My Enemy’s Neighbour
An Analysis of the Impact of Identity-Mobilised Counterinsurgent Militia on Intrastate Conflict Intensity, 1991-2015.
Benjamin D. McCann
Word count: 14,937
S2609479
2 Contents:
1. Introduction 3
2. Theoretical Perspectives on Militia and their Roles
4
3. Constructivism & Ethnic Identity 8
3.1. Ethnic Identities in Civil War 10
4. Co-Ethnic Militia: Benefits of Deployment
11 4.1. Co-ethnic Militia & Strong
State Guarantors
13
4.2. Rival Ethnic Militia 14
5. Methodology 16
5.1. The Dataset: Case Selection & Data Gathering
18
5.2. Empirical Strategy 19
5.3. Limitations 20
6. Data Results & Analysis: Multiple Regression Analysis
22 6.1. Data Analysis: Theoretical
Implications
7.1 Case Study: The Sons of Iraq 25
7.2 Case Study: The SLM-MM 29
7.3 The SOI & The SLM-MM: Identity Defection, Strong State Guarantors and Militia Strategy in Counterinsurgency Campaigns 32 8. Conclusion 35 9. Bibliography 10. Appendix 36 45
3 1. Introduction
Counterinsurgent forces recruited from the shared terrain of the insurgent group, both in literal terms and by way of human geography and demographics, are those of anti-insurgent local irregulars or militia. Such auxiliaries have a long lineage in scholarly and military-practitioner counterinsurgency literature and have been considered as advantageous strategic augmenters in civil war for many generations of military leaders. Clausewitz & Machiavelli similarly note the enhanced security in both rural and urban environments under the protection of local militias.1
Despite these seemingly primordial roots, local militias are deeply engrained in the consciousness of modern counterinsurgent logic.
In the classical theatre, the British liberally deployed units of indigenous ‘Home Guards’ when confronted with the apparent, if embellished, threat of the Mau Mau rebellion in Kenya,2 and
during the Malayan emergency, auxiliary Home Guards peaked at 250,000.3 Dixon suggests
this British tradition extended with attempts to initiate cooperation with some militia groups during the Troubles, with those experiences informing British Lt. Gen. Graeme Lamb’s decision to make the initial contact with Sunni tribal leaders that preceded the Anbar Awakening in Iraq.4
Comparably, Galula describes his own attempts to organise self-defence militias in Algeria and efforts to co-opt armed locals to serve alongside French imperial troops.5 The Janjaweed in Darfur, Kadyrovtsy in Chechnya, Citizen Armed Force Geographical
Units (CAFGU) in the Philippines and Mayi Mayi militia in the eastern regions of the D.R. Congo are indicative of the phenomenon’s continuation in the 21st century and suggestive of
its applicability as a relevant strategic approach employed by incumbent states and communities caught in the midst of intrastate conflict.
Despite the impressive pedigree of militias in Counterinsurgency operations, the 2006 edition of FM 3-24 recommends a cynical attitude should be assumed towards non-state armed groups, suggesting that local militias damage the legitimacy of the ‘Host Nation’ by undermining its claim to effectively provide security itself. The irregulars instead represent a dangerous tumour
1
Schneckener, ‘Militias and the Politics of Legitimacy’, Small Wars & Insurgencies, 28:4-5 (2017). 2
Anderson, Histories Of The Hanged: The Dirty War In Kenya And The End Of Empire (Phoenix, 2005). 3
Hack, ‘Everyone Lived in Fear: Malaya and the British Way of Counterinsurgency’, Small Wars & Insurgencies, 23:4-5, (2012).
4
Dixon, ‘’Hearts and Minds’? British Counter-Insurgency From Malaya to Iraq’, Journal of Strategic Studies, 32:3 (2009), 373-374.
