Working Paper #10
Refugee-Related
Political Violence:
When? Where? How
Much?
Sarah Kenyon
Lischer
December 2001
Center for International Studies
Massachusetts Institute of Technology
292 Main St. E38-600
Cambridge, MA 02139
slischer@mit.edu
Table of Contents
I. Challenging Popular
Misconceptions 1
II. Measuring
Refugee-Related Violence 3
III. Trends in
Refugee-Related Violence 5
A. Common and Dangerous Types of Refugee-Related Violence 6
B. Frequent Refugee-Related Violence 9
C. Persistent Violence 13
D. Intense Violence 14
E. Africa’s Disproportionate Violence 16
IV. Implications 18
List of Illustrations
Tables
Table 1: Refugee Situations
Involved in Political Violence 11
Table 2: Years of Violence
Reported—Receiving States (Appendix 1) 19
Tables 3, 4 and 5: Intensely
Violent Refugee Situations (Appendix 3) 21-22
Charts
Chart 1: Types of Political
Violence 8
Chart 2: Proportion of
Refugees Involved in Political Violence 10
Chart 3: Refugees Involved
in Political Violence 10
Chart 4: Relatively Few
Violence Affected Receiving States 12
Chart 5: Violence Involving
Afghan and Palestinian Refugees 13
Chart 6: Refugee-Related
Violence (minus Afghans and Palestinians) 13
Chart 7: Persistent and
Intense Refugee-Related Violence 16
Chart 8: Disproportionate
Refugee-Related Violence in Africa 17
Chart 9: Persistently
Violenct Receiving States (Appendix 2) 20
Chart 10: Persistently
Violent Sending States (Appendix 2) 20
Boxes
Box 1: Political Violence
Involving Refugees 6
Abstract
Lack of information about the nature and extent of refugee involvement in political violence has long hindered researchers and policymakers. This paper presents new time series data in order to analyze the frequency, persistence, intensity, and type of political violence involving refugees for the years 1987 to 1998. The analysis reveals a number of interesting, and surprising, trends that contradict the conventional wisdom about refugee militarization. Overall, while absolute numbers of refugees involved in political violence have decreased, the number of states affected remains constant. The difference results from smaller refugee populations becoming involved in political violence. Another significant finding contradicts the assumption that political violence affects most refugee areas. In fact, very few refugee situations experience political violence. In most years, over one hundred states host refugees, yet 95% of all refugee-related violence usually takes place, on average, in fewer than fifteen states. The findings from this dataset reveal trends in refugee-related violence and change the terms of the current discourse on refugees and political violence.
Refugee-Related Political Violence: When? Where? How
Much?
Sarah Kenyon Lischer[1]
I. Challenging Popular Misconceptions
Since the mid-1990s, a few high-profile instances of refugee militarization have encouraged the common assumption that rampant and increasing political violence affects most refugee camps. In discussing the Great Lakes refugee crisis, the United Nations High Commissioner for Refugees affirmed that “we are increasingly confronted, not just in this region but worldwide, with the problem of separating refugees from fighters, criminals, or even genocidaires.”[2] Kofi Annan, the Secretary-General of the United Nations concurred: “The safety of refugees has increasingly become a matter of international concern, as has the security of States hosting large refugee populations or having such populations near their borders.”[3] Like officials of international humanitarian agencies, scholars also assume that most refugees inevitably become involved in political violence. In his survey of international military interventions during the 1990s, William Shawcross claims that “in the eighties [the militarization of camps] had been the exception…In the nineties it became commonplace.”[4]
The dominant view of widespread refugee militarization is reinforced by journalists and scholars who generalize from a few notorious instances of violence: the presence of genocidal militias among the Rwandan Hutu refugees in eastern Zaire; the US backed Afghan guerillas in Pakistan; the attacks on Cambodian refugees along the Thai border; South African bombing raids against refugees and exiles in neighboring states; the massacres of Palestinian refugees in Lebanon. Such violence has led to the supposition of “an apparent growth in the number of armed attacks on refugee camps and settlements.”[5]
The conventional wisdom expressed above is not based on empirical facts but on perceptions. These perceptions are wrong. Until now, there has been no systematic analysis of refugee-related violence that could determine if the phenomenon was rising or falling and how it changed over time. Studies of notorious cases of violence do not answer several essential questions concerning refugee involvement in political violence. Is the phenomenon confined to a few major cases, like Rwanda, Afghanistan, and Cambodia? Or is the problem of refugee-related violence more pervasive? How have the security threats arising from refugee situations changed over time? What regions of the world are most affected? What are the characteristics of sending and receiving states that become involved in conflict due, in part, to refugee flows?[6] Without an understanding of those questions, theories of violence and policies for its remedy cannot advance.
