Columbia International Affairs Online

CIAO DATE: 9/5/2007

Why Iraq Partitioned Itself

Primary Oil Producing Fields

Geographical Distribution of Iraqi Oil Fields And Its Relation with the New Constitution

Iraq Index Tracking Variables of Reconstruction & Security in Post-Saddam Iraq

Why Iraq Partitioned Itself

Chaim Kaufmann
Lehigh University

August 2007

Columbia International Affairs Online


Iraq is in the process of partitioning itself into Kurdish, Sunni Arab, and Shia Arab cantons.  The separation of populations is driven by mutual fear – ethnic cleansing creates fear that begets more ethnic cleansing.  The data is uncertain, but anecdotal evidence suggests that this process may be considerably more than half complete.  The U.S. conquest of Iraq started the civil wars, but neither the U.S. nor any outside force can now assuage Iraqis' fears of each other or stop the ongoing partitions.  What we can and must do is protect continuing refugee movements and contribute to the relief of those who have already fled their homes.  The U.S. must also negotiate with Iraqi factions, neighboring powers, and world powers to improve the chances that the de facto partition of Iraq will be followed by relative peace instead of by ongoing war.

I. Why Study Iraq

Until recently, Western debates about communal civil wars have been motivated mainly by humanitarian concerns.   We wanted to know what, if anything, we could do to prevent, dampen, or stop them.   While we recognized that mistakes by well-meaning foreign powers could damage prospects for avoiding communal wars or for ending them once started, we took for granted that their main causes were internal or regional.

Iraq is different.   Many communal wars have killed more people or lasted longer than Iraq has so far, including some that are ongoing.   Iraq is, however, by far the worst civil war caused directly by a U.S. policy choice and the worst caused by any major power since the Soviet invasion of Afghanistan in 1978.    While many American and foreign policy makers and observers realized in 2003 that overthrowing the government of Iraq risked worsening communal tensions or even provoking civil war, few imagined the scale of killing that has happened.

Why did Iraq collapse into communal war after it was conquered by the United States?   Why has the war become as large and destructive as it has?   Could the disaster have been avoided by better occupation policy?  What opportunities and risks remain?  Answers to these questions are needed to formulate future U.S. policy not only toward Iraq but towards all states that are afflicted by serious communal tensions.  These in turn must be based on our best comparative understanding of the causes of outbreak, escalation, and de-escalation of communal conflicts—including, but not limited to, the causes of the Iraq disaster. 

Our three main theories of communal conflict imply different expectations for what should have happened in Iraq, different policies that we should have or should now follow, and different expectations about how future U.S. policy might cause, worsen, or ameliorate other communal conflicts.  It matters which models can provide the best accounts of what has happened in Iraq.

Iraq is a reasonably strong test of one main theory of communal conflict, namely rational choice.  This is so because, despite the widely shared belief that the U.S. administration simply failed to think about the consequences of regime change, in fact several of the most important U.S. policy decisions about Iraq were fairly close matches for prescriptions that follow from rational choice logic—especially over-optimism both ex ante and while formulating policies to contain the violence after it began.

Iraq provides only a weak test of constructivism, both because the administration did not base policies according to this logic and because the theory is designed more to explain the impact of the outcomes of debates about communal identities than to predict those outcomes.

For security dilemma theory Iraq is relatively straightforward, since the situation in 2003 resembled other cases in which the theory correctly predicted escalation to high levels of violence.  The theory's independent variables identify conditions that could avoid escalation of inter-communal security dilemmas—such as preserving rather than destroying central order—but these conditions were not present in Iraq.

For the purposes of our analysis, it is useful to separate the violence in Iraq into two distinct wars, one in the North between Kurds and several other communities, including Sunni Arabs, and one in the South between Shia and Sunni Arabs.[1]  The wars are taking place in two separate bands of communal mixed population that intersect hardly at all.  Although both Kurds and Shia claim to be upholding the Iraqi government against rebels, in fact the government has little sway over most Shia armed forces even when these are nominally army or police, and none over the Kurds.[2]  The Sunnis are sufficiently fragmented that is difficult to tell what share of Sunni mobilization on each front is purely local and what share represents coherent efforts by some factions to fight a two-front war. 

The situation is most reminiscent of World War II, which for many purposes we think of as two separate wars.  Although Japan was formally allied to the European Axis, the 'partners' objectives were unrelated and coordination and mutual assistance were minimal.  On the Allied side the United States and Britain fought on both fronts, but others—China, Russia, and many smaller powers—fought only in one theater.[3]  The situation in Iraq is analogous, with the Kurds and Shia in the role of the very loose Axis and Sunni factions in the role of Allied powers who vary in their attention to local versus global objectives.  A second reason for considering the two Iraqi civil wars distinct is that, as in the case of World War II, we can imagine one war occurring without the other; at least two of the three models of communal conflict are consistent with this possibility.

The remainder of this paper is divided into three sections.  The first lays out the expectations that can be drawn from each of the three theories for whether large-scale communal war in Iraq should have occurred.  The second assesses the performance of the three models in predicting the trajectories of the communal wars once begun, up to the present. The third lays assesses policy recommendations for Iraq today, drawn from both theory and from current U.S. policy debates, and offers suggestions.

II. Competing Expectations for the Wars in Iraq

Broadly, the three models predict different expectations for communal peace or war in Iraq, both in the period shortly after the 2003 conquest and over the years since.  While none gets everything right nor everything wrong, broadly speaking the security dilemma model performs best and rational choice worst.  The constructivist model has less ability than the others to make point predictions about the Iraq case—too much depends on developments in discourse within each community after April 2003—but what predictions the theory could make did pan out more often than not.  In addition, even when the constructivist model does not make specific predictions it identifies important sensitivities, predicting fairly well the impacts of developments in political discourse after 2003.

Rival Logics

Rational choice, constructivist, and security dilemma theories of communal conflicts emphasize different variables in order to form expectations for what might occur when central authority in a communally divided society is removed or destroyed. 

Rational choice focuses on opportunities, for both elites and ordinary individuals, to make material gains by criminal activities, even if these activities go under the name "rebellion." 

Constructivism takes a "top-down" approach, in which the primary driver of outcomes is elite competition for political support pursued mainly through discourse aimed at solidifying or transforming individual identities in ways that favor the political ambitions of those elites.  Masses operate essentially as "price takers," choosing from the menu of identities and policies presented to them by competing elites.

Security dilemma theory relies on a "bottom-up" approach in which individual security fears, based on visible mobilization of other communities, actual violence, and awareness of mutual vulnerability in communally mixed areas, determine the degree of pressure on communal elites to support mobilization and hard-line policies toward threatening communities.  In this model elites are price takers, constrained to satisfy mass fears.[4]

Rational Choice

The rational choice model, although consistent with a fair amount of banditry in Iraq due to the reduction in centrally-imposed order in April 2003, should have expected relatively little specifically communal violence.  There are three variants of the model that are most prominent in the literature on managing civil wars.  The first is the 'economic model' associated with Paul Collier and Anke Hoeffler, which focuses on opportunities for material gain.  As explained by Collier, Hoeffler, James Fearon, and David Laitin, the economic model focuses on availability of opportunities for elites and for ordinary individuals to make material gains (and attendant risks) by becoming "active rebels" versus opportunities and risks that accompany supporting the government or remaining neutral.[5]

The second is Stathis Kalyvas' model of civil wars as not actually wars but disorder providing opportunities for individual score-settling.[6]  Third is Fearon and Laitin's model, "Explaining Interethnic Cooperation," that explains how communities use information asymmetries to forestall inter-communal violence.[7]

In all three models factors such as the depth of existing communal divisions, identity construction through political discourse, population geography, and individual security concerns should all have been less important.

The economic model would expect that elites in the three main Iraqi communities should have been most concerned with obtaining material gains, both for themselves and to demonstrate to supporters their ability to deliver rewards. The likelihood of rebellion should also be heavily dependent on elite access to resources that could be used to fund rebel forces[8]—what we sometimes call 'diamond wars.'  Since, in the short run, Iraq's main exploitable resources were the oil reserves in Kurdish- and Shia-dominated regions, with virtually none in mainly Sunni regions, the likelihood of Sunni rebellion should have been low.[9]  Sunni elites' main goal should have been to obtain oil revenue sharing.  Strategies for this purpose could have included some mobilization to present a credible threat of potential costs of refusing Sunni demands, but this should have been calibrated to avoid appearing irreconcilable or inflicting so much damage as to generate incentives for large-scale suppression efforts. 

Kurdish and Sunni elites, for their part, should have been willing to pay for Sunni quiescence; their main concerns should have been how much the oil-rich communities would have to pay and how to arrange burden-sharing between them.  This should have been the dominant issue in national politics, in both formal and informal discourse between and within both communities.  After nearly four years, there has been some high-profile discussion of this issue, but no progress and little visible discussion of terms; most discourse has not got beyond disagreement over the basic principle.

Following material interest logic, Larry Diamond, a senior advisor to the U.S. occupation command in Iraq in 2003-2004, argued in February 2003 that "building democracy—or at least orderly and responsible governance—in a post-war Iraq must start in partnership with the Iraqi people and the internal community. …  Concessions for the production of future Iraqi oil would be made transparent by this emerging Iraqi authority."[10]  In April, Diamond called for speedy formation of an interim Iraqi government drawn "from all major regions and ethnic and religious groups and from both within the country and from exile…. it would give each of them a stake in the process, muting incentives to attack or undermine it.  By making existing armed groups (the regular military and the Kurdish and Shiite opposition forces) part of the process, it would go a long way toward preempting one of the most intractable problems in post-conflict situations: violent spoilers who resolve to get a better deal by force."[11]

By 2006, Diamond's views were mostly unchanged: he argued that the way to get democracy is that "over time, with good institutions and some luck, politicians will at least behave as if there were democrats [even though] many leaders may embrace it for selfish cynical, and purely tactical reasons."  Genuine commitment to democracy is not needed.[12]  Economic interests of Iraqi factions are more important than ethnic or sectarian boundaries: important stakeholders, "Iraq's oil wealth could lubricate the strains of coalition politics.  Then the ruling coalition in Iraq could resemble the one in Nigeria: diverse in ethnic, regional, and religious composition; constantly under strain; and held together by the same interests that often threaten split it apart – the desire of each group to get a fair share of the national pie."[13]

Iraq can become a functioning democracy by "drawing in to the political process "a wide range of relevant actors, [meaning] "any group that can mobilize violence."[14] "Negotiations must take place with many of the groups that are waging or supporting the insurgency [which] "will guarantee each regional group its share of national wealth and autonomy over its own affairs while preventing and one group from dominating the center."[15]

"Such an agreement [can] be reached in a manner consistent with democracy. … Many Sunni groups appear to understand that the days of Sunni hegemony in Iraq are over."  Although Sunni leaders "called for postponement of the [January 2005] elections, their reasons appeared mainly pragmatic: "they did not demand the withdrawal of American forces or dissolution of the interim government" as pre-conditions, but focused their demands on measures that would improve Sunni parties' chances, including impartial international oversight, release of political prisoners, lifting of political disabilities on ex-Ba'athists, and more time to organize politically.[16] 

Sunni parties did participate energetically in the December 2005 elections, but this did not lead to bargains of the sort that Diamond imagined. 