5
4
in the security environment and should be tracked “just like insurgent and other armed groups”.6
This tone was adjusted in the manual’s 2014 edition, likely an epiphany brought about after observing the beneficial role irregular Sunni militias played during the Anbar Awakening in Iraq. The manual now encourages its readership to consider that irregulars can play a “constructive role” and their ability to provide security can also “be a building block to help build legitimacy at the local level”.7
Many of the anti-insurgent irregular organisations mentioned in the opening paragraphs of this study spawn from the very identity groups that ethno-nationalist or religious insurgent organisations claim to represent, presenting the phenomenon of identity ‘defection’ to the counterinsurgent regime. Other counterinsurgent militia, on the other hand, are products of mobilisation by ‘rival’ identity groups that perceive the rise of identity-driven insurgencies as a threat to their own physical or political security. This thesis aims to evaluate the proposed ‘constructive role’ played by local militias by assessing the consequences of identity-based irregular counterinsurgent mobilisation on intrastate conflict. More specifically, this study attempts to distinguish how the identity of militia groups influences the intensity of intrastate asymmetrical violence.
2. Theoretical Perspectives on Militia and their Roles
Irregulars can be deployed in a variety of contexts and assigned a diverse set of roles, yet Civilian Defence Forces (CDF) are among the most thoroughly examined form of pro-government irregulars.8 Peic, who describes such militias as populated by civilians, are
characterised by their predominantly defensive duties and defined by their deployment in their immediate local communities and area.9
Other studies identify CDFs with even tighter criteria, remarking that they are exclusively restricted to their local area from which they were initially recruited, and reiterate their defensive nature; playing a potentially key role in providing security for their own communities or, at most, engaging in limited combat operations.10 Such
6
U.S. Department of the Army, Field Manual 3-24, Counterinsurgency, (US Department of the Army, 2006), 3-20. [Other depictions of the threat of local militias on pages: 3-11, 1-9].
7
US Department of the Army, FM 3-24/MCWP 3-33.5, Insurgencies and Countering Insurgencies, (US Department of the Army, 2014): 5-7.
8
Clayton & Thomson, ‘The Enemy of My Enemy is my Friend: The Dynamics of Self-Defense Forces in Irregular War: The Case of the Sons of Iraq’, Studies in Conflict and Terrorism, 37:11 (2014)
9
Peic, ‘Civilian Defense Forces, State Capacity and Government Victory in Counterinsurgency Wars’, Studies in Conflict and Terrorism, 37:2 (2014): 165.
10
5
definitions also find consensus on the composition of these forces, explaining regular military troops are absent from such militias, and are composed of civilians and, in some cases, former insurgents.11
To argue, however, that irregular forces perform only defensive duties would be misleading. A definition proposed by Jentzsch, Kalyvas & Schubiger offers a slightly broader depiction of civil militia as “armed groups that operate alongside state security forces or independently of the state, aiming to shield local populations from rebel demands or depredations and seeking to acquire its loyalty or collaboration”.12 This classification, though
still focused on a militia’s defensive traits, offers a more amicable definition insofar as it considers militias from an anti-insurgent perspective, regardless of its level of integration with the state. This offers a healthy departure from restrictive ‘pro-government militia’ (PGM) labels that prolong the narrative that depicts civil war as an affair between two distinct actors; the incumbent state and the rebel group. Such labels arguably neglect the agency of local actors and, by equating anti-rebel movements with pro-government forces, mistreat actors by aligning them to binary labels that oversimplify complex interactions between a plethora of diverse participants. Shneckener similarly critiques the pro-government label, explaining that the classification “underestimates the independent existence and long-term persistence of these armed groups.”13
The limits of the ‘pro-government’ label are further articulated by Anderson who explains that the anti-Mau Mau Kenyan homeguard were not fighting other indigenous people out of devotion for the colonial apparatus, rather:
“these people did not like colonialism. In taking a stand, these so-called loyalists were in fact motivated by more prosaic and personal concerns: by the interests of their families; by the need to protect their property; by their sense of social status and by their own values”.14
This paper contemplates counterinsurgent militia as a non-state armed group primarily recruited from the civilian population that, with varying levels of integration with the state, aims to violently expel insurgent influence and deter rebel coercion in their own local communities and areas, and commit to a variety of combat operations aimed at undermining or defeat the insurgent in their local area or, occasionally, further afield. This definition allows the full diversity of anti-insurgent counterinsurgents to be carefully considered, whether they
11
Ibid., & Peic (2014): 165. 12
Jentzsch, Kalyvas & Schubiger, ‘Militias in Civil War’, Journal of Conflict Resolution, 59:5 (2015): 756. 13
Schneckener, ‘Militias and the Politics of Legitimacy’, Small Wars & Insurgencies, 28:4-5 (2017): 800. 14
6
are static or mobile; highly integrated with the state and its command structures or occasionally violently targeted by such structures. To avoid monitoring infighting amongst fractured insurgent organisations, this definition also contends that militia must not be committed to overthrowing the state through violent means. Their primary goal should be countering the rebellion.