This paper presents new time
series data in order to analyze the frequency, persistence, intensity, and type
of political violence involving refugees for the years 1987 to 1998. The
analysis reveals a number of interesting, and surprising, trends that
contradict the conventional wisdom about refugee militarization. Overall, while
absolute numbers of refugees involved in political violence have decreased, the
number of states affected remains constant. The difference results from smaller
refugee populations becoming involved in political violence.[7]
Another significant finding contradicts the assumption that political violence
affects most refugee areas. In fact, very few refugee situations experience
political violence. In most years, over one hundred states host refugees, yet
95% of all refugee-related violence takes place, on average, in fewer than
fifteen states.
These findings do not negate
the concern expressed by policy makers and scholars about refugee related
violence, but the systematic analysis does redirect the focus of concern. The
main difference between 1980s and the 1990s is that post-Cold War
refugee-related violence was not condoned by a great power (i.e., the United
States or Soviet Union). This disengagement partly results from the increasing
proportion of refugee-related violence in Africa—a region that the United
States, in particular, does not view as a vital national security interest.
Thus, post-Cold War refugee militants usually lack superpower patrons and are
therefore less controllable. These changing patterns of violence over time have
increased threats to regional stability, to aid workers’ safety, and often to
Western security interests.
The following pages explain
the construction of the data set and present the results of the analysis. The
next section defines essential terms and explains the categories of violence
captured by the data set. The middle section of the paper presents the results
for the twelve years (1987 to 1998), comparing among the years and analyzing
overall findings. The paper concludes with the implications of the findings for
understanding the causes of violence and predicting future conflict.
II. Measuring Refugee-Related Violence
This project presents
information on refugee-related violence along four dimensions: frequency,
persistence, intensity, and type. The frequency of refugee-related violence
describes the number of refugees involved in political violence for a given
year, in absolute terms and as a proportion of all refugees. Another measure of
the frequency of violent activity is the number of receiving and sending states
affected by the violence.[8]
Looking at sending and receiving states, in addition to the number of refugees,
balances potentially skewed results due to a large, and very violent, refugee
situation. A “refugee situation” refers to a refugee population from one
sending state in one receiving state for a given year, e.g. Ethiopians in Sudan
during 1989. All refugees in that situation (e.g. all Ethiopians in Sudan
during 1989) are counted as involved in political violence if an incident is
reported for that year. Frequency can also be analyzed in sub-categories to determine the
regions or time periods most affected by violence.
In addition to frequency,
the data shed light on persistence and intensity of violence. Persistence is measured
as refugee situations that repeatedly experience political violence over time.
The most persistent cases are those that have reported violence for more than
half of the twelve years in the dataset. The term intensity refers to the level
of the violence, measured by casualty figures and narrative descriptions of the
violent incidents. This measure gives an idea of the seriousness of the
violence. Unlike, persistence, the intensity of violence presents many
measurement difficulties, as a result of poor reporting of casualty figures and
inherent biases in the data.[9] The dataset includes all incidents of
political violence, regardless of intensity—ranging from a single cross-border
raid on a camp to a full-scale invasion. For each year, however, it is possible
to determine the cases with highest levels of violence by comparing rough
casualty figures and the narrative descriptions of the violence.
In describing the type of
refugee-related violence, the analysis focuses on the five outcomes described
in the next section of the paper. These are: attacks between the sending state
and the refugees, attacks between the receiving state and the refugees,
factional conflict among refugees, internal violence within the receiving
state, interstate war or unilateral intervention. Using the measures of
frequency, persistence, intensity, and type, a picture of refugee-related
violence emerges that alters the conventional wisdom and presents a more
nuanced view of refugee involvement in political violence.
Readers are invited
to reclassify these measures as more data becomes available.
The data for this project
come primarily from three sources that cover the period 1987 to 1998: United
Nations High Commissioner for Refugees (UNHCR) Annual Protection Reports, the
U.S. Committee for Refugees (USCR) World
Refugee Survey, and New York Times
abstracts. UNHCR’s internal Annual Protection Reports, beginning in 1987,
summarize the security and protection problems for each refugee receiving
state. The reports consist of narrative answers to survey questions and include
responses on security incidents, protection problems, and refugee law in the
relevant countries. The UNHCR data are supplemented by the US Committee for
Refugees’ annual publication, World
Refugee Survey, which provides individual reports for each country. The study also analyzed New York Times abstracts from 1987 to 1998 for articles mentioning
refugees and violence.