Diamond himself identifies a main reason why not: "In today's world, nationalism and anti-colonialism run deep. … It was the failure to comprehend these dynamics …that was perhaps the single greatest mistake of the U.S. intervention."[17]  "In immediate post-conflict situations, insecure citizens often turn to nationalists and ethnic … parties [while] more moderate parties tend to get squeezed out."[18]

Stathis Kalyvas' argument that most civil war violence is actually local score-settling, not 'war' with a coherent 'master cleavage,'[19] does not match events in Iraq.  Armed groups in Iraq have engaged in simple criminal activity, such as car theft and kidnappings for ransom.  The majority of the violence, however, and especially most of the killings, have been across communal boundaries, not within communities.  Further, most of the communal violence has been indiscriminate and anonymous, with victims being murdered purely on sectarian identity, without any attempt to identify individual victims' responsibility for anything—a possibility that should be rare under Kalyvas' model.[20]   For his part, as of June 2007, Kalyvas does not grant that there are communal civil wars ongoing in Iraq.[21]

The third important contribution of rational choice theorizing to the management of communal conflict, and the most directly relevant to Iraq is Fearon and Laitin's model of 'in-group policing' that seeks to explain the mechanisms that determine conditions under which we find either intercommunal peace or violence.[22]  Simply, communities police themselves to prevent inter-communal violence.[23]  By assumption, members of each community can track each others' activities well enough to identify offenders, although outsiders cannot.[24]  Thus each community must punish its own cross-communal offenders because otherwise offended members of the other group, unable to identify their specific abusers except as members of the first group, are likely to retaliate indiscriminately.[25]  

It is unimportant whether or not there is a government, either independent of both communities or controlled by one of them.  Authority structures capable of imposing punishment must exist within communities, although their creation is normally not difficult because almost everyone has an interest in in-group policing;[26] for the same reason, communal authority(-ies) who punish cross-group offenders need not agree on anything else.  The importance of in-group policing increases the weaker the community compared to others who could retaliate massively—in Iraq, such incentives should have been strongest for Sunnis.[27]  Interestingly, the logic relies on the existence of communal groups or functional equivalents; otherwise the expected level of violence in society would be higher.[28]

Mainstream Sunni elites should have concentrated on preserving intra-communal order both to avoid provoking other groups and because no Sunni warlord could have delivered much to supporters unless able to mobilize enough weight to matter in national-level politics.  The last thing that Sunni elites should have done would have been to promote or permit unconditional escalation of violence as well as actual efforts to exacerbate sectarian tensions; there should have been no room for someone like Abu Musab al-Zarqawi or 'al-Qa'ida of Mesopotamia.'  Such extremists should have suppressed by communal rivals or informed upon.  But in-group policing failed in Iraq.

Unfortunately, none of Collier, Hoeffler, Fearon, or Laitin offered much in the way of rational choice-based policy prescriptions for Iraq before 2003 or, since then, assessment or policy analysis.[29]  The most important exception to this is Fearon's March 2007 Foreign Affairs article, based largely on September 2006 Congressional testimony.  Fearon explains that self-policing failed because of communal factionalization, both "at the national political level and at the level of neighborhood militias and gangs."[30]  Fearon blames the Bush administration for this: "The worst decision, in my view, was the failure to take account of the fact that by decapitating Hussein's regime the U.S. invasion would be introducing anarchy, so that without a very large initial force presence …. the chances of stability would be extremely low."[31] 

These statements appear to include two distinct arguments, one based on a misconception and one possibly valid.  The "large force presence" appears to be a reference to the pre-war suggestions of some experts that it would take 500,000 to 700,000 troops to maintain peace in occupied Iraq.[32]  This was never an option, however; the U.S. lacked, and still lacks, deployable ground forces much larger than roughly 150,000 that we have maintained in Iraq on average since March 2003.[33]  (Fearon might reasonably respond that in that case invading Iraq at all was a bad idea, a sentiment with which security dilemma and constructivists theorists would likely agree because of loss of central order and creation of opportunities to demonize the U.S. and local allies respectively.)

"Decapitation" may be a reference to the May 2003 decision to disband the Iraqi Army and stop soldiers' pay, as well as banning from government jobs some 30,000 to 50,000 high-ranking Ba'ath party members, amounting to much of the Sunni professional class.  This created conditions for formation of many independent armed groups among Sunnis, fragmenting authority within the community. 

As Fearon and Laitin point out, the success of in-group policing depends on feasibility, although neither of the two dangerous scenarios that they identify was or is applicable to Sunni Arabs in Iraq.[34]   Fearon may here be appealing to a third possibility, namely that cross-communal attackers might, even if few compared to the size of their own community, be costly to sanction because they are well-armed; this would fit the situation of Iraqi Sunni Arabs in May 2003. 

This does not, however, get the theory out of the woods because high retaliatory violence (expected or already occurring) also increases the incentives of all community members to sanction offenders and to build military capabilities and institutional structures strong enough to do it.  Further, even if we accept that Sunni fragmentation in 2003 made a temporary gap in in-group policing unavoidable, that does not explain why the violence should have escalated for four additional years given the unilateral incentives inside each community to police themselves.  If Sunni moderates were unable to carry out in-group policing unaided, as rational actors they should have sought aid from powerful outsiders, such as providing intelligence to help U.S. or government forces weed out extremists for them, or even bringing in other Sunni powers.[35]  Moderates should also have sought oil money from the other communities, the U.S., or both to gain more power to recruit supporters against extremists and to weaken die-hards' ability to hold onto their own sources of support.[36]  The U.S. should have been willing to pay to move toward peace, as should Kurdish and Shia elites simply to save lives in their own communities.[37]

Over time, then, the theory predicts, any fragmentation that stood in the way of in-group policing should have tended to resolve itself.  This should fail only if the costs of in-group policing are dominated not by the distribution of material power within the community but by legitimacy—if communal loyalties make it illegitimate to punish members for violence against the 'enemy' community—or by expectations of non-contingent violence from the other community.  In fact, Sunnis who advocated moderation were denounced as traitors and sometimes killed, while cooperating with the Shia-led government was unthinkable in part because Shia forces were expected to kill Sunnis indiscriminately no matter what Sunnis did.  Shia too have come to expect indiscriminate violence from Sunnis.  If most elites and masses are rational, neither Sunnis nor Shia should have behaved this way.

In fact Fearon argues that Sunnis could even now be co-opted (including, presumably, enforcing in-group policing on themselves) if not for the fact that the Shia are also fragmented, into four main factions by his count.[38]  The trouble with this argument is that, at least until Fall 2006, the important Shia factions were united in the United Iraqi Alliance (UIA) which maintained a common front toward the other communities.[39]   This should mean that under Fearon's logic the Sunnis had three years during which the conditions for communal settlement were met.[40]

Fearon and Laitin could argue that their model should not be expected to explain the continuing deterioration of Iraq by taking the position that beyond some (unspecified) threshold fragmentation can entirely overwhelm the incentives and tools for in-group policing and re-integration.  This would, however, gut much of the usefulness of the theory.  Since all collapses of central authority necessarily mean at least temporary anarchy, not only in Iraq but also in numerous other recent cases – Albania, Afghanistan, Bosnia, Congo, Gaza, Kosovo, Lebanon, Sierra Leone and Somalia to begin with – the domain of the theory would be limited, possibly to only those cases where peace is already highly likely because in-group policing is working well.


The constructivist model implies the war in the North but is agnostic about the South.  By 2003, Kurdistan had enjoyed de facto independence in parts of three provinces for twelve years, and Kurdish elites had taken full advantage of this to sharpen Kurdish separatist identity, building on prior strengthening of separatism because of repeated Kurdish rebellions before 1991 and the 1988 Anfal genocide campaign waged against them.  In Kurdistan today, only the Kurdish flag flies; the Iraqi flag is rarely or never seen.  In an unofficial 2005 referendum, 98% of Iraq Kurds favored de jure independence for Kurdistan.[41]

Accordingly, the constructivist model would predict that the Kurds would act to reclaim the territories that they had come to consider their entitled homeland, including the remainder of the three provinces, already largely under Kurdish control, Kirkuk province, and parts of Eastern Ninewa province including the partially Kurdish city of Mosul (Kirkuk and Mosul are also the two main oil centers in Northern Iraq).  This is what happened.[42]  

Enjoying, as the Kurds did in April 2003, a tremendous military advantage over their local rivals, they moved immediately and aggressively to size the claimed territories and solidify control of them by colonization together with some ethnic cleansing.  Constructivism does not have much to say, however, about the forms or scale of resistance by Sunni Arabs or others, who had not by 2003 embraced equally strong jingoistic claims against the Kurds.  These might or might not develop after 2003, depending on the outcomes of discourse.

(It should be noted that this prediction for Kurdish behavior depends on accepting "weak form" as opposed to "strong form" constructivism.[43]  Weak form constructivists, such as Stuart Kaufman, accept that at any given time the available symbols available for identity manipulation depends on the outcomes of prior rounds of identity discourse; strong form makes arguments that come closer to an ideal where all issues in discourse are always up for grabs and history matters little).[44]

For the South, the model makes few predictions.  Sunni/Shia tension had been present going back to Ottoman times at least, and the former regime did suppress a Shia rebellion in 1991 and murdered clerics and intellectuals in the 1990s.  However, Iraqi Shia fought as loyally as other Iraqis in the war against Iran and there was no widespread rage among Shia masses or being spread by Shia leaders that was just waiting for the first chance to boil over.

The theory gets one aspect of Shia response to the escalation of the war since 2003 partly wrong.  Of the three most influential Shia leaders, Ali al-Sistani, Moqtada al-Sadr, and Abdul-Aziz al-Hakim, all three have consistently called for restraint in use of violence against Sunnis.[45]  Nor did these three face serious challenges from new contenders for power advocating harder behavior.  This has not, however, prevented a dramatic hardening of ordinary Shia attitudes toward Sunnis, which Shia elites are now bound to respect.

The fact that both al-Sadr's and al-Hakim's militias have committed many sectarian murders does not overturn this evaluation.  The essence of constructivism is that public discourse matters, and cannot be rendered irrelevant by the fact that events on the ground do not match the content of public debate.  If we grant the possibility that "street truth" can overwhelm discourse then we have adopted an outlook closer to security dilemma theory.[46]

Among Sunni elites, competitors for power could have tried to construct the occupation of Iraq in any or all of several ways: as purely American aggression, as Kurdish aggression against Arabs, as an attack on mainstream Islam from the growing power of Shia schismatics, or as a co-ordinated assault by an alliance of the U.S., the Kurds, the Shia, and the 'Persians,'  and we have seen rhetoric and behavior consistent with all of these.  In practice, many Sunni resistance groups do not seem to consider combating American forces, the Shia-dominated government, and the Shia community as separate wars; whether this reflects a constructed reality much distinct from a  description of the enemies attacking them is not clear.

Security Dilemma 

The logic of security dilemma theory is that, faced with an absence of central order and at least some reason for inter-communal fear, all communities can be expected to mobilize, a process that can interrupted only if powerful central order is quickly re-established, if communal leaders on all sides are able to suppress the mobilization process early, or if the imbalance of power is so great that vulnerabilities are not mutual.[47]  Incentives for ethnic cleansing will depend on population geography and physical geography: how many people of either or both communities are living in places that make them appear both threatening and vulnerable?

Accordingly, the expectations of the theory for Northern Iraq are clear on some points but not others.  The collapse of the regime created a communal security dilemma in mixed Kurdish/Arab/Turcoman zones, especially in and around the cities of Kirkuk and Mosul which Kurds claim but where they were in the minority in April 2003.  Initially, however, this security dilemma was mild because of overwhelming Kurdish advantage in mobilized armed strength deriving from U.S. assistance since 1991.  The theory would thus predict relatively little violence at first, which is what happened; Kurdish advance initially faced relatively little resistance. 

In communal conflicts, however, the side(s) less mobilized initially normally move to redress the balance, often with some success, which intensifies security dilemmas.  Thus the theory would expect the war in the North to escalate over time, although it says relatively little about how much or how quickly since these depend on details of mobilization opportunities.  Thus the theory is consistent with the marked acceleration of Sunni violence in the North since 2005 but cannot claim credit for predicting the timing or the scale.

In the South, the existence of the deep belt of intermixed settlement around Baghdad would predict that once central authority collapsed and the communities begin mobilizing against each other, eventual escalation to full-scale communal war and ethnic cleansing would be difficult to prevent.[48]  War could have been prevented by either: immediate replacement of the Ba'ath regime with another equally capable and willing to enforce order as ruthlessly as necessary, essentially meaning leaving all but a handful of top leaders in place; or enough U.S. troops to suppress nearly all inter-communal violence—which, as mentioned, we didn't have; or energetic action by communal elites in both communities who commanded enough power to enforce restraint on co-religionists. 

Lacking these, the fact that initial Sunni resistance was aimed more against U.S. forces and against the new government than against Shia civilians did not matter.  Ordinary Shia could see that in practice this was also aimed against them as a community, and Sunnis could see equally clearly that government action against insurgents meant, in practice, targeting Sunnis.  The war has developed as the theory would expect: communally based killings led to increased mutual fears both directly and indirectly through spread of atrocity tales, leading to an upward spiral of more mobilization, more killings, more fear, perceived need for ethnic cleansing, actual ethnic cleansing, and increasing separation of the two populations.

