Midwest Political Science Association Wiley

Midwest Political Science Association
Wiley
The Externalities of Civil Strife: Refugees as a Source of International Conflict
Author(s): Idean Salehyan
Source: American Journal of Political Science, Vol. 52, No. 4 (Oct., 2008), pp. 787-801
Published by: Midwest Political Science Association
Stable URL: http://www.jstor.org/stable/25193850
Accessed: 24-10-2015 19:51 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Midwest Political Science Association and Wiley are collaborating with JSTOR to digitize, preserve and extend access to
American Journal of Political Science.
http://www.jstor.org
This content downloaded from 200.17.203.24 on Sat, 24 Oct 2015 19:51:45 UTC
All use subject to JSTOR Terms and Conditions
The Externalities
of Civil
of International
Strife: Refugees
as a Source
Conflict
Universityof NorthTexas
Idean Salehyan
Domestic strife and civil war frequently produce largepopulation dislocations and refugeeflows across national boundaries.
Mass
refugee flows
civil violence
entail
often
consequences
negative
for
across national
boundaries
extends
frequently
in developing
states, particularly
receiving
as uinternaVy
are not constrained
conflicts
countries.
by borders.
Moreover,
This
article
argues that refugeeflows between states significantly increase the likelihood ofmilitarized interstate disputes (MIDs) in that
as
states are more
to prevent further
and refugee
externalities,
likely to initiate MIDs
they intervene
Refugee-receiving
as
states initiate MIDs
borders
in pursuit
this research challenges
conventional
Moreover,
they violate
of dissidents.
on distributional
theories
states. These
are
that focus
between
exclusively
of international
conflict
bargains
propositions
dyad.
sending
tested in a quantitative analysis of the relationship between refugees andMID

refugees significantly increase theprobability of international conflict.
of civil unrest
Periods
is a growing body of literature on the secu

in general and refugee mi
rity implications of migration
in
2006; Lischer
(see, e.g., Adamson
gration
particular
and persecution
frequently
dislocations and mass refugee
generate population
across national
migration
There
Civil wars
boundaries.
and Iraq, to name

for states
repercussions
in Afghanistan,
the Balkans,
Sudan,
a few examples, have significant
in the region as people leave their homes
2005; Rudolph
1992). Recent
in search of
tic" politics?may
also have far-reaching effects as mi
flee
one-sided
violence by their government. These
grants
can
flows
be
refugee
quite large: conflicts in Afghanistan,
strong
In addition,
by
others
of war, but may serve as catalysts for conflict,

including conflict between states. Some have suggested
can provoke international hostili
that refugee migration
ties between states, including military action (Dowty and
"internal"
tional boundaries
violence
Loescher
1996; Posen 1996), although this claim has not

rigorously tested. Case studies of particular refugee
crises provide anecdotal evidence that international mil
been
system and may provoke
when
are
costs
external
as conflicts between governments

and
are frequently not contained by bor
The
author
is assistant
thanks
Peter Andreas
International
Kristian
American
e2008,
Relations,
Skrede Gleditsch
Journal
Midwest
professor
and
Political
and Derrick
the Evolution
and Tetsuya
of Political
of political
Science,
Science
science,
Frazier
of World
Matsubayashi
Vol.
Association
52, No.
University
as well
often
preceded
by
mass
refugee
mi
further refugee migration

(Newland 1995). Sim
a
ilarly, frequently heard justification for NATO military
in Kosovo was to prevent further mass migra
operations
tion out of the Balkans. Refugee-sending
countries may
prevent
state conflict.
Salehyan
are
gration. For instance, it has been argued that a major

reason for the U.S. invasion of Haiti that ousted the mili
tary junta and reinstated President Aristide in 1994 was to
na
refugees/dissidents
ders. This article argues that negative externalities
and
cross-border
violence
often provoke
tensions between
can lead to militarized
inter
states?refugee
migration
Idean
endeavors
itary
high.
across
extend
may
2006; Weiner
the intuition
products
and Rwanda have generated well over 1

Mozambique,
million
refugees each. Therefore, political turmoil at the
domestic
level can have significant spillover effects for
reactions
2003; Salehyan and Gleditsch

quantitative work confirms
that political violence and persecution

are significant de
terminants of flight (see, e.g., Davenport, Moore, and Poe
and Oberg 2006; Moore
and Shellman
2003; Melander
2004). Yet, refugees are not simply the unfortunate
by
safety elsewhere. Government

persecution of political op
human
and ethnic cleansing
ponents,
rights violations,
as
matters of "domes
understood
campaigns?normally
other states in the international
initiation, 1955-2000. Results confirm that
of North
Texas,
as
Christopher
Rudolph
Princeton
University,
Politics,"
for help
producing
4, October
2008,
Figure
Pp.
P.O. Box
305340,
Dent?n,
and
TX
76203
(idean@unt.edu).
the participants
of the conference
on
on earlier drafts. The author
for comments
"Migration,
also thanks
2.
787-801
ISSN 0092-5853
787
788
also launch military
pursuit of refugees
territories in
attacks on neighboring
and rebels among them. The Israeli
in 1982 and the Rwandan
invasion of Lebanon
of Zaire
in 1996 were
clear refugee camps

the link between refugee
has been under
invasion
by the desire to
largely motivated
that harbored militant
factions. Yet,
flows and international
theorized
and has not been

Is there amore
conflict
subjected to
general rela
systematic empirical testing.

tionship between refugee flows and international
conflict?
Despite a large body of work on the causes of war and

international disputes (e.g., Bennett and Stam 2004; Bre
mer 1992; Bueno de Mesquita
et al. 1999; Fearon 1995;
states
Russett and Oneal 2001), refugee flows between
hardly receive mention
many examples where
interventions
in the literature although there are

armed
refugee crises precipitated
and hostilities
between
scant attention
ally, there has been

political unrest within
states. More gener

to how civil wars and
states may spark disputes between

see
Gleditsch,
(but
Salehyan, and Schultz 2008).
on factors that constrain
have
focused
scholars
Typically,
states
between particular dyads such as distance,

power ratios, political liberalism, and economic ties (e.g.,
Bremer 1992; Gartzke 2007; Morrow
1999; Russett and
Oneal 2001). Others have looked at issues that may mo
the use offeree
states, particularly the dis

of territory and resources (Hensel 2001; Hensel,
and Sowers 2006). The externalities of civil war
Mitchell,
in discussions of international dis
have been overlooked
tivate armed conflict between
tribution
putes, although many have found that internal conflicts

for neighboring
have significant negative consequences
and
states (Ghobarah, Huth, and Russett 2003; Murdoch
The
burdens
and
Gleditsch
Sandier 2004; Salehyan
2006).
of refugee inflows and the diffusion of violence across bor
international
ders point to additional factors motivating
conflicts
and are specific
triggers
that lead to interstate
hostilities.
than treating civil and international war as
separate fields of inquiry, recent research blurs the bound
intra- and interstate violence; analyses of
aries between
Rather
civil war have begun

ables and vice versa
to incorporate
international
vari
and Enterline
(e.g., Balch-Lindsay
and James 1995; Gleditsch
2007; Re
2000; Carment
gan 2000; Saideman 2001; Trumbore 2003; Walter 2002;
(2003), Saide
2004). For instance, Trumbore
the persecu
that
Woodwell
and
(2004) argue
(2001 ),
can
to
international
tion of domestic minorities
give rise
conflicts when countries intervene to protect their ethnic
Woodwell
man
kin. Beyond kinship ties, Gleditsch,

Schultz, and Sale
and
civil
war?ethnic
of
all
forms
find
that
hyan (2008)
a
milita
of
raise the probability
nonethnic?significantly
rized interstate dispute. Refugee flows and other conflict
externalities may be one explanation for this link. Finally,
IDEANSALEHYAN
rebel organizations
that have sanctuaries in neighboring
states may be especially likely to foster international con
flict (Salehyan 2008).
to these studies, this article argues that
Adding
of civil
refugee flows are more than just a consequence
to
international disputes
unrest, but may also contribute
between sending and receiving countries. They may do so
in a pair of related ways. First, refugees constitute a nega
tive externality borne by receiving states, and these states
may launch military actions to seal their borders, threaten
sending regimes with violence, and even invade the send
ing state to prevent further flows. Second, refugee-sending
states may violate the sovereign territory of their neigh
bors in order to attack political and/or ethnic rivals that
have fled across the border as well as punish the states that
harbor them. Such conflicts are more difficult to resolve
to end refugee
because promises
bargaining
and attacks against
human rights violations
producing
dissidents are not credible. These claims are tested in a
through
analysis linking refugees to the initiation of

interstate disputes
(MIDs), which covers the
Even when controlling
for the pres
years 1955-2000.
ence of civil war within states in a dyad and other factors
quantitative
militarized
to be associated with MIDs, refugees are a signif

icant predictor of conflict. In addition, this study reveals
that refugees increase the probability of dispute initiation
known
by both sending and receiving countries. Finally, stronger

states are more likely to initiate conflicts, both when they
send and when they receive refugees.
Refugees
and International
Despite clear humanitarian

not merely humanitarian
Conflict
dimensions,
refugee crises are
disasters but have real secu
for states. I argue

rity implications
tion increases both (1) the likelihood
that refugee migra

that receiving states
forces against the sender, and (2) the likeli

that sending states will initiate disputes against the
host. Mass, disorderly migration
poses significant bur
conflicts
dens on receiving states and refugee-generating
will mobilize
hood
frequently extend across territorial boundaries.

