Partitioning to Peace: Sovereignty, Demography, Ethnic Civil Wars

International Security
Spring 2008

Partitioning to Peace;
Sovereignty, Demography, and Ethnic Civil Wars

Carter Johnson.

Carter Johnson is a Ph.D. candidate at the University of Maryland,
College Park, where he is writing his dissertation on population
migration during and after war. He received his M.Sc. from the London
School of Economics and was Visiting Fellow with the Civic Education
Project at the University of Uzbekistan and State Pedagogical
University of Moldova. He is the former Project Coordinator of the
Minorities at Risk Project.; For their helpful comments and
suggestions, the author would like to thank Mark Lichbach, as well as
Christian Davenport, Diana Dumitru, Chaim Kaufmann, Corinne Lennox,
Maren Milligan, Jillian Schwedler, the anonymous reviewers, and the
participants of the DC Area Workshop on Contentious Politics.

Some scholars have proposed partition as a way to solve ethnic civil
wars. Partition theorists advocate the demographic separation of
ethnic groups into different states, arguing that this is the best
chance for an enduring peace. Opponents argue that partition is costly
in terms of its human toll and that its advocates have yet to
demonstrate its effectiveness beyond a limited number of self-selected
case studies. This analysis systematically examines the outcome of
partition, highlighting the centrality of demography by introducing an
index that measures the degree to which a partition separates ethnic
groups. This index is applied to all civil wars ending in partition
from 1945 to 2004. Partitions that completely separated the warring
groups did not experience a recurrence of war and low-level violence
for at least five years, outperforming both partitions that did not
separate ethnic groups and other ethnic war outcomes. These results
challenge other
studies that examine partition as a war outcome. The results also
have direct implications for Iraq’s civil war, postindependence
Kosovo, and other ethnic civil wars.

Since the early 1950s, civil wars have been longer lasting and more
frequent than international wars, producing high levels of death and
disability. 1 Ethnic wars have been especially common, comprising 55
percent (70) to 72 percent (91) of all civil wars between 1945 and
1999. 2 Moreover, cross-national evidence suggests that ethnic wars
last longer than nonethnic wars. 3 These numbers are even more
troubling given that, during the 1990s, more than 200 ethnic
minorities and subordinate majorities throughout the world were
contesting their political status. 4 In addition to the challenge of
ending civil wars, one of the most vexing problems has been their high
recidivism rate, with postconflict countries facing up to a 50 percent
chance of experiencing renewed war within the first five years of
establishing peace. 5

Since the mid-1990s, one solution to preventing the recurrence of
ethnic civil war that has gained international policy and scholarly
attention has been partition. 6 The debate surrounding partition
emerged at the end of the Cold War, as ethnic conflicts came to the
forefront of Western policymakers’ attention and international
boundaries were once again open to large-scale change. Still, Western
governments have demonstrated a certain ambivalence toward
partition. Although they have opposed the recognition of several de
facto partitions, such as Nagorno Karabakh in Azerbaijan, they have
promoted the incorporation of partitions into the peace plans of Sudan
and Papua New Guinea. 7 The negotiations surrounding the final status
of Kosovo further reflected Western ambivalence, with the independence
of Kosovo from Serbia fiercely contested within the European Union;
there is also the possibility of a further partition of Kosovo into
majority Albanian and
Serbian regions, which is the de facto political composition of
independent Kosovo. 8 In addition, policymakers have increasingly
proposed partition as one way to solve the civil war taking place in
Iraq. 9

Scholarly debate about the relative merits of partition is not new. 10
Since the end of the Cold War, however, the debate over partition has
primarily emphasized humanitarian issues. When scholars and
policymakers have proposed partition, it has been as a last resort, to
end ethnic wars when widespread massacres and forced population
transfers have already begun to occur and where long-term military
commitments by the international community are either not forthcoming
or are unable to produce peace. 11 Partition advocates argue that
under these conditions, partitioning groups into separate states where
they can protect themselves militarily provides the best chance for
ending ethnic wars and establishing an enduring peace. They do not,
however, support the blanket application of new borders to solve
ethnic civil wars. Rather they argue for the need to separate warring
populations–with population transfers where necessary–in an effort
to create
relatively homogeneous units where ethnic groups’ security fears are
reduced and demobilization and reconstruction efforts can begin
without the need for long-term commitments of international troops.

Studies on the debate over partition have largely remained theoretical
or focused on case studies and policy prescriptions. 12 Evidence has
pointed to some successes, such as the 1974 partition of Cyprus, which
led to decades of peace, and some failures, such as the partition of
British India, which led to widespread massacres and ultimately
resulted in war. This article offers a systematic, cross-national test
of all partitions that followed ethnic civil wars between 1945 and
2004. It finds that partition is a uniformly effective tool in
preventing a recurrence of war and low-level violence, but only if it
includes the physical separation of ethnic groups. This finding
challenges that of Nicholas Sambanis. who, in 2000, produced the first
empirical study of partition using a large-n, cross-national
database. 13 Based on his results, Sambanis concluded that "partition
does not significantly prevent war recurrence [, which] suggests, at
the very least,
that separating ethnic groups does not resolve the problem of violent
ethnic antagonism." 14

Sambanis’s analysis helped to further scholarly understanding of
partition’s relationship to the recurrence of conflict, but it did not
test the core theoretical argument of partition advocates. The
Sambanis analysis suffers from a methodological error because it
identified new borders (i.e., sovereignty) as the critical independent
variable to represent partition, and not the demographic separation of
warring ethnic groups. Testing the relationship between sovereignty
and conflict recurrence does not capture, and therefore cannot refute,
the position of partition advocates. 15 This article, in contrast,
introduces an index to calculate the amount of unmixing of ethnic
groups that occurs with partition, therefore capturing partition
advocates’ core argument.

The article is divided into six sections. First, I review the
theoretical and empirical literature on partition. Second, I examine
partition’s empirical record and raise critical questions about
Sambanis’s main conclusion that partition is not particularly
effective at preventing war recurrence. Third, I propose an
alternative variable–the Postpartition Ethnic Homogeneity Index
(PEHI)–for testing whether partition is a viable solution for ending
ethnic wars. Fourth, I demonstrate that, where the index shows warring
ethnic groups were in fact separated, neither war nor low-level
violence reoccurred for at least five years, suggesting that partition
advocates are correct. Fifth, I compare complete partitions with both
incomplete partitions and other forms of ethnic war termination, such
as government victory. Sixth, I discuss some of the policy
implications that follow from the analysis, in particular for the
final status of Kosovo and proposals to
partition Iraq.

Partition as a Means to End Civil Wars
Since the mid-1990s, scholars have produced a number of theoretical
and empirical studies on how civil wars end and how to create an
enduring peace. Factors that can influence the termination of civil
wars and help to prevent their recurrence include the length or cost
of the war, the ability of a central government to make credible
commitments, the presence of mediators, the strength of state security
forces, and a willingness to address grievances. 16

In cases where long-term military commitments by the international
community are not forthcoming or do not establish interethnic peace,
some scholars have suggested partition. 17 Further, given the high
percentage of civil wars that recur within the first five years of the
establishment of peace, advocates of partition argue that it may be
the best way to bring about an enduring solution to at least some
kinds of intractable ethnic conflicts. 18

Partition theory rests on two primary principles. First, ethnic civil
wars are qualitatively different from other kinds of civil
war. Second, warring ethnic groups confront a security dilemma that
prevents them from de-escalating and demobilizing. As a result, ethnic
groups must be separated and given sovereignty to produce long-term

In the social science literature, "ethnic conflict" refers to
conflicts involving ascriptive group identities, identities that are
very difficult, if not impossible, to change and that are often based
on an individual’s descent (e.g., language, religion, or race). 19
Wars involving groups with different communal identities (e.g., Sunni
and Shiite Arabs in Iraq) or linguistic identities (e.g., Ossetian and
Georgian groups in Georgia) fall into this category. It is this
component of an ascriptive identity, and the politicization of that
identity, that distinguishes ethnic civil war from other forms of
civil war. The ascriptive component can lead militant organizations to
identify entire ethnic groups as loyal or disloyal within a country’s
population in a way that ideological conflicts cannot. 20 In fact,
ethnic group members often go to great lengths to find out who is a
member of an "enemy" group, including the use of census data in Nazi
electoral lists in Sri Lanka, and identity cards in Rwanda. 21

The concept of ethnic civil war does not assume primordial identity in
the sense that groups are fixed and unchanging, nor does it require
fixed individual identities. 22 In fact, the concept of an "ethnic
war" is still compatible with mainstream constructivist understandings
of individuals sharing multiple, overlapping identities, some of which
become more salient than others depending on the context. 23 Advocates
of partition need not accept ethnic identity as given, only that it
becomes given under certain conditions. In pre-colonial Rwanda and
Burundi, for example, Hutu and Tutsi identities were flexible, but
this was certainly not the case during the 1994 Rwandan genocide, when
some Hutus targeted all Tutsis because of their ethnic identity. 24 In
the words of one individual caught up in the Bosnian ethnic civil war,
"I am a Croat. . . . I was Yugoslavian, and now I am a Croat. I always
knew that I am a Croat, but I didn’t feel it so much. Now, you
have to be Croat, Serb, Muslim, Jewish, or whatever…. For me
personally, these identities didn’t interest me at all: my being a
Croat wasn’t important. But now, you have to be." 25

In contrast to ideological wars, where loyalties are more fluid both
during and after combat, in ethnic wars, members of one ethnic group
are far less likely to fight for the opposing side, 26 dividing
communities and making postwar reconciliation in an intermingled state
very difficult–some would argue impossible. 27 As Chaim Kaufmann
states, "War hardens ethnic identities to the point that cross-ethnic
political appeals become futile, which means that victory can be
assured only by physical control over the territory in dispute. Ethnic
wars also generate intense security dilemmas, both because the
escalation of each side’s mobilization rhetoric presents a real threat
to the other, and even more because intermingled population settlement
patterns create defensive vulnerabilities and offensive
opportunities. . . . Once this occurs, the war cannot end until the
security dilemma is reduced by physical separation of the rival
groups. 28

A security dilemma can develop when, in an anarchical system, one
state’s defensive action makes other states feel less secure, drawing
two or more states into a conflict, even where none originally sought
it. 29 Scholars have noted the presence of security dilemmas in ethnic
conflicts. 30 As empires collapse and states fail, the resulting
anarchy causes competing and distrustful ethnic groups to engage in
defensive actions that may seem threatening to other groups, thereby
heightening tension. The mixture of different ethnic populations
influences the intensity of a security dilemma and can encourage
offensive military action even when defense is the overriding
consideration: members of one ethnic group may try to save their kin
located within the territorial confines of another ethnic group or,
conversely, preemptively expel members of other ethnic groups located
inside their own territory. 31 Some scholars have identified
ethnically mixed regions as
particularly prone to violence in ethnic wars. In discussing the
former Yugoslavia, Hurst Hannum states, "Keeping ‘trapped’ Serbs
within Croatia and Bosnia and Herzegovina (and trapped Croats within
the latter) may have actually contributed to continuation of the
violence, since peaceful means of redrawing borders seem to have been
excluded." 32