The map below shows narrower bands of mixed settlement than do most pre-2003 maps.  Whether this is intended to reflect ethnic cleansing that occurred between 2003 and 2006 is not clear:[49]

III. Explanations for the Trajectories of the Wars in Iraq

We should also interest ourselves in the competing models' performance in predicting or explaining the further trajectories of Iraq's communal wars once started.  There are several questions that should be asked for each theory.  Some of the most important are summarized in the following table:

Predictions of Theories of Communal Conflict for Iraq, 2003-2007

Predictions of à


Rational Choice


Security Dilemma

Northern Front:

Initial developments

No war.  Kurds, Sunnis bargain over oil revenue.

Kurds seize land they claim.  Rivals do not initiate violence.

Low violence because of Kurdish military dominance.

Cross-communal appeals in election


No.  Sunni elites have nothing to offer Kurdish masses.

No, because of cons-truction of Kurdish identity since 1991.

No, because of mutual security fears.

Escalation of violence

Sunnis may coerce Kurds but avoid ex-cessive provocation.

Uncertain; depends on discourse after 2003.

Yes, as balance shifts; especially in mixed areas.

Southern Front:

Initial developments


bargaining over oil revenue.

Uncertain, because of competing Arab/ Iraqi/sectarian identities.

War beginning in mixed areas.

Cross-communal appeals in election


Yes, because  material interests of Shia vary.

Likely, outcomes uncertain.

No, because of mutual security fears.

Escalation of violence

Sunnis may coerce Shia but avoid ex-cessive provocation.

Uncertain; depends on discourse after 2003.

Yes, feeds on mobilization and on

atrocity tales

Linkage between fronts

No, except for Kurdish/Shia bargaining over burden-sharing.

Maybe, depending on how Sunni discourse defines enemies.

Indirectly, if Sunni mobilization for one front changes balance on other.

Unity within communities

Weak, because material interests trump identity.

Varies depending on competition between discourses.

Strong in proportion to security threats.

Could U.S. facilitate peace?

Yes, via side pay-ments or coercion. But might not be needed.

No, but might be able to assist contesting factions with money or broadcast access.

Maybe, by shifting balance of power or brokering federalism.


Beyond the main question of whether we should have expected communal wars in Iraq, the main questions that should be asked of each theory are, first, escalation: should we have expected these wars to be relatively easily dampened or at least contained?  If not, how bad should we have expected them to get?

Second, elections: should establishing electoral democracy have ameliorated or worsened communal conflicts?  Especially, how much should we have seen in the way of cross-communal political appeals and how should these have fared?

Third, oil: should the main factions, with or without the help of the United States, have been able to resolve the main disputes through an oil revenue-sharing scheme?

Fourth, U.S. options: what, if anything could the U.S. have done to stop the wars in Iraq once under way?

Rational Choice

Whether the communal wars in Iraq should have been easy to contain depends on which of the main rational choice arguments we accept.  If the critical issue is material rewards, then containment or resolution should have been – and should still be – relatively easy.  During the constitution-writing process in Summer 2005 it became unmistakable that the critical Sunni demand was oil revenue sharing.  This applied not only to those Sunni factions who were already actively involved in politics but also to many or most of those supporting the insurgency.  American officials realized this also, and have put strong pressure on Kurdish and Shia parties to respond reasonably to this demand.

It should have been in the interest of nearly all factions to reach a deal.  Further, the U.S. should have been able to facilitate a deal by compensating the Kurds and Shia for payments to Sunnis, by coercing recalcitrants or helping moderates suppress them, or by providing investment to enlarge the pie;[50] other powers friendly either to the U.S. or to any Iraqi faction should also have been willing to contribute

The same logic should apply to development aid generally – and perhaps also to simple bribery – which should have been offered fairly indiscriminately to all but the most bitter end factions, and aimed specifically at locations that those factions would be able to defend.

Second, a rational choice model should have expected free elections to promote peace by opening up wide avenues for cross-communal alliances.  Individual interests do not follow communal boundaries.  Some members of some communities in certain contested regions face great security threats; others face little or none.  Economic interests vary by class and by region, independent of sect or ethnicity.  Perhaps most important for Iraq, Shia in Basra sit on top of large oil reserves; those in Baghdad do not.  Basra oil can be exported while the Sunni insurgency continues, but that from Kirkuk mostly cannot.

There should, therefore, have been room for any or all of: an economically based coalition of Sunnis, some Shia, and perhaps Kurds; middle classes and others who preferred a relatively secular Iraq; and a coalition of religious parties across communal lines.

The first of these was not attempted.  Iyad Allawi's effort at the second got just 8% of the vote in the December 2005 elections;[51] this project is politically dead for now. Instead, 92% of votes cast in December 2005 were cast for communal parties.  As for the third, Moqtada al-Sadr made an effort in Fall 2005 to organize a  religiously-based slate of candidates across Sunni-Shia lines in one province but was dissuaded by the scale of criticism from other Shia.[52]

Third, rational choice theory should expect that an oil revenue sharing arrangement, including at least the strongest factions in all communities, should have been put in place long before now, but it has not.  It is important to note that this is not a "diamond war" situation where cashable resources prop up an insurgency,[53] but one where the resources are unavailable to the insurgents as long as the war lasts, and all sides lose economically from continued war.

 Iraq's new constitution, approved in October 2005, was written by a Kurdish-Shia coalition with virtually no Sunni input.  Two key provisions were that the country's Northern (Kurdish) and Southern (Shia) areas could form autonomous regions, and that these regions, not the central government, would control most oil production.[54]

Despite promises that Sunnis would be permitted to raise the oil issue later, there has been no action and only sporadic negotiations, much of which have concerned Kurdish-Shia efforts to guarantee revenues to themselves, not schemes for sharing with Sunnis.  One round in February 2007, offered Sunnis virtually nothing,[55] while a second in May and June 2007 mainly concerned whether promises of distributions from the center to the Kurdish Regional Government (KRG) would be worth anything in practice.[56]

An oil law has not been approved.  On July 3, 2007, the Iraqi cabinet approved a relatively unimportant part of the draft law (to authorize a government oil company) but could not agree on distribution of revenue.[57]   In practice it was not clear that the producing regions (mainly Basra and the KRG) would turn over all, or very much, of their revenues to the center.  Further, the language on distribution of what the center would control was so vague as to allow any outcome.[58]  The KRG as well as the Sunni Arab parties wanted revenues controlled by the central government to be allocated strictly by population, to prevent re-direction to their disadvantage.  Kurds also objected that last-minute changes had been made but kept from them.[59]

Thirteen of the 37 ministers protested the cabinet vote by their absence, while five more "suspended participation" on August 6.  This put all the Sunni parties, the small Shia secular party, and two of the Shia religious parties (the Sadrists and Virtue) out of the cabinet; the rump consisted two other religious parties (SIIC and al-Da'wa), and the Kurds.[60]   In August the KRG unilaterally adopted its own oil law.[61]   

A second issue is that the U.S. has been pushing for inclusion in the law of provisions for "production sharing agreements" (PSAs) apparently designed to favor U.S. oil companies,[62] provoking resistance from some Shia, especially the Sadrists, as well as from both moderate and more nationalist Sunnis.[63]

The scale of the foregone revenues resulting from this deadlock is staggering.  Iraq's average oil production for 2006 was about 2 million barrels per day; at recent prices ranging around $70/barrel this comes to about $51 billion/year.[64]  Absent the wars few industry experts doubt that Iraqi production could fairly quickly recover to its level as of July 1990, before Iraq's invasion of Kuwait and the sanctions and discord that have depressed Iraq's production since then.  This was 3.5 mbd, meaning that Iraq's civil wars are costing the sides about $38 billion per year.  If we assume that Iraq could, within several years, produce oil at the same rate, in relation to reserves, as Saudi Arabia, that would be 5-9 mbd worth $128 to $230 billion per year.  Thus the wars in Iraq are costing Iraqis annual oil revenues of between $75 to $180 billion, a foregone increase of between 85% and more than 200% of the country's GNP[65].

It is impossible that this could not be shared in a way that would leave all factions too strong to suppress better off than they are now.  Hard bargaining should be expected, as should efforts to design means of reducing cheating using IOs or whatever other means can be found.  But a rational model cannot accommodate failure of Iraqi factions to bargain seriously or U.S. behavior that hinders rather than promotes agreement.

Further, while the parties have both shared and conflicting interests, they have had two years to measure each others' bargaining strength.  The fact that today some oil cannot flow because rebels are destroying pipelines should be no barrier.  The strongest Sunni factions, rather than lose such a lucrative opportunity, should co-opt or suppress anyone in the way, or accept help to do so.[66]

The only way that rational Iraqi actors should be able to avoid resolving the oil issue would be if both Kurds and Shia considered themselves so strong that they could stiff the Sunnis without suffer either significant loss of revenue (not true for the Kurds) or of lives (not true for the Shia and, depending on pain tolerance, perhaps not the Kurds either).[67]

What would make sense of both Sunni and Shia behavior on this issue would be hatred constructed by elites or mass memories of atrocities.  Iyad Samarrai, secretary- general of the largest party in the Iraq Accordance Front, the main Sunni parliamentary coalition, explained that Sunni ministers were boycotting the government because they have gained nothing to show for their participation and their 44 votes, saying "the issue cannot be resolved with appeals alone."  This has happened despite the fact that the Sunni coalition shares common interests with Kurds over fears on Shia domination, with some Shia factions, especially the Sadrists who oppose devolution of powers to regions, and with several factions who suspect SIIC of planning to seize control of the central government, oil revenue, or both.[68]

The fact that the U.S. is apparently hindering rather than helping Iraqi factions reach a settlement is also hard to explain.  Pressure for PSAs may have seemed desirable during pre-war planning that did not envision the scale of communal fighting now afflicting Iraq.[69]

On the fourth question, Fearon argues that the presence of U.S. forces has both led the Shia to over-estimate their security and led Sunnis to overestimate their chances of regaining power once U.S. forces leave, thus stiffening both sides' negotiating positions.[70]  In effect this adopts the logic of Zartman's 'hurting stalemate:' "more civil war may be the only way to reach a point where power sharing could become a feasible solution to the problem of governing Iraq."[71]  Indeed, many observers have suggested that U.S. force presence may have emboldened both sides.

Many of the same observers have noted, however, that these misperceptions have declined over time.  Sunnis calls for U.S. help to protect them against the Shia have increased,[72] while Shia leaders are showing less interest in a long-term U.S. military presence.  On July 14, 2007, Prime Minister al-Maliki stated that Iraq could defend itself without U.S. help, although more training and weapons would still be welcome.  Amar Hakim, the new head of SIIC, countered that that U.S. withdrawal should follow further strengthening of Iraqi government forces; although this is weaker support for continued U.S. presence than expressed by his father in 2004.  Moqtada al-Sadr has opposed U.S. forces from the start.[73]  Whether or not misperceptions of the military balance mattered in 2003, they matter less in 2007.


Constructivist approaches offer little on the first question about whether we should have expected the civil wars, once started, to escalate or de-escalate.  Ongoing conflict offers opportunities to demonize not only the "enemy" community but also politicians in one's own community who can be called "soft" on the threat.  It does not, however, predict the likely outcomes of contest between "pragmatists" who promise resolution of the conflict on favorable terms and "bitter-enders."  Contestants with better access to media and other outlets for political discourse should have an advantage.

The main exception should be among Kurds, where Kurdish separatism had been constructed so successfully during the previous decade that Kurdish leaders such as Massoud Barzani and Jalal Talabani may have faced some danger of blowback when having to explain to the 98% of Kurds who voted for full independence in 2004 that they could not have it without antagonizing Turkey and endangering Kurdistan's autonomy within Iraq.  Thus far they have managed the challenge.

Second, constructivism should be agnostic on the impact of elections, which would create openings for politicians selling any of many different visions of the communities and of Iraq itself.

Third, constructivist models would not view oil revenue as nearly as central as  under a rational choice model, but this issue too would create opportunities for identity construction in any of several directions.  Much would depend on whether previous identity construction in some communities had created symbolic entitlements to oil, but the practical impact even of this would be hard to predict.  If Kurds feel attached to their oil, would that advantage a hard-liner or a pragmatist who promises to deliver oil revenue by getting Sunni rebels to stop attacking pipelines?

Fourth, constructivist models would suggest that the U.S. might be able to sway political contests within communities by providing money or access to broadcast outlets, provided that this did not undermine the favored politicians' domestic credibility.  Open endorsement should normally be avoided; there is anecdotal evidence that U.S. endorsement has harmed the domestic legitimacy of Prime Minister Nouri al-Maliki, and most observers agree that U.S. support harmed Iyad Allawi's secularist coalition in the December 2005 election even if we don't have the data to prove it.

