Security Studies Program Seminar

How to Evaluate Communal Partitions

Chaim Kauffman
Associate Professor, Lehigh University

October 11, 2006

There is a timeliness to this presentation given the risk of Iraqi ethnic partition/separation. The question becomes one of assessing whether separation of warring populations is the best – or only – way to reduce deaths in communal civil wars and, if so, should the U.S. impose such separation on Iraq.

To an extent, this is unimportant because Iraq is partitioning itself. What drives this trend?

The logic of communal partitions is straightforward: whenever communities in multiethnic states cannot rely on the central government to prevent civil strife, they mobilize for self-defense and thus become a “pocket international system.”

The security dilemma thus fully applies to explain ethnic mobilization (an idea borrowed from Prof. Posen's 1993 Survival piece). Herein, the problem is that the material and rhetorical methods used for defense of the ethnic community poses an offensive threat to others. If and when we get to the point where all sides have mobilized for war and/or started conflict, it is immaterial as to what the original intent of the mobilization focused on.

The severity of the security dilemma varies: the more intermixed populations are, the greater the offensive opportunities of each sides for both organized forces, bandits and terrorists and thus the more difficult it is to organize community defenses short of cleansing areas of other ethnic groups to create homogeneous territorial zones.

The worst possible situation is to be situated in areas more individuals of a given ethnicity both exposed and threatening (i.e. ethnic pockets).

The security dilemma viz. population geography is insufficient for causing violence (other forces apply – greed, windows of opportunity, etc.), but it is useful for explaining the overarching process of mobilization and violence. Instead, the special value of security dilemma lens is to explain how communal wars play out – where worst atrocities are committed, when, why – and how de-escalation will play out.

In the Bosnia and Israel examples, ethnic cleansing occurred in those areas posing security threats to the Serbian and Jewish communities (respectively, e.g. road junctions) and not elsewhere.

These examples demonstrate that, until the security dilemma is mitigated and thus, the populations separated, peace negotiations are irrelevant.

Communal wars are path dependent – when one ethnic group goes to war, so do the others. Subsequently, you can't remix populations after conflict terminates because identity is defined by extremists and so polarized. Also, no power on earth can talk people out of remembering atrocity memories. Extremists thus gain leverage over moderates for a long time after the conflict.

Per the study of Carter Johnson (U Maryland), there have been 18 civil wars with partitions between 1945 and 2004, 6 of these can be called “good separations,” in the sense that the warring populations were mostly separated; 12 were “bad” partitions (this characterization is mine, not Johnson's). Of the good ones, none led to full-scale war resumption within 5 years of conflict termination via separation; lower levels of violence are similarly nonexistent.

Note also that separation is not, as far as I know, a global trend; war un-mixes populations but peace mixes them as people move in search of opportunities.

There are no clear or agreed standards of evaluating the outcome of partitions. Few have proposed standards.

Need to answer a normative question: what is the value we are pursuing? I propose “minimization of loss of life.” Use partition to support this goal.

Yet if people don't want to move, do you force them out? If you need a military intervention, how do you weigh US soldier lives versus local lives? What if there is a competing national interest such that leads us to anticipate more ethnic lives would be lost in support of the national interest than would be the case without the national interest?

I propose we use a conservative standard of having saved lives if: 1) absent partition, a civil war or large ethnic cleansing campaign would almost assuredly begin; 2) such a campaign would cost more lives than intervention/partition; 3) even if partition costs more lives in the short-run, it would save lives in the long run.

We can use counterfactuals (ex post facto) and context (in the moment) – despite admitted probabilities in both – to decide on such a course of action.

Example of India using the 3 criteria. If the UK had not decided to partition, could the British have saved lives later on (i.e. what was the latest date for setting the trajectory of this case on a different path so as to avoid civil war and communal violence?) By this rule, October 1946 – when the Muslim communal militias are integrated – is the latest possible date to have avoided the violence; this may have even been a bit earlier when the Indian National Congress failed in the 1937 elections to garner Muslim support.

The spiral of communal violence was set even if we cannot formally agree as to when the last escape moment occurred. Moreover, given the ethno-political strategies of local actors in the Punjab, Kashmir, and Bengal, violence would have been more widespread post-independence without the partition that did occur because the locations of final boundaries would be determined not in advance by Congress and the Muslim League's mutual acceptance of partition, but afterwards, but the outcome of fighting over territory. Each group would have armed willy-nilly and sough to maximize their control over territory. Third, while wars would have been worse in short-run, they would ALSO be worse for long-run prospects; Muslims in India would be perceived as a real security threat, instead of merely an underprivileged minority and truce lines would not have been permanent, so increasing the risk of territorial war ad infinitum.

In Iraq, meanwhile, the South is the problem.

By eliminating order in Iraq in 2003, we created perfect conditions to bring the security dilemma to the community level. Our soldiers have slowed the escalatory process but can't stop it.

There are many reasons to be skeptical. Neither community can trust the intentions of the other. Balance of Sunni and Shi'ite capabilities are relatively even. Presence of US forces has tended to encourage both sides – Shi'ite do not realize the full costs of the war (US as protector) while the Sunnis believe they can easily defeat the Shi'ite if the US withdraws. There is very little constituency for meaningful economic-political power sharing. Sunnis believe Maliki is being overly flexible (weak) in suppressing death squads by not using force. Neither community has little authority to enforce peace agreement on their followers. Zone of intermixed settlement around Baghdad is 100+ miles deep – plenty of opportunities for violence.

War has passed the tipping point as no one in either community can stop the death squads (i.e. the polarizing groups) – this hit no later than March 2006.

Everyone in central Iraq is thus intensely concerned with personal security – level of fear is so high that even rumors of possible attacks is pushing whole villages into flight.

That most Iraqis say they don't want partition is unimportant – security dilemma is pushing a de facto separation and partition “even if part of it [the ultimate dividing line] runs through Baghdad.”

Partition, however, would save lives.

Can/should US play a role in this? We don't have to act – Iraqis doing it themselves. The people on the ground DON'T want us to negotiate a partition. There is no constituency among Shia for meaningful power sharing.

We cannot determine the future politics of Iraq.

What is left that we can usefully do is protect refugees, e.g. by identifying the most exposed neighborhoods/villages and helping people at risk of ethnic cleansing move in advance.


 Chaim Kaufmann is Associate Professor of International Relations at Lehigh University.

Rapporteur: Joshua Shifrinson

back to Wednesday Seminar Series, Fall 2006