Empirical results from Jaroslav Tir’s study of postsecession war
recurrence support this conclusion. 33 His examination of every
secession and partition in the twentieth century finds that the
presence of stay-behind minorities increases the likelihood of
conflict. 34 To build a lasting peace, therefore, warring ethnic
groups must be separated into homogeneous regions capable of
self-defense. Kaufmann argues, "Solutions that aim both to restore
multiethnic civil politics and to avoid population transfers, such as
institution building, power sharing, and identity reconstruction,
cannot work during or after an ethnic civil war because they do not
resolve the security dilemma created by mixed demography." 35

Scholars have differed on the role of sovereignty in ending ethnic
wars and ensuring long-term peace. Supporters of partition, such as
Kaufmann, initially argued that demographic separation with regional
autonomy was sufficient, as long as that autonomy protected the
group’s key interests and provided for "regional defense
capabilities." 36 Alexander Downes, on the other hand, later stressed
that sovereignty is an essential ingredient. 37 Moving beyond the
security dilemma, Downes suggests that even where populations have
been separated, autonomy alone is not enough to maintain peace. The
very process of ethnic civil war, he claims, "makes reconstructing a
multiethnic state afterwards problematic because it destroys the
parties’ ability to trust each other not to violate any agreement
negotiated." 38 Downes and others therefore maintain that to be
successful, partition must include both political sovereignty (i.e.,
independence) as well as the separation
of ethnic groups. 39

Debates over partition that had traditionally focused on the normative
goal of self-determination started to change in the mid-1990s, when
academics such as Kaufmann, John Mearsheimer, and Stephen Van Evera
began to propose partition as a way to minimize the number of deaths
and reduce the human suffering in ethnic civil wars. 40 Since borders
can rarely be drawn to create ethnically homogeneous states, advocates
of partition recommend population transfers to accomplish what would
otherwise occur under worse conditions for civilian populations. The
argument is that a third-party transfer would be better than forced
transfers perpetrated by enemy militias aiming for ethnic cleansing or
worse. 41

Population transfer is a controversial subject that has been open to
serious criticism for at least two reasons. First, even organized
population transfers, if involuntary, are a violation of fundamental
human rights enshrined in international law. 42 Second, there is
debate about the degree to which any power can conduct "humane"
population transfers. A cursory look at examples from the twentieth
century reveals a chaotic and lethal record, primarily affecting
civilians. 43

In addition to raising serious concerns over the ethics of population
transfers, critics of partition theory level two broader objections:
first, partitions transform internal wars into international ones; and
second, they do not solve ethnic antagonisms. Advocates of partition
counter that postpartition wars tend to occur where populations were
not separated (e.g., postpartition Ireland, British India, and
Palestine). 44 In addition, they note that conflicts between
postpartition states are often the subject of greater international
attention and diplomatic pressure, and thus are likely to be brought
to an end sooner than if they remained internal conflicts. Moreover,
the partitioned states are subject to international laws regulating
war, potentially rendering any conflict less inhumane. 45 As for
solving ethnic antagonisms, it is not clear whether this is within
anyone’s power. At a minimum, the separation of warring ethnic groups
reduces the security
threat, which may give moderate politicians within each group a
chance to be heard. 46

If separating warring ethnic groups through partition can be shown to
prevent the recurrence of ethnic war, then the international community
should accord it greater consideration when seeking solutions to such
conflicts. It was with this challenge in mind that Sambanis undertook
his cross-national study, arguing that "beyond a handful of
self-selected cases, partition theorists have not presented proof that
partition is the only viable and credible solution to ethnic civil
war. They have not even proven that partition outperforms other war
outcomes in terms of peace-building potential." 47 The following
section outlines both Sambanis’s evaluation and the data set used in
this study.

Cross-National Statistical Testing of Partition Theory
Sambanis compiled a data set of all civil wars between 1945 and 1999
to compare the effectiveness of partition with that of other causes of
war termination and peace building. 48 Sambanis tested each of these
independent variables, including partition, on three dependent
variables: its ability to prevent war recurrence; its ability to
reduce low-level postwar violence; and its ability to promote postwar
democratization. Based on his analysis, Sambanis concluded that
"although it may seem like a clean and easy solution, partition fares
no better than other outcomes of ethnic civil war." 49 He also
concluded that "the evidence does not support the assertion that
partition significantly reduces the risk of war recurrence." 50 He
went on, "I can point to only very weak evidence in support of the
hypothesis that partitions help end low-level ethnic
violence. . . . More importantly, the positive impact of partitions
seems fragile and extremely dependent." 51

Sambanis used a broad definition of ethnic civil war, which allowed
him to draw on a variety of civil war-related databases. 52 He based
his definition on six criteria: the war caused more than 1,000 battle
deaths; it challenged the sovereignty of an internationally recognized
state; it occurred within the recognized boundaries of that state; it
involved the state as one of the principal combatants; it included
rebels with the ability to mount an organized opposition; and it
involved parties concerned with the prospect of living together in the
same political unit after the end of the war. 53

Sambanis’s definition of low-level violence follows largely from Peter
Wallensteen and Margareta Sollenberg’s data set that coded all armed
conflicts causing 25 or more deaths but falling short of war. 54
Sambanis defined partition as "a war outcome that involves both border
adjustment and demographic changes." 55 This article follows Sambanis
and includes instances of both "partition" and "secession."
Traditionally a partition was understood as a "fresh division" of
territory, usually executed by a sovereign (often great) power that
occurred at the time of decolonization. 56 In my study, however, who
imposes partition is relatively unimportant: the critical factor is
whether dividing warring groups into separate entities can prevent war
recurrence. 57 Further, whether it is possible to accurately
distinguish between secessions and partitions is unclear: Kaufmann,
for example, codes Cyprus (1974) as a "partition" but Abkhazia
(1992-93) a "secession,"
even though both Turkish Cypriots and Abkhaz were involved in
separatist movements that were ultimately successful because of
assistance from an external power (Turkey and Russia,
respectively). 58 Moreover, given that the implications of partition
theory affect partitions and secessions equally in the minds of
academics and policymakers, it is logical to code both. Finally, it is
relatively unimportant whether a postpartitioned entity achieves de
jure sovereignty (as in the case of Bangladesh’s internationally
recognized separation from Pakistan) or de facto sovereignty (as in
the case of South Ossetia’s unrecognized separation from Georgia);
therefore both types are included. Although some scholars have begun
to include wars of decolonization in data sets of civil wars (e.g.,
Algeria’s independence from France and Mozambique’s from Portugal),
this practice remains questionable conceptually. 59 Moreover, because
I am primarily interested in
reevaluating Sambanis’s analysis, like him, I exclude such
wars. Using Sambanis’s data set, I was able to reproduce his

My cases differ slightly from those used by Sambanis. First, I
excluded Tajikistan because it did not undergo a recognizable
partition during or after its civil war, and because most experts
deemed it a regional and ideological, not ethnic, conflict. 60 Second,
I included the case of Bosnia, but where Sambanis uses the 1992
partition, I used the 1995 partition. The 1992 partition of Bosnia
from Yugoslavia did not occur at the end of the war, which raged for
three more years. 61 I coded the Dayton accord 62 as a partition of
Bosnia between Serbs, on the one hand, and Bosniaks and Croats on the
other. 63 The territory of Bosnia and Herzegovina was divided into two
de facto states in 1995, each maintaining separate armed forces that
cannot enter the other’s territory. 64 This qualifies Bosnia as a
partition. As the realist scholars John Mearsheimer and Stephen Walt
confirmed at the time, Bosnia "produced a partition
settlement. . . . The settlement is a
veiled partition but a partition nevertheless." 65 Third, I excluded
the 1992 Croatia case because of the difficulty of categorizing it as
a war end. Although there were cease-fires between the Zagreb-based
Croatian authorities and the Knin-based Republic of Serbian Krajina
(RSK), 66 the conflict between Yugoslavian/Serb and Croatian forces
continued in many regions, including the Serb siege of Dubrovnik and
the Croat siege of Bihac. In addition, serious military operations
between the forces of the RSK and Croatia’s army resumed soon after
each cease-fire. 67 I therefore excluded this case from the
analysis. Given the ongoing violence between Yugoslavian/Serb and
Croat forces between 1991 and 1995, 68 it is more appropriate to
consider this a Croatian "war of independence," ending with the
partition of Croatia from Yugoslavia in 1995, which is what I included
in my analysis. 69 Finally, I updated all relevant variables for all
cases of ethnic civil war
through mid-2004. This update includes the 1999 partition of Kosovo
from Yugoslavia.

More important, I introduce a new independent variable for
analysis. This new variable is critical because Sambanis’s definition
of partition relies on the existence of border and demographic changes
but not the degree of separation between ethnic groups. The following
section outlines my alternative approach.

The Centrality of Demography
If, as supporters of partition argue, the critical independent
variable is demography, and if partitioned countries with new
minorities increase the security threat, then demographic changes need
to be captured for use as an independent variable. Social scientists
have not developed many demographic indicators to capture degrees of
ethnic heterogeneity. Tatu Vanhanen, for example, created the ethnic
heterogeneity index to explore the general relationship between ethnic
conflict and ethnic division. 70 Daniel Posner developed an index
based on politically relevant ethnic groups. 71 Neither index,
however, can identify which groups were at war and the degree to which
they separated after the war. To address this problem, I created the
Postpartition Ethnic Homogeneity Index (PEHI).

In constructing the PEHI, I began with a state that contains a titular
ethnic group and a minority ethnic group. The two groups engage in a
civil war, and at some point, their territory is partitioned in the
hopes of ending the conflict. The result is two countries, each with
its own titular majority as well as a potentially "stay-behind"
minority from the other ethnic group. Determining the degree to which
the ethnic groups were separated requires knowing (1) the percentage
of the minority group in the original country (recorded as OSM, for
original state minority); (2) the percentage of the original minority
left in the rump state after partition (RSM, for rump state minority);
and (3) the percentage of the original titular group now found as a
minority inside the new state (NSM, for new state minority).

Given the theoretical focus on demography, with an understanding that
leaving sizable minorities on either side of a new border could
increase the chances of renewed warfare and low-level violence, this
index uses both new minorities to calculate the degree to which
partition and population transfers succeeded in separating the warring
groups. For countries with more than two ethnic groups at war, groups
were aggregated if they fought on the same side or if they were
treated as one by the opposing force; 72 if there were separate
warring ethnic dyads within a civil war, they were treated as separate

To calculate the PEHI, I subtracted the new minority percentages (RSM
and NSM) from the original minority percentage (OSM). I then divided
this percentage by the original minority percentage (OSM) and
multiplied the result by 100. This simple calculation yields the
percentage change in the size of ethnic minorities produced by
partitioning the country, thus indicating the degree of ethnic

PEHI = OSM-(RSM + NSM)/OSM x 100.