Security Dilemma

Security dilemma theory would expect that communal wars, once started, should escalate in proportion to the number of people of either community located in places that make them appear both threatening and vulnerable. In Iraq, the wars in both North and South should have expanded until they affected virtually all of the two main bands of mixed settlement, as has happened. 

There are several ways that escalation could be avoided.  Four—keeping the Ba'athists in power, direct suppression by the Coalition, military dominance by one side, and condemnation of inter-communal violence by leaders in all communities—have already been mentioned.  The first might have worked but would have been repugnant on other grounds, the second was beyond our power, the third was not present in the South and has eroded in the North, and the fourth did not happen widely enough or soon enough.  As mobilization and violence increase, the political space within in each community for calls for restraint versus the other contracts.  It was not until the Samarra mosque bombing in February 2006 that top Shia leaders made their strongest calls for restraint, which had no discernable effect even in the short run.  There has never been a consensus among Sunni authority figures for restraint versus the Shia. 

A fifth possibility is that, if the stronger side in the immediate situation is, like the Kurds in 2003, weak when a wider region or a longer time frame is considered and depends on a security guarantee from a strong power that can use that dependence to restrain the client.  But the U.S. did not make constraining the Kurds a priority, and in 2004 withdrew the 101st Airborne Division from the North, turning over military command – on behalf of the Iraqi government – to the peshmerga.   Finally, pockets where local minorities live deep inside a region dominated by another group pose little or no threat and may therefore escape attack, although harassment or worse based on revenge for atrocities committed elsewhere cannot be ruled out.

The theory says less about speed of escalation than about escalation itself, except where strategic changes make certain settlements more or less threatening or vulnerable than before.[74]  Such events have not played a major role in either of Iraq's wars, so we cannot say whether we should have expected faster or slower escalation than has occurred—with the exception of the possibility that initial Kurdish military dominance may have slowed escalation in the North compared with the South.

Last, if and when the warring populations are substantially separated behind defensible boundaries, incentives for violence should decline.  In Iraq, however, the emerging boundaries run mainly on flat land, in many places thickly settled.

Second, if elections are held during ongoing communal war, security dilemma theory would not expect significant cross-communal campaigning for the same reasons that communities at war are generally deaf to cross-communal appeals.  The election might measure the state of the war but should not be expected to affect its course for good or ill.

Third, security dilemma theory would not expect hopes for gains from oil revenue to trump security concerns.[75]  If the oil issue is settled before substantial separation of populations and effective autonomy for communally-dominated cantons, that would count against security dilemma theory.

Fourth, the theory suggests few paths by which the U.S. could improve the chances of ending the war.  The U.S. lacks the legitimacy and the power to unilaterally partition Iraq.  The U.S. could strengthen one faction—presumably the Shia—sufficiently to allow them to conquer the country, but beyond parts of the city of Baghdad, there is little in the mainly Sunni areas that will be of enough value to Shia elites or masses to be worth the ocean of blood that they would have to bleed to take it.  Decisive victory also runs at least some risk of being followed by genocide.[76]  The main thing that the U.S. can do has already been mentioned, i.e. helping to stabilize the de facto partition that is emerging anyway.  How this might be done is considered below.

IV. What Do We Do Now?

The human toll of the communal wars is tremendous and rapidly growing worse.  The most respected minimum estimates of Iraqi civilian deaths are those of the  Brookings Institution and of Iraq Body Count.  Brookings estimates that about 99,000 Iraqi civilians were killed between March 2003 and July 2007, while IBC estimates 69,000 to 75,000 deaths through June 2007.[77]  Both counts are considered conservative because they count only deaths that can be documented either by multiple western news reports of by the Iraqi government; the real totals may be higher.  Probably less than 20% of the victims were killed by U.S. forces.[78]

During the last year the killings have accelerated – according to IBC, four-fifths as many Iraqis were killed in the fourth year of the war than in the previous three years combined.[79]  Since IBC counts run somewhat behind events (two months behind as of August 12, 2007), the actual toll for the 4th year was likely even worse.  Brookings estimates show a slight fall-off in June and July 2007 compared with the same months in 2006, although the toll for the whole period of the "surge" (January 2007 to present) is more than 20% higher than for the same months in 2006.[80]

Well over 4 million Iraqis were displaced as of April 2007, including 1.9 million internally displaced persons (IDPs), over 2 million refugees in neighboring countries, and 200,000 further afield, mainly in Europe.[81]  Many of these, however, were already displaced before 2003.  Estimates of the number of new displacements are further complicated by the fact that many refugees and IDPs returned home as a result of the U.S. conquest of Iraq, especially Kurds whom the Ba'ath regime had forcibly relocated further South, Shia who had been planted as forced colonists in their places, and Shia returnees from Iran.[82]  

The best estimates that we have are from Brookings, which estimates 1,135,000 new IDPs in Iraq since April 2003,[83] and from the United Nations' International Organization for Migration (IOM), which estimates about 1,400,000.[84]   Rates of internal displacement may have accelerated since the U.S. military surge began, especially in Baghdad, according to IOM and the Iraqi Red Crescent, while Brookings estimates about 60,000 new IDPs per month during January to July 2007.[85] 

According to Brookings, 911,000 people fled Iraq entirely during 2006, or about 75,000 per month.[86]  If we assume that this rate has continued into 2007, at least  135,000 Iraqis are being displaced per month, perhaps more if IOM or IRC are correct.

There is little question that the main cause of ongoing displacement in Spring and Summer 2007 is sectarian violence and the fear of it.[87]  According to Dana Graber Ladek of IOM, in the first years of the war American offensives were a main cause of displacement, but now "sectarian violence is the biggest driving factor — militias coming into a neighborhood and kicking all the Sunnis out, or insurgents driving all the Shias away."[88]  The UNHCR estimates that only 5% of new IDPs are due to military operations.[89]

As might be expected in communal wars where ethnic cleansing is occurring, the new IDPs represent several distinct exchanges of populations:  First, between KRG-controlled areas in the North and provinces slightly further South, apparently including both majority-Sunni as well as Baghdad; second, in Baghdad and surrounding provinces where Sunnis and Shia alike are fleeing vulernable towns and neighborhoods; and in the South, where Sunnis are fleeing Shia majority areas.[90] 

On present trends, nearly all the potential victims of ethnic cleansing in Iraq will become actual victims, and the numbers suggest that we may already be more than halfway to that point – but we cannot say whether this will take months, more than a year, or longer.  

U.S. debate about how to stabilize Iraq centers on three main options – cut and run, gradual drawdown, and stay the course (or "surge," in its most recent incarnation).  It has been difficult for proponents of any of these to articulate clearly how and why these policies can be expected to resolve Iraq's communal wars because none of them are based on coherent theories of communal conflict.

Cut and Run

"Cut and run" has some political support, based on the deep unpopularity of the war among American voters,[91] but few have made arguments that it could be good for Iraq.  A prominent recent example is Nicholas Kristof's recommendation in December 2006 that the U.S. should be completely out of Iraq by November 2007, including renouncing any intention to retain bases.  His logic is that U.S. forces are inflaming Iraqi nationalism, generating more and more extremists, not only anti-American but also making heroes of sectarian mass murderers and torturers.  Thus we cannot win the hearts and minds of either Sunnis or Shia, and the longer we stay the greater the risk of getting dragged into a wider Sunni/Shia regional confrontation.[92]

The reason that cut and run does not have more support from foreign policy elites is that most recognize that an unconditional, rapid exit will trigger a fight over the Iraqi central government, with some factions seeking to control it and others to destroy it; this could cause escalation of existing intra-Shia divisions to (a third) full-scale civil war.  It would also reduce barriers and increase incentives for neighboring powers to intervene and, finally, it would increase both incentives and opportunities for all sides to step up ethnic cleansing in areas where U.S. forces have provided at least a degree of security that has slowed this process.[93]


Most Americans,[94] and at least a plurality of prominent politicians, including the major Democratic presidential candidates, favor a gradual-drawn of U.S. forces in Iraq.  The most prominent and detailed proposal of this kind is the December 2006 report of the Iraq Study Group,[95] which recommends re-orienting U.S. military effort from direct combat missions to improving the training and professionalism of the Iraqi Army and police, combat corruption, increase oil production by improving physical security as well as opening the industry to foreign investment,[96] increase and improve coordination of reconstruction aid, encourage 'national reconciliation,' and seek cooperation from Iran and Syria.[97]  Most U.S. combat forces should be out of Iraq by the first quarter of 2008, and "even if the Iraqi government [does] not implement [the] planned changes [called for in the report], the United States must not make an open-ended commitment to keep large numbers of American troops deployed in Iraq."[98]

The U.S. has, however, already invested more than four years in trying to create security forces that would be loyal to a legitimate government of Iraq instead of to their communities, tribes, or warlord affiliations.  This effort has been a near-total failure.  No new evidence or logic has emerged that suggests that this can be done, although the deepening of both communal hatreds and fears are reason to expect that chance would be even slimmer than in 2003. As discussed above, there is also no government of Iraq widely honored as legitimate.

Stay the Course and the "Surge"

While this is U.S. policy, the evidence is that it is failing.  On July 12, 2007 President George W. Bush reported that the Iraqi government was making satisfactory progress on 8 of 18 benchmarks defined by Congress in May.[99]  A closer look at the benchmarks shows that at most one of the 8 "satisfactory" grades could be considered related to bridging communal divisions, while all 10 of the "mixed" or unsatisfactory" results concerned failures to reduce communal mistrust, discrimination, and violence.[100]  An assessment provided to the House Armed Services Committee by Thomas Fingar, the deputy director for analysis at the National Intelligence Council, painted a bleaker picture of the benchmarks, saying there were ''few appreciable gains.''[101] 

On August 2nd Secretary of Defense Robert Gates said that the Bush administration had overestimated the chances of gaining "a political truce" in Iraq: "I think [that] we probably all underestimated the depth of the mistrust."  Gates also testified before Congress that he would like to see the surge end by December 2007.[102]  An August 2007 National Intelligence Estimate foresaw little chance of sectarian reconciliation within six to twelve months.[103]

Outside the White House and the American Enterprise Institute, the most prominent supporters of the surge are Michael O'Hanlon and Kenneth Pollack.  They argued in July 2007 that "we are finally getting somewhere in Iraq, at least in military terms," and "can win, if not 'victory,' a sustainable stability that both we and the Iraqis could live with."[104]

O'Hanlon and Pollack offer three categories of evidence: first, their personal impressions that U.S. troop morale is higher than they saw on previous visits, in part because of the surge; the troops "feel now they have the numbers needed to make a real difference."  Second, "the American high command assesses that more than three-quarters of the Iraqi Army battalion commanders in Baghdad are now reliable partners (at least for as long as American forces remain in Iraq);"[105] and third, that we have lately made huge strides in gaining Iraqi co-operation in many outlying provinces:  "In the last six months Iraqis have begun to turn on the extremists and turn to the Americans for security and help. … Anbar Province [has] gone from the worst part of Iraq to the best (outside the Kurdish areas).  Today the Sunni sheiks there are close to crippling Al Qaeda and its Salafist allies."  They also claim that around Mosul, which has been subject to waves of terrorist bombings as Kurdish and Sunni forces contest control of the city, "reliable [Iraqi] police officers man the checkpoints in the cities, while Iraqi Army troops cover the countryside."  In other provinces, "Provincial Reconstruction Teams are working. Wherever we found a fully staffed team, we also found local Iraqi leaders and businessmen cooperating with it to revive the local economy."[106]

Unfortunately, it is not clear that any of O'Hanlon and Pollack's claims reflect much about realities in Iraq.  In an interview with Glenn Greenwald of, O'Hanlon admitted that he and Pollack had spent only 7½ days in Iraq, never more than 2-4 hours at a time outside the Baghdad 'Green Zone,' and only 2 hours in Mosul.  Except for a handful of chance, mostly brief, encounters within the Green Zone, all of their interviews were scheduled for them by U.S. military handlers.  O'Hanlon admitted that "I don't claim any great sense of what the Iraqi public or Iraqi leadership is thinking" and that the morale claims were not based on unobserved interviews with U.S. enlisted troops.[107]  For his part, Pollack admitted to the New Yorker's George Packer that he spoke with very few Iraqis and "could independently confirm very little of what he heard from American officials."  The claims about weeding out bad Iraqi commanders and about progress in Mosul "came from American military sources."[108] 

Pollack and O'Hanlon bury at the bottom of their article an admission that unless sectarian divides in Iraq are bridged, an issue on which they do not claim progress, military gains would ultimately avail nothing.[109]