The resulting data tables
categorize the universe of refugee populations by receiving state and by
national origin of the refugees. This means that, for each year, there exists
an observation for each receiving state and, within the receiving state, a
separate observation for all refugee populations (of over 2,000 persons) by
country of origin (e.g. Guatemalans in Mexico in 1989). Each observation notes (where available) the
ethnicity of the refugees, the primary living situation (e.g. camps, urban,
etc.), and any incidents of political violence. Where possible, the effects of
the violence are quantified with casualty statistics. The type of violence is
categorized according to one of the five outcomes listed in Box 1. The data do
not encompass instances of criminal violence, such as assault, rape, or theft.
They also do not address other forms of threat, for example environmental
degradation caused by refugee camps. Since these data encompass both violent
and non-violent populations, they do not suffer from the selection biases that
occur if one focuses solely on high-profile conflict situations.
Any project that undertakes statistical analysis of
refugees will encounter the well-known problems involved in enumerating refugee
populations.[10] Refugee
experts agree that “all aggregate statistics on refugee flows should be
interpreted with care” due to the difficulty of counting these mobile
populations and the many incentives to distort the numbers for political
reasons.[11] The data for this study used the population
figures provided by USCR in the World
Refugee Survey. Although this volume is considered one of the most reliable
sources of data available, disparities continue to exist in the population
statistics put forward by USCR, UNHCR, and refugee receiving states. Thus, this
dataset cannot escape the more general difficulties that plague refugee statistics.
The methodological problems associated with survey data
also affect the data analysis. This study relies on documents from UNHCR that
have been collected over a twelve year period. One cannot claim that these
Annual Protection Reports are free of all problems of validity and reliability.[12]
Like all survey data, the responses may contain hidden flaws resulting from
human error or institutional biases. The study has attempted to correct for
institutional bias by also analyzing published news reports (New York Times abstracts) and
independently gathered data from the US Committee for Refugees (the World Refugee Survey). This allows for
cross-checking and corroborating the data using a variety of sources.[13]
Despite imperfections in the available data, the results of the analysis
provide information on hitherto unmeasured phenomena—the nature and extent of
refugee involvement in political violence.
III. Trends in Refugee-Related Violence
The data presented here add
new dimensions to the discourse about refugees and security and contradict
current conceptions of refugee-related violence. One important addition to the
discourse on refugee-related violence is the categorization of types of
violence. Most current studies and policies treat refugee-related violence
indiscriminately. Until now, there was no systematic information on which types
of violence occurred more often and with what degree of lethality. The dataset
confirms that violence between the sending state and the refugees occurs more
often than the other types. More importantly, attacks between the sending state
and refugees tend to be more persistent, intense, and more likely to lead to
international war.
In adding new dimensions to
current debates, the findings correct common misperceptions that distort
popular understanding of refugee-related violence. The first misconception is
that the level of refugee-related violence has increased dramatically since the
end of the Cold War. It has not. The number of refugees involved in political
violence has declined by half over the twelve year period, from 8 million to 4
million refugees. This drop is not due to any overall improvement in refugee
security, but mostly reflects the decline in violence among the Afghan and
Palestinian situations.
More significant than any
numeric change after the Cold War is the alteration in the political context of
the violence. The end of the superpower conflict has meant that major donors
did not sanction the violence (for the most part), thus aid workers have faced
increased security risks even though overall levels of violence have not risen.
In particular, a greater proportion of refugee-related violence now occurs in
Africa, which receives much less security assistance from the West than other
regions (e.g. the Balkans).
A second misconception is that violence is rampant in
nearly all refugee situations. It is not. Nearly all refugee-related violence
affects an average of only 15 refugee situations each year. Most refugee camps
do not experience political violence, although other forms of violence and
insecurity may exist. The essential puzzle raised by this finding is how scores
of refugee situations manage to remain relatively peaceful.
A. Common
and Dangerous Types of Refugee-Related Violence
The new data presented here capture information about refugee-related political violence.[14] Political violence, as distinguished from criminal violence, consists of organized violent activity for political goals. Political violence involving refugees manifests itself in five possible types (see Box 1). The first, and most common, outcome is a violent cross-border attack between the sending state and the refugees.[15] Examples include the repeated bombing raids by South Africa against suspected African National Congress (ANC) refugees in Angola and Botswana during the 1980s and the Rwandan Hutu militia raids on Rwanda from their bases in the refugee camps of eastern Zaire.[16] The second type of violence arises due to conflict between the refugees and the receiving state such as the fighting between Palestinian refugees and the Jordanian government, which nearly led to civil war in the early 1970s.