The higher the PEHI number, the greater the degree of separation
achieved by partition. The maximum score a partition can receive is
+100, indicating a complete separation of the warring ethnic
groups. This number falls as the size of the stay-behind minorities
grows relative to the original minority percentage. 73

Timely data on minority populations in the aftermath of ethnic civil
wars proved difficult to find. For coding, I relied on a staple set of
books and encyclopedias. 74 The guiding principle in gathering the
data was to have at least two credible sources provide the same
numbers; when these numbers were close but not exact, an average was
taken. Where two sources could not be found among the staple, I
consulted case-specific academic publications and news reports gauging
refugee flows of ethnic groups. 75 Where data were unavailable for the
year immediately after partition, I used the first available data.

Table 1 presents the PEHI component figures from the seventeen cases
of partition that occurred after ethnic civil wars between 1945 and
2004. For example, in Azerbaijan the OSM–in this case, the
Armenians–formed 5.8 percent of Azerbaijan’s population before the
civil war. After the civil war, approximately 20,000 Armenians
remained in rump Azerbaijan, creating an RSM of 0.25 percent. The
number of Azeris found in the new state of Nagorno Karabakh after the
war ended was negligible (NSM < 0.01). The following equation reflects
the PEHI for the case of Azerbaijan:

PEHI (Azerbaijan-Nagorno Karabakh) = 5.8 -(0.25 + 0)/5.8 x 100 =

Table 1. Calculation of PEHI Values for Partitions after Ethnic Civil

Original Rump New Postpartition
State State State Ethnic
Country Minority Minority Minority Homogeneity Index
Azerbaijan-Nagorno Karabakh 5.8 0.25 <0.01 95.69
Bosnia (1995) 31.2 2.30 3.20 86.40
Cyprus (1963) 18.2 11.90 <0.01 34.60
Cyprus (1974) 12.3 <0.01 <0.01 100.00
Ethiopia-Eritrea (1991) 6.4 0.12 <0.01 98.13
Georgia-Abkhazia (1993) 1.8 <0.01 0.08 99.83
Georgia-South Ossetia (1994) 3.0 <0.01 0.05 98.33
India-Kashmir (1965) 10.4 10.40 3.00 -28.85
India-Kashmir (1989-94) 10.4 10.40 3.00 -28.85
India-Pakistan (1947-48) 24.4 10.40 1.60 50.82
Israel-Palestine (1948) 33.3 <0.01 13.80 58.56
Moldova (1992) 31.0 24.00 40.50 -108.06
Pakistan-Bangladesh (1971) 46.0 0.30 0.20 98.91
Russia-Chechnya (1996) 0.6 0.30 2.50 -366.67
Somalia (1992) 27.4 25.00 28.00 -93.43
Yugoslavia-Croatia (1995) 19.7 1.10 4.49 71.62
Yugoslavia-Kosovo (1999) 14.0 0.70 6.00 52.14

NOTE: Scores of <0.01 assume a value of 0 for calculation of the
Postpartition Ethnic Homogeneity Index.

The PEHI shows what was achieved with partition. Rather than a simple
binary code indicating if de facto sovereignty was achieved, the PEHI
captures the degree to which minorities were separated. For example,
the 1963-64 partition of Cyprus, where Turks migrated into small
defensive enclaves during intense interethnic war, failed to
significantly divide the populations, leaving a large number of Turks
outside the defensible enclaves. According to the PEHI, this partition
homogenized the territories by a paltry 34.6 percent, reducing the
security dilemma only marginally. Partition theory would expect a high
likelihood of war recurrence under these conditions, which is what
took place. In contrast, Azerbaijan’s partition succeeded in
separating Azeris and Armenians, with a PEHI of close to 100
percent. As predicted by partition theory, there has been no
recurrence of war.

Examining the PEHI
I added the PEHI to the Sambanis data set to assess its impact on the
most crucial dependent variable: war recurrence. Using binary probit,
the variable warend2 (no war recurrence for at least two years after
the end of the civil war) was regressed on the continuous variable
PEHI only for ethnic wars that experienced partition, controlling for
the original minority percentage. The results demonstrated a positive
coefficient (i.e., the higher the ethnodemo-graphic separation, the
less likely war will recur within the first two years), significant at
the 0.1 level for a one-tailed test. Due to a small n, however, these
results are at most suggestive, given the statistical problem that, as
yet, there have been too few partitions. For these statistical
results, see the appendix.

The PEHI indicates whether any one partition selected from the
database would be considered a "complete" partition or an "incomplete"
partition by partition advocates. A complete partition is one in which
the warring minorities are fully separated, leaving negligible
stay-behind minorities; an incomplete partition is one in which the
minorities are not separated, leaving sizable stay-behind minorities
in either of the two emerging states. For this study, any partition
that succeeded in separating the warring parties by a PEHI of 95
percent or more is considered a complete partition. The threshold of
95 percent is not fixed, but rather should be seen as a guide to
indicate partitions where ethnic groups have been effectively
separated in their entirety, a critical demand by partition
advocates. 76

Table 2 compares complete and incomplete partitions against the two
main criteria established by Sambanis: recurrence of war either two or
five years after the end of a civil war, and recurrence of low-level
violence within two or five years. 77 This five-year threshold is
particularly significant given World Bank data suggesting that "the
typical post-conflict country faces a 50 percent risk of renewed
conflict within the first five years of reaching peace." 78

Table 2. Complete and Incomplete Partitions

Low- Low-
War Level Level
Postpartition Ended War Violence Violence
Ethnic for Ended Ended Ended
Homogeneity Complete Two for Five for Two for Five
Country Index Partition Years Years Years Years
Azerbaijan-Nagorno 95.69 Yes Yes Yes Yes Yes
Karabakh (1994)
Bosnia (1995) 86.40 No Yes Yes Yes Yes
Cyprus (1963) 34.60 No No No No No
Cyprus (1974) 100.00 Yes Yes Yes Yes Yes
Ethiopia-Eritrea 98.13 Yes Yes Yes Yes Yes
Georgia-Abkhazia 99.83 Yes Yes Yes Yes Yes
Georgia-South 98.33 Yes Yes Yes Yes Yes
Ossetia (1994)
India-Kashmir -28.85 No Yes No No No
India-Kashmir -28.85 No Yes No No No
India-Pakistan 50.82 No Yes Yes Yes Yes
Israel-Palestine 58.56 No Yes Yes Yes Yes
Moldova (1992) -108.06 No Yes Yes No No
Pakistan- 98.91 Yes Yes Yes Yes Yes
Russia-Chechnya -366.67 No No No No No
Somalia (1992) -93.43 No No No No No
Yugoslavia-Croatia 71.62 No Yes Yes Yes Yes
Yugoslavia-Kosovo 52.14 No Yes Yes No No

As the results in Table 2 indicate, for all partitions achieving a
PEHI separation score of 95 percent or higher, there were no
recurrences of war for at least five years, nor were there recurrences
of low-level violence for five years. In fact, no partitioned state
achieving a PEHI score above 70 percent experienced a recurrence of
war or low-level violence, suggesting the threshold of 95 percent
could even be lowered. For partitions with lower PEHI scores, the
results are mixed, with most countries experiencing either war
recurrence or a return of low-level violence. This suggests that a
partition that successfully separates warring ethnic groups produces
substantially different results from partitions that do not separate
the groups, which is what partition advocates predict. This further
underscores the need to disaggregate partitions into those that
separate the warring ethnic groups and those that do not. Although the
number of cases is
small–there have been only six cases of complete partition–the
results are consistent and unambiguous. Given the small number,
however, these results must be treated with caution. While partition
advocates cannot be faulted for the lack of complete partitions since
1945, it is important to recognize the limits of currently available

One case that stands out is the Ethiopia-Eritrea conflict. Although
this partition meets the criteria established by Sambanis, with no
recurrence of war or low-level violence between Ethiopia and Eritrea
for at least five years after their partition, these countries did
return to war in 1998, after a seven-year peace. Partition advocates
do not claim, however, that separating warring ethnic groups will
always prevent a return to war or low-level violence forever into the
future; rather they claim that it is often the best option to give
peace a chance. In addition, for this particular case, had more
population transfers occurred at the time of partition, the tens of
thousands of Eritreans remaining in Ethiopia would not have faced the
horrific expulsions that occurred during the 1998-2000 war. 79

Further, as mentioned earlier, partition advocates argue that any
future war between partitioned states will be an improvement over a
return to civil war, because the two sovereign states will be
subjected to greater international attention and diplomatic pressure,
increasing the likelihood of the war ending quickly. Ethiopia and
Eritrea exemplify this logic: the civil war the two sides fought
lasted more than fifteen years, whereas the interstate conflict of
1998 ended within two years following heavy international
pressure. This positive result must be tempered, however, by the fact
that international wars, though relatively infrequent in today’s
world, can be very lethal. 80

There are several countries that experienced incomplete
partitions–that is, partitions that do not completely separate the
warring ethnic groups–and yet did not experience war recurrence or
low-level violence within the first five years of the end of their
civil wars. Although this indicates that demographic separation is not
the only way to prevent war recurrence, a close look at the cases of
states that did not experience war recurrence for five years reveals
troubling insights. The conflict over India-Pakistan (1947-48) did not
recur in the first five years, but the incomplete partition, which
left substantially intermixed populations in place, was followed by
three wars over the proceeding half century. The incomplete partition
between Israel and Palestine (1948) was similarly followed by
low-level violence and war recurrence over subsequent
decades. Moreover, it was arguably the reintroduction of significant
ethnic intermingling after Israel’s
occupation of the West Bank and Gaza in 1967 that has led to
heightened conflict. Croatia’s "war of independence" (1991-95) also
ended with an incomplete partition. Although the degree of ethnic
unmixing was not enough to be deemed a complete partition by the
strict criteria outlined in the article. The final Croatian military
operation of the war in 1995 forced approximately 200,000 Serbs to
flee Croatian territory, reducing the percentage of Serbs in Croatia
by almost two-thirds by the end of the war, and therefore
substantially unmixing the populations. 81

A potential concern with the results of this analysis may be over the
issue of endogeneity, or whether a selection bias has taken place
where cases of complete partition occurred in states where ethnic
minorities were already compact and homogeneous, and thus relatively
easy to separate after a war without ethnic cleansing or large
population transfers. Few communities are ethnically homogeneous,
however, and even those ethnic groups that are territorially
concentrated typically have a significant minority in their midst. In
this analysis, all of the complete cases involved large-scale forced
population transfers during the countries’ wars, with the possible
exception of Bangladesh. 82 Militias and government armed forces
displaced hundreds of thousands of people during the two ethnic wars
in Georgia, during the war over Nagorno Karabakh, and during the
ethnic war in Cyprus. In the other partition reaching a high
PEHI–Bosnia (86.4 percent)–armed forces
displaced hundreds of thousands of civilians based on their ethnic
identity in what had been an ethnically intermixed territory.

