 1991

Uppsala University published Casualties of Conflict,

2 6

which

contained the claim that ‘nine out of ten victims (dead and

uprooted) are civilians’. On the back cover of the book, how-

ever, the parenthetical words were dropped, leaving only

the statement that ‘nine out of ten victims of war and armed

conflict today are civilians’.

For Uppsala, the category of ‘victim’ included refugees

as well as war dead. But some readers wrongly equated ‘vic-

tim’ with ‘fatality’. What the Uppsala data suggested was far

less dramatic: approximately 67% of those killed in con-

flicts during 1989 were civilians. Today the figure is likely

much lower.

The other contemporary source of the myth—also

from 1991—is Ruth Leger Sivard’s World M ilitary and Social

Expenditures.

2 7

Sivard wrote that ‘in 1990 [the proportion of

civilian to combatant death s] appears to have been close

to 90 %’. But Sivard’s estimate included fata lities from war-

related famines, which is not what most people have in

mind when they talk about civilians being killed in war.

Moreover, there are no global data on deaths caused by

war-related famine and (more importantly) disease—so

it is not clear what sources Sivard used to arrive at her

conclusion.

What then can be said about civilian fatalities in war.

Prior to 1989 information was so poor that it was virtually

impossible to make even crude estimates of the global civil-

ian death toll. Even today, our estimates of civilian deaths

are based on information that is never complete and is rarely

accurate. Data collected by the Uppsala Conflict Data

Program suggest that between 30% and 60% of fatalities in

2002 were civilians.

It is precisely because it is so difficult to distinguish be-

tween combatant deaths and civilian deaths that Uppsala

embraces both in its ‘battle-related deaths’ category.

Indeed, the only claim we can make with any confi -

dence is that the oft-cited 90% civilian death rate for the

1990s is a myth.

THE MYTH OF CIVILIAN WAR DEATHS

In World War I, 5% of fatalities were civilian; in World War II, fatalities rose to 50%; and in the 1990s,

90% of war deaths were civilian.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

when countries are ranked by their death rates. In 2002

the countries with the highest death rates were Liberia

(21.2 deaths per 100,000 of the population), Israel and the

Palestinian Territories (13.2), Burundi (12.0), Nepal (11.0)

and Uganda (9.2).

And in 2003 the countries with the highest death rates

were Liberia (59.4), Iraq (35.1), Burundi (16.2), Sudan (8.5)

and Uganda (6.5).

For state-based conflict, the five countries with the

most reported battle-related deaths in 2002 were (in order)

Nepal, Sudan, India, Colombia and Uganda; in 2003 they

were Iraq, Sudan, India, Liberia and the Philippines.

For non-state conflict, the five countries with the most

reported battle-related deaths in 2002 were the DRC, India,

Colombia, Somalia and Nigeria; and in 2003 they were the

DRC, Somalia, Uganda, Sudan and Nigeria.

And for one-sided violence, the five countries show-

ing most reported deaths in 2002 were Uganda, the DRC,

India, Burundi, and Israel and the Palestinian Territories;

in 2003 they were Uganda, India, Liberia, Sudan, and Israel

and the Palestinian Territories.

The findings of the new dataset suggest that govern-

ments kill far fewer civilians than do rebel groups. In 2002,

23% of those who died in one-sided political violence were

killed by governments, while 77% were killed by non-state

groups. In 2003, 32% were killed by governments and 68%

by non-state groups. These figures should be viewed with

some caution, however, as they may reflect government

control over local media as much as real differences in

death rates.

Conclusion

The new Uppsala/Human Security Centre dataset has

already generated a number of important and surprising

findings, but its full potential will not be realised until an-

nual data have been collected for some years and clear

trends can be ascertained.

The Human Security Report 2006

will publish the 2004 and 2005 data, enabling the

presentation of four-year trends in all categories for the

first time.

As the previous discussions have shown, we can be

confident about the accuracy of data on the number of

armed conflicts and cases of one-sided violence, but ob-

taining good data on the number of battle-related deaths

and deaths from one-sided violence will always pose a

much greater challenge.

Data collection on the human costs of war remains a

complex and contested business.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Measuring human rights abuse

Despite growing attention to human rights, no

government, international organisation or NGO

collects quantifi ed global or regional data on

human rights abuse. Determining whether

such abuse is increasing or decreasing is

extremely diffi cult.

There is more to security than counting conflict num-

bers, war deaths or terrorist attacks. Core human rights vi-

olations—such as torture; extrajudicial, arbitrary and sum-

mary executions; the ‘disappearance’ of dissidents; and the

use of government-backed death squads—are also an inte-

gral part of the human security agenda. But mapping trends

is problematic.

The UN’s Commission on Human Rights is mandated

to monitor and report on the human rights situation in

member states, but it has signally failed to do so in any

consistent and impartial manner.

Major human rights organisations provide detailed ac-

counts of human rights abuses in individual countries, but

they have long resisted any attempt to provide quantified

measures of violations. They point to the inherent unreli-

ability of much of the data and argue that brutalising 10

citizens is as unacceptable as brutalising a thousand. Any

system of ranking that might imply that the former is more

tolerable than the latter should be rejected, they argue.

But without quantitative annual audits, neither gov-

ernments, international agencies, nor the human rights

community can determine global or regional trends in core

human rights abuse. And knowing whether abuse is in-

creasing or decreasing is a necessary condition for evaluat-

ing the impact of human rights policies.

The Political Terror Scale

A little-known dataset developed at Purdue University

some 20 years ago, and now maintained at the University

of North Carolina, Asheville, by researchers Linda Cornett

and Mark Gibney, goes a significant way toward address-

ing the measurement challenge.

The Political Terror Scale (PTS) uses annual reports

from Amnesty International and the US State Department

to measure the human rights situation in individual coun-

tries.

The higher a country ranks on the five-level scale,

the worse its human rights record.

Level 1. Countries operate under a secure rule of law.

People are not imprisoned for their views, and torture

is rare or exceptional. Politically motivated murders are

extremely rare.

Martin Adler / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Level 2. There is a limited amount of imprisonment for

non-violent political activity. Few persons are affected,

and torture and beatings are exceptional. Politically

motivated murder is rare.

Level 3. Imprisonment for political activity is more ex-

tensive. Politically motivated executions or other po-

litical murders and brutality are common. Unlimited

detention for political views, with or without a trial, is

also commonplace.

In five regions the human rights

situation has improved since 1984.

Level 4. The practices of Level 3 affect a larger portion

of the population. Murders, disappearances and tor-

ture are a common part of life. But in spite of the per-

vasiveness of terror, it directly affects only those who

interest themselves in politics.

Level 5. The terrors characteristic of Level 4 affect the

whole population. The leaders of these societies place

no limits on the means they use, or the thoroughness

with which they pursue, personal or ideological goals.

This scale assumes a constant interval between each

level—so a Level 4 score is taken as equivalent to two Level

2 scores. This assumption allows researchers to sum the

scores for all the counties in a region. The total score is then

divided by the number of countries to arrive at an average

score for the region.

What do the PTS trend data reveal.

The PTS indicates that the level of human rights abuse in

the developing world appears to have remained relatively

constant for the 24 years that the researchers have been

collating data. (There are not enough data on human rights

abuses in the developed world to chart trend lines.) There

is certainly nothing comparable to the dramatic rise and

fall in armed conflicts for the same period, or to the long-

term downward trend in battle-deaths.

In five out of the six regions in the developing world

for which data have been collected the human rights situ-

ation has improved modestly since 1994.