More War

Fearon's main policy advice for stopping the communal wars in Iraq follows the standard economic model arguments: "The basis for an Iraqi state is the common interest of all parties, especially the elites, in the efficient exploitation of oil resources.  Continued civil war could persuade Shiite leaders that they cannot fully enjoy oil profits and political control without adequately buying off Sunni groups, who can maintain a costly insurgency.  And civil war could persuade the Sunnis that a return to Sunni dominance is impossible (as discussed, both of these facts have already been amply demonstrated to Iraqis).  Kurdish leaders have an interest in the autonomy they have already secured but with access to functioning oil pipelines heading South."[110]

Fearon argues that what Iraq needs is more war, which he hopes would clarify the Iraqi factions' bargaining strength[111] as well as compel military and therefore political consolidation within communities:  "The best outcome for Iraqis that is still available would be "a power-sharing agreement among a small number of Iraqi actors who [each] actually commanded a military force and controlled territory," or "the rise of a single dominant military force whose leader [could] cut deals with [weaker] warlords." [112]  Such outcomes, however, cannot be enforced by the United States.  The only way to get there is "more fighting [that] holds the prospect of creating pressures for consolidation on both sides."[113]

This mistakes both the situation in Iraq and the logic of military consolidation in communal wars.  Communal wars can cause either fragmentation or consolidation depending on local conditions.  Individuals' strongest concerns are usually their own homes and safety, not the fortunes of the whole community.  National forces may intervene in local battles, but local threats do not guarantee that those threatened will join or support any national-level force.[114] 

Pressure for co-ordination is strong only when nearly the whole community, no matter where they live, perceive themselves as all under extreme threat from the same opposing force—which is arguably the case for many Iraqi Sunnis but not for Shia, who are in far greater danger in some neighborhoods and provinces than those in others.  While there has been considerable combat among Shia factions, most of this has occurred in the far South, which is secure from a communal point of view, not in mixed areas in and around Baghdad.[115] 

In Iraq, war is causing both consolidation and fragmentation in different communities for different reasons.  Among Sunnis, a rebellion that began as "scattered, erratic, and chaotic [is] increasingly dominated by a handful of large groups that … to a surprising degree, coordinate their words and deeds."[116]

The Shia began relatively united but have fragmented over time, perhaps in part because they have far greater mobilization potential and perceive that they can afford it.  During 2003-2005, parties that could field militias to fight Sunnis—mainly SIIC and the Sadrists—gained at the expense of others.  As the communal war has become still the Sadrists and the Virtue Party have split from SIIC and al-Da'wa and from each other.  Moqtada al-Sadr has apparently lost control of at least some of the Mahdi Army, which has become a collection of effectively independent armed groups.  There has also been a practically uncountable proliferation of local Shia and Sunni militias; it is hard to see how more war could reverse that.

One cannot say, however, whether Shia fragmentation is hindering or promoting settlement of the key inter-communal issues such as oil.  As discussed, a unified Shia authority should, if rational, favor paying off Sunnis, but so should all of the main actual factions.  Shia fragmentation increases the number of actors in the bargaining, but also means that not all are needed for settlement to be possible.  As noted, the actual state of Shia splits yields, as of August 2007, a narrow majority against meeting Sunni demands, but that was not the only possible outcome.

Finally, Fearon recommends that the U.S. move away from its "unconditional military support for the Shiite-dominated government" so that "Sunni insurgent groups would [see] the United States less as a committed ally of the 'Persians' and more as a potential source of financial or even military backing."[117]  As discussed, most of this is already happening—the main effect of the U.S. "surge" has been to shift effort to hunting Shia militias (mainly the Mahdi Army) instead of Sunnis.

The novel part of Fearon's suggestion—that the U.S. consider funding or arming Sunnis as well as (or in place of?) Kurdish- and Shia-dominated government forces—would likely do great harm, even according to Fearon's own logic.  Strengthening the Sunnis or weakening the Shia before a general settlement would introduce uncertainty into the military balance, undoing part of what Fearon argues would be the virtue of more war.  Such a maneuver would also destroy the U.S. relationship with Iraqi Shia and escalate incentives for the strongest regional power—Iran—to intensify its intervention.

Identity Reconstruction

The solution implied by constructivist theory would be to take advantage of the fact that people have some degree of attachment to identities at many levels of scale—family, sometimes clan or tribe, ethnic group or nation, sect or religion, and sometimes culture or civic loyalty to state institutions[118]—not to mention cross-cutting identities such as profession or class.

The goal would be to shift Iraqis' primary identities from the narrower sectarian and ethnic levels that dominate Iraqi politics now to wider levels, such as 'Iraqi' or 'Muslim,' that could include all of them.  The communal boundaries that separate Iraqis cannot be made to disappear, but it would help mightily if they became politically unimportant compared with the Sunni Arab/Shia Arab/Kurdish division that dominates Iraqi politics now.  Nations and nationalism cannot exist without national boundaries, which would make it difficult to maintain nationalist wars across boundaries that few people on either side think of as their primary identity boundary. 

At least in the South, this would mean only regaining a situation that existed not much more than four years ago, when the identity boundary between Shia and Sunni Arab Iraqis was not obviously much stronger—or perhaps any stronger—than the two communities' common identities as Arabs and as Iraqis.  (Most Kurds as well as Arab Iraqis have long agreed that Kurds are neither Arab nor Iraqi.)

The problem is how to get there.  Jawboning through discourse has become such a clear non-starter that we now rarely see serious proposals.  The other option is material incentives but, as discussed, the situation itself already offers incentives for peace that would be hard to top.  While identity reconstruction might have helped head off Iraq's wars before they began and might, once they are ended, help avoid recurrence, it can't do much right now.


As the communal wars in Iraq have grown worse, increasing numbers of politicians and academic analysts have come to advocate de facto partition of Iraq as a way to end the sectarian killing.[119]  Almost no one recommends de jure partition because it is understood that Turkey would use force to prevent establishment of a Kurdish sovereign state and because all important Sunni and Shia Arab figures oppose it.  Accordingly, proposals go under names like "federalism" or "soft partition."    
Opponents of partition argue that it would generate yet more refugees and might not end the war.

Both sides in this debate are asking the wrong question.  The question is not: Should we partition Iraq? It is: Can anyone prevent Iraq from partitioning itself?[120]

Probably not.  Communal wars have tipping points: when mass support for ethnic cleansing—or, at least, for the militias that do the ethnic cleansing has become so strong that no authority in either community can muster a constituency determined and strong enough to suppress them.  Past this point, the war cannot be stopped until the warring communities are substantially separated.  The aftermath of the Samarra mosque bombing in February 2006 showed that the Sunni/Shia civil war had by then passed its tipping point.  Prominent leaders of both communities made energetic efforts to dampen the upsurge in sectarian killing but had no measurable impact.

This means that it no longer matters how the war started or whether most members of both communities want to wage communal war.  The fact that virtually all Sunnis and most Shi'ites still oppose partition is not relevant; the structure of the situation is causing them to act in ways that are partitioning the country anyway.  The ethnic cleansing will continue until nearly all mixed urban neighborhoods, towns, and rural districts have become unmixed, as forces representing whichever community is stronger in that locality kill or frighten away most members of the other.  The eventual result will be a de facto partition.

A September 2006 poll found that 82 percent of Iraqi Shia said that we are damaging, not improving security; 71 percent of Shia wanted us out within a year, and 62 percent thought attacks on US forces were justified.[121] In their minds, many Shia have already put us in the dustbin of their history. Sunni attitudes toward the United States are even more negative, but matter less because they are the weaker side.

America has only one remaining essential military mission in Iraq: refugee protection. Our 160,000 heavily-armed troops are more than enough to protect, transport, and resettle those Iraqis who have not yet become refugees but likely will as the civil war grinds toward completion.  As discussed above, one cannot tell from UNHCR data just how many likely future refugees remain, but U.S. commanders on the ground – and local Iraqi community leaders – know.  We should identify the 150 to 200 towns, villages, and urban districts that are most at risk for ethnic cleansing - and sit on them for days, weeks, or months until we can organize well-defended transport for those who wish to move.  The U.S. should also contribute mightily to refugee aid; we broke Iraq and we broke these people's lives.  We should pay the bill, or as much of it as can be paid in money.

Eventually there will be borders separating three regions, each dominated by one community – even if a border runs right through Baghdad.  Self-preservation will cause both sides to make sure that that line will be very difficult to cross; the barriers and check points that US forces have recently pulled down in parts of Baghdad will go back up and will be multiplied many-fold.

According to U.S. intelligence, to some extent separation of communities in Iraq is reducing violence,[122].but the partition borders will be stable only to the extent that they are defensible, which means that they must be negotiated by the factions either on or off the battlefield, with or without mediation by other powers but in any case not imposed by the U.S. 

Beyond this, we must rethink how to stabilize the region's future.  There is reason to fear that the partition of Iraq may do less to reduce violence than do most communal partitions.  Even with Sunni and Shia populations largely separated, Iraq will remain unstable: its flat terrain, relatively good roads, and the world's third largest oil reserves will provide both opportunity and motivation for losers in one round of fighting to try again later. Neighboring powers, including Turkey, Iran, and Saudi Arabia will also have powerful motives to intervene.

It would be desirable to shift the balance of military strength somewhat further in favor of the Shia and/or the Iraqi government; borders are more stable when the more satisfied side is also the stronger side.  It would also be useful to shift the offense-defense balance in favor of defense: lots of barbed wire and anti-tank rockets for everyone, but no tanks or fighter-bombers.  This aid should be provided as part of the political settlement but, ideally, not by the United States, which has no interest in further angering anyone in the region.

The wider Persian Gulf region has also been destabilized. Iran and the Shia rump of Iraq will be the region's natural dominant powers, while current U.S. allies on the Southern side of the Gulf are weak not only externally but internally. If they survive it will be by reform or by accommodation.  American extended deterrence based on military power alone will no longer suffice; regular consultation among all interested powers will be necessary.

Two prominent additional proposals for federalism or "soft partition" in Iraq are those of Senator Joseph Biden and Michael O'Hanlon (despite the fact that O'Hanlon also supports the surge).[123]

Biden's recommendations are not very different from mine: first, three autonomous regions with a limited central government responsible for functions such as defense, foreign policy, and oil.  This would not end the Sunni insurgency, but would, Biden argues, make it easier for Sunni elites to marginalize extremists.  Further, much as many Sunnis and Shia dislike federalism, they will accept it: "The Shia know that they can dominate the government, but they can't defeat a Sunni insurrection. … Until recently, the Sunnis sought a strong central government because they believed they would retake power. Now, they are beginning to recognize that they won't and that the greatest danger to their interests is a highly centralized, Shia-run state."[124]

Second, the other communities must "guarantee the Sunnis a fair share of Iraq's … oil reserves," administered through the central government with international supervision: "Iraq's oil wealth would be the glue that keeps the country together."  This will require amending the constitution; a second amendment will also be needed to authorize a Sunni autonomous region in addition to the Kurdish and Shia ones already provided for.  All communities stand to gain: "Petroleum experts agree that the Iraqi oil industry will attract more desperately needed foreign capital if it is run as a unified whole. Shi'a and Kurds will get a slightly smaller piece of a much larger pie. That's a better deal than they would get by going it alone, which will not attract the needed investment. At the same time, guaranteeing Sunnis a piece of this pie will reduce the incentive of insurgents to attack the oil infrastructure. That would be good news for everyone."[125]

Third, the U.S. should increase investment in economic reconstruction of Iraq.

Fourth, there must be a regional conference "where Iraq's neighbors, including Iran, pledge to respect Iraq's borders and behave cooperatively."

Fifth, although most U.S. forces would be withdrawn by the end of 2008, some 20,000 would be permanently stationed in Kurdistan.

Biden's plan includes one requirement –constitutional change to guarantee oil revenue-sharing – that the record suggests may not be achievable.  Second, it makes no provision for protecting the refugees who would be generated by partition, but should.  Third, keeping U.S. forces in Kurdistan would create a moral hazard that could bring disaster on the Kurds in the long run.  At some point—no one can now guess when or in what circumstances—those forces and the security guarantee may be withdrawn.  If the Kurds have not spent the intervening time improving relations to prepare for that day, they will find themselves horribly exposed.