Complete Partition versus the Alternatives
A comparison of countries that experienced complete partitions with
those that experienced other outcomes between 1945 and 2004 (i.e.,
incomplete partition or no partition) reveals the benefits of
separating warring ethnic groups. Table 3 shows a cross-tabulation of
countries whose ethnic wars ended for at least two years. Seventy-one
percent of these wars did not recur. Nevertheless, in cases of
complete partition, no country experienced a return to war (100
percent). The chi-square test produced a statistic of 3.92 for a
probability of 0.14, although three cells have expected counts of less
than 5.00.

Table 3. Comparison of Alternatives for War Recurrence

Did the War End for Complete Incomplete No
At Least Two Years? Partition Partition Partition Total
Yes 6 6 45 57
(100%) (55%) (71%) (71%)
No 0 5 18 23
(0%) (45%) (29%) (29%)
Total 6 11 63 80

Chi-square = 3.92 (df=2), Pr = 0.141

In 68 percent of the cases, countries did not experience a recurrence
of war for at least five years, while all cases of complete partition
(100 percent) avoided a recurrence. The chi-square statistic is 6.07
for a probability of 0.048, statistically significant at the 0.05
level; again, three cells have an expected count of less than 5.00.

Turning now to low-level violence, an even greater contrast is evident
between complete partition and the alternative of incomplete partition
or no partition (see Table 4). In 60 percent of the cases, low-level
violence did not end for the first two years. Strikingly, for those
civil wars that ended with a complete partition, none experienced
further low-level violence during that period. The chi-square value is
10.06 for a p-value of 0.007, statistically significant at the 0.01
level, although three cells have expected counts of less than 5.00.

Table 4. Comparison of Alternatives for Recurrence of Low-Level

Did Low-level Violence End Complete Incomplete No
for At Least Two Years? Partition Partition Partition Total
Yes 6 3 23 32
(100%) (27%) (37%) (40%)
No 0 8 40 48
(0%) (73%) (63%) (60%)
Total 6 11 63 80

Chi-square = 10.063 (df=2), Pr = 0.007

In addition, countries in 60 percent of the cases were unable to
prevent the outbreak of low-level violence for at least five years
after ending their ethnic wars, while all cases of complete partition
proved successful in preventing the recurrence of low-level violence,
producing a chi-square of 9.24, with a p-value of 0.01; again, three
cells have expected counts of less than 5.00.

These figures strongly support the position of scholars who advocate
partition. Complete partitions that separated warring ethnic groups
prevented a return to war for at least five years. For the period
under review, complete partition was a sufficient
condition. Partitions that separated warring ethnic groups have also
terminated low-level violence for at least five years. This, too, was
a sufficient condition. This finding is all the more significant given
that a majority of post-civil war countries continue to experience
low-level violence, a plague that haunts civilian populations for
years after combat operations formally conclude.

This study’s results suggest that partition should be considered by
policymakers under certain conditions. This is particularly relevant
to postindependence Kosovo as well as to considerations to partition

Policy Implications
If the international community wants to end ethnic civil wars but it
is not prepared or not able to invest the long-term resources
necessary to achieve this militarily, then partition may be an
option. Partition should be considered, however, only where
populations are already largely separated at the time of intervention,
or where interveners are prepared to separate groups using mass
population transfers. If neither of these conditions holds, partition
will provide no increased protection against war recurrence or other
forms of violence. The saliency of this point is evident in debates
over Kosovo’s final status. Post-1999 Kosovo is an example of an
incomplete partition, and based on the results here, a final status
agreement that does not transfer Serb-controlled regions (or Serbs
themselves) back to Serbia will not provide the much-desired peace
dividend offered by complete partition. 83 Indeed, in spite of the
presence of international
peacekeepers, clashes in 2004 led to large-scale interethnic violence
and the ethnic cleansing of Serb villages. 84 Therefore, despite
Kosovo’s declared independence, the international community should
give serious consideration to the further partitioning of Kosovo into
homogeneous Serb and Albanian regions before withdrawal. 85 As it is,
the northern, Serb-dominated region effectively controls its own
affairs, and does not recognize the authority of Pristina. 86

Regarding Iraq, my results may appear to support calls for
partitioning the country, especially given the large amount of
communal unmixing that has already taken place. The accelerating speed
with which religious targeting and displacement occurred between 2005
and 2007 strongly suggested the emergence of a security dilemma
between the Shiite and Sunni communities, where families whose homes
were in the "wrong" area were threatened with death if they did not
leave. 87 The question faced by the international community, and the
United States in particular, is whether to permit this separation to
be organized by unregulated communal militias under conditions of
large-scale violence, or whether to facilitate this process under less
inhumane conditions through organized population transfers.

Implementation of a partition strategy, however, would face huge
challenges given the conditions in Iraq, and would likely worsen the
situation for civilians. Although the data from my study suggest that
partitioning Iraq along communal lines would have a strong chance of
creating a lasting peace within the Shiite-Sunni conflict, 88 Iraq is
not divided into homogeneous regions, despite claims to the
contrary. 89 There are large geographical regions of the country that
are relatively homogeneous, but there are nevertheless several densely
populated, multiethnic regions in the center of the country that would
need to be demographically separated under any partition plan, not to
mention multiethnic flash-point cities elsewhere in the country, such
as Kirkuk. 90 How to accomplish this in a way that protects civilian
lives would be a major challenge.

Because partition without the separation of ethnic groups does not
increase the likelihood of securing peace, population transfers become
necessary. This poses at least two problems. First, the implementation
of such transfers may sound procedural, but the reality would be far
from it. Let us imagine, for example, that considerable numbers of
minority group members refuse to move to their new home state. Would
the U.S. military be prepared to use force to deport those civilians,
the very civilians that partition is supposed to protect? Unmixing
populations can require great force, as the twentieth century
demonstrated. One solution to that problem is to keep transfers
"voluntary," where civilians are given the choice to move to "their"
new state or remain a minority within the other group’s new
state. Given the uncertainty of life under a new state dominated by an
enemy group following intense warfare, most members of the minority
group would likely
move. 91 If such "voluntary" transfers do not materialize, then
U.S. military force against civilians would become necessary. If the
United States is not prepared to use deliberate force against Iraqi
civilians, then partition should not be considered.

A second, more troubling difficulty for the situation in Iraq is that
partition advocates call for international interventions when a
long-term commitment of troops is not forthcoming, and this is
decidedly not the situation in Iraq. U.S. and other coalition forces
have had a huge military presence in Iraq for years and, if anything,
have demonstrated their inability to ensure even basic security for
many Iraqi civilians, especially in central regions of the
country. Any announcement of a decision to partition the country would
almost certainly lead to a large increase in the number of families in
mixed regions seeking refuge in "their" new states, emboldening local
militias that are pressing for separation. While U.S. and Iraqi forces
could attempt to minimize the violence that such a mass migration
might encounter, the ability of the military to ensure the safety of
hundreds of thousands of Iraqis on the move, given the difficulty in
providing even
basic security, is highly questionable. 92 In this case, the decision
to partition could make the situation considerably worse for
civilians, thus taking away the humanitarian rationale of partition

Partition could also be considered for other ethnic civil wars, such
as Sri Lanka’s war with Tamil separatists, Sudan’s war in Darfur, and
the Philippines’ war against Moro separatists. The key is to consider
this as a solution only where borders can be drawn around relatively
homogeneous groups, or where an intervener can be sure that population
transfers will occur under less inhumane conditions than the war
itself produces. As I stated above, this latter condition may be too
difficult to achieve in practice, suggesting partition simply may not
be viable in some cases.

This article has examined partition as a way to prevent the recurrence
of ethnic war and low-level violence. After reviewing theoretical
issues involving the dynamics of ethnic war, I reexamined the first
large-scale, cross-national empirical study of partition conducted by
Nicholas Sambanis. Sambanis relied on the presence of a new political
border as his indicator for partition, concluding that partition does
not significantly prevent war recurrence and should not be promoted by
the international community. Using the Sambanis data set, I tested the
variable suggested by partition advocates–demographic separation–for
all partitions that ended ethnic civil wars between 1945 and
2004. Introducing a new index to evaluate ethnodemographic separation,
the Postpartition Ethnic Homogeneity Index, I reanalyzed these
partitions and found that in all cases where the PEHI showed a
complete separation of warring minorities, there were no war
recurrences and no
occurrences of low-level violence for at least five years after the
end of the ethnic civil war. These results trump the alternatives of
incomplete partitions and no partitions, providing strong evidence for
advocates of partition.

I considered some of the policy implications of the results of this
study for countries experiencing ethnic warfare today and in the
recent past, in particular Kosovo and Iraq. I concluded that partition
should be considered in ethnic civil wars only where populations are
already separated demographically or where the intervener is able to
implement population transfers safely. This suggests that partitioning
Kosovo into majority Serb and Albanian districts may have long-term
benefits for regional peace. Partitioning Iraq, however, given its
highly populated, multiethnic regions, would likely increase the level
of conflict and human suffering, and therefore may not be a viable

This study suggests some promising areas for future research. First,
the PEHI could be extended to incorporate not only the presence of
minorities, but also their location and territorial concentration
after partition. Several scholars have identified group concentration
as a powerful predictor of rebellion, and Barry Posen has argued that
the location of minorities may increase the security dilemma risk. 93
For example, minorities living near a state’s borders might increase
the likelihood of renewed violence. Second, the frequency of
militarized interstate disputes among postpartitioned states could be
compared with that in all other states to see if these new dyads are
more or less at risk of interstate warfare. Third, an examination of
ethnic reintegration after ethnic war could be made to test partition
advocates’ claims that reintegration after war is either impossible or
likely to lead to renewed conflict. Although all wars produce high
levels of
displacement, the degree to which displaced populations return, even
after ethnic cleansing, is surprisingly variable both across cases and
temporally in the years following the end of ethnic wars. Many in the
international community are normatively committed to the idea of
multiethnic societies and yet are confounded by the realities of
protracted refugee problems and intransigent postwar communities long
after the war has ended. Does lack of reintegration stem from a
top-down process led by political elites, or a bottom-up process led
by local communities? Do returns increase or decrease the risk of war
and violence? Answers to these questions are not only of great
theoretical interest but extreme practical value.

Appendix: Statistical Examination of the PEHI
I added the PEHI to the Sambanis data set to check for significance on
war recurrence. Using binary probit analysis, I regressed the variable
warend2 (no war recurrence for at least two years after the end of the
civil war) on the continuous variable PEHI only for ethnic wars that
experienced partition. The PEHI is affected by the prewar minority
percentages; as a control, therefore, the prewar minority variable has
also been included in the model. The results show a positive
regression coefficient for the PEHI, as one would expect based on the
theory, with a p-value significant at the 0.1 level (see Table 1).