There is good reason to believe that human rights

abuses have been under-reported in the past and that this

under-reporting is reflected in the PTS data. If this is indeed

the case then the human rights situation in the developing

world has likely improved over the last 20-plus years.

The data for the four regions in Figure 2.6 show a

small net increase in human rights abuse over the whole

24-year period.

But in the Middle East and North Africa,

sub-Saharan Africa, and South Asia, the data reveal a mod-

est decline in abuse from the early to mid-1990s to 2003.

In East/Southeast Asia and the South Pacific the human

rights situation worsened from the mid-1990s.

In Eastern Europe and Central Asia, and in Latin

America and the Caribbean, the trend data indicate a net

improvement in the human rights situation over two de-

cades (Figure 2.7).

How reliable are the PTS trend data.

The measurement methodology used in the Political Terror

Scale confronts a number of challenges. First, the method

0 0

7 5

5 0

2 5

0 0

7 5

5 0

2 5

0 0

East/Southeast Asia and the South Pacific

Middle East and North Africa

Sub-Saharan Africa

South Asia

Source: Cornett and Gibney, 2004

In four developing country regions the data

indicate that human rights abuse rose slightly

over the 24-year period. But in three of the four

regions, there has been a modest improvement

in the past decade.

Figure 2.6 Political repression 1980–2003:

A net increase in four regions

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

used to determine a country’s annual PTS ranking on the

five-point scale is inherently subjective, relying heavily on

coders’ judgments of the scope and intensity of human

rights abuses.

Second, the Amnesty International and State Depart-

ment reports are far more comprehensive today than they

were two decades ago. Both organisations rely on media

reports. One recent study found that the percentage of

articles mentioning ‘human rights’ in the Economist and

Newsweek more than doubled between 1980 and 2000.

This suggests that abuse levels were likely under-reported

in the past.

Better reporting today means that fewer human rights

abuses go unrecorded than in the past.

This in turn means

that the human rights situation today is almost certainly

better than the trend data suggest.

Third, the Political Terror Scale coding practices have

likely undergone subtle changes over the years, according

to Cornett and Gibney. For example, levels of abuse in Latin

America that would be scored as Level 3 today may well

have been counted as Level 2 in the early 1980s when gross

human rights abuses were more common, and countries

were generally held to lower standards. If such an uncon-

scious coding shift has indeed occurred, this again suggests

that abuse levels in the past were likely higher than the trend

data indicate.

Fourth, the PTS fails to ‘capture the all-pervasive threat

of violence in a totalitarian state’.

When governments rule

by fear, they may not need to resort to physical coercion.

Fifth, the dataset is not population-weighted: data

from Nepal are given the same significance as data from

China, which has over 50 times more people.

There may well have been more

human rights abuse in the past

than the trend data suggest.

Sixth, data for developing countries are not repro-

duced here because the Amnesty and State Department

reports have not always covered them consistently or

in depth.

Taken together, these data challenges suggest we

should approach the PTS findings with some caution. But

notwithstanding the qualifications, in the absence of any

other data, the Political Terror Scale sheds much-needed

light on a murky corner of human insecurity.

0 0

7 5

5 0

2 5

0 0

7 5

5 0

2 5

0 0

Eastern Europe and Central Asia

Latin America and Caribbean

Source: Cornett and Gibney, 2004

In Eastern Europe and Central Asia and in Latin

America and the Caribbean the data indicate a

net decline in human rights abuse.

Figure 2.7 Political repression 1980–2003:

A decrease in two regions

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Tracking criminal violence

Why include criminal violence in a global human

security audit. Because violent crime kills far

more people than war and terrorism combined.

And while less than one-sixth of states are cur-

rently afflicted by armed conflict, all suffer from

criminal violence.

Political violence and murder share obvious similari-

ties: both involve intentional killing. They are also very

different. While war is a quintessentially political activity,

homicide is not. Armed conflicts engage large numbers

of fighters and require considerable organisation. Most

criminal violence, by contrast, involves individuals or small

groups, and requires little complex organisation.

Some scholars argue that the traditional distinction

between war and violent crime breaks down in many of

today’s civil wars. Ohio State University’s John Mueller, for

example, contends that many armed conflicts are now little

more than collective criminal violence waged by gangs of

thugs for private gain. Similarly, Paul Collier, former direc-

tor of the World Bank’s research department, has described

rebellion as a ‘quasi-criminal activity’.

The traditional distinction between crime and war

breaks down in other ways. In some cases, rebels

resort to violent crime to generate funds for military ends.

In others, governments label armed resistance directed

against them as ‘criminal violence’ in order to delegiti-

mise it.

A nexus of criminal and political violence

There is another reason to include criminal violence in a

human security audit. Political violence and violent crime

can be causally related.

One recent study found that homicide rates increase

by an average of 25% for some five years following the end

of civil wars.

This is not surprising. War erodes the legal

and normative restraints on violence that prevail in times

of peace. And wartime habits of violence can carry over

into peacetime.

Civil wars are often followed by revenge killings, while

demobilised soldiers frequently wind up unemployed,

turning to violent crime in order to support themselves and

their families.

Despite the obvious connections between criminal

and political violence, both governments and researchers

have traditionally dealt separately with violent crime and

armed conflict.

In government, violent crime is the responsibility of

justice departments, while war remains the province of

Martin Adler / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

foreign ministries and defence departments. Among re-

searchers, criminologists deal with violent crime, while

political scientists focus on war. These divisions of labour

make sense much of the time, but not where political and

criminal violence overlap so much that they become vir-

tually indisting uishable.

The international community is now beginning to

pay more attention to the links between political and

criminal violence. This is particularly true in post-conflict

reconstruction programs, and is evident in the strong

emphasis being placed on reintegrating former fighters

into society and on reforming the police, judiciary and se-

curity forces.

In addition, rising international concern about terror-

ism following September 11, 2001, has led to much greater

cooperation between crime fighters and security services in

counterterrorist operations around the world.

The data challenge

It is far from easy to determine global trends in criminal

violence.

First, definitions of violent crime vary. For example,

deaths from terrorism, war and genocide are not usu-

ally defined as homicides. But in some countries, some

of the time, they are. Definitions of sexual assault also

vary considerably.

The international community is now

beginning to pay more attention to

the links between political and crim-

inal violence.

Second, bureaucracies in developing countries have

fewer resources and less experience in collecting and

collating data. They may be more susceptible to politi-

cal interference. Sometimes there is no clear distinction

between the military and the police, which increases the

likelihood that deaths from terrorism or armed conflict

will be recorded as criminal acts.

Third, international agencies use varying data col-

lection methods, which can lead to widely divergent es-

timates. The World Health Organization, for example,

estimated the homicide rate for Africa in 2000 at 22 per

100,000 people.

Interpol’s estimate for the same year was

less than one-third of WHO’s: 6 per 100,000.

Fourth, major problems arise from under-reporting

and under-recording. Rape is universally under-reported

and in some countries under-recorded.

Fifth, statistics are often simply unavailable—in most

years fewer than 50% of governments provide homicide

and rape data to Interpol.

Finally, the countries that fail to provide data are dis-

proportionately poor, which means they are much more

likely to have been involved in civil wars. Both poverty and

a recent history of warfare are associated with higher than

average homicide rates.

It is likely, therefore, that global

and regional violent crime rates significantly understate

the real level of violence.

All of these factors, taken together, hinder the effec-

tive collection and analysis of regional crime data.

And

this in turn makes attempts to track trends in violent

crime inherently problematic.