Box 1 Political Violence Involving
Refugees ·
Attacks between the sending state and the refugees ·
Attacks between the receiving state and the refugees ·
Ethnic or factional violence among the refugees ·
Internal violence within the receiving state ·
Interstate war or unilateral intervention
Third, ethnic or factional violence that erupts among refugees can spread conflict to the receiving state. For example, fighting between rival Burundian Hutu groups in the camps in western Tanzania has threatened Tanzania’s security in the 1990s. Fourth, receiving states may fear that the arrival of refugees will spark internal conflict by creating an unstable ethnic balance that encourages a previously oppressed minority to confront the state. During the NATO war in Kosovo, many observers predicted that the presence of thousands of ethnic Albanian refugees in Macedonia could lead to civil war between Slavs and Albanians. The fifth type of violence occurs when refugees become catalysts for interstate war or unilateral intervention. For example, the 1994 United States intervention in Haiti occurred, in part, to prevent the arrival of thousands of refugees on Florida’s shores.[17]
Each type of violence has
its own dynamic; the different types are not necessarily comparable.[18]
Attacks between the sending state and refugees occur most often, closely
followed by attacks between the receiving state and refugees (See Chart 1). In
the later years of the dataset, attacks between refugees and the sending state
increased as a proportion of all violence. Interstate war and unilateral
intervention, although infrequent, also
occurred more often in the last five years of the dataset than the first five
years. The category of “internal violence within the receiving state” either
occurs infrequently or is not easily captured by reports of refugee violence. A
category of “Uncertain/Other” is used for the small number of incidents in
which the reporter could not determine the identity of the attacker.[19]
In most cases, attacks
between the sending state and the refugees entail the most intense violence,
such as bombing and shelling of camps (see Appendix 3).[20]
This finding, while not surprising, was previously unknown due to the dearth of
systematic study of refugee-related violence. Cross-border invasion, either by
the sending state or refugees, is the most extreme form of violence between the
refugees and the sending state, and is most likely to pull the receiving state
into an international war. Unlike attacks involving only the receiving state
and the refugees, attacks involving the sending state present a greater threat
to the sovereignty of the receiving state and may be viewed as a national security
threat by both the sending and receiving state. For example, attacks between
refugees and the sending state escalated into international war in Central
Africa, when Rwanda attacked Zaire and the Hutu refugees under the pretext of
eliminating the security threat posed by the camps. The data show that violence
between the refugees and the sending state usually involves a greater number of
casualties and a more sustained period of conflict than any other type of
political violence except international war.
Violence between refugees
and the receiving state often involves police actions or riots between locals
and refugees. Examples of this include the continuing violence involving
Burmese refugees in Bangladesh. The Bangladeshi police and/or military often
use violence to encourage repatriation. Local villagers sometimes join in
police attacks against refugees, leading to riots and even more severe police
action. Another type of violence occurs when rebel groups in the receiving
state (often supported by the sending state) attack refugees. This has occurred
numerous times in northern Uganda, where Sudanese-funded groups attacked
southern Sudanese refugees. Violence between refugees and the receiving state
often erupts when refugees protest their conditions. For example, Vietnamese
refugees in Hong Kong rioted many times, resulting in scores of deaths, to
protest forced return to Vietnam.[21]
Factional or ethnic violence
among refugees is the third most common phenomenon. This often occurs when
refugee groups include members of different ethnic groups or competing
political parties. The ramifications of factional or ethnic violence include
lawlessness in the refugee camps and endangerment of the staff of humanitarian
aid groups. Factional violence is not likely to engulf the sending and
receiving states. One exception would be cases in which a faction or ethnic
group has supporters within the receiving state. In that case, violence could
lead to a broader civil conflict in the receiving state. In many situations,
factional or ethnic violence does not occur in isolation but accompanies one of
the other manifestations of violence. Afghan refugees in Pakistan, for example,
experienced conflicts with the sending state, the receiving state and among
refugee factions.
International war or
unilateral intervention because of refugees occurs rarely. The most recent
occurrence was the 1996 invasion of Zaire by Rwanda, which combined civil war,
international war, and attacks on refugee camps that killed thousands of Rwandan
Hutu. Other refugee-related wars include the 1979 war between Tanzania and
Uganda and the 1971-72 war between India and Bangladesh (formerly East
Pakistan).[22]
The war involving Zaire and
Rwanda presents a different pattern from that of the earlier refugee-related
wars. In this case, the cross-border attacks between the Hutu exiles and the
Rwandan government escalated into international war when the sending state
invaded the receiving state. In the earlier cases, the receiving state invaded
(or intervened in) the sending state in order to reduce a perceived threat.