it is useful to note the scale and
Before proceeding,
extent of global refugee migration.
Figure 1 displays the
of
in
the
number
trend
refugees, 1961-2002. The
global
rose steadily during the
number of refugees worldwide
the rise in the number of armed
Cold War?reflecting
during that period (see Gleditsch et al. 2002)?
in the early 1990s. The
and peaked at roughly 18million
number
of refugees have been
the
and
number of conflicts
as the superpowers
declining since the end of the Cold War
conflicts
OF CIVILSTRIFE
EXTERNALITIES
1 Number
Figure
789
of Refugees Worldwide,
1961-2002
20000
18000
16000
14000
12000
(0
$
o>
?g
10000
8000
6000
4000 H
2000
,# ^
N#N<p?> /
Source:
United
Nations
High
<$>
?> /
Commissioner
?? N*NN?jfN#/
N#/
N# N#NN# N#/
for Refugees.
were no longer funding many insurgencies and were more

operations. At its peak,
willing to approve peacekeeping
of the
the number of refugees exceeded the population
not be entirely borne by the country itself, but may create

significant externalities, or spillover effects, for others.
Refugees may contribute to several types of negative
and if "refugees" were a single country, it

Netherlands,
state in the world. Yet,
the
would be
59th most populous
across states. Some
refugees are not evenly distributed
nomic
and Jordan hosted over

were also
1million
refugees each; refugee populations
some
in
countries
such
the
least
of
quite large
developed
countries
as Tanzania,
such as Iran, Pakistan,
Zaire,
Refugees
and Malawi.
as a Negative
Externality
The terms "internal," "domestic," and "civil" conflict im

ply that violent interactions between states and dissidents,
and the negative consequences
of war and human rights
are
the state. Rather than
contained within
violations,
treating states as independent units, itmay be more ap
in dense net
propriate to view them as interconnected
works
of social interactions
state have significant

region and elsewhere
where
repercussions
processes within one

for other states in the
(Balch-Lindsay and Enterline 2000;

Gleditsch
and Garrett 2006).
2002a; Simmons, Dobbin,
The human costs of civil war and state persecution may
can inflict an eco

First, refugee migration
on
some inter
countries.
burden
host
Although
externalities.1
national
aid may be available, refugee hosts?including

some of the world's poorest nations?often
bear much
of
the cost of maintaining

refugees. Refugees require human
itarian assistance and public services supplied by their
hosts and may compete with locals for jobs and scarce re
sources, bringing them into conflict with domestic actors
to voluntary
labor migrants,
(Martin 2005). Compared
are
to
to productive
also
less
contribute
refugees
likely
economic
in
their
hosts
(Cortes 2004). As op
activity
to
economic migrants,
posed
refugees are not selected for
em
their skills, may have suffered war trauma making
ployment difficult, and may have lost their assets prior
to flight. In a related study, Murdoch
and Sandier (2004)
find that civil wars in one country have a significant neg
ative impact on GDP
1
Refugees
may not
to their
positively
cial integration
However,
immediate
and
growth
in neighboring
a negative
burden
always be
in the long
societies
host
effective
policies
short-term
they often have

care and may have
and can contribute

run.
can
impacts.
so
Successful
yield positive
costs associated
distributional
states; they
outcomes.
with
their
790
IDEANSALEHYAN
as a potential causal factor,

propose population migration
but do not test this explicitly.
Second, refugees may entail negative public health
for their host countries. Refugee camps are
consequences
can seriously jeopardize interstate

relations. Receiving countries may confront the sender to
compel them to end civil wars and human rights viola
often
similarities
crowded
and unsanitary, creating conditions

ripe
for infectious disease; refugee inflows stretch domestic
resources thin; they divert health resources away
medical
from normal
needs
care; and refugees may have specific health

the host. A
related to war trauma that overwhelm
body of empirical research finds

significantly increase the rate of morbidity
growing
that civil wars
and mortality
as
in
country
neighboring
states (Ghobarah, Huth, and Russett 2003). Studies have
to the spread
shown that refugee camps have contributed
itself as well
in the affected
such as HIV/AIDS,
malaria, cholera, and diar

other
infectious
diseases
rhea, among
(Collier et al. 2003;
Rowland and Nosten 2001; Toole andWaldman
1997).
Third, refugees may upset the ethnic balance in their
of diseases
a
through what may be thought of as "de
and
immigrants maybe
externality. Refugees
mographic"
seen as unwelcome
foreigners and a cultural "threat" to
host countries
the host community

(Weiner 1978, 1992). Ethnic ten
sions may become especially pronounced when refugees
possess ethnic ties with groups already present in the host
en
society. In countries where ethnic cleavages are deeply
inflows may tilt the
trenched, large, unexpected migrant
in the host society and spark in
As
Brown
conflict.
writes, "The sudden influx of
tergroup
can
aggravate ethnic problems and further com
refugees
ethnic balance
delicate
the picture by changing the domestic balance of

(1998, 25) point
power" (1996, 576). Lake and Rothchild
to similar dynamics in their discussion of the diffusion of
plicate
ethnic
conflict
across
states.
These
spillovers
tions that produce refugee burdens. Such bargains share

in the eco
with conflicts over externalities
realm (see Coase 1960). As the "Coase Theorem"

suggests, those that suffer a negative externality will be
nomic
at
to pay some cost to prevent
its production
willing
the source. Thus, negative externalities become part of
the bargain between states. Receiving countries can issue
a
to the sending state, demanding
threats and warnings
change in that state's domestic behavior. They may also
close border crossings and move troops near the inter
and force de
national boundary?military
operations
In
ployment near the border may be seen as provocative.
some cases, they may become willing
to prevent nega
invasions of the source
tive externalities through military
state.
not be a forgone conclusion.

(see, e.g., Fearon
perspective
bargaining-and-war
Powell
Slantchev
2003; Wagner
1995;
1999;
2000) em
phasizes that war results when parties cannot agree on
The use of force may
The
an acceptable settlement. This framework was developed

in the context of dyadic disputes over the distribution
of some good such as territory. Conflicts about refugee
flows resulting from civil wars and human rights vio
lations are different
the political
tion are a function
because
from bargains
over divisible
goods
processes generating forced migra
of internal unrest and introduce a
third actor, the domestic opposition

(see Schultz 2007 for
a related discussion).
framework
Yet, as the bargaining
states
to
reach
be
when
difficult
suggests, agreements may
cannot credibly commit to a deal. The refugee-receiving
secu
Finally, refugee flows may directly affect the
to
rity and stability of the host country by contributing
on the host's territory. Sale
armed
conflict
organized
(2006) find that refugee inflows from
hyan and Gleditsch
country may demand that the sending state cease attacks

the
on rebels and dissidents?and
perhaps accommodate
across
stem
to
of
the
flow
order
refugees
opposition?in
to do so are difficult to
the border. Credible promises
increase the risk of civil

states significantly
neighboring
war. Along with the refugees themselves, foreign fight
ers, arms, and ideologies that contribute to violence may
secure because
refraining from such repressive policies

the
threatens
very survival of the regime and the benefits
of rule. The sending state may view the risk of external
stream across the border. "Refugee warrior" communities

(see Lischer 2005; Salehyan 2007; Zolberg, Suhrke, and
1989) can expand rebel networks to encompass
Aguayo
establish bases on external
the host state when militants
it hopes to
to be acceptable given the benefits
crackdowns against domestic op
achieve by continuing
ponents; or, itmay simply be unable to contain domes
tic violence and control rebel actors. In addition, Jenne
(2007) and Thyne (2006) argue that external signals in
support of rebel aims and threats of intervention often
they form social ties with domestic

groups of a similar ethnicity or political ori
opposition
entation. Therefore, at the extreme, refugee inflows may
lead to violent turmoil on the host country's territory. For
territory
and when
instance, Jordan was involved in a bloody armed conflict

to expel Palestine Liberation Or
in 1970 when itmoved
on its
ganization fighters operating within refugee camps
soil.
attacks
the oppo
by emboldening
more
extreme
demands
sition and leading them to make
(see also Crawford and Kuperman 2006).
invasions in Haiti and Kosovo were partly
Military
motivated by the desire to stem further refugee migration.
in
In a televised address shortly before the U.S. operation
exacerbate
domestic
conflict
EXTERNALITIES
OF CIVILSTRIFE 791
in 1994, President Clinton remarked: "Today, more
than 14,000 refugees are living at our naval base in Guan
Haiti
tanamo. The American
people have already expended al

to
million
$200
support them... We will continue
a
to face the threat of mass exodus of refugees and its
constant threat to stability in our region and control of
most
our borders."2
In these cases, the refugee-sending

country
could not commit to ending repressive policies because
leads
doing so would threaten the regime. This discussion
to a first hypothesis:
for
caught up in cross-border attacks. No state welcomes
eign incursions and military violations of its sovereignty,
even though the physical ability to firmly control bor
ders maybe
1987).
lacking (Andreas 2003; Herbst
1989; Jackson
Refugees may not be passive actors, however. While

in
the vast majority of refugees never directly participate
rather
victims
than
of
unfortunate
war,
violence,
being
some have mobilized
into insurgent organizations
and
refugee
camps
serve
often
a double
as
purpose
sanctu
aries for militant