The global homicide rate

The following discussion draws on an analysis of global

crime trends produced for the Human Security Centre

by Graeme Newman, editor of the UN publication Global

Report on Crime and Justice.

It focuses on two major threats

to human security—homicide and rape.

Figure 2.8 shows the reported global homicide rates

from 1959 to 2001. The trend line is relatively stable, but

with two significant peaks. It is likely, however, that the

two peaks reflect either misreporting, or political violence

in certain countries being counted as homicides.

The 1975 to 1976 peak appears to be the result of a

huge and inexplicable increase in the number of re-

ported homicides in Nigeria (from about 1500 in 1974

to more than 42,000 in 1975) and in Peru (from just

over 400 to nearly 6000 between 1975 and 1976). In nei-

ther of these countries were there then reports of major

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

armed conflict or genocide. Since real homicide rates

would not increase so steeply in such a short period of

time, this increase has no obvious explanation—other

than misreporting.

And the 1994 peak likely reflects the decision by the

Rwandan government to categorise as homicides the es-

timated 800,000 deaths from the genocide.

If these anomalous spikes are ignored, then the global

homicide rate appears to have been relatively stable for

some 40 years. Certainly, there have been no changes re-

motely comparable to the dramatic increases and decreases

in battle-deaths during the same period.

Homicide, North and South

Comparing the reported homicide rates for developing and

industrialised countries further illustrates the difficulty of

analysing the data (Figure 2.9).

The volatility shown in the trend line for developing

countries reflects the fact that some of these countries ex-

perienced civil war and genocide, and that deaths from

these causes were being reported as homicides. In Figure

2.9, the peaks caused by the Rwandan genocide and the

increased ‘homicide’ counts in Nigeria and Peru become

even more obvious.

The sharp upsurge in 2001 is mostly due to a seven-

fold increase in the Africa-wide homicide rate from 2000

to 2001. Since ‘normal’ homicide rates never increase so

steeply in so short a period, this spike can almost certainly

be attributed to recording war deaths as homicides—or

to misreporting.

A world view of rape

Tracking global rape trends (Figure 2.10) became possible

only after 1976, when Interpol began analysing the statis-

tics for rape and other sexual offences separately. Although

more common than homicide, rape remains chronically

under-reported. In 1996 the United Nation’s Office on

Drugs and Crime (UNODC) conducted a crime victim sur-

vey in 10 industrialised countries and found that, on av-

erage, only one in five cases of sexual violence was ever

reported to authorities.

The reasons for under-reporting vary considerably.

Chief among them is the desire of many rape victims to

avoid the social stigma associated with sexual assault, and

the personal trauma of recounting the ordeal and partici-

pating in the police investigation and in any subsequent

trial. Many victims also fear, often with good reason, that

their accusations will not be taken seriously.

Industrialised countries

Developing countries

Source: Graeme Newman, 2003

Reported homicide rates in industrialised coun-

tries are stable compared with those in the de-

veloping world.

Figure 2.9 Homicide rates in industrialised

and developing countries, 1959–2001

Source: Graeme Newman, 2003

4 4

The peaks shown here may be the result of war

and genocide deaths being counted as homi-

cides. The 1994 peak, for example, is likely

attributable to the Rwandan genocide.

Figure 2.8 World homicide rates, 1959–2001

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

As with homicide statistics, fewer than 50% of coun-

tries around the world publish rape statistics in any given

year. However, as the UN’s Global Report on Crime and

Justice has pointed out, the number of countries (not indi-

viduals) reporting sexual violence increased in the 1990s,

reflecting the growing seriousness with which rape is be-

ing viewed internationally.

Establishing statistical trend data on sexual offences is

further complicated by different definitions of rape. Marital

rape, for example, is counted as a crime in some countries

but not in others.

Part of the reason for the volatility in the reported

world rape rate is that the incidence of rape—like that of

homicide—can increase dramatically during civil strife or

genocide. For example, the spikes in the global rape rate

between 1992 and 1997 correspond to periods of political

violence in Rwanda and the former Yugoslavia, where ex-

tremely high levels of rape were reported.

But this explanation does not account for other peaks

and troughs: the doubling between 1998 and 2001, for ex-

ample. Much of this volatility may be due to inaccurate re-

porting and recording.

The regional rape statistics offer some insight into the

problems of getting accurate data. For example, the so-

called New World countries of the United States, Canada,

Australia and New Zealand had the highest reported in-

cidence of rape in the world for most of the period under

review (1977 to 2001)—well over 10 times that of Asia.

Is this because rape really is more prevalent in these

countries. Or do the relatively low reported rape rates in

Africa (prior to 1993), Central and Eastern Europe, and

Asia reflect serious under-reporting of sexual Farmed.

The latter is certainly part of the answer. In many coun-

tries in these regions, victims of sexual violence have little

legal recourse and may become the victims of reprisals if

they report the rape to authorities.

Under-reporting and under-recording are also greater

where the police and the judiciary fail to take sexual vio-

lence seriously, or fail to act on, or even to record, com-

plaints from victims. In many industrialised countries, pub-

lic information campaigns have increased the sensitivity of

policing bureaucracies to the problem of rape, and have led

to an increase in reporting.

Much of the volatility in the reported

rape rate may be due to inaccurate

reporting and recording.

But none of this can explain why reported rape rates

are several times higher in the New World states than

in Western Europe. We do know from U NODC’s 1996

crime victim survey that the difference between Western

Europe and North Amer ica is not a result of differences

in reporting rates—these are not great enough to explain

the large differences in the recorded rape rates between

Europe and North America.

If the disparities between

Europe and the New World countries are not a function

of differences in the rate of reporting , how can they be

explained.

Part of the answer may be that some states in Australia

and some provinces in Canada still subsume the crime of

rape into the broader categor y of ‘sexual assault’, which

also includes lesser of fences that occur more frequently.

Source: Graeme Newman, 2003

The reported global rape rate more than dou-

bled between 1977 and 2001. This may reflect

increased reporting rather than more rape; it

could reflect both.

Figure 2.10 World rape rates, 1977–2001

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

An estimated 5000 to 6000 youths and adolescents

were involved in Rio de Janeiro’s drug trade in 2003.

4 8

Employed and armed by the main drug factions, they take

part in violent confrontations with rival groups and with se-

curity forces, in much the same way that child soldiers fight

in rebel armies.

The drug gangs seek to control the Rio slums through

territorial and paramilitary domination.

4 9

There are frequent

armed disputes over territory, often leading to violent show-

downs with police.

Rio’s annual criminal death toll is so high that it some-

times exceeds the death toll in Colombia, where a violent

civil war has been waged for decades. There were an esti-

mated 60,000 conflict-related deaths in Colombia from the

1980s to 2003.

During the same period, Rio de Janeiro re-

ported 49,913 firearms fatalities, 70% of them attributed to

drug-related violence.

From December 1987 to November 2001, 467 Israeli

and Palestinian youth w ere killed in Israel and the occupied

territories.

During the same period, 3937 under-18-year-

olds were killed by gunfire in the municipality of Rio de

Janeiro alone.

The similarities between children recruited into the

drug trade and into rebel armies are striking. Although

joining a drug ring in Rio is by and large voluntary, many

poor children have few other options and may get involved

when they are as young as eight years old. The majority are

15 to 17 years old, as is true in many documented cases of

child soldiers.

5 4

Like child soldiers, youth working for Rio’s drug car-

tels also function within a hierarchical structure maintained

through orders and punishment, including summary execu-

tions. Drug lords provide the youth with arms, including

assault rifles, machine guns and grenades, and the youth

openly display these in the communities they patrol.