Table 1. Probit Results for No War Recurrence after Two Years

Variable z-value >|z|
Postpartition Ethnic 0.01 1.56 0.06
Homogeneity Index
Prewar minority -0.03 -0.91 0.18
Constant 0.93 1.35 0.09

NOTE: N = 17. is an unstandardized coefficient; z is a z-test of ; and
is the -value for a one-tailed z-test.

The results suggest that the greater the separation of warring
minorities produced by a partition (i.e., the higher the PEHI), the
greater the expected likelihood is of not experiencing a return to war
for at least two years. Given the small n (17), however, these results
are only suggestive. It should also be noted that if any other control
variables are entered into the probit analysis, all results become
insignificant; this is almost certainly due to the small n.

1 James D. Fearon and David D. Laitin write, "Between 1945 and 1999,
about 3.33 million battle deaths occurred in the 25 interstate wars
that killed at least 1,000 and had at least 100 dead on each
side. These wars involved just 25 states that suffered casualties of
at least 1,000 and had a median duration of not quite 3 months. In
contrast, in the same period there were roughly 127 civil wars that
killed at least 1,000, 25 of which were ongoing in 1999. A
conservative estimate of the total dead as a direct result of these
conflicts is 16.2 million, five times the interstate toll. These civil
wars occurred in 73 states–more than a third of the United Nations
system–and had a median duration of roughly six years." Fearon and
Laitin, "Ethnicity, Insurgency, and Civil War," American Political
Science Review, Vol. 97, No. 1 (February 2003), pp. 75-90, at
p. 75. The number of ongoing civil wars has declined since its peak in
the late 1980s and early 1990s,
but it remains high (ranging between 20 and 30 wars since 1993), and
recent data show a second upward trend beginning in 2004. See Joseph
Hewitt, Jonathan Wilkenfeld, and Ted Robert Gurr, Peace and Conflict,
2008 (Boulder, Colo.: Paradigm, 2007). For a discussion of the toll of
civil wars on civilians, see Hazem Adam Ghobarah, Paul Huth, and Bruce
Russett, "Civil Wars Kill and Maim People–Long after the Shooting
Stops," American Political Science Review, Vol. 97, No. 2 (May 2003),
pp. 189-202. See also Paul Collier, Lani Elliott, Håvard Hegre, Anke
Hoeffler, Marta Reynal-Querol, and Nicholas Sambanis, Breaking the
Conflict Trap: Civil War and Development Policy (Washington, D.C.:
World Bank, 2003).
2 In his study of civil war duration, James D. Fearon codes his cases
as "ethnic," "nonethnic," and "ambiguous." Ethnic wars made up 55
percent of all civil wars between 1945 and 1999, and ambiguous wars 17
percent. See Fearon, "Why Do Some Civil Wars Last So Much Longer Than
Others?" Journal of Peace Research, Vol. 41, No. 3 (May 2004),
pp. 275-301.
3 Roy Licklider, "The Consequences of Negotiated Settlements in Civil
Wars, 1945-1993," American Political Science Review, Vol. 89, No. 3
(September 1995), pp. 681-690.
4 Ted Robert Gurr, Minorities at Risk: A Global View of Ethnopolitical
Conflicts (Washington, D.C.: United States Institute of Peace Press,
1993); and John McGarry and Brendan O’Leary, "Introduction: The
Macro-Political Regulation of Ethnic Conflict," in McGarry and
O’Leary, eds., The Politics of Ethnic Conflict Regulation (London:
Routledge, 1993), pp. 1-40.
5 See Paul Collier, Anke Hoeffler, and Mans Soderbom, "On the Duration
of Civil War," Policy Research Working Paper, No. 2681 (Washington,
D.C.: World Bank, September 2001); Paul Collier and Nicholas Sambanis,
"Understanding Civil War: A New Agenda," Journal of Conflict
Resolution, Vol. 46, No. 1 (February 2002), pp. 3-12; and Licklider,
"The Consequences of Negotiated Settlements in Civil Wars." For other
war recurrence statistics, see Barbara F. Walter, "Does Conflict Beget
Conflict? Explaining Recurring Civil War," Journal of Peace Research,
Vol. 41, No. 3 (May 2004), pp. 371-388.
6 John McGarry and Brendan O’Leary present a taxonomy of
macro-political forms of ethnic conflict regulation identifying
partition as one of eight. See McGarry and O’Leary, "Introduction";
and Nicholas Mansergh, The Prelude to Partition: Concepts and Aims in
Ireland and India (Cambridge: Cambridge University Press, 1978).
7 Sudan’s comprehensive peace agreement that ended the war between the
Khartoum government and the Sudan People’s Liberation Movement/Army
(SPLM/A) included a provision for a referendum on independence for the
south. See International Crisis Group, "The Khartoum-SPLM Agreement:
Sudan’s Uncertain Peace," Africa Report, No. 96 (Brussels:
International Crisis Group, July 25, 2005). Similarly, the 2001 peace
agreement between Papua New Guinea’s government and the separatists in
Bougainville included a referendum on independence. See Conciliation
Resources, Bougainville Peace Agreement, August 31, 2001,
8 On August 28, 2007, the Dutch foreign minister, Maxime Verhagen,
became the first European foreign minister to state that partition of
Kosovo into majority Albanian and Serb political territories would be
an acceptable outcome. This was followed by a similar comment by
Sergei Lavrov, Russia’s foreign minister, on August 31, 2007. See
Radio Free Europe/Radio Liberty, "Dutch Minister Says Partition of
Kosova Acceptable," RFE/RL Newsline, Vol. 11, No. 160, Pt. 2 (August
29, 2007); and "Russia Rules Out Crossing ‘Red Line’ on Kosova, but
Would Accept Partition," RFE/RL Newsline, Vol. 11, No. 163, Pt. 2
(September 4, 2007). As far as I am aware, no state actively lobbied
for a further partition of Kosovo, though some had suggested that such
an outcome would be acceptable if Kosovo and Serbian leaders accepted
this compromise. See Dan Bilefsky, "In a Divided Kosovo City, a
Resounding Vow to Remain Part of Serbia," New York Times, February 19,
9 For suggested partitions of Iraq, see Peter W. Galbraith, "Iraq’s
Salvation Lies in Letting It Break Apart," Sunday Times, July 16,
2006; Chaim Kaufmann, "Separating Iraqis, Saving Iraq," in "What to Do
in Iraq: A Roundtable," Foreign Affairs, Vol. 85, No. 4 (July/August
2006), pp. 156-160; and Leslie H. Gelb, "Last Train from Baghdad,"
Foreign Affairs, Vol. 85, No. 4 (July/August 2006), pp. 160-165. One
of the earliest proposals on partition came from Leslie H. Gelb, "The
Three-State Solution," New York Times, November 25, 2003. See also the
symposium "Ethnic Conflict, Ethnic Partition, and U.S. Foreign
Policy," sponsored by the Ethics and Public Policy Center, Washington,
D.C., January 15, 2003. On September 27, 2007, the U.S. Senate passed
a nonbinding measure calling for Iraq to be divided into federal
regions, with the likely outcome of separate Kurdish, Shiite, and
Sunni states. See Alissa J. Rubin, "In Iraq, Repeated Support for a
State," New York Times, October 1, 2007.
10 In 1985 Donald L. Horowitz stated, for example, "Separating the
antagonists–partition–is an option increasingly recommended for
consideration where groups are territorially concentrated." Horowitz,
Ethnic Groups in Conflict (Berkeley: University of California Press,
1985), p. 588.
11 Chaim Kaufmann, "Possible and Impossible Solutions to Ethnic Wars,"
International Security, Vol. 20, No. 4 (Spring 1996), pp. 136-175; and
Chaim D. Kaufmann, "When All Else Fails: Ethnic Population Transfers
and Partitions in the Twentieth Century," International Security,
Vol. 23, No. 2 (Fall 1998), pp. 120-156. Jaroslav Tir has also
suggested that peaceful partitions are extremely unlikely to produce
the onset of war. See Tir, "Keeping the Peace after Secession:
Territorial Conflicts between Rump and Secessionist States," Journal
of Conflict Resolution, Vol. 49, No. 5 (October 2005),
pp. 713-741. Although Tir’s addition to the literature is useful, it
does not address the problem of how to prevent the recurrence of civil
12 Alexander B. Downes, "The Problem with Negotiated Settlements to
Ethnic Civil Wars," Security Studies, Vol. 13, No. 4 (Summer 2004),
pp. 230-279; John J. Mearsheimer, "The Case for Partioning Kosovo," in
Ted Galen Carpenter, ed., NATO’s Empty Victory: A Postmortem on the
Balkan War (Washington, D.C.: Cato Institute, 2000), pp. 133-138;
Alexander B. Downes, "The Holy Land Divided: Partition as a Solution
to Ethnic Wars," Security Studies, Vol. 10, No. 4 (Summer 2001),
pp. 58-116; Kaufmann, "When All Else Fails"; Kaufmann, "Possible and
Impossible Solutions to Ethnic Wars"; Radha Kumar, Divide and Fall?
Bosnia in the Annals of Partition (New York: Verso, 1997); Radha
Kumar, "The Troubled History of Partition," Foreign Affairs, Vol. 76,
No. 1 (January/February 1997), pp. 22-34; Mansergh, The Prelude to
Partition; Robert Schaeffer, Warpaths: The Politics of Partition (New
York: Hill and Wang, 1990); Jan Tullberg and Brigitta S. Tullberg,
"Separation or
Unity? A Model for Solving Ethnic Conflicts," Politics and the Life
Sciences, Vol. 16, No. 2 (September 1997), pp. 237-248; John
J. Mearsheimer and Stephen Van Evera, "When Peace Means War: The
Partition That Dare Not Speak Its Name," New Republic, December 18,
1995, pp. 16-21; John J. Mearsheimer and Robert A. Pape, "The Answer:
A Partition Plan for Bosnia," New Republic, June 14, 1993, pp. 22-28;
and Peter W. Galbraith, The End of Iraq: How American Incompetence
Created a War without End (New York: Simon and Schuster, 2006).
13 Nicholas Sambanis, "Partition as a Solution to Ethnic War: An
Empirical Critique of the Theoretical Literature," World Politics,
Vol. 52, No. 4 (July 2000), pp. 437-483, at p. 479.
14 Quoted in ibid., p 479.
15 Alexander Downes notes this flaw in the Sambanis analysis in
Downes, "The Holy Land Divided."
16 Licklider, "The Consequences of Negotiated Settlements in Civil
Wars"; T. David Mason and Patrick J. Fett, "How Civil Wars End: A
Rational Choice Approach," Journal of Conflict Resolution, Vol. 40,
No. 4 (December 1996), pp. 546-568; T. David Mason, Joseph
P. Weingarten Jr., and Patrick J. Fett, "Win, Lose, or Draw:
Predicting the Outcome of Civil Wars," Political Research Quarterly,
Vol. 52, No. 2 (June 1999), pp. 239-268; Barbara F. Walter, Committing
to Peace: The Successful Settlement of Civil Wars (Princeton, N.J.:
Princeton University Press, 2002); and Barbara F. Walter and Jack
Snyder, eds., Civil Wars, Insecurity, and Intervention (New York:
Columbia University Press, 1999). Walter examines the difficulty of
states in credibly committing to peaceful coexistence with rebel
organizations after war, as it applies to all civil wars. See Walter,
Committing to Peace. For a discussion of the commitment problem as it
applies to ethnic conflicts in
particular, see James D. Fearon, "Commitment Problems and the Spread
of Ethnic Conflict," in David A. Lake and Donald Rothchild, eds., The
International Spread of Ethnic Conflict: Fear, Diffusion, and
Escalation (Princeton, N.J.: Princeton University Press, 1998),
pp. 107-126. For grievance-based arguments of ethnic minorities, see
Ted Robert Gurr, Peoples versus States: Minorities at Risk in the New
Century (Washington, D.C.: United States Institute of Peace Press,
2000). For a focus on individual motivations for joining rebel forces
and reducing these motivations, see Walter, "Does Conflict Beget
Conflict?" For the relationship between security force strength and
civil war, see Jeffrey Herbst, "African Militaries and Rebellion: The
Political Economy of Threat and Combat Effectiveness," Journal of
Peace Research, Vol. 41, No. 3 (May 2004), pp. 357-369, at
p. 358. Although Herbst does not explicitly claim to focus on war
recurrence, the countries within
his regional focus of Africa have experienced renewed warfare, and
therefore his conclusions about the need to strengthen state security
forces to prevent civil war are relevant to this debate.
17 Kaufmann, "Possible and Impossible Solutions to Ethnic Wars"; and
Kaufmann, "When All Else Fails."
18 The dangers of leaving both sides to "fight it out" are immense:
from a security standpoint, there is a real danger that the war could
spread, through diffusion or contagion, into a wider regional war;
from a moral standpoint, there is a threat of the mass killing of
civilians and potential genocide. For arguments on this topic, see
Lake and Rothchild, The International Spread of Ethnic
Conflict. Third-party peacekeeping has proven viable where a third
party has vested interests to commit troops and resources. Although