HI:
between a dyad will increase the

Refugee migration
the
that
probability
receiving country will initiate
a militarized
interstate dispute (MID) against the
country.
sending
groups (Lischer 2005; Salehyan 2007;

Stedman and Tanner 2003; Zolberg, Suhrke, and Aguayo
into a rebel organization may provide
1989). Recruitment
refugees a better alternative to life in a camp and provide
individuals a sense of purpose. These "refugee warrior"
are facilitated by hosts that are too weak
communities
to prevent
Refugees
and the Externalization
of Conflict
All forms of international
the fre
underscore
migration
of the state
between the boundaries
quent incongruence
and its citizenry (Fox 2005; Guarnizo, Portes, and Haller
and
2003; Shain 1989; Shain and Barth 2003; Waldinger
Fitzgerald 2004). States are formally sovereign over a given
territory and claim to govern the affairs of their citizens;
yet, migration
implies that not all citizens of the state
are physically present within
its territorial jurisdiction.
The state's laws and regulatory capabilities do not reach
citizens
its geographic
domain. But when mi
beyond
to the homeland
the
leave
attachments
and
state,
grants
activism in home-country
at
end
the
bor
politics rarely
der, and may persist for generations
(Collier and Hoeffler
2004; Lyon and Ucarer 2001; Shain and Barth 2003; Shef
fer 2003; Tatla 1999). Therefore, interactions between the
state and the "polity" often extend beyond formal/legal
state boundaries and give rise to transnational social pro
cesses
involving
migrant
such flows
activities
threaten
across na
tional boundaries.
flee civil war and state per

Migrants
secution for safety in countries of asylum, and most of
the world's refugees flee to areas in close proximity
to
their country of origin. However, particularly when bor
ders are porous, they are not necessarily
immune from
continued attacks by the state. Governments may pursue
their ethnic and political rivals across the border, and such
attacks necessarily violate the sovereignty and territorial
incursions
integrity of others. States will object tomilitary
on their soil,
if
local
also
become
particularly
populations
2"In the Words

Invade Haiti."
of
New
the President:
York Times.
The
Reasons
September
Why
16, 1994.
the US May
of camps or by hosts that

activities as a way to under
actively encourage dissident

mine
their international
rivals. Thus, violence
the country of origin and refugee communities
between
may not
civil
takes the form of transnational
be one-sided,
but
wars that span international boundaries.
Sending states
on
their
launch
attacks
may
neighbor's territory in order
to strike at external rebel bases and to clear refugee camps
close to the border; these attacks may also jeopardize the
in the neighbor.
safety of local populations
More directly, the sending state may accuse the host
country
Accepting
of harboring
dissidents and political opponents.

act, but
refugees ismore than a humanitarian
the sender in committing
human rights vio
implicates
lations, jeopardizing bilateral relations (Rosenblum and
states may directly at
Salehyan 2004). Refugee-sending
tack the host state in retaliation for protecting their politi
cal opponents and for encouraging refugee militarization.
activities and the existence
Thus, support for militant
of
external
diasporas.
In the case of refugee migration,

to extend violence and opposition
the militarization
especially
rebel
bases
among
likely to escalate
refugee
communities
into international
(Salehyan 2008).
As before, bargaining and negotiations
cult in these circumstances.
Cross-border
are
hostilities
may be diffi
strikes against
external
rebel bases and "hot pursuit" raids may jeopar

dize bilateral relations, but the costs of confrontation with
the host state may be outweighed by the tactical need to
strike rebel sanctuaries. Moreover, disputes about
alleged
are difficult to re
foreign support for rebel organizations
solve through bargaining because itmay be difficult for
the host state to credibly commit to
reversing its policies
Schultz
In
2006;
(see Bapat
2007).
many instances, host
countries deny supporting
rebels despite home country
and it is difficult to prove that the state is
allegations,
not engaging in a pernicious
policy; it is hard to prove
a negative fact.
to
Verifying
compliance with demands
IDEANSALEHYAN
792
rebel support is difficult since rebel assistance
states supporting rebel
may
covertly. Moreover,
organizations may find that the benefits of promoting
the risk of retaliatory
instability in neighbors
outweigh
discontinue
India's Invasion
of East Pakistan
continue
strikes. Finally, weak hosts may find it difficult to comply

to evict rebel organizations
demands
because it is
with
to forcibly move
too costly and dangerous
against
such
groups.
suggests that refugee origin states may

disputes against refugee hosts. Such
This discussion
launch militarized
dynamics have recently been apparent during the refugee

crisis in Darfur. Refugees from the Sudanese region of
Darfur have fled to eastern Chad and have been attacked
allies
forces alongside their paramilitary
by government
to
inva
Israel's
eliminate
rivals.3
ethno-political
seeking
in 1982 was
the desire
in large part motivated

by
camps of Palestine Liberation
tion of southern
in a military occupa
for two decades. This leads to
sion of Lebanon
to clear refugee
forces. This resulted
Organization
Lebanon
a second hypothesis:
H2:
between a dyad will increase the

Refugee migration
the
that
sending country will initiate
probability
a militarized
interstate dispute (MID) against the
receiving
country.
state of Bangladesh
The modern-day
seceded from Pak
istan in 1971 following years of perceived
second-class
status. Provincial
in 1970 brought
elections
the pro
independence Awami Lague to power in East Pakistan,
and Bengali nationalists began to more vocally advocate
for secession by taking to the streets. InMarch of 1971, the
of Pakistan, seeking to restore law and order
government
to the region, arrested leaders of the pro-independence
movement,
including the popular leader, Sheikh Mujibur
Rahman. Following
the arrest, Bengali units in the Pak
istani army revolted and, under the leadership of Ziaur
on
of Bangladesh
Rahman, declared the independence
March 26. This led to open clashes between the Pakistani
Army and a new rebel force, known as the Mukti Bahini
(Sisson and Rose 1990).
The fighting in Bangladesh
led to the largest ever out
to the UN High Commis
pouring of refugees. According
10 million
sioner for Refugees (UNHCR),
Bangladeshis
fled for India, at a rate of over 100,000 per day (United
for Refugees 2000). These
Nations High Commissioner
Indian regions ofWest
fled to the conflict-prone
and
Assam,
Tripura, where politi
Meghalaya,
Bengal,
cal control and relations with New Delhi were already
strained. The refugee influx presented a severe financial
migrants
and fell par

toll on the Indian government
on
in
the
border
hard
local
regions.
populations
ticularly
India suffered several externalities as a direct result
and human
Illustrative
Case Narratives
estimated
in Bangladesh. The government
that the cost of caring for the refugees for a six-month
some inter
period was roughly $175 million. Although
of the conflict
to the large-N quantitative
analysis,
proceeding
this section presents a pair of illustrative cases that will
be
the theorized causal connections
help to underscore
Before
conflict. Both refugee

tween refugees and international
states may initiate
receiving states and refugee-sending
actions. The Indian invasion of Bangladesh
militarized
how concerns over
(East Pakistan) in 1971 demonstrates
refugee burdens can elicit a forceful response. Clearly, In
international
dia and Pakistan had an enduring
rivalry
but the revolt in East Pak
since independence/partition,
the
istan and the subsequent
refugee exodus provided
inva
Rwanda's
conflict.
to
of
escalation
the
led
that
spark
invades a
sion of Zaire offers a case where a government
state in order to attack refugee encampments
neighboring
and flush out militants
among them. Along with back
Sese Seko in 1996, the
Mobutu
forces
rebel
against
ing
Rwandan armed forces attacked refugee camps in Eastern
national aid was offered, much of the cost fell on India

itself; on October 22, the government
imposed new taxes
on rail and air travel, postage, and periodicals to cover the
cost of the refugee flow.4 In addition, the refugee crisis
led to amajor outbreak of cholera in the camps, severely
(United Nations
jeopardizing public health conditions
for Refugees 2000, 64). One doc
High Commissioner
tor in the region indicated that "... many of the refugee
of
camps were flooded in ankle-deep water and millions
people spent the night
itation is non-existent
the
spread cholera and other diseases throughout
with
frictions
caused
also
inflow
The
refugee
region."5
as refugees sought work to supplement
local populations
soon
4Keesings
Zaire.
Record
of World
into India"
Refugees
Aid to Refugees."
3
For an account
of
the Sudan/Chad
Human RightsWatch (2006).
conflict
involving
Darfur,
see
squatting in mud and slush. San

and it is feared that the rain will
and
"Mass
Events.
"Effect
on
wvyrw.keesings.com
5
Quote
Influx
of East
(access
date March
from
obtained
Keesings
Refugees
10, 2007).
of East
Pakistani
Economy-International
10,2007).
(access date March
Record
Pakistani
Influx
Indian
into
of World
India."
Events.
www.keesings.com
"Mass
EXTERNALITIES
OF CIVILSTRIFE 793
to
(Sisson and Rose 1990, 180). According
are
refugees ".. .living outside camps
being
as
on
much
landowners
farm
laborers
wages
by
their rations
one report,
employed
lower than those that must
is creating
This
locality...
be paid to laborers from the