Children involved in the drug trade are sometimes tar-

geted by police for summary execution. In 2001 offi cers killed

a total of 52 under-18-year-olds during police operations.

5 5

Military solutions to Rio de Janeiro’s drug trafficking

disputes are unlikely to work. Wars end but the drug trade

doesn’t—and this is where the parallel to the plight of child

soldiers breaks down. Criminal gangs will continue to com-

pete for control of the drug trade as long as the drugs re-

main illegal and people continue to buy them. While demo-

bilisation programs make sense for child soldiers once peace

agreements are signed, peace agreements have no coun-

terpart in the constant violent struggle to control the drug

trade. And to categorise children working in drug gangs as

soldiers may serve only to legitimise the already high levels

of lethal state force used against them.

Efforts to ease the plight of children affected by, and

involved in, organised armed violence is not, of course,

restricted to Rio. Viva Rio, a Brazilian NGO working to re-

duce violence against children, is collaborating with the

International Action Network on Small Arms to coordinate

research in other countries where children are involved in

gang violence.

Additionally, the Children and Youth in Organized

Armed Violence (COAV) program is working with local

partners in Colombia, Ecuador, El Salvador, Haiti, Honduras,

Jamaica, Nigeria, Northern Ireland, the Philippines, South

Africa and the United States. Its focus is on children and

youth employed or otherwise participating in organised

armed criminal violence where there are elements of a com-

mand structure and power over territory, local population

or resources. This defi nition helps distinguish COAV’s work

from that of researchers dealing with child soldiers or violent

youth crime perpetrated by individuals. Violent youth groups

encompassed by this mandate range from ‘institutionalised’

street gangs in El Salvador, Honduras and the United States,

to politically motivated armed groups known as ‘popular or-

ganisations’ in Haiti, to vigilante groups and ethnic militia

in Nigeria. Through comparative analysis, COAV is seek-

ing to understand the causes of youth-organised armed

violence and to identify creative solutions and best-practice

policy responses.

CHILDREN , DRUGS AND VIOLENCE IN RIO

Violence and illegal narcotics often go hand in hand. Nowhere is this more so than in Rio de Janeiro,

where criminal drug gangs use armed children to help run the trade. Thousands of children have

been killed in violent shoot-outs with the police or rival cartels.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

This explanation cannot, however, account for the ex-

traordinary volatility in reported rape rates in the New

World nations, where the average reported rape rate for

the four countries more than halved, and then more than

doubled, twice, between 1977 and 1983. Indeed, there is

no obvious explanation for these changes—it is impos-

sible to believe that real rape rates could change so dra-

matically and quickly.

No firm conclusions

Global political violence data for the post–World War II era

show unmistakable and highly significant trends. There are

no comparably clear trends in global criminal violence, at

least not for homicide and rape.

The most serious problem with the homicide statis-

tics is the lack of data for many countries and the erratic

reporting in many others. With respect to sexual violence,

determining whether the apparent increase in the world

rape rate is due to better reporting and recording, or is due

to a real increase in the incidence of rape, or to both, is

simply impossible.

As long as these problems continue to compromise the

collection and collation of violent crime statistics, it will be

impossible to track trends in global or regional homicide

and rape rates with any real degree of confidence.

We can reliably document trends in political violence

but not for violent crime—global statistics on murder

and rape are simply too problematic.

Andrew Testa / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Human trafficking

The trafficking of human beings has burgeoned

into a multi-billion-dollar industry that is so

widespread and damaging to its victims that it

has become a cause of human insecurity.

Associated with the worldwide liberalisation of trans-

port, markets and labour, the breakup of the Soviet Union,

and a doubling in world migration over the past 40 years,

human trafficking has become a major source of revenue

for organised crime.

According to the US State Department, ‘Human traf-

ficking is the third largest criminal enterprise worldwide,

generating an estimated $9.5 billion in annual revenue.’

The State Department’s estimate is just for the rev-

enue generated by trafficking itself. The illicit

for the traf fi c kers generated by the victims after they

arrive in the country of destination are many times

higher. A recent report by the International Labour

Organization (ILO) estimated these to be some US$ 32

billion a year.

The US State Department claims that ‘at least’ 600,000

to 800,000 individuals are trafficked across national

frontiers each year. Of these, it is believed that approxi-

mately 80% are women and girls and up to 50% are

minors.

If those who are trafficked within borders are

included, the total number of victims could be as high

as 4 million.

But the illegal nature of the trade, the low priority

given to data collection and research, and the frequent re-

luctance of victims to report crimes, or to testify for fear of

reprisals, combine to make gauging the numbers trafficked

each year extremely difficult.

According to an analyst from the UN’s Global

Programme against Trafficking in Human Beings, ‘Even

though some high-quality research exists most of the data

are based on “guesstimates”, which, in many cases, are

used for advocacy or fund-raising purposes’.

According to the US State

Department, ‘human trafficking

is the third largest criminal

enterprise worldwide’.

Recent research by the International Organization for

Migration’s (IOM) Counter-Trafficking Service shows that

between 2001 and 2003 the number of victims of interna-

tional trafficking referred to the IOM decreased in Kosovo

Mikkel Ostergaard / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

by 67%, in Macedonia by 46% and in Moldova by 35%.

In Albania the decline was 90% between 2000 and 2003.

Numbers of victims receiving ‘assistance’ from the IOM in

Bosnia and Herzegovina declined by 75% between 2001

and 2003.

These findings do not necessarily mean that trafficking

has declined—there are other possible explanations—but

at the very least they raise questions about the convention-

al wisdom that it is steadily increasing.

What is human trafficking.

These findings do not necessarily mean that trafficking has

declined—there are other possible explanations. But at the

very least these data raise questions about the claim that

the trade in human beings is steadily increasing.

In November 2000, the UN General Assembly adopted

a new protocol that defines trafficking (including traffick-

ing within countries) as:

The recruitment, transportation, transfer, harbouring

or receipt of persons, by means of the threat or use of

force or other forms of coercion, of abduction, of fraud,

of deception, of the abuse of power or of a position of

vulnerability or of the giving or receiving of payments

or benefits to achieve the consent of a person having

control over another person, for the purpose of exploita-

tion. Exploitation shall include, at a minimum, the ex-

ploitation of the prostitution of others or other forms

of sexual exploitation, forced labour or services, slavery

or practices similar to slavery, servitude or the removal

of organs.

Cross-border trafficking should not be confused with

people-smuggling. Traffickers seek to exploit their victims

for long-term profit. People-smuggling ends once human

cargoes are delivered across borders.

However, as the ILO has noted, in practice ‘it is often

difficult to distinguish ... between workers who have en-

tered forced labour as a result of trafficking and those who

have been smuggled.’

Trafficking takes place in three stages: recruitment,

transportation and exploitation.

Recruitment

Cross-border traffickers frequently use false promises

of well-paid overseas employment to recruit their vic-

tims. Positions are advertised in legitimate employment

agencies, in mainstream magazines, in newspapers or on

the Internet.

Women are the major victims of trafficking, and most

are trafficked into some form of prostitution. Some are

aware that they will be employed as sex workers, but few

understand the degree to which they may be indebted, in-

timidated, exploited and controlled.

These findings raise questions

about the conventional wisdom

that trafficking is steadily increasing.

An IOM study reported that among trafficked females

who had been interviewed by researchers, 10% had been

kidnapped.

Another study estimated that 35% of mi-

nors trafficked from Albania were abductees.