O'Hanlon suggests the more limited objective of trying to protect individual Iraqis from ethnic cleansing or potential genocide, and that this should be done without preconditions as to how many political units might comprise Iraq or what their boundaries or powers might be. "We should offer individuals who want to protect themselves and their families the chance to move to an Iraq territory more hospitable to their ethnicity and/or religion. … The international community and Iraqi government could help offer housing and jobs to those wishing to move, as well as protection en route. … Obviously, this idea would only work if Iraq's government, through a strong consensus of its Sunni Arabs, Shiites and Kurds, endorsed it," which O'Hanlon grants does not seem likely soon.[126]  This goal is certainly worthwhile, but the idea that nothing can be done without universal agreement is probably wrong.

Two prominent opponents of partition are Juan Cole and James Fearon.  Cole opposes partition, including the "soft partition" implied by a nine-province Shia autonomous region, on two grounds.  First, Cole rejects security dilemma logic entirely, predicting that partition "would probably not reduce ethnic infighting.  It might produce more.  The mini-states that emerge from a partition will have plenty of reason to fight wars with one another, as India did with Pakistan in the 1940s and has done virtually ever since."  The fallacy here is that one cannot the human cost of a partition to a hypothetical peaceful settlement that was not available in the Indian case rather than to the actual.alternatives in 1947.  The actual alternative was large-scale civil war that would have been more destructive than partition.[127]

Second, the war could spread: "The Sunni Arab mini-state [might] commit atrocities against the Shiites, [which] might well bring in the Iranian Revolutionary Guards. They in turn would be targeted by Saudi and Jordanian jihadi volunteers.  Worse, "a break-up of Iraq might not stop at Iraq's borders.  The Sunni Arabs could be picked up by Syria [or] could become a revolutionary force in Jordan.  A wholesale renegotiation of national borders may ensue, [which] cannot be depended on to occur without bloodshed."  Related, a Shia autonomous region in Southern Iraq would likely "fall into the orbit" of Iran, a dangerous development for Saudi Arabia and the United States.[128]  These particular dangers may not be very likely—Iraqi Shia will be capable of responding to any Sunni provocation without Iranian troops, and few observers propose dissolving Iraqi sovereignty.  The larger point, however, is that absent active international co-operation by states that do not now have good relations there will be a substantial risk of regional conflict escalation in some form.[129]  But avoiding acknowledgement of Iraq's collapse won't reduce the risk.

Fearon objects to partition on three grounds.[130]  First, that the border separating the Sunni and Shia regions would not be very defensible.  This is a valid point, as mentioned above; manipulation of the military balance and engagement of regional powers can reduce this threat, but it is unclear how much.

Second, partition would likely increase rather than reduce the rate of killing.  Whether this is so is hard to guess, since both Sunni and Shia ethnic cleansers are killing as fast as they can.  Also, so many people have already fled dangerous mixed-community areas that the number of potential additional refugees is not known.  Most important, 160,000 U.S. troops can do a lot to relocate people safely.

Finally, dividing Iraq into three parts would create economic inefficiencies.  Since the option of a unified, peaceful Iraq is not available, this is irrelevant.

Final Thoughts

The situation in and around Iraq today is reminiscent in many ways of the Balkans in the late 19th century.  Faced with a deeply unstable configuration of competing local nationalisms and major power interests, the European major powers sought to manage the region through a series of conferences, beginning with the famous 1878 Congress of Berlin.  In the end they failed to avert World War I, but it is not obvious that the effort was doomed from the start.  Iraqi factions, regional powers, and global powers must manage their conflicting interests in the new situation in the Gulf to minimize risk of an even larger war that no one wants.  All parties must explain to others what policies they can expect, and must learn from rivals what they cannot tolerate.

Something like a "Congress of Amman" is called for now. Admittedly, few of the parties show signs of being ready to talk constructively—least of all the two most important, the United States and Iran. Perhaps a sufficient achievement for the First Congress of Amman would be agreement that there will be a Second.

[1]. One could identify a third war in which Coalition forces and the Iraqi government confront insurgents in the mainly Sunni provinces of Western Iraq, but this has always overlapped with the Sunni/Shia civil war.  Insurgents attack the government in large part because it is Shia-dominated and U.S. forces in part because they are seen as allied to that government and to the Shia community.  In addition, there are rivalries within the both Sunni and Shia communities which have sometimes been violent, although these have not led to significant cross-ethnic alliances or accommodation and thus do not effect the essentially communal natures of both wars. 

[2]. Iraqi law does not run in Kurdistan, nor is the Iraqi flag flown.  A limited effort in 2007 to bring Kurdish forces to assist with population security in Baghdad was undermined by the failure of most of the assigned troops to show up.

[3]. Leaving aside Russia's eight-day contribution to the Pacific War.

[4]. In rational choice neither elites nor masses are price takers; they bargain hard with both friendly and hostile factions, although within the bounds of possible deals limited by their capabilities and the possibilities inherent in the situation.

[5]. "Rebellions are motivated by greed, which is presumably sufficiently common that profitable opportunities are not overlooked."  David Collier, Anke Hoeffler, and Nicholas Sambanis, "The Collier-Hoeffler Model of Civil War and the Case Study Project Research Design," in Collier and Sambanis, eds., Understanding Civil War: Evidence and Analysis, Volume 1: Africa (Washington, D.C. World Bank, 2005), pp. 1-35 at 3.  "Our starting point finds the rationale for groups in the problem of social order [due to the] great variety of human transactions and interactions [that] involve the possibility of opportunism, [such as] cheating, shirking, malfeasance, fraud, exploitation, embezzlement, extortion, robbery and rape."  James D. Fearon and David D. Laitin, "Explaining Interethnic Cooperation," American Political Science Review vol. 90, no. 4 (December 1996), pp. 715-35 at 717. See also Fearon and Laitin, "Weak States, Rough Terrain, and Large-Scale Ethnic Violence since 1945," paper presented at the annual meeting of the American Political Science Association, Atlanta, 1999; and "Ethnicity, Insurgency, and Civil War," American Political Science Review vol. 97, no. 1 (March 2003), pp. 75–90.

[6]. Stathis N. Kalyvas, The Logic of Violence in Civil War (Cambridge, UK: Cambridge University Press, 2006).

[7]. Fearon and Latin, "Explaining Interethnic Cooperation."

[8]. Collier, Hoeffler, and Sambanis, op. cit.; Fearon and Laitin, "Weak States, Rough Terrain," and  "Ethnicity, Insurgency, and Civil War."

[9].  While Sunni rebels did receive small outside funding in the early years of the rebellion and have received more as the rebellion has grown over the years, these flows have been dwarfed by U.S. and other aid to the government and by the additional sources of aid (e.g., from Iran) that could be made available to the Shia.

[10]. Diamond, "End Game: Why the United States Should Not Go it Alone," Hoover Digest, January 1, 2003.  On Diamond's policy involvement see his Squandered Victory: The American Occupation and the Bungled Effort to  bring Democracy to Iraq, With a New Afterword (New York: Henry Holt, 2006), pp. 9, 11.

[11]. Diamond, "Can Iraq Become a Democracy?"  Hoover Digest and Orange County Register April 6 and 13, 2003.

[12]. "If democracy can be initiated only when all the major stakeholders believe in free and fair elections, justice, liberty, and the rule of law, few democracies would exist in the world today."  Diamond, Squandered Victory, p. 318.

[13]. Diamond, p. 334.

[14]. Diamond, p. 320.

[15]. Ibid., pp. 322-23.

[16]. Ibid., pp. 322-23.

[17]. Ibid., p. 313.

[18]. Ibid., pp. 79-80.

[19]. Kalyvas, The Logic of Violence in Civil Wars, pp. 389-91.

[20]. "Civilians use political actors to settle their own private conflicts.  Put otherwise, civilians turn political actors into their own private contract killers."  Ibid., p. 14.  See also pp. 89, 127, 143-44, 151-53, 172, 176.

[21]. "Sectarian violence is hardly the only (and possibly not the dominant) form of violence in Iraq."  Stathis N. Kalyvas and Matthew Adam Kocher, "Ethnic Cleavages and Irregular War: Iraq and Vietnam," Politics & Society vol. 35, no. 2 (June 2007), pp. 183-223 at 187.

[22]. Fearon and Latin, "Explaining Interethnic Cooperation," p. 715.

[23]. Ibid., p. 719.   Fearon and Laitin further emphasize, p. 723, that in their main scenario "individuals have a positive incentive to sanction people [of their own group] who have defected, no matter against whom."

[24]. Pp. 718-19.

[25]. Pp. 721-23.  Although not mentioned specifically, a second, implied, reason why elites should stop community members from cross-communal misbehavior is that this could undermine the pursuit of material gain by many individuals in both groups.  Diamond appears to agree on the importance of in-group policing: "Some of the actors responsible for the insurgent and terrorist violence cannot be [co-opted] at any price. …If Iraq is to become a democracy, these die-hard elements … must be isolated – separated from their networks of support."  Diamond, p. 320.

[26]. Pp. 728-29.  "People who exploit the trust of others can be identified as individuals and sanctioned with relative ease by the response of the ethnic community," p. 719.

[27]. P. 726. 

[28].  "The more often individuals interact outside their own group the less effect is the threat of in-group sanctions;" p. 722; see also p. 718.

[29]. According to Collier in 2005, "The key priority [in Iraq] will probably not be infrastructure, but the building of credible institutions of social inclusion, and checks and balances that restrain the corrupt politics of patronage."  Paul Collier, "Iraq: A Perspective from the Economic Analysis of Civil War," Oxford University, June, 2005.  This recommendation hasn't couldn't have been implemented because, as normal for the economic model, it ignores the barriers placed by communal fears and mistrust.

[30]. Fearon, "Iraq's Civil War," Foreign Affairs vol. 86 no.2 (March/April 2007, pp. 2-15 at 10.  Fearon also points out that Shia politics are divided into, by his count, four major factions.  Most observers count two dominant armed forces, al-Sadr's Mahdi Army and the Badr Corps of the Supreme Islamic Iraqi Council (SIIC, formerly known as SCIRI).  Whatever our count, intra-Shia divisions should matter to in-group policing only if some strong factions favor violence against Sunnis while others oppose it, which is not the case.

[31]. Quoted in Michael Wilkerson, "No Bright Side in Iraq," The Stanford Daily, October 4, 2006.  Although Fearon does not unpack the concept in this article, under his theory "anarchy" should mean not the absence of government but rather the absence of authorities within groups capable of policing their own communities.

[32]. E.g., James Dobbins, "America's Role in Nation-Building," (Washington, D.C.: RAND, 2003).

[33]. The Army and Marines together have roughly 300,000 deployable personnel, of whom several tens of thousands are needed for Korea and other commitments, leaving well under 150,000 for Iraq even if every eligible soldier serves a full-year rotation every second year. Maintaining an average of about 150,000 has required use of extended deployments, "stop-loss" orders, and the National Guard.  Robert Pape, "The 500,000-Man Myth" (April 2006, unpub.); see also An Analysis of the U.S. Military's Ability to Sustain an Occupation of Iraq (Washington, D.C. Congressional Budget Office, September 3, 2003).

[34].  One is where there is so much "noise" in inter-communal relations that cross-communal attacks and within-group violence can't be distinguished from each other; the other where the community is so small that a large fraction of all the individuals making it up might agree to jointly prey on the other group.  "Explaining Interethnic Cooperation," pp. 723-25; 723. 

[35]. Kalyvas' model would also expect this, and some has occurred, but not enough to de-escalate the war.  Dawood Salman, "Sunnis Fight Off Insurgents," Institute for War & Peace Reporting, December 1, 2006.

[36]. Rational choice cannot explain why there should be anyone who cannot be bought any price, and must accept the existence of such people in any real situation as exogenous to the theory.

[37].  Even if extremists received financial support, e.g. from Saudi Arabia, moderates should have been able to obtain even greater financial and military support from the U.S. 

[38]. Fearon, "Iraq's Civil War," p. 10.  Actually, fragmentation of the Shia into a handful of strong factions would weaken but not remove incentives for in-group policing.   An offense committed by members of one Shia faction could, if unpunished, lead to Sunni retaliation spread across several factions or even bypass the offending faction.  Given the immense scale of Shia casualties, however, all Shia factions should have sufficient incentive to police themselves regardless of what others do.  Alternatively, in the hypothetical situation where the key difference among the main Shia factions was that one – irrationally – preferred all-out communal war, the others should have banded together to suppress the extremists.

[39]. The breakdown of Shia unity is commonly dated to the Sadrists' defection on the parliamentary vote on regional autonomy in October 2006.