Before reviewing literature relevant to role of identity in counterinsurgent militia deployment, it is vital that the more general foundations of the potential consequences of irregular counterinsurgent deployment are recognised. Firstly, militia forces offer to improve security at a local level, nominally in the form of CDFs. Though not exclusively defensive, the central role irregular groups play in providing local security from rebel violence is vitally important to counterinsurgent operations. Whilst the ‘Provision of Security’ element is a crucial first step in the greater ‘Hearts and Minds’ (H&M) dogma, many forces lack the regular manpower to contest rebel influence in the often-remote terrain favoured by insurgents. Local forces, regardless on their level of integration with the PCO, can play an important part in the counterinsurgent effort, acting as bulwarks against insurgent pressure over civilian populations that the PCO is either unable or unwilling to counter.15
This thereby alleviates the strain of defensive duties for the official security forces as the state can refocus its own forces to offensive operations against the rebels.16
Forces that arise from mutual human and physical environments as insurgents are also more likely to possess a greater understanding of both the local geographical landscape and the human terrain. This expertise can enrich the counterinsurgency effort in several ways. Firstly, scholars identify that the topography of a given conflict environment can have important effects on the onset, endurance and outcome of insurgency campaigns. In Fearon & Laitin’s study, more mountainous terrain correlated positively with higher chances of civil war occurrence17
and comparable conclusions or analysis determining terrain as a key ingredient for prolonged conflict are echoed in the research of Buhang et. al.18, Hendrix,19 and appear in Collier &
Hoeffler’s models of intrastate violence.20 Much of this work suggests that guerrilla tactics are
15
Peic (2014): 165. 16
Jentzsch, Kalyvas & Schubiger (2015): 758. 17
Fearon & Laitin, ‘Ethnicity, Insurgency & Civil War’, The American Political Science Review, 97:1 (2003): 85.
18
Buhaug, Gates, & Lujala, “Geography, Rebel Capability, and the Duration of Civil Conflict,” Journal of Conflict Resolution, 53:4 (2009).
19
Hendrix, ‘Head for the Hills? Rough Terrain, State Capacity and Civil War Onset’, Civil Wars, 13:4 (2011). 20
Collier, Hoeffler, & Rohner, ‘Beyond Greed and Grievance: Feasibility and Civil War’,Oxford Economic Papers, 61/1 (2009).
7
especially suited to rough terrain as mountainous regions diminish the conventional advantages of regular, mechanised militaries, whilst also offering insurgents sanctuary that is inaccessible to state forces.21
Local militia present an opportunity to destabilize an insurgent’s solace found amongst rugged terrain. Contrary to expeditionary or indigenous state forces, local groups from rugged areas possess a deeper level of familiarity with their surroundings that have been infiltrated by rebels.
Secondly, scholars note the dependency of successful population-centric counterinsurgency operations on sufficient understanding of the local human terrain. A greater awareness of local socio-cultural norms and community networks can more aptly inform a counterinsurgent regimes ability to craft policy that targets local grievances, and avoid generating new ones. Duyvesteyn, whilst admitting that an over-emphasis on cultural properties can mislead practitioners towards assumptions veering on cultural determinism, recognises that “a lack of cultural awareness can have far reaching consequences” on COIN campaigns.22 Indeed,
Kilcullen’s appeal that officers involved in COIN operations should “know the people, the topography, economy, history, religion and culture. Know every village, road, field, population group, tribal leader and ancient grievance”23
further implies the benefits of local mobilisation, with local irregulars well-suited to relate and respond to these environmental features. This discussion on cultural awareness will resume, with additional relevance, upon reviewing co-ethnic militia.