Abductors

are often acquaintances, relatives or friends of the family.

In some cases, children are simply sold by their parents

or guardians.

Transportation

The second stage in cross-border trafficking is trans-

portation to the target country—which is usually, but

not always, the promised destination. Here, traffickers

often use people-smuggling networks, relying on cor-

rupt police and bribed border guards and customs of-

ficials to help move their human cargoes expeditiously

across frontiers.

But increasingly, trafficked persons travel openly, on

legal or forged travel documents, which are often obtained

with help from their traffickers.

Exploitation

The final stage in the trafficking chain is ‘employment’ in

a wide variety of businesses that seek cheap, compliant

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

workers, and whose operators ask few questions about the

origin of their employees.

A study by the United Nation’s Office on Drugs and

Crime (UNODC) found that 85% of women, 70% of

children and 16% of trafficked men are used for sexual

exploitation.

The ILO, on the other hand, estimates that

only 43% of victims are trafficked into commercial sexual

exploitation.

The fact that two separate UN organisations

come out with such different figures serves again to remind

how uncertain all estimates in this area are.

Male victims tend to be exploited as forced labour,

drug vendors and beggars—even as combatants in

armed conflicts.

Controlling victims

To realise their profits, traffickers must ensure that their

‘investments’ are safeguarded, making the control of vic-

tims a top priority.

Debt bondage is one of the commonest methods.

Victims are often grossly overcharged for transportation,

and when the costs of accommodation, food and cloth-

ing are added to this debt and exorbitant interest rates

are charged, escape from indebtedness can become virtu-

ally impossible.

Wartorn countries may be used as

transit routes. Mass displacement

and loss of livelihoods create a huge

potential supply of victims.

Violent coercion—rape, beatings and threats to fam-

ily and loved ones—is also often used to intimidate the

victims. One study of trafficked women assisted by

the IOM found that some 55% had been beaten and

sexually abused.

To discourage escape, traffickers may withhold their

victims’ identification and travel documents, reduc-

ing the prospect of successful escape and return home.

Victims without legal residency rights in their destina-

tion country are reluctant to appeal to the authorities for

help, in case doing so puts them at risk of prosecution,

deportation, or both.

Human trafficking and conflict

Armed conflicts create new opportunities for traffickers.

Wartorn countries may be used as transit routes, while

mass displacement and loss of livelihoods create a huge

potential supply of victims.

Women and girls are frequent-

ly trafficked within and across borders to provide sexual

services to combatants and to work as cooks, cleaners and

porters. The longer a war persists, the more extensive the

displacement, and the more prolonged and widespread

the suffering and poverty, the greater the opportunities

for traffickers.

A 2002 report from the US State Department described

abuse by government-backed militias in the long-running

civil war in impoverished southern Sudan. This included

’capture through abduction (generally accompanied by

violence); the forced transfer of victims to another com-

munity; subjection to forced labor for no pay [and] denial

of victims’ freedom of movement and choice’.

The aftermath of war also provides opportunities for

traffickers. In Bosnia and Herzegovina, Sierra Leone and

other societies struggling with post-war reconstruction,

peacekeepers and humanitarian and aid workers generate

a strong demand for sex workers, one that traffickers have

been quick to meet. Involvement of UN peacekeepers in

trafficking has become a major source of concern for the

UN’s Department of Peacekeeping Operations.

Human trafficking and poverty

The victims of trafficking mostly originate in countries

plagued by weak or corrupt governments, economic de-

cline, poverty, social upheaval, organised crime and vio-

lent conflict. In failing states and some former communist

countries, weak law enforcement has permitted the or-

ganised crime networks involved in trafficking to flourish

largely unchallenged.

Poverty appears to be the single most significant driv-

er of human trafficking. In a 2001 report on nine coun-

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

tries in West and Central Africa, the ILO’s International

Programme on the Elimination of Child Labour found that

‘countries that have widespread poverty, low education

levels and high fertility rates tend to be those from which

children are trafficked.’

In Eastern Europe and the for mer Soviet Union, high

unemployment, low wages and the disappearance of so-

cial safety nets following the collapse of communism have

made the blandishments of traffickers more attractive to

potential victims, while governments have done little to

try to control the trade. Trafficking of women and girls is

also driven in part by sexual and ethnic discrimination.

Responsibility for trafficking does not rest solely with

the traffickers. Without the strong demand for trafficked

labour from brothel owners, sweatshop operators and

others, there would be no trade in human beings. And

without demand from consumers for cheap sex and the

goods made by ultra-cheap labour, there would be no

brothels or sweatshops.

Responsibility is also shared by governments in recipi-

ent countries, which have done little to combat the trade.

Combatting trafficking

Over the last few years, in part as a result of increased me-

dia attention, there have been signs that the problem is

being taken more seriously. In Europe, for example, there

has been a concerted effort to create and harmonise legis-

lation aimed at bringing down the criminal organisations

that run the big trafficking networks.

Recognition of the need to combat trafficking is grow-

ing in the rest of the world as well. By August 2005, 117

countries had signed, and 87 countries had become par-

ties to, the new UN trafficking protocol, which entered into

force in December 2003.

This new protocol makes a strong call for the pro-

tection of victims. But Europol, in its 2003 report, Crime

Assessment : Trafficking of Human Beings into the European

Union, notes that, notwithstanding this commitment, it

is still the victims of trafficking that too often bear the

brunt of legal censure, while their ex ploiters walk free.

‘Women soliciting in public are criminalised, initially

for offences related to the selling of sexual services and

subsequently for of fences arising from their illegal entry

to and residence in the country and lack of valid docu-

mentation.’

The case for a more victim-sensitive approach to com-

batting trafficking is pragmatic as well as moral. Prosecuting

traffickers—and dismantling the networks that support

them—is far more likely to succeed with the cooperation

of victims. No one is better positioned than the victims to

identify—and testify against—the exploiters.

Kosovo: women forced into prostitutions have often

been traffi cked from the former Soviet Union.

Teun Voeten / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Creating a human security index.

It is not possible at present— and may not even

be desirable—to produce a reliable human secu-

rity index. But it is possible to determine which

countries are most threatened by political vio-

lence, human rights abuse and instability.

Every year the much-cited Human Development Report

produced by the United Nations Development Programme

(UNDP) ranks countries around the world according to

their citizens’ quality of life. It draws on data about life

expectancy, educational achievement and income to cre-

ate a single composite measure that it calls the Human

Development Index (HDI).

UNDP’s annual ranking exercise generates intense in-

terest, a great deal of media coverage—and considerable

controversy. In 2004 Norway, Sweden and Australia ranked

on top of the index, while Sierra Leone, Niger and Burkina

Faso took the bottom places.

Composite indices like the HDI can serve a number of

purposes. They can:

Encapsulate in a single measure a range of

complex data.

Facilitate comparisons between countries over time.

Stimulate public discussion.

Be used to bring pressure to bear on governments.

In recent years there has been considerable debate

about creating a human security index, and two such in-

dices have already been developed. Both focus primarily

on development issues; neither includes any measure of

violence; neither is regularly updated.

There are a number of possible measures of human

insecurity that might be combined to make a composite

index, including battle-related death rates, ‘indirect’ death

rates, and homicide and rape rates.

In recent years there has been

considerable debate about creating

a human security index.

Is it possible to combine indicators of this kind into a

single composite human security index.

The short answer is that it is certainly not currently

possible, and that it is probably not desirable.