many analysts have argued the case for greater intervention, most
recognize that the preponderance of obstacles, national and
international, leaves little hope for such commitments even in
ethically justified interventions. For a normative argument in support
of greater intervention, see Stanley Hoffmann, "The Politics and
Ethics of Military Intervention," Survival, Vol. 37, No. 4 (Winter
1995-96), pp. 29-51.
19 For detailed discussion, see Horowitz, Ethnic Groups in Conflict,
pp. 41-54. See also Ashutosh Varshney’s discussion of broader and
narrower usages of the term "ethnic violence," in Varshney, Ethnic
Conflict and Civic Life: Hindus and Muslims in India (New Haven,
Conn.: Yale University Press, 2002). Varshney refers to "communal"
groups as "religious" groups, whereas Ted Robert Gurr defines them as
"cultural and religious identity groups." See Gurr, "Why Minorities
Rebel: A Global Analysis of Communal Mobilization and Conflict since
1945," International Political Science Review, Vol. 14, No. 2 (April
1993), pp. 161-201, at p. 161. See also James D. Fearon’s discussion
of ethnicity with its emphasis on descent and "social relevance" in
Fearon, "Ethnic Mobilization and Ethnic Violence," in Barry
R. Weingast and Donald A. Wittman, eds., The Oxford Handbook of
Political Economy (New York: Oxford University Press, 2006),
pp. 852-868.
20 Numerous definitions of ethnic war exist. The Political Instability
Task Force (formerly the State Failure Task Force), a
U.S. government-sponsored research project to build a database on
major domestic political conflicts, focuses on the political
mobilization of ethnic groups as the key factor, defining ethnic wars
as "secessionist civil wars, rebellions, protracted communal warfare,
and sustained episodes of mass protest by politically organized
communal groups." See Daniel C. Esty, Jack A. Goldstone, Ted Robert
Gurr, Pamela T. Surko, and Alan N. Unger, "Working Papers State
Failure Task Force Report" (McLean, Va.: Science Applications
International Cooperation, November 30, 1995). Ethnic wars may also
involve different goals, recruitment patterns, and in some cases,
forced population migration based on ethnicity (sometimes called
"ethnic cleansing"). Coding ethnic civil wars typically involves the
ethnic groups seeking changes in their status or
in government policies directed toward them, whether that be an end
to repression, increased power at the center, or secession. For a
comprehensive discussion, see Nicholas Sambanis, "Do Ethnic and
Nonethnic Civil Wars Have the Same Causes? A Theoretical and Empirical
Inquiry (Part I)," Journal of Conflict Resolution, Vol. 45, No. 3
(June 2001), pp. 259-282; and Nicholas Sambanis, "What Is an Ethnic
War? Organization and Interests in Insurgencies," unpublished paper,
Yale University, 2006.
21 Helen M. Hintjens, "Explaining the 1994 Genocide in Rwanda,"
Journal of Modern African Studies, Vol. 37, No. 2 (June 1999),
pp. 241-286; Kaufmann, "Possible and Impossible Solutions to Ethnic
Wars"; and William Seltzer, "Population Statistics, the Holocaust, and
the Nuremberg Trials," Population and Development Review, Vol. 24,
No. 3 (September 1998), pp. 511-552.
22 Some scholars do not agree. Kaufmann, for example, states that
"ethnic identities are fixed at birth." See Kaufmann, "Possible and
Impossible Solutions to Ethnic Civil Wars," p. 140.
23 Rogers Brubaker has been at the forefront of challenging his
colleagues to move beyond "constructivist cliches" that characterize
identities as multiple, unstable, contingent, and so on, and to
examine the ways in which the practice of reification works: "As
analysts, we should . . . try to account for the ways in which–and
the conditions under which–this practice of reification, this
powerful crystallization of group feeling, can work." See Brubaker,
Ethnicity without Groups (Cambridge, Mass.: Harvard University Press,
2004), p. 10. See also Frederik Barth, ed., Ethnic Groups and
Boundaries: The Social Organization of Culture Difference (Boston:
Little, Brown, 1969).
24 Hintjens, "Explaining the 1994 Genocide in Rwanda"; and René
Lemarchand, Burundi: Ethnic Conflict and Genocide (Washington, D.C.:
Woodrow Wilson Center Press, 1997).
25 Michael Fahy and Jonathan Mogul, "An Interview with Lidija Fekeza:
An Archeologist in Sarajevo: Culture under Siege," Journal of the
International Institute, Vol. 3, No. 1 (Fall 1995),
.109. Similar quotes can be
found in endless news reports, where some claim not to know which
ethnic group they belonged prior to the conflict. One such individual
stated, "Before this crisis I didn’t even know if I was a Serb or a
Croat." Quoted in Paul McGeough, "Fortress of Fear: First Battleground
of a Civil War?" Sydney Morning Herald, May 18, 1991. See also
Slavenka Drakulic, The Balkan Express: Fragments from the Other Side
of War (New York: W.W. Norton, 1993), pp. 50-52, quoted in Rogers
Brubaker, Nationalism Reframed: Nationhood and the National Question
in the New Europe (Cambridge: Cambridge University Press, 1996),
p. 20.
26 This is not to say that cross-ethnic appeals are impossible. As
other scholars have pointed out, even during ethnic wars it is
possible to find individuals from the rebellious ethnic group working
for the government. Kaufmann argues this may occur when there is an
"extreme power imbalance," which would be consistent with Stathis
N. Kalyvas’s argument that civilians will provide support to whichever
actor has military dominance locally, although most of his cases come
from colonial wars. I thank an anonymous reviewer for this last
observation. See Kaufmann, "When All Else Fails," p. 140; and Kalyvas,
The Logic of Violence in Civil War (New York: Cambridge University
Press, 2006). This may explain how the Russian government was able to
co-opt local rebel leaders and their followers (e.g., Akhmad Kadyrov
and the so-called Kadyrovtsy) only after taking territorial control of
most of Chechnya’s urban centers.
27 Kaufmann is the most vociferous about the impossibility of
reintegration. See Kaufmann, "Possible and Impossible Solutions to
Ethnic Wars." Ideological conflicts may sometimes display fixed or
inflexible identities. For example, during the Russian civil war of
1917-21, members of certain classes (e.g., "kulaks") were targeted for
deportation or death. For evidence of the extreme class-based nature
of the early Soviet state, see Nicolas Werth, Istoria Sovetskogo
Gosudarstva [History of the Soviet state] (Moskva: Izdatelstvo Ves
Mir, 2003), pp. 145-146. In the Colombian civil war of 1948-62, some
people believed that being a liberal or a conservative was genetic:
"They [Conservatives] cut the genitals off other men so that they
wouldn’t procreate any more Liberals." Former priest Walter
j. Broderick, quoted in Bert Ruiz, The Colombian Civil War (Jefferson,
N.C.: McFarland, 2001), p. 59. I thank an anonymous reviewer for
sharing this quotation.
28 Kaufmann, "Possible and Impossible Solutions to Ethnic Wars,"
p. 139.
29 Robert Jervis, "Cooperation under the Security Dilemma," World
Politics, Vol. 30, No. 1 (January 1978), pp. 167-214.
30 Barry R. Posen, "The Security Dilemma and Ethnic Conflict,"
Survival, Vol. 35, No. 1 (Spring 1993), pp. 27-47.
31 Some scholars have applied a similar logic to all states with
ethnic minorities, maintaining that the greatest potential threat to
any group is its own state, given the state’s capacity to kill. As
Stephen M. Saideman and his colleagues state, "The search for security
motivates groups in divided societies to seek to control the state or
secede if the state’s neutrality cannot be assured." Indeed, as the
advocates of partition argue, in an ethnic civil war the biased nature
of the state has already been demonstrated, and all threatened groups
must mobilize for self-defense. Saideman, David J. Lanoue, Michael
Campenni, and Samuel Stanton, "Democratization, Political
Institutions, and Ethnic Conflict: A Pooled Time Series Analysis,
1985-1998," Comparative Political Studies, Vol. 35, No. 1 (February
2002), pp. 103-129, at pp. 106-107 and p. 122. While the security
dilemma has received broad support as an explanation within the ethnic
war literature, its
advocates do not claim that it explains all ethnic violence. For
other explanations of ethnic violence, see Rogers Brubaker and David
D. Laitin, "Ethnic and Nationalist Violence," Annual Review of
Sociology, Vol. 24 (1998), pp. 423-452.
32 Hurst Hannum, "Territorial Autonomy: Permanent Solution or Step
toward Secession?" in Andreas Wimmer, Richard J. Goldstone, Donald
L. Horowitz, Ulrike Joras, and Conrad Schetter, eds., Facing Ethnic
Conflicts: Toward a New Realism (Lanham, Md.: Rowman and Littlefield,
2004), p. 280.
33 Tir, "Keeping the Peace after Secession."
34 Tir uses the territorial dispute data set developed by Paul K. Huth
and Todd L. Allee, which includes a variable for whether territory is
ethnically valued, that is, when a minority group shares ethnic traits
similar to those of the largest ethnic group within the challenging
state (including speaking the same language). See Huth and Allee, The
Democratic Peace and Territorial Conflict in the Twentieth Century
(Cambridge: Cambridge University Press, 2002).
35 Kaufmann, "When All Else Fails," p. 122.
36 Kaufmann states, "The critical causal factor is separation of
people into defensible enclaves, not partition of sovereignty."
Kaufmann, "Possible and Impossible Solutions to Ethnic Wars," p. 162.
37 Downes, "The Holy Land Divided"; and Alexander B. Downes, "More
Borders, Less Conflict? Partition as a Solution to Ethnic Civil Wars,"
SAIS Review, Vol. 26, No. 1 (Winter/Spring 2006), pp. 49-61.
38 Ibid., p. 61. Downes’s explanation moves partition beyond the
confines of the security dilemma alone.
39 Kaufmann has implicitly supported this notion of sovereignty by
examining only case studies of partition that included at least de
facto sovereignty, and more recently, has explicitly supported this
notion. See Kaufmann, "When All Else Fails," especially pp. 124-126;
and Chaim Kaufmann, "’Partition Theory’ in the Marketplace of Ideas,
and in Iraq," in Mia Bloom and Roy Licklider, eds., Living Together
after Ethnic Killing: Exploring the Chaim Kaufmann Argument (New York:
Routledge, 2007). The requirement for political sovereignty and
demographic separation negates some of the recent criticisms leveled
against partition. David D. Laitin, for example, relies on data from
the Minorities at Risk (MAR) data set focusing on the conflict-prone
nature of group concentration within a unified state, which suggests
that higher demographic separation would likely lead to increased
rebellion. Laitin does not, however, look at separation with
sovereignty and does
not look at full demographic separation–the MAR data force him to
include only higher and lower levels of group concentration within a
country. Laitin, "Ethnic Unmixing and Civil War," Security Studies,
Vol. 13, No. 4 (Summer 2004), pp. 350-365.
40 See Tullberg and Tullberg’s roundtable article where they outline a
"rational model" for solving ethnic conflicts that involves the
democratic choice by an ethnic group to secede. The authors argue that
"migration over the border between the newly formed states should be
part of such a solution." Tullberg and Tullberg, "Separation or
Unity?" p. 237. For a brief earlier overview of the history of the
partition debate, see Kumar, "The Troubled History of Partition." See
also Schaeffer, Warpaths, especially pts. 1 and 2.
41 Kaufmann wrote, "The international community should endorse
separation as a remedy for at least some communal conflicts; otherwise
the process of war will separate the populations anyway, at much
higher human cost." See Kaufmann, "When All Else Fails,"
pp. 122-123. Mearsheimer and Van Evera argue, "U.S. policymakers must
be willing at times to decide that states cannot be sustained and
should instead be disassembled. Only if we accept this reality
honestly and promptly will we have a reasonable chance of managing
their disassembly and keeping it relatively peaceful. Partition should
remain a last resort, but, regrettably, we still live in a world where
it is sometimes necessary." Mearsheimer and Van Evera, "When Peace
Means War," p. 21.
42 Population transfers are considered a violation of such human
rights as the freedom of movement (UDHR 13), the right to own property
(UDHR17), and the right to a family life (UDHR 12). For an examination