A local irregular’s acute awareness of local political structures and more organically formed social networks (relative to national or international COIN forces) also presents opportunities to build intricate local intelligence capabilities. Good intelligence is habitually recognised as a cornerstone of fruitful counterinsurgency and represents one of the few relatively uncontested norms of counterinsurgency theory. Kitson frankly asserted “the only general point which can be made about intelligence is that it is of the greatest importance”24
. A plethora of cases, transcending the neo-classical/‘post-classical’25, democratic/authoritarian,26 and
population-centic/enemy-centric divides, tend to inform a consensus of the significance of intelligence in
21
Hendrix (2011): 348-349. 22
Duyvesteyn, ‘Hearts and Minds, Cultural Awareness and Good Intelligence: The Blueprint for successful Counter-insurgency?’, Intelligence and National Security, 26:4 (2011): 452.
23
Kilcullen, Twenty-Eight Articles": Fundamentals of Company-Level Counterinsurgency, (2006) and mirrored in Kilcullen, ‘Counter-insurgency Redux’, Survival, 48:4 (2006): 124.
24
Kitson, Low Intensity Operations (Faber & Faber, 1972): 76-77. 25
Kilcullen, ‘Counter-insurgency Redux’, Survival, 48:4 (2006): 124. 26
Ucko, ‘’The People are Revolting’: An Anatomy of Authoritarian Counterinsurgency’, Journal of Strategic Studies, 39:1 (2016).
8
all forms of counterinsurgency operations. Often coinciding with aforementioned principles of situational awareness, the core of COIN intelligence gathering focuses on the human component of intelligence gathering, or HUMINT, relying on networks of informants and collaborators imbedded in local communities or in the very insurgent organisations themselves.27 There are innumerable examples of local forces, operationalising their existing
social networks in their local communities or region, generating detailed intelligence to more adequately challenge insurgent groups.28 The Kamajori militia that engaged in extensive
operations against the RUF in Sierra Leone, employed a special military police section that built elaborate clandestine intelligence networks composed of civilians and former insurgents, tasked to general intelligence gathering, target selection and monitoring the terrain for RUF activity.29 Clayton & Thompson point to the case of the Son’s of Iraq as an instance wherein
militia contributed to key intelligence infrastructure that ultimately transformed the nature of the conflict; especially building upon the primary COIN operator’s ability to distinguish between insurgents and civilians (thanks to local militiamen’s familiarity with the local population and their activities), encouraging the US-led coalition to exercise more discriminate targeting methods. 30
These consequences of militia deployment could be seen as pathways towards reducing the influence of insurgent groups, as well as their ability to engage in violence. The greater understanding of the environments the insurgents operate ensure irregular counterinsurgents can more efficiently disrupt insurgent operations, whilst a superior familiarity with local communities is likely to better inform counterinsurgent targeting and undermine the vicious circle of violence often generated by indiscriminate targeting by counterinsurgent regimes.
3. Constructivism & Identities
In 2005, Montgomary McFate lambasted the hypocrisy of the anthropologist field and its inertia to rise to the demands of a Pentagon that, entrenched in the quagmire of Iraq and Afghanistan, was in dire needs of cultural enlightenment. McFate concluded that “unfortunately, anthropologists, whose assistance was urgently needed in time of war, entirely
27
Duyvesteyn, (2011). 28
Staniland, ‘Between a Rock and a Hard Place: Insurgent Fratricide, Ethnic Defection and the Rise of Pro-State Paramilitaries’, Journal of Conflict Resolution, 56:1 (2012).
29
Mauna, ‘The Kamajoi Militia: Civil War, Internal Displacement and the Politics of Counter-insurgency’, Africa Development / Afrique et Développement, 22:3/4 (1997): 90.