There are a number of practical challenges. The most

serious is that the existing datasets used to measure

human insecurity are not comprehensive enough—

Ahikam Seri / Panos Pictures

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

and many are not updated annually. Data on homi-

cide and rape are missing for most of the least secure

countries in the world and there are no global data on

indirect deaths—those deaths caused by disease and

the lack of food, clean water and health care that result

from war.

Moreover, even if it were possible to create a sin-

gle composite human security index, it is not clear that

doing so would be desirable. While composite indices

have distinct advantages, simplicity also has a downside.

Composite indices can conceal more information than they

convey. Presenting the data from individual human secu-

rity datasets separately, rather than aggregating them into

a single index, conveys more information—and conveys it

more clearly.

Aggregating very different measures—death rates and

rankings of human rights violations, for example—also

raises difficult questions about how to weight the different

measures when combining them.

And while providing useful insights into the least se-

cure countries, such measures are not very useful for de-

termining the most secure countries, the majority of which

are found in the developed world.

Here the difficulty is that there is not much insecurity

to measure. By definition, highly secure countries rarely ex-

perience warfare, so very few suffer battle-related deaths.

And human rights measures for some of the most secure

countries are also missing from the Political Terror Scale

dataset. The World Bank instability indicator does include

the more secure industrialised countries, but it is too nar-

row to serve as a useful measure on its own. Differentiating

among countries that are not afflicted by war, are highly

stable politically and have very low levels of political re-

pression is extraordinarily difficult.

Nevertheless these two datasets, along with the

Uppsala/Human Security Centre dataset, measure impor-

tant dimensions of human insecurity. The statistics they

provide form the basis of Figure 2.11, and give three paral-

lel measures of the world’s least secure countries:

T he Uppsala/Human Security Centre dataset. The

figures shown are the ‘best estimates’ of death

rates from political violence in 2003. They include

both battle-related deaths and deaths from one-

sided violence.

The Political Terror Scale from the University of North

Carolina, Asheville, which measures core human rights

abuse. Countries are scored on a scale from 5 (worst) to

1 (best), based on human rights violations in 2003.

The World Bank’s composite Political Instability and

Absence of Violence Index,

a measure that gauges the

probability that a government ‘will be destabilised

or overthrown by possibly unconstitutional and/or vi-

olent means, including domestic violence and terror-

ism.’

Countries are ranked on a scale from 0 (worst)

to 100 (best).

Are these the world’s least secure countries.

Figure 2.11 reveals a remarkable overlap between these

three measures of human insecurity. Countries plagued by

high levels of political violence and human rights abuse

tend also to be politically unstable, and vice versa.

But the absence of any measure of criminal violence

is a major concern. A number of countries that experience

neither wars nor political instability nevertheless have very

high levels of criminal violence.

The Ipsos-Reid survey finding that many people fear

criminal violence more than they fear political violence

(see Part I) gives further weight to the argument that

criminal violence data should be included in any compos-

ite index of human security.

An even bigger omission is the absence of any data on

indirect deaths. As we show in Part IV of this report, war-

related disease and malnutrition kill far more people than

combat does.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

Figure 2.11 The world’s least secure countries.

Source: Human Security Centre, 2005

Three different measures of human insecurity give three separate ‘least secure’ rankings. There is a high

degree of overlap between the rankings.

8 3

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

part ii

ENDNOTES

1. UNESCO Bangkok, ‘Trafficking Statistics Project’, UNESCO Bangkok website, www.unescobkk.org/index.php.id=1022

(accessed 1 August 2005).

2. ‘UNICEF Official Cites “Largest Slave Trade In History”’, UN Wire 20 February 2003, http://www.unwire.org/UNWire

/20030220/32139_story.asp (accessed 1 August 2005).

3. In 2002 the US and its allies were fighting al-Qaeda, which is not a state.

4. For Uppsala’s analysis of the 2002 and 2003 political violence data, see the Human Security Report website at www.human

securityreport.info.

5. For more detailed information on definitions, see the Human Security Report website at www.humansecurityreport.info. See

also the Uppsala Conflict Data Program’s database at www.pcr.uu.se/database/.

6. These data are current as of September 2005 but may be subject to slight revision as new information becomes available.

Note that coding decisions mean that there are sometimes slight differences in the information contained in Figures 2.1–

2.5. For example, Uppsala argues that since the ultimate goal of al-Qaeda is the overthrow of the US, the conflict should be

coded as a conflict over control of the government of the US. The conflict is consequently counted in the Americas conflict

total in Figures 2.1 and 2.2. In Figure 2.3, which lists the numbers of cases of armed conflict and one-sided violence by

location, the USA vs. al-Qaed a conflict is counted in Afghanistan’s conflict total because that is where the fighting oc-

curred. I n Figures 2.4 and 2.5, battle-related deaths and deaths from one-sided violence are counted in the country where

they occur.

7. Due to coding rules, the total number of conflicts in each region in Figure 2.3 may not exactly match the total number of

conflicts in each region in Figure 2.1. The interstate conflict between India and Pakistan in 2002 and 2003 is counted once

in the Asia total in Figure 2.1. However, it is counted twice in Figure 2.3 (which lists the number of cases of armed conflict

and one-sided violence by country): once in India’s total and once in Pakistan’s total (fighting occurred in both countries).

Similarly, the US vs. al-Qaeda conflict is included in Afghanistan’s total (and Asia’s) in Figure 2.3, whereas in Figure 2.1 it

is included in the Americas regional total.

8. Ted Robert Gurr, ‘Containing Internal War in the Twenty-First Century’, in Fen Osler Hampson and David M. Malone, eds.,

From Reaction to Conflict Prevention: Opportunities for the UN System (Boulder: Lynne Rienner, 2002).The author notes that

the sample was not comprehensive since it only examined intercommu nal conflicts among groups that were also involved

with conflicts with a government.

9. Professor Marshall was p reviously at the Center for International Development and Conflict Management at the University

of Maryland.

10. Monty G. Marshall, ‘Global Conflict Trends’, Figure 3, Center of Systemic Peace website, 1 February 2005, http://mem

bers.aol.com/CSPmgm/conflict.htm (accessed 19 June 2005). For a comparison with trend data on the number of states

with armed conflicts that is derived from Uppsala’s state-based data, see Monty G. Marshall, ‘Measuring the Societal

Impact of War’, in Fen Osler Hampson and David M. Malone, eds., From Reaction to Conflict Prevention: Opportunities for the

UN System.

11. Although this information is not published in this report, it is available on the Human Security Report website at www

.humansecurityreport.info.

12. Small Arms Survey, Small Arms Survey 2005, (London: Oxford University Press, 2005).

13. For information on the Correlates of War dataset, see www.correlatesofwar.org/. For information on the International Institute

of Strategic Studies’ Armed Conflict Database, see http://acd.iiss.org/armedconflict/MainPages/dsp_WorldMap.asp.

14. For information on the Factiva database operated by Dow Jones and Reuters, see www.factiva.com. The database draws on

some 9000 media sources around the world, including local, national and international newspapers, leading business maga-

zines, trade publications and newswires. Uppsala also draws on other sources of data, including primary source material.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

15. For an account of the methodology used in these studies, see the American Association for the Advancement of Science

website at http://shr.aaas.org/hrdag/.