resentment
among the local
India sought a solution

issue was certainly not
sion, yet itwas amajor factor that contributed to the war

and a trigger that prompted the Indian response.
working force."6
As Sisson and Rose
(1990,179-81)
argue, the refugee
influx severely disrupted local politics in affected regions.
Relations between
the Indian central government
and
West Bengal?where
the Communist
party was recently
in the regional governing coalition?had
previ
been
tense, and the refugee crisis was a "major factor
ously
in India's decision to dismiss the West Bengali coalition
included
in mid 1971" (179). Regions

in the north
government
east were also seriously affected as ethnic minorities
inAs
sam, Tripura, and elsewhere felt threatened by the influx
of foreigners (Sisson and Rose 1990, 181; see also Weiner
1978). India called on the government of Pakistan to end
the conflict and accept repatriation. In a letter to Pakistan
issued on May 14,1971, the Indian government declared,
"This deliberate expulsion of such a large number of peo
ple from their homes has created a human problem of
which
is capable of producing
unparalleled magnitude
serious
peace
in the area, leading to a threat to

repercussions
in the region."7 Yet Pakistan found it difficult to
comply with Indian demands as recognizing the opposi

tion and ending the imposed state of emergency would
mean resigning itself to
independence. Rela
Bangladeshi
tions between the two states became even more strained
after Pakistan
India of harboring Mukti

and launched several counterinsurgency
militants
accused
Bahini
strikes
in the border
region.8
the prospect of a protracted rebellion and a
3, 1971, India
long-term refugee situation, on December
sent an invasion force into
in
support of the
Bangladesh
Fearing
movement.
independence
On
December
16, Pakistan
sur
rendered and Bangladesh won its independence. Anxious

to ease the refugee burden, India then declared that all
refugees must return to Bangladesh by the end of Febru
ary 1972, and began
in conjunction with
to make
the necessary preparations

Thus, although rela
India and Pakistan had long been tense,
tions between
the UNHCR.
the influx of 10million

cipitated
a major
6Keesings
Record
7Keesings
Record
into
of World
(access
www.keesings.com
Refugees
refugees from East Pakistan pre

conflict between the two countries as
of World
India."
Events.
date March
Events.
"The Refugee
10, 2007).
"Mass
www.keesings.com
Influx
(access
Situation."
of East Pakistani
date March
to the refugee crisis. The refugee

the only reason for India's inva
Rwanda's
Invasion
of Zaire
in 1994 shocked the world for

genocide
its level of brutality, but beyond the immediate human
to long-term
toll, it contributed
instability across the
Great Lakes region of Africa. The Arusha accords signed
The Rwandan
between
the Tutsi-dominated
Rwandan
Patriotic
Front
(RPF) and the Hutu-led

government was viewed by some
members
of the Hutu elite as granting too much power
to the RPR Yet, before the accords could be implemented,
was assassi
Rwandan
President
Juvenal Habyarimana
the excuse Hutu extremists needed to
nated, providing
begin a campaign of genocidal violence
against Tutsis
and moderate Hutus. To put an end to the genocide, the
RPF reinvaded
the country from their bases in Uganda

the
in Kigali, which
in turn
government
toppled
an
enormous
as
exodus
millions
of
prompted
refugee
and
Hutus
fled the country fearing RPF reprisals. This mass

emigration was encouraged and orchestrated by the ex
Rwandan Armed Forces (ex-FAR) and the Interahamwe
militias, who sought to establish a government-in-exile
and regroup
as a rebel force across
the border
(Lischer
2005).
At
the peak of the refugee crisis, Zaire and Tanza

over 1 million
Rwandan
refugees each. As
Lischer (2005) explains, while Tanzania took steps to pre
vent refugee militarization,
Zaire's President Mobutu Sese
nia hosted
Seko was neither willing nor able to prevent the ex-FAR

forces from taking control of the camps. This posed a
humanitarian
dilemma for aid agencies as the necessity
of feeding and sheltering legitimate refugees was
compli
cated by the fact that the camps were largely run by the
of the genocide
1998; United
(Gourevitch
perpetrators
Nations High Commissioner
for Refugees 2000). Officials
on the ground understood
the gravity of the situation.
Shahyar Khan, the UN Secretary General's representative
to Rwanda,
exclaimed,
"we
are
sitting
on
a volcano...
we
must
separate the wolves from the sheep."9

From inside Zaire, the ex-FAR/Interahamwe
forces
began to mobilize militarily, while gathering recruits and
supplies among the refugees. News reports estimated that
roughly 40,000 fighters were present in and around the
10,
2007).
Record
of World
Events.
"Indo-Pakistan
Border
8Keesings
sion." www.keesings.com
10, 2007).
(access date March
Ten
in Bonner,
"Rwanda
Faces New War Unless
In
9Quoted
Raymond.
ternational
Force Is Sent, UN Aides
Say." New York Times. Novem
ber 6, 1994. Late Edition.
Section
1, page 12.
794
Zairean
town of Goma.10 Clashes
between Hutus
and lo
cal Zairean Tutsis ensued with the complicity and support

of President Mobutu, whose forces were responsible for
their own human rights violations
against Tutsis. RPF
leader Paul Kagame viewed the refugee situation in East
ern Zaire with alarm. As the international
community
was unwilling to demilitarize
the camps, Rwanda decided
to act on its own. As Kagame would relate in an interview,
and handle three things:
"We were ready to hit them...
first to save the Banyamulenge
(Zairean Tutsis) and not
the camps, return the
then to dismantle
let them die...
to Rwanda,
and destroy the ex-FAR and militias;

refugees
and third, to change the situation in Zaire" (quoted in
Gourevitch
1998, 296).
In October
its offensive
of 1996, Rwanda began
2006) and amilitary

attack the refugee camps, combat
of Democratic
Forces for the Liberation
of the
and
(ADFL) to lead the charge against Mobutu,
Congo
later, to govern the country. With support from Rwandan
troops, rebels took over the town of Goma by November.11
army quickly crumbled as the ADFL contin
towards the capital; by May
ued its offensive westward
1997, Kinshasa was taken by Kabila's forces. Meanwhile,
Rwandan troops invaded eastern Zaire to clear the refugee
The Zairean
to the Mu
of the refugees had been moved
to
dismantle.
gunga camp, which Rwandan forces began
The Rwandan army began to fire upon the refugees in or
der to force people back into Rwanda and worked to cut
aid and disperse the ex-FAR militants
off humanitarian
camps; most
for Refugees 2000;

(United Nations High Commissioner
had no choice
cut
the
aid
food
With
off,
10).
refugees
chap.
the largest
but to return to Rwanda. By mid-November,
over
and
been
cleared
had
700,000
refugees re
camps
for Refugees,
turned home.12 As UN High Commissioner
"The link between
Sadako Ogata commented,
refugee
more
is
nowhere
and
and
peace
security
perhaps
problems
evident than in the Great Lakes region in Africa" (United
for Refugees 2000, 262).
High Commissioner
was
not simply a humanitar
Thus, the refugee crisis
Nations
ian issue, but spread conflict between the Rwandan gov

states. It
ernment and armed dissidents into neighboring
led to both instability in Zaire (see Lischer 2005; Salehyan
10
"Rwandan
The Economist,
1, 1995. Page 34.
April
invasion by Rwanda to
ex-FAR militants,
and
for his support. Yet, the 1996-97
and Gleditsch
retaliate against Mobutu

invasion of Zaire would
Rwandan
not be the first such
later turn on his former patron and

begin to support Hutu refugee warriors, reorganized un
Forces for the Liberation
der the banner of the Democratic
attack. Kabila would
(FDLR). Rwanda again sent in troops in 1998

the FDLR and Hutu refugees; yet this second
invasion was not successful in toppling the government
of Rwanda
to attack
in Kinshasa
and prompted
tervene as well,
sparking
states (see Clark 2002).
First, it armed Zairean Tutsis to fight

against Mobutu.
forces.
and Mobutu's
back against Hutu refugee militants
Second, it armed and supported Laurent Kabila and the
Alliance
IDEANSALEHYAN
states across the region to in

involving several African
a war
Quantitative
or not
To ascertain whether
I conduct
patterns,
analysis of militarized
empirical
Analysis
these cases reflect broader
a time-series
cross-section
interstate disputes (MIDs)