[40]. We also have the problem that the point of the Sadrist defection was that they preferred a friendlier policy toward Sunnis. 

[41]. Peter Galbraith, "The Way to Go in Iraq," New York Review of Books, August 16. 2007.

[42]. It is worth noting that Kurdish claims that they had been in the majority in these areas before being driven out by the prior regime should have been of little relevance under either of the other two models.

[43]. On these categories, see Chaim Kaufmann, "Rational Choice and Progress in the Study of Ethnic Conflict: A Review Essay," Security Studies vol. 14, no. 1 (January–March 2005), pp. 178–207.

[44]. Stuart J. Kaufman, Modern Hatreds: The Symbolic Politics of Ethnic War (Ithaca: Cornell University Press, 2001).

[45]. For an example by Hakim, generally the least friendly toward Sunni interests, see John Ward Anderson and Saad Sarhan, "Another Shiite Site Is Bombed," Washington Post, April 9, 2006.

[46].  It remains possible that the public statements of Shia leaders are, and are intended to be, understood by their constituencies as meaning the opposite of what the words say, but there is little evidence for this.  Further, if this were the situation we should observe competitors willing to say openly what the existing elites are willing to say only in coded fashion, and we do not.

[47]. Barry R. Posen, "The Security Dilemma and Ethnic Conflict," Survival vol. 35, no. 1 (Spring 1993), pp. 27–47; Chaim Kaufmann, "Possible and Impossible Solutions to Ethnic Civil Wars," International Security vol. 20, no. 4 (Spring 1996), pp. 136-75. 

[48]. Sabrina Tavernise, "Cycle of Revenge Fuels a Pattern of Iraqi Killings," New York Times, November 20, 2006; John F. Burns, "Shiites Rout Sunni Families in Mixed Area of Baghdad," New York Times, December 10, 2006.  Even religiously mixed groups had difficulty remaining neutral.  Mazzen Abdulhameed, "Iraqi Tribes Dragged into Sectarian Strife,", October 25, 2006; Marc Santora, "Sectarian Ties Weaken Duty's Call for Iraq Forces," New York Times, December 28, 2006. 

[49]. Map from Iraq Support Unit CASWANAME Iraq Update, November 3, 2006, p. 2,  A June 2006 demographic map of Baghdad shows considerable segregation since 2003.  "Ethnic and Sectarian Tensions in Iraq: Baghdad," New York Times, June 25, 2006.

[50].  Diamond, p. 334.

[51]. Certification of Council of Representatives Elections Final Results (Baghdad: February 10, 2006), p. 2; actual certification took place January 20, 2006.

[52].  Juan Cole, Informed Comment,, October 27, 2005,

[53]. Fearon and Laitin, "Ethnicity, Insurgency, and Civil War."

[54]. Although in principle oil revenues would be shared, the central government would control revenues from "existing" wells, while the autonomous regions would retain revenues from "new" wells.  It was understood that in practice "new" wells would be construed to include virtually all production.  Michael O'Hanlon, "The Iraqi Constitution: Potentially Fatal Flaw," Washington Times, September 2, 2005; James Glanz, "Constitution or Divorce Agreement?" New York Times, October 9, 2005.  In October 2006 the Iraqi Parliament approved procedures for establishing these regions over the objections of the Sunni members and the followers of Moqtada al-Sadr (Sadr's support is strongest in Baghdad, which would not be included in the Southern region).  Kirk Semple, "In Victory for Shiite Leader, Iraqi Parliament Approves Creating Autonomous Regions," New York Times, October 12, 2006.

[55]. The draft law would have given the central government control of about one third of existing production and no new production.  Richard A. Oppel Jr. and Qais Mizher, "Shiite Leader Rejects Big Charter Changes, Frustrating Sunnis," New York Times, January 12, 2006.  Edward Wong, "Main Iraqi Blocs Reach an Accord on Oil Revenues," New York Times, February 27, 2007; Antonia Juhasz, "Whose Oil is it, Anyway?" New York Times op-ed, March 13, 2007.

[56]. The June draft would have given the KRG 7% of Iraq's oil from local production, plus 17% of distributions from a central pool – but other obligations were to come first.  The top three distribution priorities were to be: "A - Funding sovereign expenditures of the Federal Government and strategic projects of benefit to all to be agreed with the governments of the Regions and Governorates, provided that this does not impact the balance and needs of the governments of the Regions and the Governorates which are not organized in a region; B- Funding the Future Fund [an economic development fund] account in accordance with Article (7) of this law; [and] C - Funding the quota of the region of Kurdistan which amounts to (17%) of the remaining revenues after subtracting expenditures mentioned in (A) and (B) of this Article and until population census is held by the State." "Law of Financial Resources," draft, June 20, 2007, pp. 2-3,; KRG charts of annexes to June 20, 2007 draft law,; "Law of Financial Resources." ''Sectarian Fighting Overshadows Oil Law Debate in Iraq,'' Power and Interest News Report., May 23, 2007,

[57]. Joshua Partlow, "Iraqi Cabinet Approves Draft Oil Legislation," Washington Post, July 4, 2007; Alissa J. Rubin, "Iraqi Cabinet Moves Forward on Oil Measure," New York Times, July 4, 2007.  Nothing passed Parliament before it adjourned on July 30.  Ned Parker, "Key Laws or No, Iraq's Parliament Takes a Break; Legislators Fail to Act on Measures that the U.S. Sees as Benchmarks of Progress, Including the Sharing of Oil Revenue," Los Angeles Times. July 31, 2007.

[58]. The key articles of the Iraqi constitution are 115, which says that "All that is not written in the exclusive powers of the federal authorities is in the authority of the regions. In other powers shared between the federal government and the regions, the priority will be given to the region's law in case of dispute" and 110, which lists the exclusive powers of the central government; these do not include oil or other natural resources.

[59]. Parker, "Iraq's Parliament Takes a Break;" "Focus; Geopolitics; Iraq: Division Continues as Bush Feels the Heat," Petroleum Economist, August 3, 2007.  Christopher M. Blanchard, Iraq: Oil and Gas Legislation, Revenue Sharing, and U.S. Policy (Washington, D.C.: June 26, 2007), pp. 6, 9-12.  The current oil minister, Hussain al-Shahristani, is an independent member of the UIA.  "Iraq's Shrinking Government,", August 11, 2007,\Story.cfm\sidANA232219115053.

[60].  Virtue withdraw in March to protest sectarianism in government, the Sadrists in May to protest government cooperation with the U.S. "surge," the Sunni religious parties in June to protest the removal of (Sunni ) Parliament Speaker Mahmud al-Mashhadani as well as failure to resolve the constitutional and oil issues (they returned briefly in July, then left again), and the secular parties on August 6 to government ''built on the philosophy of sectarianism.'  Kathleen Ridolfo, "Former Premier Pushing New Plan For Reconciliation," RFE/RL, March 16, 2007; Sumedha Senanayake,"Iraq: Sunni Ultimatum Rocks Al-Maliki's Position," RFE/RL, May 9, 2007; and "Iraq: Sunnis Fear Being Forced Out of Government," RFE/RL, June 28, 2007; Stephen Farrell, "Bloc of 5 Ministers Threatens to Leave Iraq's Cabinet," New York Times, August 7, 2007.

[61]. Michael Sung, "Iraqi Kurd government adopts regional oil law," Jurist: Legal News and Research, August 8, 2007.

[62]. "The U.S. State Department's Oil and Energy Working Group, meeting between December 2002 and April 2003, also said that Iraq 'should be opened to international oil companies as quickly as possible after the war.' Its preferred method of privatization was a form of oil contract called a production-sharing agreement."  Antonia Juhasz, "It's Still All About Oil in Iraq," Los Angeles Times. December 8, 2006.  See also Philip Thornton, "Iraqis Could Lose $200bn of Their Money, Says Report; Western Companies to Cash in on Oil Supply Plan," The Independent (London), November 22, 2005.

[63]. "Occupiers Must Not Get Iraqi Oil: Sadr Bloc," Agence France Presse, July 5, 2007; Edward Wong and Sheryl Gay Stolberg, "A Draft Oil Bill Stirs Opposition from Iraqi Blocs," New York Times, May 3, 2007.  Among the effects of production-sharing agreements can be loss of control of the level of production (and of influence on price) and loss of sovereignty, including immunity of foreign oil companies from domestic courts and inability to legislate environmental or other regulation.  PSAs are relatively uncommon; they cover 12% of world oil reserves.  Greg Muttitt, Production Sharing Agreements: Surrendering Iraq's Resource Sovereignty, July 22, 2006, pp. 11-16, 19,

[64]. Iraq's oil reserves, the world's 3rd largest, are various estimated at 114-215 billion barrels.  Estimated costs of developing this production vary, but most estimates are negligible in comparison to a $70/barrel oil price.  Saudi reserves are 260 bn and production capacity is 10.5-11 mbd.   There are additional uncertainties in the calculation deriving from factors such as Iraqi domestic oil consumption (0.5 mbd in 2006) which presumably cannot be sold at world market prices.  Natural gas is omitted.  Energy Information Administration, Iraq Energy Data, Statistics and Analysis - Oil, Gas, Electricity, (Washington:  Department of Energy, August 2007);; and EIA, Saudi Arabia Energy Data, Statistics and Analysis - Oil, Gas, Electricity, (Washington, D.C.: Department of Energy, February 2007);

[65]. 2006 GNP estimate. World Factbook: Iraq (Washington, D.C., Central Intelligence Agency, 2007),  There would also be benefits of peace for other sectors of the economy as well as multiplier effects of the increased oil revenue.

[66]. Such discussions as have occurred seem to assume allocation of revenues by population, although if all actors were rational chances of agreement might be increased by considering a smaller shares for the Sunnis.  I can't find records of discussion of the other issues.

[67].  Imaginably, the Kurds could perceive other non-economic interests (i.e., hope of a long-term U.S. security guarantee) that could dominate their choices, this would make them more willing to accommodate Sunni demands as that is what the U.S. wants.

[68].  Peter Spiegel and Alexandra Zavis, "'Depth of Mistrust' in Iraq Unforeseen," Los Angeles Times, August 3, 2007.

[69].  Whether this behavior can be considered rational depends on what we assume about the U.S. administration's hierarchy of interests in Summer 2007.  If oil profits for friends of officials ranked above stability in Iraq, then PSAs might still be worth pursuing even if they reduce the odds of agreement and of peace, and even if they would likely be repudiated before long.

[70]. Fearon, "Iraq's Civil War," p. 10.  See also Barry Posen, "Exit Strategy: How to Disengage from Iraq in 18 Months," Boston Review (January/February 2006). 

[71]. Ibid., Fearon, p. 10.

[72]. Ann Scott Tyson, "Sunni Fighters Find Strategic Benefits in Tentative Alliance With U.S.," Washington Post, August 9, 2007.

[73].  Ned Parker, "Iraq Ready for Pullout, Premier Says," Los Angeles Times, July 15, 2007; Babak Dehghanpisheh, "Q&A: Iraq Shiite Leader on U.S. Withdrawal,", July 15, 2007.  The Kurds, of course, favor continued U.S. occupation.  "Iraq Warns of Civil War if U.S. Troops Withdraw," RFE/RL, July 10, 2007.

[74]. Chaim Kaufmann, "What Have We Learned About Ethnic Conflict?  What Can We Do in Iraq?" Harvard International Review vol. 28, no. 4 (Spring 2007), pp. 44-49 at 45.

[75]. Kirkuk Province has four times the oil reserves of Ninewa Province, only some of which is in the partly Kurdish region around Mosul, but this has not had a discernable effect on the intensity of communal fighting.  Kamil al-Mehaidi, Geographical Distribution of Iraqi Oil Fields and its Relation with the New Constitution (May 27, 2006),

[76]. Fearon, "Iraq's Civil War," also names humanitarian concerns among reasons for not arming the Shia.

[77].;; Iraq Index: Tracking Variables of Reconstruction & Security in Post-Saddam Iraq (Washington, D.C.: Brookings Institution, March 29, 2007)\fp\saban\iraq\index.pdf;

[78].  Since nearly 50% of killings by U.S. forces occurred during the initial conquest, the proportion of killings since May 2003 caused by U.S. forces would be lower.