Those wars or interventions can be classified as defensive in nature. By
contrast, in eastern Zaire, the sending state (Rwanda) invaded the receiving
state because it perceived an opportunity to eliminate the security threat
posed by the militant exiles. The war in Congo/Zaire should be seen more as an
opportunistic invasion that was designed to take advantage of the collapse of
President Mobutu’s regime.[23]
Arguably, during the Cold
War, patterns of conflict like the opportunistic interventions in the Great
Lakes were contained (or at least controlled) by the interests of the great
powers. In spite of American or Soviet support for militant exiles in
Afghanistan, Thailand, and southern Africa, those conflicts did not escalate
into regional wars. Observers fear that post-Cold War political dynamics could
encourage more conflicts like the Congo war, given the passivity of the Cold
War superpowers, as well as the disengagement of former colonial masters. If
the Congo war in the Great Lakes represents a new trend in refugee-related
violence, then the growth of violence between
refugees and the sending state presents a greater risk of war than it
did in the past.
B.
Frequent Refugee-Related Violence
The results of the data
analysis clearly demonstrate that—despite public rhetoric to the contrary—the
post-Cold War period has not seen a dramatic upsurge in refugee-related
violence. The proportion of refugees involved in violence declined from 60% in
1987 to 32% in 1998, with a sharp drop to 13% in 1997 (See Chart 2). [24]
Viewed in absolute terms, the data also show a decline in involvement in
violence (See Chart 3). The number of refugees involved in political violence
has dropped from nearly 8 million in 1987 to 4.3 million in 1998.[25]
Viewed in isolation, the
drop in refugees affected by violence presents a misleading picture of the
overall trends. Surprisingly, the number of receiving and sending states
involved in refugee-related violence did not decrease, and even experienced a
slight increase on average in the last six years of the dataset (See Table 1).
In the first half of the dataset (1987 to 1992), an average of 16 receiving
states reported refugee-related violence each year, whereas the same statistic
was 19 states in the second half of the dataset (1993 to 1998). The trend for
sending states shows a similar constancy with the number of affected sending
states fluctuating between 10 and 18 states. The average number of sending
states involved per year in the first half of the dataset (1987 to 1992) was 13
states. The same statistic was 15 states between 1993 and 1998. However, since
1995, there has been a decline in the number of sending states affected from 18
to 13. These results indicate that the decline in the number of refugees
involved in violence has not resulted in a significant reduction in the number
of states affected.
Table 1
Refugee
Situations Involved in Political Violence
Year #
Receiving # Sending
States States
Involved Involved
1987 17 15
1988 14 13
1989 21 15
1990 15 12
1991 13 10
1992 14 13
1993 18 14
1994 23 17
1995 19 18
1996 20 16
1997 17 14
1998 15 13
The most surprising fact to
emerge from the analysis is how few states actually experience refugee
related-violence. Most refugee-related violence occurs in only a handful of
states. Over the twelve year period, 11 receiving states, on average, hosted
97% of the refugees involved in political violence each year (see Chart 4).[26]
During that period an average of 82 states each year hosted 2,000 or more
refugees. The findings are similar when analyzed according to sending
states. For the same period, refugee
groups from only 12 sending states accounted for 96% of refugees affected by
political violence.[27]
An average of 40 states each year produced 2,000 refugees or more. These statistics demonstrate that a small
proportion of sending and receiving states account for nearly all of the
refugee-related violence.
>2,000 refugees
hosted
Violent
states >100,000
refugees hosted > 2,000 refugees
hosted All
receiving states
The
relative constancy of the number of states affected by refugee-related violence
seems to clash with the dramatic reduction in the proportion of all refugees
involved in political violence. By looking more closely at the data, one finds
that the precipitous decline in the total number of violence-affected refugees
derives, in large part, from the reduction of hostilities between the Soviet
Union and Afghanistan and in the Arab-Israeli conflict (see Chart 5). At their height, in 1987, Afghan and
Palestinian refugees together comprised nearly 8 million refugees (over 60% of
all refugees for that year). During the late 1980s and early 1990s, intense
violence involving those refugees led to much higher statistics on refugee
militarization. Removing those two situations shows
a different pattern of violence, which peaked between 1994 and 1997 (see Chart
6).
C.
Persistent Violence
Political violence involving
refugees is not endemic, except in a handful of receiving and sending states.
Although usually fewer than 20 receiving states report refugee-related violence
each year, a total of 55 receiving states have reported such violence
throughout the time span of the dataset. A few of those states report violence
nearly every year, but the vast majority of receiving states report
refugee-related violence for only one to four years (see Table 2, Appendix 1).
Only 8 of the 55 receiving states reported political violence for more than
eight of the twelve years studied.[28]
Over the twelve year period,
41 sending states have produced refugees affected by political violence,
although for each year the number is usually fewer than 15 states. Like the
receiving states, most sending states report intermittent violence. Of the 41
states, over half were affected by violence for only one to four of the years
under study. Only nine of the 41 sending states were involved in
refugee-related violence for more than six of the twelve years.[29]
There are a number of
explanations for this low level of persistence. In some cases, a violent refugee
situation became less violent over time, such as in the case of the Palestinian
refugees in Jordan. In other cases, low measures of persistence occurred
because the refugee situations existed for fewer years. Examples of that
phenomenon include the Liberian refugees in Ivory Coast, who suffered
persistent violence, but who stayed in Ivory Coast for less than six years.