16. Note that some commissions are truth commissions while others are truth and reconciliation commissions.

17. Small Arms Survey, Small Arms Survey 2005.

18. The treatment of Uppsala’s data by the Small Arms Survey (SAS) is problematic for a number of reasons. Before 2002

Uppsala did not collect death toll data on non-state conflicts or on one-sided violence, so large numbers of deaths from

political violence were not being counted. But deaths in these categories were being measured in the epidemiological sur-

veys and the in-depth historical investigations referenced in the SAS. By the time of the Iraq war, Uppsala was collecting

data on death tolls from non-state conflicts and one-sided violence, but here the SAS’s comparison is problematic for a

different reason. The SAS’s analysis compares Uppsala’s

‘best estimate’ of reported and codable battle-related deaths in Iraq

in 2003 with those of a much-cited epidemiological survey published in the British medical journal the Lancet (L. Roberts,

R. Lafta, R. Garfield et al., ‘Mortality Before and After the 2003 Invasion of Iraq: Cluster Sample Survey’, Lancet 364 (20

October 2004): 1857–1864). Uppsala’s estimate of deaths is 8494, while the Lancet estimate for the year is 22,980—a figure

more than twice as large. The Lancet survey estimate is for all deaths from political violence from the date the war started

to the end of 2003. But Uppsala’s dataset compilers were unable to code any of the killings during the post-war insurgency,

primarily because the perpetrators could not be identified, so the only battle-related deaths Uppsala codes for 2003 are for

the short period in March and April when the conventional war was being waged. In the Iraq case, when ‘like’ really is be-

ing compared with ‘like’—when war-related deaths from violence are compared over the same period of time—there is no

appreciable difference between Uppsala’s death count and that of the Lancet survey. T his single example doesn’t refute the

claim that in general the Uppsala’s data collection approach under-counts battle-related deaths—as noted previously the

very nature of Uppsala’s methodology makes this inevitable. The SAS’s evaluation of the Uppsala data suggests that more

case studies—and more detailed analysis—are necessary to get a better estimate of the degree to which under-counting is

taking place in the report-based methodologies.

19. Note that the Middle East figures should be considerably higher, since Uppsala was unable to code the fatalities in Iraq

during the post-war insurgency.

20. The Uppsala/Human Security Centre dataset records ‘best’, ‘low’ and ‘high’ estimates for each category of political violence

each year. The ‘best estimate’ is the figure that Uppsala regards as being most credible, based on the most authoritative

available information. In Figures 2.4 and 2.5 only the ‘best estimates’ are published, but the ‘high’ and ‘low’ figures are avail-

able on the Human Security Report website at www.humansecurityreport.info.

21. Note that the numbers recorded in this figure and the following figure do not represent the total number of deaths from

political violence but rather the number of reported and codable deaths. These totals are almost certainly lower than the true

death toll.

22. Note that the death toll for Iraq is only for the period of the conventional war (20 March 2003 to 9 April 2003), not the

subsequent insurgency.

23. United Nations Development Programme, Human Development Report 1998: Consumption for Human Development (New

York: United Nations Development Programme, 1998).

24. Graça Machel, Impact of Armed Conflict on Children (New York: United Nations and United Nations Children’s Fund, 1996),

http://www.unicef.org/graca/graright.htm (accessed 10 August 2005).

25. European Union Institute for Security Studies, A Secure Europe in a Better World (Paris: European Union Institute for Security

Studies, 2003), http://www.iss-eu.org/solana/solanae.pdf (accessed 10 August 2005).

26. Christa Ahlström and Kjell-Åke Nordquist, Casualties of Conflict—Report for the World Campaign for the Protection of Victims

of War (Uppsala: Uppsala University, Department of Peace and Conflict Research, 1991).

27. Ruth Leger Sivard, World Military and Social Expenditures, 14th ed. (Washington, DC: World Priorities, 1991).

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

28. See endnote 22.

29. Uppsala has collected data on one-sided violence back to the mid-1990s, which will be used to graph trends in the 2006

Human Security Report. Su bject to funding, the same backdating exercise will be undertaken for non-state conflicts.

30. See the background document commissioned by the Human Security Centre, Linda Cornett and Mark Gibney, ‘Tracking

Terror: The Political Terror Scale 1980–2001’, at the Human Security Report website, www.humansecurityreport.info.

31. The Political Terror Scale counts human rights abuses by any group—government or non-government. Most abuse, how-

ever, is by governments.

32. The definitions of regions correspond approximately with the categories used by the US Department of State. The exact

definitions are spelled out in Linda Cornett and Mark Gibney, ‘Tracking Terror: T he Political Terror Scale 1980–2001’. See the

Human Security Report website, www.humansecurityreport.info.

33. For a discussion of this issue, see James Ron, Howard Ramos and Kathleen Rodgers, ‘Transnational Information Politics:

Amnesty International’s Country Reporting, 1986–2000’, Department of Sociology, McGill University, 11 August 2004,

http://www.gwu.edu/~psc/news/ISQ%20submission2.pdf (accessed 1 August 2005).

34. Ibid.

35. Linda Cornett and Mark Gibney, ‘Tracking Terror: The Political Terror Scale 1980–2001’.

36. John Mueller, The Remnants of War (Ithaca: Cornell University Press, 2004); Paul Collier, ‘Rebellion as a Quasi-Criminal

Activity’, Journal of Conflict Resolution 44 (2000): 839–853.

37. Paul Collier and Anke Hoef.er, ‘Murder by Numbers: Socio-Economic Determinants of Homicide and Civil War’, Centre for

the Study of African Economies Working Paper Series, no. 2004–10, http://www.csae.ox.ac.uk/workingpapers/pdfs/2004-

10text.pdf (accessed 10 August 2005). The authors found no evidence that the causal relationship went the other way: that

high rates of homicide increased the risk of civil war.

38. In July 2004, for example, the 9/11 Commission Report argued for ‘integrated all-source analysis’ of international terror-

ism. National Commission Report, 22 July 2004, http://www.9-11commission.gov/report/911Report.pdf (accessed 20 June

2005).

39. World Heath Organization, The World Health Report: 2004: Changing History (Geneva: World Heath Organization, 2004),

www.who.int/whr/2004/en/ (accessed 20 June 2005).

40. See Interpol data in the background document commissioned by the Human Security Centre, Graeme Newman, ‘Human

Security: A World View of Homicide and Rape’, at the Human Security Report website www.humansecurityreport.info.

41. Paul Collier and Anke Hoef.er, ‘Murder by Numbers: Socio-Economic Determinants of Homicide and Civil War’.

42. There are several methodological difficulties in compiling and presenting this data over time. During the past 40 years there

has been a dramatic increase in the number of countries in the international system, and many borders have changed. Not

all countries keep accurate records. Some of the countries that do keep records compile and report them annually; others

do so less frequently.

43. UN Office for Drug Control and Crime Prevention, Global Report on Crime and Justice (New York: Oxford University Press,

1999).

44. Graeme Newman, ‘Human Security: A World View of Homicide and Rape’.

45. Cited in Bree Cook, Fiona David and Anna Grant, Sexual Violence in Australia, Research and Public Policy Series no. 36

(Canberra: Australian Institute of Criminology, 2001), http://www.aic.gov.au/publications/rpp/36/ (accessed 24 June 2005).

46. UN Office for Drug Control and Crime Prevention, Global Report on Crime and Justice.

47. Bree Cook, Fiona David and Anna Grant, Sexual Violence in Australia.

48. Louise Rimmer, ‘Cool Hand Luke’, The Scotsman, 16 August 2003, http://news.scotsman.com/index.cfm.id=887492003 (ac-

cessed 7 June 2004).

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

49. Josinaldo Aleixo de Souza, Socibilidades emergentes—Implicações da dominação de matadores na periferie e traficantes nas favelas

(PhD diss., Universidade Federal do Rio de Janeiro, 25 September 2001).