[79]. "Year Four: Simply the Worst," Iraq Body Count Press Release 15, March 18, 2007,

[80]Iraq Index (Washington D.C.: Brookings Institution, August 16, 2007), p. 13,

[81]. UNHCR, Statistics on Displaced Iraqis Around the World (April 2007),

[82]. The UNHCR estimates that more than 300,000 people returned to Iraq between 2003 and 2006, mainly in 2004, but that most of these have subsequently become displaced within Iraq.  CASWANAME Iraq Update, p. 7; UNHCR Update on the Iraq Situation, November 1, 2006, pp. 2, 5;

[83]. Iraq Index, August 16, 2007, p. 33.  250,000 new IDPs in Iraq from 2003 to 2005 and just short of 900,000 more in 2006 and 2007 (through July).

[84]. Iraq Displacement: 2007 Mid-Year Review, International Organization for Migration, July 19, 2007,  402,000 new internal displacements during 2003-2005, and about 1,000,000 more between February 2006 and 2007.  These figures exclude the four Kurdish-controlled provinces.  The UNHCR has estimates of new IDPs since February 2006 but not totals since April 2003.  See also Hassan M. Fattah, "Meeting on Aiding 2 Million Iraqi Refugees Highlights Divisions," New York Times, July 27, 2007.

[85]. James Glanz and Stephen Farrell, "More Iraqis Said to Flee Since Troop Rise," New York Times, August 24, 2007; Iraq Index; Iraq Displacement: 2007 Mid-Year Review..

[86]. Iraq Index, August 16, 2007, p. 33.

[87]. Iraqis often describe communal militias as necessary for security even when complaining of banditry and intra-communal warfare.  On Shia opinion after several days of intra-communal fighting in Amara. See Sabrina Tavernise, "Iraqi Shiites in Favor of Mahdi Militia," New York Times, October 20, 2006.

[88].  This is also the view of the Iraqi government and the IRC.  According to an IOM survey, 63% of Iraqi DPs said they had fled because of direct threats to their lives and 25% said that they had been expelled by force.  Glanz and Farrell, "More Iraqis Said to Flee."  Rents in Karrada, a relatively secure Shia neighborhood of Baghdad, have increased in the past two years as those who can afford it have fled there from more dangerous parts of the city, although Karrada itself is not fully safe from car bomb attacks.  Richard A. Oppel Jr. and Qais Mizher, "Car Bombs Kill 12 People in One of Baghdad's Safer Neighborhoods," New York Times, July 24, 2007.  Sectarian cleansing was likely the main cause of displacement in 2006 as well.  "Shia, Sunnis Forming Iraq Ghettos," Al Jazeera, July 30, 2006; see also Ashraf al-Khalidi and Victor Tanner, Sectarian Violence: Radical Groups Drive Internal Displacement in Iraq (Washington, D.C.: Brookings/University of Bern, October 2006).

[88].  Map from CASWANAME Iraq Update, p. 3.

[89].  Presumably meaning U.S. and Iraqi Army operations.  CASWANAME Iraq Update, p. 6.

[90]. The UNHCR says that 47% of the new IDPs since February 2006 are in the South, which it assumes are mainly Shia, 36% in Central Iraq (Sunni), and 16% in KRG-controlled areas.  CASWANAME Iraq Update, p. 6.  According to IOM, 64% of 2006-2007 IDPs in Iraq are Shia, while UNHCR figures for destination countries of refugees show roughly 90% in Sunni-majority countries, suggesting that overall there are probably more displaced Sunnis than Shia, which accords with conventional wisdom on scale of ethnic cleansing activity.[90]

[91].  In April 2007, 37% of Americans favored complete withdrawal by the end of the year.  Gallup Poll, Roper Center, University of Connecticut, conducted April 23-26, 2007, released May 1.

[92]. Nicholas D. Kristof, "Cut and Walk," New York Times, December 5, 2006.

[93]. Kristof points out that more than three-fourths of Iraqis believe that U.S. departure would reduce violence in Iraq.  This is telling regarding the chances of U.S. forces "stabilizing" Iraq, but is a better measure of Iraqis' anger than it is a prediction of real consequences.

[94]. 59% of respondents to the April 2007 Gallup Poll favored complete withdrawal from Iraq by various dates between mid-2008 and the end of 2009.  Only 4% favored staying longer.

[95]. James A. Baker, III, and Lee H. Hamilton, co-chairs, The Iraq Study Group Report, December 6, 2006.

[96]. An additional advantage of rapid passage of a comprehensive oil law would be legal clarity for both Iraqis and investors.  Baker and Hamilton, Iraq Study Group Report, p. 56.

[97]. Baker and Hamilton, esp. pp. 36-38, 48-59.

[98]. Baker and Hamilton, pp. 6-7.  Reversal of draw downs should be ruled out: "As redeployment proceeds, military leaders should emphasize training and education of forces that have returned to the United States in order to restore the force to full combat capability."

[99]. "President Discusses Points Contained in His Report to Congress Concerning Iraq," New York Times, July 13, 2007.

[100].  Six of the "satisfactory" grades were for: 1) establishing committees to assist the U.S. 'Baghdad Security Plan;' 2) providing three brigades for operations in Baghdad, albeit  undermanned;  3) not allowing the Baghdad Security Plan 'to become a safe haven for outlaws;' 4) establishing joint U.S.-Iraqi security stations in Baghdad; 5) progress toward regional autonomy as provided for in the 2005 constitution; and 6) allocating most of, and spending some of, $10 billion in reconstruction aid.  One passing grade was unjustified, namely a claim of progress toward resolving the dispute over the 2005 constitution.  One was for a phony issue: ensuring rights of minority political parties in parliament – i.e., rights to speak and to introduce bills, not assurance of influence on outcomes.   The ten "mixed" and "unsatisfactory" grades were for 1) failure to make progress on scaling back de-Ba'thification; 2) on an oil law; 3) on an electoral law; 4) on an amnesty law to 'bring in' insurgents; 5) on disarmament of militias; 6) on reducing sectarian pressures on Iraqi military commanders; 7) on reducing sectarian bias in law enforcement; 8) on reducing sectarian violence and sectarian control of local security; 9) on increasing Iraqi forces that can operate independently; and 10) ceasing factionally and communally motivated accusations against members of Iraqi security forces (mainly Sunnis).

[101]. David S. Cloud and John F. Burns, "Bush to Declare Progress in Iraq on Some Fronts," New York Times, July 12, 2007.

[102]. Although the state aim of the "surge" had been to reduce violence to allow reconciliation at the national level, Gates implies that the focus would be shifted to reforms at local levels where possible.  Spiegel and Zavis, "Depth of Mistrust."

[103]. Director of National Intelligence, National Intelligence Estimate: Prospects for Iraq's Stability:

Some Security Progress but Political Reconciliation Elusive (Washington, D.C.: National Intelligence Council, August 2007).

[104]. Michael E. O'Hanlon and Kenneth M. Pollack, "A War We Just Might Win," New York Times, July 30, 2007.

[105].  In the same vein, "the Iraqi Army's highly effective Third Infantry Division [which] started out as overwhelmingly Kurdish [is now] 45 percent Shiite, 28 percent Kurdish, and 27 percent Sunni Arab." O'Hanlon and Pollack, "A War We Just Might Win."

[106]. O'Hanlon and Pollack, "A War We Just Might Win."

[107]. "Glenn Greenwald Busts the Pollack-O'Hanlon Trip to Iraq," BROADCATCHING, August 12, 2007,  For the full transcript, see Glenn Greenwald, "Interview with Michael O'Hanlon," UT Documents, August 11, 2007.

[108]. Pollack, quoted in Packer, George Packer, "Interesting Times: O'Hanlon and Pollack (2)," New Yorker, August 01, 2007.

[109]. "Iraqi politicians of all stripes continue to dawdle and maneuver for position against one another when major steps towards reconciliation -- or at least accommodation -- are needed. This cannot continue indefinitely. Otherwise, once we begin to downsize, important communities may not feel committed to the status quo, and Iraqi security forces may splinter along ethnic and religious lines."  O'Hanlon and Pollack also report as positive a development that one would expect them to count as pessimistic: "In Baghdad's Ghazaliya neighborhood, which has seen some of the worst sectarian combat, we walked a street slowly coming back to life [because] American soldiers and a mostly Kurdish Iraqi Army company [were] patrolling the street" in place of Shia police and Sunni militiamen.  "A War We Just Might Win."  They make no suggestion that Sunni/Shia relations in Ghazaliya have become more trusting or that 'outside policing' of this sort can be replicated across Iraq.  Pollack also told Packer that "everywhere he went the line [he] heard was that the central government in Baghdad is broken and the only solutions that can work are local ones."  Packer quote, "Interesting Times."  

[110]. Fearon, "Iraq's Civil War," pp. 13-14.  Kalyvas and Kocher offer no policy recommendation for Iraq except that they oppose partition and that "the seeming intractability of contemporary Iraq's ethnic conflicts is less a permanent state of affairs and more a temporary configuration of forces occasioned by war, state collapse, and foreign occupation.  Scholars and policy makers should pay less attention to supposedly intractable cleavages and more to the internal dynamics of violent conflict."  Kalyvas and Kocher, p. 217.

[111]. Edward Luttwak advocates this as a solution to communal wars generally.  Luttwak, "Give War a Chance," Foreign Affairs vol. 78, no. 4 (July/August 1999), pp. 36-44.

[112]. Fearon, "Iraq's Civil War," p. 15.

[113]. Ibid., pp. 10-11.  There is a contradiction here.  Fearon points to Shia fragmentation as a main barrier to power-sharing, but his vision just above of a small number of territorially-based actors should require no more than an intra-Shia truce.  Under a rational model it should not matter if more than one of the actors sharing power were of the same religion.

[114].  During the 1992-1995 war in Bosnia, the isolated Muslim enclave of Bihac was under little threat at any point; the local warlord, Fikret Abdic, did not recognize authority of the government in Sarajevo.  Inability to mobilize many troops for mobile service is common in communal wars, and often leaves one or both sides vulnerable to defeat in detail.  Israel won the first (civil war) stage of its War of Independence in part because it had a tiny mobile force, the Palmach, while the Palestinians did not.  Chaim Kaufmann, "Intervention in Ethnic and in Ideological Civil Wars: Why One Can Be Done and the Other Can't," Security Studies vol. 6, no. 1 (Autumn 1996), pp. 62-103.

[115]. In August 2007, two province governors were assassinated in what was said to be an SIIC-Sadrist dispute, both in far-Southern provinces.  Stephen Farrell, "Governor of Iraqi Province Assassinated," New York Times , August 21, 2007.  Even if Iraqi Shia perceived far stronger and more uniform danger than they do, several Shia armies might fight as well or better than one by allowing avoidance of contests over authority and spoils.  In 1948, Israel intentionally delayed integration of the Irgun into the Israeli Army until after the main battles had been won because the government expected, rightly, that integration could not be done without shooting.

[116]. Robert Malley and Peter Harling, "Iraq: The Enemy We Hardly Know", Boston Globe, March 19, 2006,

[117]. Fearon, "Iraq's Civil War," p. 13.

[118]. The last commonly called "civic nationalism" to distinguish it from "ethnonationalism."

[119]. For one of the earliest, see Leslie H. Gelb, "The Three-State Solution," New York Times, November 25, 2003.

[120]. Senator Joseph Biden also makes this point: Iraq "is on the verge of violently partitioning itself." Joseph R. Biden, Jr., "Breathing Room," The National Interest, Fall 2006.

[121]. "Most Iraqis Want U.S. Troops Out Within a Year," Program on International Policy Attitudes/Saban Center poll, September 27, 2006.

[122]. "Where population displacements have led to significant sectarian separation, conflict levels have diminished to some extent because warring communities find it more difficult to penetrate communal enclaves."  Prospects for Iraq's Stability, p. 9.

[123]. Amitai Etzioni, "Plan Z: A Community Based Security Plan for Iraq" (undated), summarizes several partition proposals as well as some criticisms.

[124]. Biden, "Breathing Room."

[125]. Ibid.

[126].  Michael E. O'Hanlon, "Voluntary Ethnic Relocation in Iraq?" Los Angeles Times, August 27, 2006.

[127]. Chaim Kaufmann, "Was the Partition of India Avoidable?" (2007, unpub.).

[128]. Juan Cole, "Partitioning Iraq: Would Dividing the country decrease ethnic infighting or lead to more fighting and inflame the Middle East?", October 30, 2006.

[129]. Helene Cooper, "Saudis Say They Might Back Sunnis if U.S. Leaves Iraq," New York Times, December 13, 2006.

[130]. Fearon, "Iraq's Civil War," p. 14.