The patterns of persistent
violence confirm other findings about the nature of refugee-related violence.
The statistics on frequency showed that a small number of states (about 20% of
the total) account for nearly all refugee-related violence. The data on
persistence confirms that most states do not experience continuous violence
over the years. Although 55 receiving states reported refugee-related violence
for varying numbers of years, 100 receiving states reported no violence at all
during the twelve year period. In persistence, as well as frequency, African
states are over-represented; six of the ten most repeatedly violent receiving
states are found in Africa, as well as six of the nine sending states. The
Palestinian and Afghan refugee situations also figure prominently in the
persistently affected list, especially during the late 1980s and early 1990s.[30]
D. Intense
Violence
In addition to the persistence of violence, it is useful to know which refugee situations experienced high level, or intense, violence. Some groups are not repeatedly involved in violence, yet when violence occurs it is extremely bloody. Other groups experience both persistent and intense violence. Still other groups appear more violent, due to ongoing conflict, yet closer examination reveals a relatively low level of intensity in which few deaths occur. Using this data set, it is possible to highlight the situations with the most intense violence (See Tables 3-5, Appendix 3).
Assessing statistics on intensity of violence is somewhat subjective, due to a lack of comparable casualty statistics. One can measure intense violence by the number of incidents within a year and the level of casualties (when given). The analysis here treats casualty figures relative to the population, not just as absolute numbers. Rough categorizations of intense violence and illustrative examples appear in Appendix 3.
Fifteen refugee situations
have experienced both persistent and intense violence over the twelve year
period (see Chart 7).[31] With the exception of
Sudanese refugees in Uganda (who continually experience high levels of
violence), most of the refugee situations show variation over time in levels of
violence. The seven refugee groups most
affected by political violence over the twelve years are: Palestinians,
Rwandans, Afghans,
Sudanese, Liberians,
Burmese, and Sierra Leoneans. In many cases, populations from these countries
have experienced violence in multiple receiving states. An additional level of
variation—that of non-violent situations—is not included in Chart 7. For
example, Rwandan refugees (both Hutu and Tutsi) living in Tanzania do not
appear on this chart because their levels of violence have been much lower.
While the chart suggests that Rwandans have a high propensity for involvement
in violence (in Zaire, Burundi, and Uganda), it helps to understand the causes
of that violence, and possible solutions, by including study of the relatively
non-violent situation in Tanzania.
Chart 7:
Persistent and Intense Refugee-Related Violence
1987-1990 1991-1994 1995-1998
Palestinians
|
|
in Gaza/W. Bank
in Lebanon
Rwandans
in Zaire
KEY: High violence Low violence None reported
in
Burundi
in Uganda
Afghans
|
|
|
|
|
|
in Pakistan
Sudanese
In Uganda
In Ethiopia
Liberians
in Guinea
in Ivory Coast
in Sierra Leone
|
|
Burmese
|
|
in Thailand
in Bangladesh
Sierra Leoneans
|
|
in Guinea
in Liberia
E.
Africa’s Disproportionate Violence
When examining the above
chart (Chart 7), it is immediately noticeable that ten of the fifteen most
persistent and intense situations occurred in Africa. A closer look at the
regional distribution of violence reveals an increase in the proportion of
African states reporting refugee-related violence between 1987 and 1998. In the
first 8 years of the dataset (1987 to 1994), the proportion of receiving states
in Africa (36%) nearly matched the proportion of violence-affected receiving
states in Africa (40%). From 1994 to 1998, however, the percent of affected
receiving states increased to 53% whereas the percent of all receiving states
in Africa dropped to 31%. This has led to a disproportionate number of African
receiving states that report refugee-related violence.
One might hypothesize that Africa’s disproportionate
violence arises because of the relatively greater number of refugees on the
continent; if Africa has more refugees than other regions, one would expect
more refugee-related violence in Africa.
Surprisingly, however, as Africa’s proportion of the world’s refugees
has decreased over time, Africa’s proportion of the world’s refugee-related
violence has increased. The distribution of refugees and violence became
markedly skewed between 1995 and 1998 (See Chart 8). In the last four years of
the dataset, the proportion of African refugees in the world dropped to 26%
whereas the proportion of
refugee-related violence involving African refugee populations rose to
59% for the same period.