of population transfers in international law, see, for example, Alfred
M. de Zayas, "International Law and Mass Population Transfers,"
Harvard International Law Journal, Vol. 16, No. 2 (Spring 1975),
pp. 207-258; and Christopher M. Goebel, "Population Transfer,
Humanitarian Law, and the Use of Ground Force in U.N. Peacemaking:
Bosnia and Herzegovina in the Wake of Iraq," New York University
Journal of International Law and Politics, Vol. 25, No. 3 (Spring
1993), pp. 627-698.
43 Stories and images of the millions injured and dead that
accompanied the population transfers in postpartition Cyprus, India,
and Palestine are enough for many to condemn this method as
barbaric. Critics charge that this experience should be enough to
dismiss the idea that even "organized" population transfers are
possible. See Kumar, Divide and Fall? especially chaps. 1-3; and
Schaeffer, Warpaths, especially chap. 9. Advocates of partition
counter that the cases selected as evidence to refute its
effectiveness did not deal with the type they are recommending; they
either did not involve the necessary demographic separation or were
not implemented as solutions to ethnic civil war. For example, in each
of the cases highlighted by Kumar, organized population transfers were
not an integral part of the advanced decisionmaking process for
partition, arguably leading to a greater degree of bloodshed and
suffering. In addition, the outcome from the 1923
population exchange between Greece and Turkey, though far from
perfect, does suggest that not all transfers must result in the
catastrophic deaths seen after the partition of British
India. Kalliopi K. Koufa and Constantinos Svolopoulos present one
side, but for a critical assessment of the long-term difficulties of
the integration of refugees in this case, see Renée Hirschon, Heirs of
the Greek Catastrophe: The Social Life of Asia Minor Refugees in
Piraeus (New York: Oxford University Press, 1989); and Koufa and
Svolopoulos, "The Compulsory Exchange of Populations between Greece
and Turkey: The Settlement of Minority Questions at the Conference of
Lausanne and Its Impact on Greek-Turkish Relations," in Paul Smith,
ed., Ethnic Groups in International Relations (Hanover, N.H.:
Dartmouth, 1991), pp. 275-308.
44 Downes, for example, writes, "Implementing partition without
separating the groups in conflict to reduce or eliminate the number of
minorities left behind is sure to see them cleansed, or for conflict
over the intermingled region to continue. Examples of this problem
include Kashmir in India . . . and Northern Ireland." Downes, "The
Holy Land Divided," p. 74.
45 See, for example, Ward Thomas, The Ethics of Destruction: Norms and
Force in International Relations (Ithaca, N.Y.: Cornell University
Press, 2001), which includes an examination of aerial bombing of
civilian targets during war. For a review of literature examining
state compliance with international norms and treaties, including the
regulation of force during war, see Harold Hongju Koh, "Why Do Nations
Obey International Law?" Yale Law journal, Vol. 106, No. 8 (January
1997), pp. 2599-2659. For a dissenting view with empirical evidence
demonstrating no effect of international treaties on state behavior
for civilian targeting during war, see Benjamin A. Valentino, Paul
K. Huth, and Sarah Croco, "Covenants without the Sword: International
Law and the Protection of Civilians in Times of War," World Politics,
Vol. 58, No. 3 (April 2006), pp. 339-377.
46 Kaufmann, "Possible and Impossible Solutions to Ethnic Wars,"
pp. 173-174.
47 Sambanis, "Partition as a Solution to Ethnic War," p. 439.
48 These variables include gross domestic product per capita, cost of
the war as measured by deaths and injury, and the war’s outcome
(government victory, rebel victory, etc.). Ibid., p. 469.
49 Ibid., p. 439.
50 Ibid., p. 473.
51 Sambanis states that the outcome of partition depends on the
following criteria: whether the war ended in a treaty, the war’s
intensity, the number of people displaced by the war, and the number
and size of ethnic groups. Ibid., p. 478.
52 For consistency, I follow Sambanis’s coding for ethnic war. He
categorized such wars with reference to as many sources as he could
consult. See ibid., p. 455 n. 49.
53 See ibid., p. 444. The definition is relatively uncontroversial
except for its "1,000 deaths," which does not require an annual death
threshold, but rather "1,000 [battle] deaths for the duration of the
war." See Nicholas Sambanis, "Appendix B: Data-Set Notes" (Washington,
D.C.: World Bank, 2000), p. 2. For a detailed discussion about the use
of battle deaths in the quantitative, cross-national data set, see
Nicholas Sambanis, "What Is Civil War? Conceptual and Empirical
Complexities of an Operational Definition," Journal of Conflict
Resolution, Vol. 48, No. 6 (December 2004), pp. 814-858.
54 Peter Wallensteen and Margareta Sollenberg, "Armed Conflicts,
Conflict Termination, and Peace Agreements, 1989-1996," Journal of
Peace Research, Vol. 34, No. 3 (August 1997), pp. 339-358; and
Sambanis, "Appendix B."
55 Sambanis, "Partition as a Solution to Ethnic War," p. 445. These
partitions deliberately exclude instances of peaceful partition, such
as Czechoslovakia. For further examination of partition under
conditions of peace, see Tir, "Keeping the Peace after Secession."
56 Schaeffer, Warpaths, p. 5.
57 Debates regarding differences between secession and decolonization
also exist, but they do not advance the current debate on partition:
to suggest that the "separation" of Nigeria and the United Kingdom is
similar to the "separation" of Azerbaijan and Nagorno Karabakh
conflates two very different events and only obfuscates the partition
process in the current debate on ethnic civil wars. Further, the word
"colonization" in ethnic civil wars is highly contested. For example,
Chechen insurgents claim to be waging a war of liberation against the
"colonizing" center of Moscow, whereas Moscow claims the uprising is a
secession and sees Chechnya as an integral part of the Russian
Federation. In the military campaign beginning in 1999, Russia labeled
the Chechen insurgents no longer as secessionists but as bandits,
criminals, or Wahhabi radicals. For purposes of analysis, many
academics put partition, secession, and decolonization in the same
McGarry and O’Leary lump "partition and/or secession (self
determination)" together in their taxonomy, and include decolonization
within it. McGarry and O’Leary, The Politics of Ethnic Conflict
Regulation, pp. 11-16.
58 Kaufmann, "When All Else Fails," p. 126. The role of Turkey in
facilitating the de facto independence of the Turkish Republic of
Northern Cyprus has been well documented. For the critical role of
Russia in enabling Abkhazia’s de facto independence, see Monica Duffy
Toft, The Geography of Ethnic Violence: Identity, Interests, and the
Indivisibility of Territory (Princeton, N.J.: Princeton University
Press, 2003), pp. 87-106.
59 Some civil war databases, such as those used by Fearon, "Why Do
Some Civil Wars Last So Much Longer Than Others?" include wars of
decolonization. Others, such as Licklider, "The Consequences of
Negotiated Settlements in Civil Wars," do not. The Correlates of War
project separates these into "internal wars" and "extrasystemic wars."
Fearon and Laitin run their analysis both with and without wars of
decolonization when testing for causes of civil war onset, recognizing
conceptual and theoretical problems for both inclusion and
exclusion. Fearon and Laitin, "Ethnicity, Insurgency, and Civil War."
60 It was not clear from the Sambanis article or notes in Appendix B
why Tajikistan was coded as a partition or an ethnic civil war;
Tajikistan’s separation from the Soviet Union occurred before its war
began. Sambanis recognizes Tajikistan as a coding error in Nicholas
Sambanis, "Partition and Civil War Recurrence: A Re-Examination of the
Evidence," Yale University, 2006. For Tajikistan as a regional and
ideological conflict, see Payam Foroughi, "Tajikistan: Nationalism,
Ethnicity, Conflict, and Socio-Economic Disparities–Sources and
Solutions," Journal of Muslim Minority Affairs, Vol. 22, No. 1 (April
2002), pp. 39-62; and Dov Lynch, "The Tajik Civil War and Peace
Process," Civil Wars, Vol. 4, No. 4 (Winter 2001), pp. 49-72.
61 Sambanis states, "Bosnian partition from Yugoslavia in 1992."
Sambanis, "Appendix B," p. 43. He recognizes some of these issues in
Sambanis, "Partition and Civil War Recurrence."
62 The General Framework for Peace in Bosnia and Herzegovina was
initialed in Dayton, Ohio, on November 21, 1995, and signed in Paris
on December 14, 1995.
63 Inclusion of the Republika Srpska partition from Bosnia also means
that I added the Bosniak-Croat dyad as a case of ethnic war ending
without the partition of sovereignty. This case does not appear in
Tables 1 and 2, which include such partitions, but it does appear in
the later comparison between partitions and other war outcomes.
64 The two republics are Republika Srpska and the Federation of Bosnia
and Herzegovina. This is not a case of territorial autonomy because
there are separate governments with armed forces that cannot enter
each other’s territory. The primary conflict was between Serb forces,
on the one hand, and Croat and Bosnian forces on the other, although
the Croat and Bosnian forces also fought each other from mid-1993
until the signing of the Washington treaty of March 18, 1994, after
which they fought together against Serb forces. Other possibilities
therefore include separate Bosniak-Serb and Croat-Serb codings for
partition, but the figures for separation are virtually the same and
do not affect the results, except to provide an additional
"partition." Further, given the conflict between Croat and Bosniak
forces, one could include this as an ethnic war without partition as
the outcome. Again, these data do not affect the final results when
comparing partition to
65 Measheimer and Walt, "When Peace Means War," p. 16.
66 RSK had a separate government and armed forces.
67 See, for example, military operations in both the Lika region of
RSK (Operation Medak Pocket, September 1993) and the Maslenica and
Zadar regions of RSK (Operation Maslenica, January 1993). By the time
of the next cease-fire, in 1994, Croatian forces were already
preparing Operation Flash, which began in May 1995.
68 The Yugoslav National Army was heavily involved in the wars for the
RSK, as was Slobodan Milosevic.
69 This is, in fact, how most Croats understand the war (commonly
labeled "Domovinski Rat" in Croatian) from 1991 to 1995. In "Partition
and Civil War Recurrence," Sambanis excludes Croatia altogether, which
I find surprising given that the definition he uses for partition in
this paper is "an outcome of a civil war that . . . leads to the
formation of a new state out of a part of another state." This is what
occurred in Croatia, where war began in 1991. See Sambanis, "Partition
and Civil War Recurrence."
70 Tatu Vanhanen, "Domestic Ethnic Conflict and Ethnic Nepotism: A
Comparative Analysis," Journal of Peace Research, Vol. 36, No. 1
(January 1999), pp. 55-73.
71 Daniel N. Posner, "Measuring Ethnic Fractionalization in Africa,"
American Journal of Political Science, Vol. 48, No. 4 (October 2004),
pp. 849-863.
72 The Minorities at Risk Project follows a similar guideline when
aggregating groups vis-A-vis the government. In Darfur today, for
example, MAR codes the "Black Muslims of Darfur" as a group, even
though it comprises three different groups: Fur, Masalit, and
Zaghawa. The same formula is used for the "Southerners" group in
Sudan, which comprises Anuaks, Azande, Dinkas, Equatorians, Latukas,
Madi, Moru, Nuers, Shilluks, Taposas, and Turkans. See Minorities at
Risk Project, "Assessment for Southerners in Sudan" and "Assessment
for Darfur Black Muslims in Sudan" (College Park: Center for
International Development and Conflict Management, University of
Maryland, 2005),
73 There are different ways to calculate the PEHI. One alternative is
to look at the separation from both sides by including an indicator of
the percentage of the original majority found in the minority region
prior to the war (e.g., ethnic Russians in Chechnya before 1994),
which I label MiM (majority in minority region), and then to calculate
the index as:
[(MiM+OSM) – (RSM+NSM)]/(MiM+OSM).