9 neglect U.S. forces”.31
5 years later, Sluka responded with the assertion that the field of cultural anthropology must resist “the current efforts to weaponize our discipline”, pointing to ethical codes and norms and echoing the discontent vocalised by the American Anthropological Association for McFate’s stance.32 Both sides of this fascinating debate, deserving of its own
research, take as fact that there was a necessity for a healthier understanding of indigenous peoples, identity, cultures and intra-ethnic relations in America’s expeditionary counterinsurgency campaigns. Considering that most conflicts between 1946 and the present-day are drawn along ethnic fissures,33
it is vitally important to examine the role of irregular counterinsurgents mobilised along lines of identity, and their deployment and performance in the contemporary COIN arena.
The immensity and nuance of the debate on the definitions and changing aspects of ethnicity are impossible to summarise in a brief paragraph. As one researcher frankly noted: “defining ethnicity is a minefield”.34 Nonetheless, an understanding of the significance of ethnic
mobilisation on conflict dynamics demands some comprehension of the concept of ‘ethnicity’ itself. This research will swiftly recite two conflicting strains of theory depicting the nature of ethnic identities that may better illuminate our understanding of the consequent arguments: primordial and constructivist theories. Primordialist theories argue that ethnic and national identities are stationary, inherited characteristics built on shared traditional linguistic, religious, cultural and, in some cases, genetic components.35 Importantly, the endurance of these
apparatuses harden identities to the point that they are no longer subject to redefinition or adjustment.36 Conversely, constructivist thinkers contend that ethnic identities are “best
understood as a dynamic, constantly evolving property of both individual identity and group organisation”.37
In this light, ethnic identities are socially constructed, artificial phenomenon. Where primordialists insist ethnic identities are static and immutable, constructivists propose
31
McFate, ‘Anthropology and Counterinsurgency: The Strange Story of their Curious Relationship’, Military Review, 85:2 (2005): 37.
32
Sluka, Jeffrey A., ‘Curiouser and Curiouser: Montgomery McFate’s Strange Interpretation of the Relationship Between Anthropology and Counterinsurgency’, Political and Legal Anthropology Review, 33:1 (2010): 109. 33
Denny & Walter, ‘Ethnicity and Civil War’, Journal of Peace Research, 51:2 (2014): 199. 34
Green, ‘What is an Ethnic Group? Political Economy, Constructivism and the Common Language Approach to Ethnicity’, DESTIN Working Paper Series, (London School of Economics, 2005): 2.
35
For further discussion on debates over nature of ethnicity see Ozkirimli, Theories of Nationalism: A Critical Introduction, (Palgrave, 2010): 63.
36
Harvey, ‘Primordialism, Evolutionary Theory and Ethnic Violence in the Balkans: Opportunities and Constraints for Theory and Policy’, Canadian Journal of Political Science /Revue Canadienne de Science Politique, 33:1 (2000), 41 and Stein & Ayalet, ‘Ancestral and Instrumental in the Politics of Ethnic and Religious Conflict’, Ethnic and Racial Studies, 40:12 (2017): 1984.
37
Nagel, ‘Constructing Identity: Creating and Re-creating Ethnic Identity and Culture’, Social Problems, 41:1 (1994).
10
they are consistently fragmented and vulnerable to manipulation by political entrepreneurs.38
Despite primordialism’s minority position in contemporary debates on the topic, its relevance persists as an important theoretical bookend to track the diverging levels of engagement with constructivist ideas.