50. Stockholm International Peace Research Institute, SIPRI Year Book, 2004: Armament Disarmament and International Security

(Oxford: Oxford University Press, 2004).

51. DATASUS, ‘Sistema de informações sobre mortalidade’, Ministério de Saúde, Secretaria da Saúde do Governo do Estado do

Rio de Janeiro, http://tabnet.datasus.gov.br/cgi/deftohtm.exe.sim/cnv/obtrj.def (accessed 7 June 2004).

52. Israeli Information Center for Human Rights in the Occupied Territories, ‘Minors Killed Since 9 December 1987’, http://

www.btselem.org/English/Statistics/Minors_Killed.asp (accessed 7 June 2003; site discontinued).

53. DATASUS, ‘Sistema de informações sobre mortalidade’.

54. Rachel Brett and Margaret McCallin, Children: The Invisible Soldiers, 2nd ed. (Stockholm: Rädda Barnen, 1998).

55. Secretaria de Segurança Pública do Estado do Rio de Janeiro (SSP-RJ), SSP-RJ website, http://www.ssp.rj.gov.br/ (accessed

10 August 2005).

56. Details of COAV’s research and a regularly updated news service focusing on children and armed violence can be found at

www.coav.org.br.

57. The US Department of State, ‘Trafficking in Persons Report’, 14 June 2004, http://www.state.gov/g/tip/rls/tiprpt/2004/34021.

htm (accessed 7 February 2005).

58. International Labour Office, A Global Alliance Against Forced Labour (Geneva: International Labour Office, 2005), http://

www.ilo.org/dyn/declaris/DECLARATIONWEB.DOWNLOAD_BLOB.Var_DocumentID=5059 (accessed 3 June 2005).

59. The US Department of State, ‘Trafficking in Persons Report’, 3 June 2005, http://www.state.gov/g/tip/rls/tiprpt/2005/46606.

htm (accessed 3 June 2005).

60. Kristiina Kangaspunta, ‘Mapping the Inhuman Trade: Preliminary Findings of the Database on Trafficking in Human

Beings’, Forum on Crime and Society 3 (December 2003): 81–103, www.unodc.org/pdf/crime/forum/forum3_note1.pdf (ac-

cessed 3 June 2005).

61. International Organization for Migration, Changing Patterns and Trends of Trafficking in Persons in the Balkan Region (Geneva:

International Organization for Migration, 2004), www.iom.int/iomwebsite/Publication/ServletSearchPublication.event=

detail&id=3831 (accessed 5 June 2005).

62. Article 3(a) of the protocol to Prevent, Suppress and Punish Trafficking in Persons, Especially Women and Children, supple-

menting the United Nations Convention against Transnational Organized Crime, http://www.unodc.org/pdf/crime/a_res_

55/res5525e.pdf (accessed 20 January 2005).

63. The UK-based NGO Anti-Slavery International describes the difference between the two as follows: ‘Human trafficking in-

volves deceiving or coercing someone to move—either within a country or abroad through legal or illegal channels—for the

purpose of exploiting him or her. Smuggling is assisting someone for a fee to cross a border illegally.’ See Human Trafficking

Q&A, Ant i-Slavery International, htt p://ww w.antislavery.org/homepage/antislavery/t rafficking.ht m#qanda (accessed 20

January 2005).

64. International Labour Office, A Global Alliance Against Forced Labour.

65. L. Kelly and L. Regan,

‘Stopping Traffic: Exploring the Extent of, and Responses to, Trafficking of Women for Sexual Exploitation

in the UK’, Police Research Series 125 (London: Home Office Policing and Reducing Crime Unit, 2000).

66. Elizabeth Kelly, Journeys of Jeopardy: A Review of Research on Trafficking in Women and Children in Europe (Geneva: International

Organization for Migration, 2002).

67. Cited in Elizabeth Kelly, Journeys of Jeopardy: A Review of Research on Trafficking in Women and Children in Europe.

68. Ibid.

69. International Organization for Migration, Changing Patterns and Trends of Trafficking in Persons in the Balkan Region.

H U M A N S E C U R I T Y R E P O R T 2 0 0 5

70. Kristiina Kangaspunta, ‘Mapping the Inhuman Trade: Preliminary Findings of the Database on Trafficking in Human

Beings’.

71. International Labour Office, A Global Alliance Against Forced Labour. The difference may be due to the fact that UNDOC’s

data come mostly from European countries, while the ILO data are drawn from all regions of the world.

72. Elizabeth Kelly, Journeys of Jeopardy: A Review of Research on Trafficking in Women and Children in Europe.

73. Elisabeth Rehn and Ellen Johnson Sirleaf, Women, War and Peace (New York: United Nations Development Fund for Women,

2002), http://www.unifem.org/resources/item_detail.php.ProductID=17 (accessed 19 January 2005).

74. US Department of State, Slavery, Abduction and Forced Servitude in Sudan (Washington: US Department of State, Bureau of

African Affairs, 2002), http://www.state.gov/p/af/rls/rpt/10445.htm (accessed 21 February 2005).

75. International Labour Organization, Combating Trafficking in Children for Labour Exploitation in West and Central Africa

(Geneva: Intenational Labour Orgainization, 2001), http://www.ilo.org/public/english/standards/ipec/publ/field/africa/cen

tral.pd f (accessed 19 January 2005).

76. Protocol to Prevent, Suppress and Punish Trafficking in Persons, United Nations Office on Drugs and Crime website, http://

www.unodc.org/unodc/en/crime_cicp_signatures_trafficking.html (accessed 30 August 2005).

77. European Law Enforcement Organisation, ‘Crime Assessment: Trafficking of Human Beings into the European Union’, The

European Police Office website, http://www.europol.eu.int/index.asp.page=publ_crimeassessmentTHB (accessed June 4

2005; document no longer available).

78. Philip Alston, ‘Toward a Human Rights Accountability Index’, Journal of Human Development 1 (2000): 250.

79. AVISO, ‘The Index of Human Security’, Aviso, January 2000, http://www.gechs.org/aviso/avisoenglish/six_lg.shtml (ac-

cessed 15 June 2005); Gary King and Christopher J.L. Murray, ‘Rethinking Human Security’, Political Science Quarterly 116

(2002–2002).

80. The 2003 Political Terror Scale data are used here. The full data table is available in the Political Science Department of

the University of North Carolina at http://www.unca.edu/politicalscience/faculty-staff/gibney.html (accessed 10 September

2004).

81. World Bank, ‘Governance Indicators: 1996–2002’, World Bank Institute website, http://www.worldbank.org/wbi/gover

nance/govdata2002/index.html (accessed 10 September 2004); Daniel Kaufmann, Aart Kraay and Massimo Mastruzzi,

‘Governance Matters III: Governance Indicators for 1996–2002’, World Bank Institute website, http://www.worldbank.org/

wbi/governance/pubs/govmatters3.html (accessed 10 September 2004). Note that the data for the ‘political stability and

absence of violence’ measure are for 2002. Since this measure does not change greatly from year to year, we feel comfortable

using the 2002 data. Iraq would be an obvious exception since 2003 saw a war and subsequent violent occupation.

82. World Bank, ‘Questions and Answers: Governance Indicators’, World Bank Institute website, http://info.worldbank.org/

governance/kkz2004/q&a.htm (accessed 3 June 2005).

83. All countries that experienced deaths from political violence in 2003 are listed. The 29 countries with the highest levels of

human rights abuses are listed, as are the 29 countries with the worst political instability and violence scores.