International conflict
the
1955-2000.13
period,
spanning
data come from the dyadic version of the MID dataset
at
produced by Maoz (2005). MIDs are events that involve
use
or
a
least two states inwhich there was
threat, display,
probit
force; this variable is coded one for cases where

there was aMID and zero otherwise. Because many pairs
of military
conflicts
of states are unlikely to be involved in military
with one another (e.g., Peru and Tanzania),
following
relevant
I restrict the analysis to politically
convention,
are either contiguous or involve at
states
the
where
dyads
least one major power.14 Also, as I am mainly interested in
after the first year
conflict onset, subsequent MID-years
are excluded from the analysis; this practice also excludes
that join aMID after it begins.15 All
involve directed-dyads, where
below
regressions reported
the dyad A-B is also included as B-A; this allows us to
ascertain which state in the dyad initiates the MID. The
third states (ormore)
variable is coded one if the first state listed

dependent
(the initiator) starts aMID against the second state (the
13
These
are chosen
years
because
of data
availability
for
the
full
panel.
Refugees:
Crime
and Nourishment."
14In results
11
Take a Town with
James. "Zaire Rebels
McKinley,
November
New
York
Times.
Rwandan
3,1996.
Army."
Section
1, page 1.
in Mass
Alia. "700,000
12Thompson,
Star. November
16, 1996. Page Al.
Exodus
from
the Help
of
Late Edition.
Zaire."
Toronto
not
noncontiguous
the magnitudes
I split the sample between

shown,
results are roughly
dyads. These
are somewhat
of the coefficients
i.e., those that

dyads,
contiguous
sults are available
upon
request.
15In practice,
may
major
powers.
aMID
breaks out between
why
states join in.
from why
subsequent
the reasons
be different
include
and
contiguous
similar, although
for non
stronger
These
re
two states
OF CIVILSTRIFE 795
EXTERNALITIES
target). If there was aMID
is treated as missing.16
in the A-B dyad, the B-A dyad

variable
The main
is the number
of
independent
refugees in a dyad. These data come from the United Na
for Refugees17 (UNHCR) and
tions High Commissioner
is supplemented with data from the United Nations Relief
tional conflict
for reasons unrelated
to forced migration
and Schultz 2008), omitting this
(see Gleditsch, Salehyan,

I
variable may lead to biased estimates. In the models,
include separate dummy variables indicating whether the
target or initiator had an internal armed conflict; the list
Re
of civil wars comes from the Uppsala University/Peace
search Institute of Oslo Armed
refugees from the country of origin in the country of asy

lum in a given year, provided that there were at least 100
num
refugees.19 Thus, there are separate variables for the
ber of refugees in the initiator and the number of refugees
Other control variables

include the liberal peace
"triad" (Russett and Oneal 2001 ). First, I include informa
tion on regime type from the Polity IV dataset (Marshall
demo
and Jaggers 2005) to account for the well-known
that both send

the initiator to test the hypotheses
The mean
and
countries
launch
MIDs.20
receiving
ing
was
of the refugee variable (for nonzero observations)
from
94,000; Pakistan hosted the largest number of refugees,

with over 3.2 million Afghan refugees hosted in 1989. To
deal with skewness and diminishing marginal
effects, I
use the natural
log of refugees (in c000s) after adding one

to the base. Finally, since international conflict may pro
duce refugees, I use lagged rather than contemporaneous
values of the independent variables. This approach is not

costs, as estimates will not reflect MIDs that oc
cur during the first year of a refugee crisis.21 Yet, reverse
without
is likely to be a bigger source of bias and so this
causation
conservative
is preferred.
approach
In estimating
to
it is important
these regressions,
war
control for civil
within a state in the dyad. Since civil
wars produce refugees and may also contribute to interna
16There were
initiations
1,227 MID
observations.
directed-dyad/year
or
in the data,
1.2%
of
the
17Ithank Bela Hovy of the Population Data Unit at the UNHCR

for providing
these data.
18The UNHCR
does
not maintain
on Palestinians.
data
data are not perfect and are sometimes

for
(see Crisp
1999). In addition,
states,
reported
refugees between
the value of this variable
to be zero although
some of
is assumed
these cases may
in fact be missing
rather than nonevents.
Despite
are the best available
these caveats,
the UNHCR
data.
figures
19I acknowledge
that refugee
to political
where
dyad-years
subject
20
manipulation
there were no
an estimate
I develop
of refugee flows rather
models,
The UNHCR
does not collect figures on annual
refugee
a proxy measure
I develop
therefore
flows;
by taking the first dif
in annual
at zero. This method
ference
truncation
has
stocks, with
In alternative
than
stocks.
flaws
since
ation
and
are
it is not
a true
third-country
similar when
largely
to the initiator
loses
flow measure
resettlement
using
some of
and
since
can affect
this measure;
its statistical
refugee
repatri
totals. The results
however,
refugee
flows
change
significantly
includes
both minor
(Gled
and major
cratic peace effect. I create dummy variables indicating

the target, initiator, or both states were demo
cratic as indicated by a Polity score of 6 or higher on the
?
10 to +10 Polity scale. Several cases in the Polity data
whether
have missing values indicating regime instability. Many

of these cases of missing data stem from civil wars and/or
them would not be
regime upheavals, and so dropping
Instead of dropping cases or interpolating
appropriate.
values, observations
including transitional or interregnal
are
as
coded
"undemocratic"
and receive a value
polities
of zero on the democracy
a dichotomous
variable
In addition, I include
coded one for cases of regime
Variables
for
instability.
instability in the target, initiator,
and both are included.
indicator.
to
Second, I include a variable for trade dependence
account for the pacifying effect of economic
ties. This is
created by determining,
for each state, the sum of im
from
the
other and dividing it by that
ports plus exports
state's GDP. Trade data are from Gleditsch
(2002b) and
GDP data come from Goldstein, Rivers, and Tomz (2007).
I include a variable
in inter
for joint membership
a
count
is
This
of
(IGOs).
governmental
organizations
IGOs of which both states in the dyad were members
Third,
and Warnke
(Pevehouse, Nordstrom,
Other control variables
include
for territorial
are connected
miles
of water
2006).
a dummy
variable
is coded one if the states

contiguity,
or
land
by
separated by no more than 400
et
al.
(Stinnett
2002). Another dichotomous
which
variable for colonial
contiguity, taken from the Correlates

(COW) data, is included and coded one if the
states were contiguous
through colonies or dependen
cies. Because power preponderance may deter hostilities,
I also include data on relative capabilities, based on the
of War
COW project's Composite
Index of National
Capabilities.
significance.
21
For instance,
the Bangladesh
discussed
above is excluded
example
this method.
The refugee flow began
inMarch
of 1971 and
following
the Indian
invasion
in December
occurred
of the same year. This
said, the results do not
values are used instead.
itsch et al. 2002) and

internal conflict.22
Conflicts
Dataset
for Refugees
Agency and the U.S. Committee
and Immigrants for figures on Palestinian refugees.18 The
data are given as annual dyads reporting the stock of
and Works
if contemporaneous
22
All
were also run

ofWar
Intrastate
regressions
using the Correlates
dataset
(Sarkees 2000) as well as the Fearon and Laitin
(2003)
list of civil wars. Results
do not change
when
significantly
using
War
these
data.
796
is the initiator's capabilities over the total capabili
ties in the dyad. Finally, because allied states may be less
likely to fight one another, I include a control variable for
This
Results
(1999).23
In estimating
since
I include a count of years

these models,
to
in the dyad (or independence)
the last MID
effect of refugees given mean

(i.e., near parity).
interpreted as the marginal

levels of relative capabilities
the similarity of the two states' alliance portfolios using

the weighted
S-score developed by Signorino and Ritter
IDEANSALEHYAN
Table
1 presents
the results of the probit regressions.
1 reports the results for the basic regressions while
2 includes the interaction effects, testing the sec
in the data,
for possible duration dependence
a
with
with
three
knots
cubic
(see Beck,
spline
along
Katz, and Tucker 1998). Also, because the dyadic obser
of one another, I employ
vations are not independent
robust standard errors clustered on the dyad. Finally, all
Model
of the independent variables are lagged one year to limit

biases caused by potential reverse causation, although re
sults do not change significantly with contemporaneous
space, Iwill not discuss the control variables in depth here;

most of the results confirm the findings of previous stud
has an unexpected
ies although shared IGO membership
wars
in the initiator and in the target
positive sign. Civil
account
variables.
model
ondary hypotheses
regarding relative power. Each model
robust standard errors,
lists columns for the coefficient,
tests. To save on
and p-values of two-tailed significance
hypotheses about the conflict

effects of refugee inflows and outflows, itmay
producing
also be the case that given the presence of refugees, certain
Sale
raise the probability of aMID
(see Gleditsch,
reduces
and
while
Schultz
2008),
joint democracy
hyan,
than others. The reasons

dyads are more conflict-prone
for flight, the absorption capacity of the state, the ethnic
makeup of the refugees and host society, and the avail
1 pro
Turning to the variables of interest, model
vides evidence in support of both hypotheses.
Refugees
that the host country will initi
increase the probability
In addition
to themain
assistance, among other factors,

ability of international
some
may explain why
dyads with refugees experience
not.
the available
while
others
do
conflict
Unfortunately,
both
it.
ate aMID
the sending country; in addition, they

that the sending country will initi
ate aMID against the host. The size of both coefficients
against
data do not allow us to test all possible hypotheses about

such effects.24 However, when sending as well as receiving
to be more likely to
refugees, I expect stronger countries
to intervene
are
states
unable
often
launch attacks. Weak
are similar and aWald
to end forced migration

from stronger countries, and in
In addition,
stead must quietly suffer refugee burdens.
the interaction
weak
states
may
across national
frontation
with
not
be
able
to pursue
their
own
boundaries when doing so risks a con

a much more powerful state. I test these
were
of the trade, GDP,

the exception
the EUGene
created using
software,
IGO data, these data

3.1 (see Bennett
version
and
and Stam 2000).