3.1. Ethnic Identities & Defection in Civil War
There is a tendency, especially when researching civil war, for commentators in the field of security studies to interpret ethnic groups as unflinching unitary bodies. Ethnically charged rebel groups supposedly have an easier time mobilising popular support amongst their constituent communities. Denny & Walter, for example, argue “ethnic identity tends to be less-elastic than other types of identity”,39
a portrayal also favoured by Kaufmann, who insists ethnic identities are “rigid” and “fixed by birth”.40
Though Kaufman concedes that ethnic groups are open to some level of modification, suggesting ethnic identities may ‘harden’ in the face of conflict, resulting in the conclusion that “hardly anyone fights for the opposing ethnic group”.41
Kaufmann’s description of ethnic groups fails to envisage the reality of cases where individuals sharing ethnic identities with the insurgent group launch armed campaigns against their own kin, crossing these supposedly impassable boundaries and shattering the primordialist depiction of internal ethnic blockades. The Homeguards in Uganda’s Acholiland, the
Kadarovsky in Chechnya, Moro peoples in CAFGU units in the Phillippines and the Kurdish Village guards assisting the Turkish campaign against the PKK are examples that illustrate the phenomenon of ‘ethnic defection’ is not limited to a particular group or region. Kalyvas explores the topic extensively in his seminal paper ‘Ethnic Defection in Civil War’.42 Kalyvas
disputes the assertion that ethnic identities are especially ‘sticky’ and intractable relative to ideologically-mobilised insurrections, instead pointing to a wide-ranging set of cases where ‘ideological defection’ has proven to be significantly challenging. He suggests that anti-rebel militia formed of individuals that share an ethnic identity with insurgents reaffirm the value of constructionist theories, demonstrating “multidirectional identity transformations…[are] more widespread than commonly assumed”,43 in contrast to unidirectional ‘hardening’ processes.
Though indicative of the broader relevance of constructivist theory, Kalyvas specifies that Civil
38
Brass, Ethnicity and Nationalism: Theory and Comparison, (Sage, 1991): 69-70. 39
Denny & Walter (2014): 199. 40
Kaufmann, ‘Possible and Impossible Solutions to Ethnic Civil Wars’, International Security, 20:4 (1996): 140. 41 Ibid.: 141. 42 Kalyvas, (2008). 43 Ibid.: 1045.
11
War poses an especially threatening environment to existing social networks and allegiances. Staniland further expounds the relevance of ethnic defection, suggesting the phenomenon alleviates primordialist arguments, whilst also exploring the processes that lead to ethnic fracturing.44
4. Co-ethnic Militia: Benefits & Consequences
The potential benefits of deploying co-ethnic militia (C-EM) anti-rebel forces are largely aligned with the previously described perks of non-ethnic militia (N-EM) deployment. Thanks to their knowledge of the population, terrain and greater cultural awareness, co-ethnic armed groups are far better equipped to tackle the demands of a counterinsurgency campaign. Lyall’s important work examining the impact of Chechen forces enlisted under the command of Russian armed forces in the 1999-2009 war informs much of the subsequent literature on co-ethnic militia and irregular forces more generally. Lyall observed a 33% decrease in insurgent attacks after Chechen-manned units conducted sweep operations compared to a 7% rise upon Russian-led sweeps. 45 He goes on to suggest co-ethnic units may impact conflict dynamics in
three ways. Firstly, greater affinity with the aggrieved population and greater concern about their welfare result in a fewer incidents of civilian targeting. This argument is reinforced by Stanton’s findings that militias that recruit from the same distinct ethnic group as the insurgents are less likely to target civilians.46
Secondly, a reduction of uncertainty over the strategic action taken by COIN forces and civilians, as ethnicity “acts as a kind of shorthand” that permits individuals to build informed decisions on how the other party will act. This poses an opportunity to remove rash, uniformed decisions that increase the likelihood of intensified violence. Furthermore, the Chechen COIN forces were manned by former insurgents. Their experience as insurgents provided such forces with better familiarity of how to identify, coerce and eliminate rebels more competently than Russian state forces.47
In a more recent study Lyall, Shiraito and Imai found that, in the case of militia mobilised against the Taleban in Afghanistan, civilians were more likely to cooperate in informing against insurgents when the non-combatant and the counterinsurgent shared an ethnic identity.48 Lyall’s work has
44
Staniland, ‘Between a Rock and a Hard Place: Insurgent Fratricide, Ethnic Defection and the Rise of Pro-State Paramilitaries’, Journal of Conflict Resolution, 56:1 (2012).
45
Lyall, ‘Do Coethnics More Effective Counterinsurgents? Evidence from the Second Chechen War’, American Political Science Review, 104:1 (2010).
46
Stanton, ‘Regulating Militias: Governments, Militias and Civilian Targeting in Civil War’, Journal of Conflict Resolution, 59:5 (2015): 913.
47
Lyall (2010): 14-16.