Violence affected African refugees as a % of all
violence affected refugees
African refugees as a % of all refugees
The disproportionate
violence in Africa raises puzzles for policy makers and scholars. The
phenomenon indicates that perhaps something about African refugee situations
differs from other regions and makes African situations more violence-prone. An
alternate explanation is that the increase in African violence merely reflects
the decrease in the Afghan and Palestinian refugee-related violence. A third
explanation argues that refugee-related violence occurs because Africa’s
refugee crises receive far less humanitarian aid, per capita, and less
diplomatic and military attention than many European or Asian crises. Further
papers in this project will examine the effects of humanitarian assistance and
international political and/or military intervention on refugee-related
violence. The results of this analysis will shed light on the trend of
increasing violence in African refugee situations.
IV. Implications
The new data presented here
change the terms of the discourse on refugees and political violence. Policy
makers and scholars have no excuse to continue speaking in vague terms about
the rising threat of armed refugees or the ubiquity of attacks against
refugees. The scope of refugee-related violence is much clearer now. About 30%
of refugee groups became involved in political violence, as of 1998. This represents
a marked decrease since 1987, not an increase. Roughly 15 receiving states
account for nearly all of the refugee-related violence that occurs in a given
year. The vast majority of refugee sending and receiving states do not become
involved in political violence. African states experienced a disproportionate
level of refugee-related violence in the later years of the dataset. The most
common, and lethal, type of violence is attacks between the sending state and
refugees.
The findings from this dataset
reveal current trends in refugee-related violence, but cannot predict the
future. For now, a number of potential scenarios offer a mixture of hope and
caution. One possibility is that the proportion of refugees involved in
political violence will continue to decrease as the great powers lose interest
in arming various exile groups. The reduction of great power support for
militant refugees, combined with a new trend toward international humanitarian
intervention, could vastly decrease refugee-related violence.
However, a paradoxical
result of superpower disengagement is that neglect could lead to more
situations like eastern Zaire, in which militant refugees engage in military
activity unhindered while reaping the benefits of international humanitarian assistance.
When it occurs in areas that have little strategic significance to the major
powers, wealthy states will expend few resources to prevent refugee-related
violence. In the mid-1990’s, the United Nations Secretary General approached 40
member states for help in demilitarizing Rwandan refugees in eastern Zaire.
Only one state volunteered troops.
It is also possible that the
lull in great power support for refugee groups constitutes a temporary
phenomenon. Considering past trends, in which a few large refugee groups (such
as Afghans, Palestinians, Rwandans, and Cambodians) experienced persistent and
intense violence, one could expect the emergence of similar groups in coming
years, for example in conflict-ridden West Africa or the Balkans. In addition,
recently dormant situations, such as the Palestinian and Afghan crises, are not
fading away but appear to be re-igniting. Since fall 2000, the Palestinian
refugee situation has once again contributed to an alarming increase in
refugee-related violence. Since September 2001, the Afghan situation in
Pakistan also seems increasingly unstable. The arrival of new refugees from the
American war and the radicalization of the existing two million refugees could
spark renewed violence.
Further analysis is needed
to more accurately predict and prevent refugee involvement in political
violence. However, whatever the future trends, the new information described in
this paper improves understanding of refugee-related violence, and serves as a
building block for further research on the spread of civil war.
Appendix 1
Table 2
Years of Violence Reported—Receiving States
9-12
years 5-8
years 1-4 years
|
1.
Ethiopia |
1.
Angola |
1.
Afghanistan |
|
2.
Gaza/West Bank |
2. Bangladesh
|
2.
Armenia |
|
3.
Kenya |
3.
Guinea |
3.
Benin |
|
4.
Lebanon |
4.
Hong Kong |
4.
Botswana |
|
5.
Pakistan |
5.
Iran |
5.
Burundi |
|
6.
Sudan |
6.
Iraq |
6.
Cent. African Rep. |
|
7.
Thailand |
7.
Liberia |
7.
Croatia |
|
8.
Uganda |
8.
Rwanda |
8.
Djibouti |
|
|
9.
Tanzania |
9.
Dominican Rep. |
|
|
10.
Zaire/DRC |
10.
Germany |
|
|
11.
Zambia |
11.
Guatemala |
|
|
|
12.
Guinea Bissau |
|
|
|
13.
Honduras |
|
|
|
14.
India |
|
|
|
15.
Ivory Coast |
|
|
|
16.
Jordan |
|
|
|
17.
Lesotho |
|
|
|
18.
Malawi |
|
|
|
19.
Malaysia |
|
|
|
20.
Mauritania |
|
|
|
21.
Mexico |
|
|
|
22.
Nepal |
|
|
|
23.
Papua New Guinea |
|
|
|
24.
Saudi Arabia |
|
|
|
25.
Senegal |
|
|
|
26.