24
This
effect
have a significant
is able to test if refugees
analysis
conflict.
of international
the probability
Yet, future research?
to exploring
the
be devoted
and quantitative?should
qualitative
violence
some
states
do
experience
given refugees, why
question:
on
while
others
do not?
This
requires
which
additional
data
collection
and
the
complicate
unduly
is to show the primary
of this article
here. The purpose
analysis
to un
work
should be devoted
additional
effect of refugees, while
effects.
conditional
derstanding
a different
research
design,
lesser effect than refugees to a country. Model 2 includes

terms between the initiator's share of total
The
and each of the refugee vari
(mean-centered)
capabilities
ables.
refugees
to and from
corollary hypotheses by interacting refugees
the initiator with the dyadic capabilities variable. In do
was mean-centered
in
ing so, the capability share variable
order to make the constituent coefficients more easily in
the coefficient on refugees is
terpretable; in these models,
23
With
test of equivalence of coefficients

fails to reject the null (x2<.000);
thus, there is little ev
a
idence that refugees from
country have a greater or
coefficients
indicate
that
relative
strength
exac
are more
erbates the effect of refugees. Stronger countries

likely to attack weaker states, both when they send and
when they receive refugees. The coefficients on refugees
are statistically significant at mean capabilities, and taken
together,
the constituent
coefficients
pass
tests of joint
significance at the .05 level.

Because probit coefficients are difficult to interpret,
of conflict by setting a
I compute predicted probabilities
the
number of refugees
and
baseline
varying
hypothetical
and other variables
of interest while
holding
from model
other val
l.25 These
based on estimates
results are reported in Table 2. The baseline case is a con
dyad, with no civil wars; neither
tiguous, nondemocratic
is transitional, and all other values are set at their means;
ues constant
moreover,
case.
there are no refugees in this hypothetical
a
a dyad experiencing
of
such
probability
The predicted
would
25
Predicted
using
and their standard

probabilities
the Clarify
software
(Tomz, Wittenberg,
errors were
and King
computed
2003).
797
OF CIVILSTRIFE
EXTERNALITIES
Probit Regressions
Interstate Disputes,
1 Refugees and Militarized
Table
(2)
(1)
Coef.
0.019
0.057
Refugee Stock IN Initiator

Refugee Stock FROM Initiator
Ref. in Initiator * Capability
Std. Err.
Robust
0.013
0.056
Civil War
in Target
Initiator
Democratic
0.046
0.022
0.036
0.000
0.059
0.013
0.000
0.092
0.063
0.143
0.049
0.049
0.318
0.224
0.040
0.000
0.139
0.045
0.002
0.050
0.625
Target
Both Democratic
Initiator
Transitional Target
Both Transitional
0.000
0.135
0.045
0.003
0.049
0.633
0.055
0.004
Share
Capability
Alliance S-Score
0.004
0.079
0.000
-0.418
0.079
0.103
0.067
-0.191
0.102
0.060
-0.030
0.094
0.746
-0.030
0.094
0.753
0.249
0.184
0.333
0.251
0.185
0.000
0.764
0.064
0.000
0.764
0.064
0.309
0.076
0.000
0.309
0.076
0.000
0.275
0.052
0.000
0.248
0.054
0.000
0.092
0.000
0.091
0.000
0.023
0.130
0.858
0.395
0.089
-0.404
0.003
-0.639
0.197
0.989
0.391
0.102
0.002
0.002
0.002
0.002
-0.205
0.014
0.000
-0.206
0.014
0.000
-0.001
-0.001
0.006
Years
0.055
-0.417
-0.672
Shared IGOMembership
0.159
-0.189
-0.406
Initiator's Trade Dep.

Target's Trade Dep.
-0.024
0.000
0.331
Contiguity
Colonial Contiguity
Peace
0.040
0.157
Democratic
Transitional
0.228
-0.024
P>|z|
0.003
Ref. from Initiator * Capability

||Controls\\
Civil War in Initiator
Std. Err.
Robust
Coef.
P>|z|
0.006
0.000
0.000
0.000
0.000
_spline2
0.000
0.000
0.000
0.000
0.000
0.000
_spline3
0.000
0.000
0.001
0.000
0.000
0.001
0.102
0.000
0.103
0.000
.spline 1
-1.846
Constant
86,497
1409.4
Wald Chi-squared
86,497
1631.68
-4410.54
Log pseudo-likelihood
MID
is approximately
1.2%. I then increase the refugee
stock in the initiator to 100,000 refugees (in logged terms),
which is roughly the mean refugee stock for all nonzero
observations.
Such an increase raises the predicted
prob
-4407.79
initiation to 2.3%; while this may be

ability of a MID
a low absolute probability,
it represents a 96.55% in
crease in probabilities
over the baseline case. A similar
increase
predicted
Table 2 Predicted Probabilities

Prediction
Std.Err. % Change
0.012
0.002
0.023
0.006
96.55
0.023
0.004
90.34
Civil war in initiator
0.021
0.003
77.46
Civil war in target

Joint democracy
0.017
0.002
40.46
0.006
0.001
Baseline*
100k refugees in initiator
100k refugees from target
*
Baseline:
transitional,
no
refugees,
contiguous
-1.832
no
civil wars,
dyad,
all other
neither
democratic,
at mean.
variables
-53.16
neither
in the refugee stock from a country raises the

probability of a dispute by over 90%. By com
a civil war in the initiator raises the probability
parison,
of a conflict by almost 78%, while a civil war in the target
raises the probability by 40.46%. The pacifying effect of
reduces the estimated probability of a
joint democracy
MID by nearly 53%. Clearly, inmany cases refugee flows,
and civil wars coincide with one an
authoritarianism,
so
other and
these effects compound one another. Thus,
in statistical
dyadic
and
substantive
refugee variables
terms, the effects of the

are quite large, confirming
the
hypotheses.
Using a contour plot, Figure 2 graphically
the substantive effects of the interaction between
displays
refugees
798
IDEANSALEHYAN
Figure 2 Refugees
Interacted with Capabilities

Target to initiator
Initiator to target
O)
.2
<*
-0.4
-02
0.0
0.2
0.4
Capabilities
and capabilities, given refugees to and from the initia

tor. Curves closer together indicate a sharper increase in
predicted probabilities while the curvature of the lines
displays the interaction. These images display roughly
similar plots, although with somewhat
stronger effects
for refugees from the target to the initiator.
on linkages between civil and international

to significantly
also promises
expand our under
of
conflict
standing
bargaining. Triangular negotiations
involving the state, dissident groups, and foreign powers
are evident inmany conflicts and are a clear next step in
Research
wars
work
on
and
bargaining
sis suggests, conflicts

successful bargaining
Conclusion
or unwillingness
groups often make
war.
First,
as
a "nested"
analy
at the domestic
level may impede

at the international
level. Inability
to accommodate
domestic
opposition
it difficult
for states to find
solutions
to their international
This article has argued that refugee migration may spark

conflicts between sending and receiving states. While mi
scholars
have
links between
suggested possible
tensions, such dynamics
refugee flows and international
have been largely ignored in the extensive literature on the
gration
causes of war and interstate disputes. Yet, as suggested by

Gleditsch,
Salehyan, and Schultz (2008), civil unrest can
also lead to conflicts between states. This article looks at
disputes. Second, principal-agent

problems may be apparent when states delegate conflict
to rebel organizations
rather than fight directly. Third,
verifiable
credible,
promises to end rebel patronage may
be difficult when
such support continues
covertly. Fi
nally, applying a "Coasian" bargaining framework to the
spillover effects of civil war can yield insights into the
conditions under which states will work to stem internal
conflicts.
This
one such connection
and finds that both refugee-sending

states are more
and refugee-receiving
likely to initiate
issues and ac
More
the
militarized
disputes.
generally,
wars
tors in civil
frequently span national boundaries and
part of a regional security nexus, blurring firm
distinctions between domestic and international conflict.
become
article also has
and pol
important normative
from
vulnerable
populations
Protecting
icy implications.
and abuse should be of paramount
persecution
impor
tance and efforts to use refugees as political pawns should
be resisted. It is important to note that while refugee flows
may
of international
conflict, not all
OF CIVILSTRIFE 799
EXTERNALITIES
refugee crises lead to an escalation of violence, and future
research should take into account how policy responses
can mitigate
security risks. The international
potential
Bennett,
community has tended to respond to refugee disasters in

terms, yet greater attention to the
largely humanitarian
states
is
needs
and the refugees themselves
of
security
Bennett,
(Lischer 2005). Rather than reacting to refugee

to keep domestic political
crises, preventative measures
unrest from escalating to civil war, ethnic cleansing, and
for Security and
and the Organization
the African Union
can help.
those in
conflicts, particularly
Many contemporary
and Sudan, have generated substan
Iraq, Afghanistan,
tial refugee flows. Syria and Jordan, for instance, have
accepted hundreds of thousands of Iraqi refugees, plac
social services, and in
ing strains on their economies,
ternal stability. Afghanistan
and Sudan have witnessed
the transnational
of
civil war and the escala
spread
tion of interstate disputes with neighbors. Yet, powerful
countries?most
been
notably the United States?have
reluctant
refugees from these conflicts, even

relieve the burden on regional hosts.
contrasts with the approach taken af
to resettle
this could
though
This significantly
ter the Vietnam War, for example, when large numbers

of refugees were resettled as a way to prevent regional
destabilization.
Similar resettlement efforts could signifi
in these
cantly improve economic and security conditions
volatile regions as well. Thus, appreciating
the complex
and conflict can
relationship between forced migration
not only improve academic research, but also yield im
proved policy outcomes.
gration
165-99.
Fiona.
2006.
and National
Borders:
Security."
and
Dylan,
Balch-Lindsay,
Time:
The World
Politics
the Line:
Mi
International
31(1):
Security
Borders
International
Andrew
A.
1992.
E.
Michael
Brown,
"Dangerous
Affect
Conditions
Dyads:
1996.
"The Causes
and Regional
Dimensions
of Internal Conflict." In The International Dimensions of