I conducted a sensitivity test using this formula, and others, and
found no substantive differences in the results: those cases with high
degrees of unmixing scored highly using all formulas.

74 The staple set consisted of Encyclopedia Columbia, 2001;
Encyclopedia Britannica, 2003; Patrick Brogan, The Fighting Never
Stopped: A Comprehensive Guide to World Strife since 1945 (New York:
Vintage, 1990); Guy Arnold, Wars in the Third World since 1945
(London: Cassell, 1995); Economist and the Economist Intelligence
Unit, ; CIA World Factbooks; Lexis-Nexis
Academic; and International Crisis Group reports.
75 For some conflicts, I needed to use refugee flows to calculate
minority percentages, in which case prewar minority percentages were
used to obtain absolute numbers of the minority, and refugee numbers
were subtracted from the total to arrive at an approximation of the
minority remaining in the territory. Where large refugee movements
occurred–many of these conflicts forced hundreds of thousands, of
people from their homes–exact numbers were impossible to obtain, so
approximations were required.
76 This accepts the inevitability of small, residual minorities that
do not alter the value of the results. The average size of the largest
residual minorities found after complete partitions amounted to a mere
0.33 percent. Kaufmann argues, "While peace requires separation of
groups into distinct regions, it does not require total ethnic
purity. Rather, remaining minorities must be small enough that the
host group does not fear them as either a potential military threat or
a possible target for irredentist rescue operations." Kaufmann,
"Possible and Impossible Solutions to Ethnic Wars," p. 163.
77 Sambanis uses postwar democratization as a third criterion and
finds postpartition states associated with higher levels of
democracy. This article does not address these results because they do
not form the core of the partition theory argument. Sambanis,
"Partition as a Solution to Ethnic War," pp. 459-464.
78 Collier, Hoeffler, and Soderbom, "On the Duration of Civil War."
79 On the civilian expulsions, Minority Rights Group reported that by
early 2000 more than 54,000 Ethiopians of Eritrean origin had been
deported. Kjetil Tronvoll, "Ethiopia: A New Start?" (London: Minority
Rights Group, 2000). Amnesty International reported in 1999 that "the
expulsion of people of Eritrean origin was often carried out in an
inhumane manner that amounts to cruel, inhumane, and degrading
treatment." Amnesty International, "Ethiopia and Eritrea: Human Rights
Issues in a Year of Armed Conflict" (New York: Amnesty International,
1999), p. 27.
80 The Ethiopian-Eritrean war of 1998-2000 resulted in more than
100,000 battle deaths.
81 Operation Storm led 200,000 Croatian Serbs to flee into neighboring
Serbia and Bosnia. See Amnesty International, "Croatia: Operation
‘Storm’–Still No Justice Ten Years On" (New York: Amnesty
International, August 4, 2005), pp. 1-3.
82 The case of Bangladesh is deceiving due to the large Bengali
population that was largely separate from the rest of West
Pakistan. Nevertheless, Urdu-speaking Biharis were the targets of
violence, resulting in tens of thousands of deaths. A Pakistani white
paper on the topic estimated that more than 60,000 Urdu-speaking
Biharis were killed during the brief conflict. Government of Pakistan,
White Paper on the Crisis in East Pakistan, August 5, 1971,
83 Because the objective is to minimize human suffering, population
transfers should be kept to a minimum, which would suggest redrawing
borders around the concentrated Serb populations in the north of
Kosovo. The other Serb enclaves could be given the option to move, but
their small size would not significantly influence the PEHI, leaving
Kosovo as a complete partition.
84 The events of March 2004 led to widespread violence, death, and the
"ethnic cleansing of entire minority villages and neighbourhoods." See
International Crisis Group, "Collapse in Kosovo," Europe Report,
No. 155 (Brussels: International Crisis Group, April 22, 2004).
85 Proposals for such a further partition were floated in the summer
and autumn of 2007 among European states. See Radio Free Europe/Radio
Liberty, "Dutch Minister Says Partition of Kosova Acceptable"; and
Radio Free Europe/Radio Liberty, "Russia Rules Out Crossing ‘Red Line’
on Kosova, but Would Accept Partition."
86 Bilefsky, "In a Divided Kosovo City, a Resounding Vow to Remain
Part of Serbia."
87 Kaufmann, "Separating Iraqis, Saving Iraq"; Patrick Cockburn, "Iraq
Is Disintegrating as Ethnic Cleansing Takes Hold," Independent, May
20, 2006; Kathleen Ridolfo, "Iraq: Displacement Crisis Worsened by
Violence," Radio Free Europe/Radio Liberty, April 21, 2006; and the
special issue of Forced Migration Review, "Iraq’s Displacement Crisis:
The Search for Solutions," June 2007.
88 Most analysts agree that at least three different civil wars were
ongoing in Iraq in 2007; partition relates only to the intercommunal
89 Peter W. Galbraith states, "The case for the partition of Iraq is
straightforward: It has already happened." Galbraith, "The Case for
Dividing Iraq," Time, November 5, 2006,
, 9171,1555130,00.html. More
generally, there is a lack of solid information about the degree of
homogenization that has occurred. On the trouble reporting even body
counts in the war, see Clark Hoyt, "The Reality in Iraq? Depends on
Who’s Counting," New York Times, October 7, 2007.
90 The Iraqi Red Crescent and the International Organization on
Migration show a complicated picture. As journalists James Glanz and
Alissa J. Rubin stated in September 2007 after reading those
organizations’ reports, "Displacement in the most populous and mixed
areas is surprisingly complex, suggesting that partitioning the
country into semiautonomous Sunni, Shiite, and Kurdish enclaves would
not be easy." Glanz and Rubin, "Future Look of Iraq Complicated by
Internal Migration," New York Times, September 19, 2007. On Kirkuk,
see, for example, International Crisis Group, "Iraq and the Kurds:
Resolving the Kirkuk Crisis," Middle East Report, No. 64 (Brussels:
International Crisis Group, April 19, 2007).
91 Kaufmann advocates voluntary transfers for Iraq. See Kaufmann,
"Separating Iraqis, Saving Iraq."
92 Kaufmann has suggested that some militias may support the
transfers, thus reducing the burden on Iraqi and U.S. forces, but this
is far from certain given the large number of actors involved. He has
also pointed out that some militias might oppose the transfers. Ibid.,
p. 159.
93 See, for example, Posen, "The Security Dilemma and Ethnic
Conflict"; Toft, The Geography of Ethnic Violence; and Gurr,
Minorities at Risk.