12
encouraged the notion that there is important a relationship between co-ethnic militia and counterinsurgency effectiveness.
Furthermore, co-ethnic militia formation can have significant consequences on the ability of the insurgent group to build cohesiveness in ethnic groups towards secessionist ambitions, as ethnic defections expose the fractured nature of ethnic identities.49 They further damage the
legitimacy of rebel groups by contesting the claim that insurgent groups are the singular representatives of a specific identity group and, in cases where co-ethnic militia are integrated with the state, potentially enhance the legitimacy of the state in the eyes of the aggrieved population by providing platforms of responsibility and participation in security governance.50
Though these benefits can bring about tidal changes in the dynamics of the conflict, co-ethnic militias formation can have undesired impacts on the nature of violence in civil conflicts. The previously described ‘revenge’ thesis proposes that individuals may be incentivised to join militia based on a desire to seek revenge for abuses against friends and family. Ethnically mobilised insurgency groups have an especially limited manpower pool, as their goals to secede or secure greater autonomy for their identity group often fail to gather wider support outside their constituent groups. Therefore, concerned that co-ethnic militia may challenge their grip over the prime commodity of manpower, insurgents are likely to indiscriminately target local communities suspected of collaborating with co-ethnic irregulars.51
This ultimately leads to a vicious cycle of increasingly violent civilian targeting and ethnic defection. The case of the LRA in Uganda marks an especially poignant example of this phenomenon. Kony’s forces began launching increasingly brutal attacks against the very Acholi peoples they claimed to be representing as Acholi enrolment in informal and state-run militia programs accelerated.52
Additionally, it has been argued that co-ethnic militia are especially unreliable and cannot be depended on by the state to remain loyal to the larger counterinsurgency regime. Already proven to be capable of defecting in one direction, Abbs, Clayton & Thomson suggest this capacity could be multidirectional, with defectors prone to side-switching back to the rebellion or deserting when threatened with prolonged periods of fierce violence.53
49 Jentzsch, Kalyvas & Schubiger (2015), 764.
50 Abbs, Clayton & Thomson, ‘The Ties that Bind: Ethnicity, Pro-government Militia and the Dynamics of Violence in Civil
War’, Journal of Conflict Research, 64:5 (2020), 908.
51Jentzsch, Kalyvas & Schubiger (2015), 759.
52Branch, (2005), 7, and Van Acker, ‘Uganda and the Lord’s Resistance Army: The New Order No One Ordered’, African
Affairs, 103:412 (2004), 335-336.
13 4.1. Co-ethnic Militia & Strong State Guarantors
If the benefits of co-ethnic militia are largely similar to those of local non-ethnic militia, should researchers expect to see that co-ethnic militias also form along along similar patterns? Kalyvas argues that, perhaps surprisingly, stronger states or powerful expeditionary forces are more likely to encourage ethnic defection than weaker states.54 Though Kalyvas does little to expand
on this hypothesis, I offer three suggestions that may clarify this proposal. The first suggestion, partly implied by Kalyvas, is that ‘strong’ states, unlike some post-colonial states, tend to have better consolidated power structures, are not dominated by a specific ethnic group at all, or are governed by a generally aligned coalition of ethnic groups. In these cases, devolving violence to a potentially rival ethnic group does not pose the same risk it would for teetering, ethnically dominated regime. To illustrate this point further, the example of the Kadyrovsky in Chechnya proves useful. If Kadyrov’s forces of ethnic Chechens eventually turned on Moscow, dominated by ethnic Russians, the relative size and influence of Russia’s armed forces, political stature and ethnic homogeneity would mean the Russian state would not be directly threatened, at least on an existential level, by such a rebellion. If, however, the state was an ethnically heterogenous, post-colonial polity dominated by one specific ethnic group, arming a politically relevant rival group poses the threat of both legitimising a rival ethnic group and posed the risk of unbalancing the structure of ethnic hierarchies that the incumbent group enjoyed.