Internal Conflict, ed. Michael E. Brown. Cambridge, MA:
MIT
571-602.
Press,
Bueno
de Mesquita,
son, and Alastair
of
Bruce,
lames
D. Morrow,
Smith.
1999.
"An
the Democratic
Peace."
American
Patrick
James.
Siver
Randolph
Institutional
Explanation
Science Review
Political
93(4): 791-808.
Carment,
David,
and
and
Interstate
straints
Ethnic
1995.
Conflict:
Con
"International
a Crisis-Based
Toward
Assessment of Irredentism." Journal of Conflict Resolution

39(1): 82-109.
Clark, John F., ed. 2002. The African Stakes of the Congo War.
New
Coase,
York:
Palgrave
Ronald.
1960.
Law
MacMillan.
"The
and Economics
Problem
of Social
Cost."
Journal
of
1-44.
3(1):
Collier, Paul, V. L. Elliott, Havard Hegre, Anke Hoeffler, Marta

and Nicholas
Reynal-Querol,
Sambanis.
2003.
the
Breaking
Conflict Trap: Civil War and Development Policy. Washing

ton, DC: World
Collier,
Paul,
Bank.
and Anke
2004.
Hoeffler.
and Grievance
"Greed
in Civil War." Oxford Economic Papers 54(4): 563-95.

2004.
Kalena.
Cortes,
"Are Refugees
Different
on
Evidence
Empirical
Some
Immigrants?
ity of Immigrant Groups in the United

Economics and Statistics 86(2): 465-80.
and Alan
Crawford,
Timothy,
Humanitarian
London:
from
Economic
the Heterogene
States." Review of
on
2006. Gambling
Kuperman.
Moral
Rebellion
Hazard,
Intervention:
New
and
Routledge.
Enterline.
of Civil War
28(2):
2000.
Duration."
"Killing
Interna
Will
Research.
Paper
Working
H. Moore,
and Stephen
Poe.
to Leave: Domestic
Threats
2003.
Just Have
1964-1989."
Migration,
International
and
Interactions
29(1): 27-55.
and Security
Security
in Refugee
Issues
Christian,
Davenport,
You
"Sometimes
and Gil
1996.
Loescher.
Alan,
Dowty,
"Refugee
Grounds
for International
Action."
International
as
Flows
Security
21(1): 43-71.
James D.
Fearon,
1995.
"Rationalist
Explanations
for War."
In
ternational Organization 49(3): 379-414.
tional Studies Quarterly 44(4): 615-42.

2006.
"State Bargaining
with
Transnational
Bapat, Navin.
rorist Groups."
International
Studies Quarterly
50(1):
Ter
Fearon,
213?
James
surgency,
D.,
and
David
and Civil War."
D.
Laitin.
American
2003.
Political
In
"Ethnicity,
Science Review
97(1): 75-90.
30.
Beck,
Stuart
for Refugees.
12.
International
Century."
Behavi
of Michigan
Crisp, Jeff. 1999. "Who Has Counted the Refugees? UNHCR and
the Politics of Numbers." United Nations High Commissioner
"Crossing
"Redrawing
The
2004.
Stam.
ing the Likelihood of InterstateWar, 1816-1965." Journal of

Conflict Resolution 36(2): 309-41.
Forced
Peter. 2003.
Andreas,
in the Twenty-First
78-111.
C.
179
26:
Interactions
Press.
Civil War.
References
Adamson,
Allan
and
Scott,
"EUGene: A
2000.
Stam.
International
Manual."
D.
Bremer,
in Europe
Cooperation
and Allan
Scott,
oral Origins of War. Ann Arbor: University
needed
genocide should be emphasized and promoted. Strength

institutions such as
ening regional conflict management
D.
Conceptual
204.
Nathaniel,
"Taking
Time
lonathan
Seriously:
N.
Katz,
and Richard
Time-Series-Cross-Section
Tucker.
1998.
Analy
siswith a Binary Dependent Variable." American Journal of

Political Science 42(4): 1260-88.
Fox,
Jonathan.
2005.
"Unpacking
Transnational
Citizenship."
Annual Review of Political Science 8: 171-201.

Gartzke,
Erik.
2007.
"The
Capitalist
of Political Science 51 ( 1 ) : 166-91.
Peace."
American
Journal
800
Ghobarah,
Paul
Hazem,
and
Huth,
Bruce
Russett.
2003.
"Civil
Wars Kill andMaim People Long After the Shooting Stops."

American Political Science Review 97(2): 189-202.
Gleditsch, Kristian Skrede. 2002a. All International Politics Is
Local: The Diffusion of Conflict, Integration, and Democrati
Ann
zation.
Arbor:
of Michigan
University
Press.
Gleditsch, Kristian Skrede, Idean Salehyan, and Kenneth A.

Schultz. 2008. "Fighting at Home, Fighting Abroad: How
Civil Wars Lead to International Disputes." Journal of Con
flict Resolution 52(4): 479-506.
Gleditsch, Nils Petter, PeterWallensteen, Mikael Eriksson, Mar
and H?vard
Sollenberg,
Strand.
2002.
"Armed
Con
flict 1946-2001: A New Dataset." Journal of Peace Research

39(5): 615-37.
Judith, Doug Rivers, and Michael Tomz. 2007. "In
Goldstein,
and
stitutions
International
Relations:
the
Understanding
Effects of the GATT and theWTO on World Trade." Inter

national Organization 61(1): 37-67.
Gourevitch, Peter. 1998.We Wish to Inform You That Tomorrow
We Will Be Killed with Our Families: Stories from Rwanda.
New
Picador.
York:
Guarnizo,
2003.
and
Portes,
Luis,
Alejandro
"Assimilation
and
William
J. Haller.
Determinants
Transnationalism:
of Transnational Political Action among Contemporary

American Journal of Sociology 108(6): 1211?
48.
Center
for Global
Policy,
George
2005.
and Host
Refugee
"Environmental
Communities."
Conflict
between
of Peace
Journal
Research
42(3): 329-46.
Zeev.
2005.
MID
"Dyadic
Dataset
Version
2.0."
Univer
http://psfaculty.ucdavis.edu/zmaoz/dyadmid.html.
sity of California, Davis.

Melander,
Erik,
tion Dependence
and Magnus
2006.
Oberg.
in Forced Migration."
to Go? Dura
"Time
Inter
International
actions 32(2): 129-52.

Will
Moore,
and
H.,
Morrow,
1999.
James.
"How
2004.
Shellman.
Stephen
secution: Forced Migration

Resolution 48(5): 723-45.
"Fear
Per
of
1952-1995." Journal of Conflict
Could
Trade
Affect
Conflict?"
Jour
nal of Peace Research 36(4): 481-89.

Murdoch, James, and Todd Sandier. 2004. "CivilWars and Eco
nomic Growth: Spatial Dispersion." American Journal of Po
litical Science 48(1): 138-51.
Newland, Kathleen. 1995. "The Impact of US Refugee Policies
on US Foreign Policy: A Case of the TailWagging the Dog?"
In Threatened Peoples, Threatened Borders: World Migration
and US Policy, ed.Michael S. Teitelbaum andMyron Weiner.
York: W. W.
New
190-214.
Norton,
Jon, Timothy
and Kevin
Nordstrom,
"Intergovernmental
A New
relates of War Data Set." Conflict Management

Science 21(2): 101-19.
Posen,
1996.
Barry.
to
Responses
"Military
2006.
Warnke.
1815-2000:
Organizations,
Cor
and Peace
Disasters."
Refugee
International Security 21(1): 72-111.
Hensel, Paul. 2001. "Contentious Issues andWorld Politics: The

Management of Territorial Claims in the Americas, 1816?
1992." International Studies Quarterly 45(1): 81-109.
Hensel, Paul, SaraMcLaughlin Mitchell, and Thomas Sowers II.
2006. "Conflict Management of Riparian Disputes." Political
Geography 25(4): 383-411.
1989.
Herbst,
Jeffrey.
Boundaries
tional
Human
in Africa."
Organization
2006. "Violence beyond Borders: The

in Eastern
Chad."
New
Human
York:
June 2006.
Rights Watch,
Robert
International
H.
1987.
"Quasi-states,
Dual
and Neo
Regimes,
classical Theory: International Jurisprudence and the Third

World."
International
41(4):
Organization
519-49.
Ithaca,
NY:
Cornell
Sarah.
In the Shadow
of Power:
States
and
Strate
gies in International Politics. Princeton, NJ: Princeton Uni

Press.
versity
Regan,
2000.
Patrick.
Civil
Wars
Powers:
and Foreign
Interven
tions and Intrastate Conflict. Ann Arbor: University ofMichi

gan
Press.
Marc
Rosenblum,
in US
Interests
R.,
and
Asylum
2004.
Idean Salehyan.
Enforcement."
Princeton
Princeton,
NJ:
2005.
Dangerous
Press,
University
Sanctuaries:
Rowland,
and
M.,
F. Nosten.
in Refugee
Control
2001.
Journal
Refugee
Civil War, and theDilemmas of Humanitarian

NY: Cornell University Press.
3-32.
Camps,
Aid. Ithaca,
Lyon, Alynna, and Emek Ucarer. 2001. "Mobilizing Ethnic Con

flict: Kurdish Separatism in Germany and the PKK." Ethnic
and Racial Studies 24(6): 925-48.
Marshall, Monty G, and Keith Jaggers. 2005. "Polity IV Project:
Political Regime Characteristics and Transitions, 1800-2004,
"Malaria
and Complex
Camps
"Norms
and
of Peace
Re
Epidemiology
and
Emergencies."
An
nals of TropicalMedicine and Parasitology 95(8): 741-54.