Sierra Leone |
|
|
|
27.
Somalia |
|
|
|
28.
Swaziland |
|
|
|
29.
Sweden |
|
|
|
30.
Switzerland |
|
|
|
31.
Syria |
|
|
|
32.
Turkey |
|
|
|
33.
USA |
|
|
|
34.
Yemen |
|
|
|
35.
Yugoslavia |
|
|
|
36.
Zimbabwe |
Appendix 2
KEY: Years reporting violence
Chart 9: Persistently Violent Receiving States
87 88
89 90 91 92 93
94 95 96 97 98
Lebanon
|
|
Kenya
Ethiopia
Thailand
Uganda
Gaza/WB
Pakistan
Angola
Sudan
Guinea
Chart 10: Persistently Violent Sending States
87 88 89 90 91 92 93
94 95 96 97 98
Sudanese
Afghans
Somalis
Palestinians
Rwandans
Ethiopians
Burundians
Burmese
Liberians
Appendix 3
Tables 3, 4 and 5: Intensely Violent Refugee Situations
1987 to 1990
|
Refugee Group |
Receiving State |
Political Violence |
|
Palestinians |
Gaza/ West Bank |
Intifada; attacks between refugees and Israeli forces |
|
Lebanon |
Israel
bombs camps; factional violence between PLO and rivals; attacks between
refugees and Lebanese forces. |
|
|
Afghans |
Pakistan |
Afghan/Soviet
forces shell camps; cross-border attacks by Afghan mujahedin based in camps;
factional fighting among refugee/rebel groups. |
|
Sudanese |
Uganda |
Sudanese
rebels in camps. Refugee camp bombed.
Cross border attacks between refugees and Sudan. |
|
Ethiopia |
Cross
border attacks by SPLA on refugees. Sudanese rebels forcibly recruit
refugees. Ethnic riots by locals and refugees near camps. |
|
|
Rwandans |
Uganda |
Refugees
form an army of 7,000 and invade Rwanda. |
|
Mozambicans |
Zambia |
Cross-border
raids by RENAMO and counter-attacks by Zambian forces. |
|
Zimbabwe |
RENAMO
incursions against refugees and locals. Zimbabwe forces retaliate against
refugees. |
1991 to 1994
|
Refugee Group |
Receiving State |
Political
Violence
|
|
Rwandans |
Burundi |
Burundian
Hutu attack Tutsi refugees after assassination of Burundi’s president; Tutsi
refugees attack new Hutu refugees; RPF crosses border to attack Hutu
refugees. |
|
Zaire |
Ex-army
and militias control camps and conduct cross border attacks on Rwanda;
conflict between refugees and Zairean forces; factional fighting among
refugees. |
|
|
Uganda |
Refugee
army continues invasion of Rwanda; |
|
|
Palestinians |
Gaza/ West Bank |
Intifada; attacks
between Palestinians and Israeli forces; factional fighting among
Palestinians. |
|
Lebanon |
Camps
under siege by Lebanese forces; air raids on camps by Israeli forces;
factional fighting among Palestinian militias in camps. |
|
|
Liberians |
Sierra Leone |
Liberian
rebels cross border and attack refugees and locals; local retaliation against
refugees |
|
Ivory Coast |
Liberian
NPFL rebels attack refugees and locals; militias in camps recruit refugees to
fight in Liberia. |
|
|
Guinea |
ULIMO
rebels attack refugees and locals; refugees recruited to join ULIMO and
attack Liberia. |
|
|
Sudanese |
Uganda |
Sudanese
rebels attack camps |
1995 to 1998
|
Refugee Group |
Receiving State |
Political Violence |
|
Rwandans |
Zaire |
50,000
former military/militia in camps; military training in camps; cross border
attacks on Rwanda; RPF cross border attacks against camps; RPF and Zairean
rebels bomb camps; refugees attack Zairean Tutsi; Zaire arms refugees to
fight rebels |
|
Sierra
Leoneans |
Guinea |
Sierra
Leone rebels attack camps and local villages. |
|
Liberia |
Sierra
Leone government shells refugee settlement |
|
|
Liberians |
Ivory Coast |
Liberians
attack across border; Ivoirians attack refugees in revenge |
|
Guinea |
Liberian
rebels attack refugees and locals; reprisal attacks on refugees by Guineans |
|
|
Burmese |
Thailand |
Burmese
government and dissident rebels attack refugees in dozens of incursions;
shelling of camps. |
|
Bangladesh |
Violent
clashes between police and militant refugees; factional fighting within camp. |
|
|
Sudanese |
Uganda |
Anti-government
Ugandan rebels massacre refugees; dozens of rebel attacks on camps; Sudanese
rebels active in camps. |