Rudolph, Christopher. 2003. "Security and the Political Econ
omy
of
International
Russett,
Bruce
Democracy,
tions. New
M.,
American
Migration."
Political
Science
603-20.
and Oneal,
John.
Press.
University
Lake, David A., and Donald Rothchild. 1998. "Spreading Fear:

The Genesis of Transnational Ethnic Conflict." In The Inter
national Spread ofEthnic Conflict, ed. David Lake and Donald
Rothchild.
1999.
Robert.
Review97(4):
Jenne, Erin K. 2007. Ethnic Bargaining: The Paradox ofMinority

Empowerment.
Powell,
search 41(6): 677-97.
Crisis
Rights
of Na
and Maintenance
Creation
"The
43(4): 673-92.
Human Rights Watch.
Lischer,
Manual."
Adrian.
Martin,
Pevehouse,
Migrants."
Jackson,
Users'
University.
Maoz,
Gleditsch, Kristian Skrede. 2002b. "Expanded Trade and GDP

Data." Journal of Conflict Resolution 46(5): 712-24.
Gleditsch, Kristian Skrede. 2007. "Transnational Dimensions of
Civil War." Journal of Peace Research 44(3): 293-309.
gareta
Dataset
Mason
IDEANSALEHYAN
Peace:
Triangulating
International
Organiza
2001.
and
Interdependence,
York: Norton.
Saideman, Stephen M. 2001. The Ties That Divide: Ethnic Pol

itics, Foreign Policy, and International Conflict. New York:
Columbia
University
Press.
Salehyan, Idean. 2007. "Transnational Rebels: Neighboring

States as Sanctuary for Rebel Groups." World Politics 59(2):
217-42.
Salehyan,
Idean.
2008.
"No
Shelter
Here:
Rebel
Sanctuaries
and
International Conflict." Journal of Politics 70(1): 54-66.

Salehyan,
Idean,
and
Kristian
Gledtisch.
2006.
"Refugees
and
the Spread of Civil War." International Organization 60(2):

277-97.
OF CIVILSTRIFE 801
EXTERNALITIES
Meredith
Sarkees,
2000.
Reid.
"The Correlates
of War
War: An Update to 1997." Conflict Management

Science 18(1): 123-44.
2007.
Kenneth.
Schultz,
"War
as an Enforcement
Data
on
and Peace
61.
In
Problem:
terstate Conflict over Rebel Support in Civil Wars." Type

Stanford
script.
University.
Shain, Yossi. 1989. The Frontier of Loyalty: Political Exiles in the

of the Nation-State.
Press.
sity
Shain,
Univer
2003.
Barth.
and
"Diasporas
Interna
tional Relations Theory." International Organization 57(3):

449-79.
Sheffer, Gabriel. 2003. Diaspora Politics: At Home Abroad. Cam

bridge: Cambridge University Press.
and
Curtis,
Signorino,
1999.
Jeffrey Ritter.
or Not
"Tau-b
Tau-b:
Measuring the Similarity of Foreign Policy Positons. Inter

national Studies Quarterly 43(1): 115-44.
Beth
Simmons,
Frank
A.,
The
"Introduction:
and Geoffrey
Diffusion
Dobbin,
International
Garrett.
2006.
of Liberalism."
International Organization 60(4): 781-810.

Sisson,
Leo
and
Richard,
Pakistan,
India,
and
of California
University
Branislav.
Slantchev,
John,
and
2003.
E. Rose.
1990.
the Creation
War
and
of Bangladesh.
Secession:
Berkeley:
Fred
Power
"The
to Hurt:
Costly
Conflict
Tanner.
2003.
Seattle:
R.
1997.
J.Waldman.
of Complex
2002.
University
Peter
Trumbore,
F.
2003.
"The
Public
and Refugee
Emergencies
Health
Situations."
18(1): 283-312.
or
"Victims
Ethno
Aggressors?
political Rebellion and Use of Force in Militarized Inter

state Disputes." International Studies Quarterly 47(2): 183
201.
United Nations High Commissioner for Refugees. 2000. The

State of the World's Refugees: Fifty Years of Humanitarian
Action.
Wagner,
New
York:
Oxford
R. Harrison.
University
2000.
Press.
and War."
"Bargaining
American
Journal of Political Science 44(3): 469-84.

Roger,
Waldinger,
and
David
2004.
Fitzgerald.
"Transnation
1177-95.
Walter,
Barbara.
2002.
to Peace:
Committing
The
Successful
Res
Press.
Weiner, Myron. 1978. Sons of the Soil: Migration and Eth

nic Conflict in India. Princeton, NJ: Princeton University
Manipulation:
"The Correlates
of War
Project
Direct Contiguity Data, Version 3." Conflict Management

and Peace Science 19(2): 58-66.
Tatla, Darshan Singh. 1999. The Sikh Diaspora: The Search for
Statehood.
J., and
Press.
Refugee
Stinnett, Douglas M., Jaroslav Tir, Philip Schafer, Paul F.Diehl,

Gochman.
M.
Aspects
olution of Civil Wars. Princeton, NJ: Princeton University
Press.
War, Politics, and theAbuse ofHuman Suffering.Washington,

DC: Brookings Institution.
and Charles
Toole,
alism in Question." American Review of Sociology 109(5):
with Completely Informed States." American Political Science

Review 97(1): 123-33.
Stedman,
Tomz, Michael, JasonWittenberg, and Gary King. 2003. "CLAR

IFY:Software for Interpreting and Presenting Statistical Re
sults." Journal of Statistical Software 8(1).
Annual Review of Public Health
and Aharon
Yossi,
CT: Wesleyan
Middletown,
Age
Thyne, Clayton. 2006. "Cheap Signals with Costly Con

sequences." Journal of Conflict Resolution 50(6): 937
of Washington
Press.
Weiner,
Myron.
gration."
1992.
International
"Security,
Stability,
Security
17(3):
and
International
Mi
91-126.
Woodwell, Doug. 2004. "Unwelcome Neighbors: Shared Eth

nicity and International Conflict during the Cold War." In
ternational Studies Quarterly 48(1): 197-223.
Zolberg, Aristide, Astri Suhrke, and Sergio Aguayo. 1989. Escape
from Violence: Conflict and theRefugee Crisis in theDeveloping
World. New York: Oxford University Press.

Midwest Political Science Association Wiley

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Midwest Political Science Association Wiley

Uploaded by

Copyright:

Available Formats

Midwest Political Science Association

tested in a quantitative analysis of the relationship between refugees andMID

is a growing body of literature on the secu

and Iraq, to name

of war, but may serve as catalysts for conflict,

1996; Posen 1996), although this claim has not

system and may provoke

as conflicts between governments

further refugee migration

gration. For instance, it has been argued that a major

and Rwanda have generated well over 1

2003; Salehyan and Gleditsch

that political violence and persecution

safety elsewhere. Government

other states in the international

initiation, 1955-2000. Results confirm that

clear refugee camps

and has not been

systematic empirical testing.

Despite a large body of work on the causes of war and

in the literature although there are

ally, there has been

states. More gener

states may spark disputes between

between particular dyads such as distance,

states, particularly the dis

putes, although many have found that internal conflicts

and are specific

that lead to interstate

civil war have begun

kin. Beyond kinship ties, Gleditsch,

analysis linking refugees to the initiation of

to be associated with MIDs, refugees are a signif

by both sending and receiving countries. Finally, stronger

Despite clear humanitarian

for states. I argue

that refugee migra

forces against the sender, and (2) the likeli

frequently extend across territorial boundaries.

were no longer funding many insurgencies and were more

not be entirely borne by the country itself, but may create

and if "refugees" were a single country, it

and Jordan hosted over

such as Iran, Pakistan,

The terms "internal," "domestic," and "civil" conflict im

state have significant

processes within one

(Balch-Lindsay and Enterline 2000;

can inflict an eco

aid may be available, refugee hosts?including

the cost of maintaining

they often have

and can contribute

as a potential causal factor,

can seriously jeopardize interstate

and unsanitary, creating conditions

care; and refugees may have specific health

body of empirical research finds

that civil wars

malaria, cholera, and diar

the host community

the picture by changing the domestic balance of

tions that produce refugee burdens. Such bargains share