The second and third arguments relate not to the incentives of the state to adopt defection programs, but the decision making of groups from aggrieved populations. Secondly, a simpler argument is one that aligns in part with Kaufmann’s perspective, suggesting ethnic groups are particularly ‘sticky’. If this is true, the enticements for ethnic defection must be significantly more attractive than they are for non-ethnic militia formation. In line with ‘opportunist’ perspectives of militia formation, where strong states may possess the necessary financial resources to lure potential co-ethnics into collaborationist organisations, where weak states lack the required economic manoeuvrability to effectively entice co-ethnic collaborators. Lastly, the incentives to side with strong COIN regimes may coincide with the simple desire to be on the winning side. If members of the aggrieved population doubt the incumbent regime’s aptitude to defeat the insurgent organisation, they may be more hesitant to defect in fear of being ostracised by their community. Counterinsurgency regimes composed of stronger states, or states with more comprehensive control over their territory, are perceived to be less likely to
14
lose the conflict, limiting the risk of insurgent victory and reparations against government collaborators. Siding with the perceived ‘front-runner’ could also ensure co-ethnics militia are granted positions of power in the post-war political configuration. To relate this to the Chechen case for example, Kadyrov was awarded the office of President in Chechnya for his loyalty to the federal state. While these arguments are especially linked to state-formed co-ethnic irregulars, community-formed ethnic militia are likely to arise in a similar context as local militia: in cases where insurgent groups are especially abusive towards their constituent group and the state fails to adequately provide forces to secure these populations.
4.2. Rival Ethnic Militia (R-EM)
There are a variety of scenarios where individuals from other identity groups will take up arms against the insurgency; from identity groups that have traditionally dominated a state’s power structures to ethnic minorities typically excluded from local and national decision-making; there appears no fixed profile for groups that engage in organised resistance. Explanations for this phenomenon vary as widely as the cases themselves. Primordialism offers a surface level justification for why identity communities come into conflict in a general sense. Fearon critiques how primordialist ideas became deeply embedded in the media’s discourse on ethnic violence in former Soviet republics. Such concepts were used to justify the surge of ethnically motivated conflict in the late 20th
century as the inevitable resurfacing of ancient ethnic feuds that had been recently inhibited throughout the dominance of communist political and social structures.55 What Fearon dubs the “age-old hatreds” theory, feeds off primoridalist
assumptions of hardened identities and inherited ancient grudges. Meanwhile, instrumentalist interpretations from Brass or Gagnon depend on the constructivist interpretation of ethnicity as malleable identities, susceptible to elite manipulation. Elites may broadcast socially constructed symbols, myths or discourses linked to a group’s perceived shared history to mobilise members of an identity group to violent action, often as a means to consolidate their grip on political or economic resources.56 Peterson suggests intergroup ‘hatred’ “need not reach
into ‘ancient’ times. All that is needed is enough time and tradition to establish a coherent
55 Fearon, ‘Ethnic War as a Commitment Problem’, Annual Meetings of the American Political Science Association, (August 30, 1995).
56
See Brass, (1994) and Gangon Jr, ‘Ethnic Nationalism and International Conflict’, International Security, 19:3 (1995).
15 schema”.57
It is also true that conflict itself can generate or deepen intergroup ‘hatred’. Further clarification of the ‘rival’ label is provided in the methodology section.
In many ways, the potential benefits of deploying irregulars composed of rival ethnic groups mirror the benefits of other types of co-ethnic and non-ethnic militia. Rival ethnic militia still possess a greater familiarity with local terrain and local population whilst offering to provide greater degrees of security for minorities targeted by insurgent groups intent on ethnic cleansing. However, in societies where intergroup mingling and socialising is less common, other ethnic groups may struggle to successfully discriminate between insurgent targets and non-combatants from an insurgent’s constituent population or successfully infiltrate the constituent population in order to extract intelligence.
Incumbent elites may also depend on militias comprised of the same identity group to covertly ‘outsource’ scandalous violent tactics. A series of scholars have suggested militias represent an opportunity for states to maintain plausible deniability for more nefarious violence committed in attempts to quell government opposition.58 Carey, Colaresi & Mitchell conclude
that the lack of culpability generated by militia presence is of strategic importance for developing states that depend on international aid from democratic powers who may be less willing to provide such assistance if it were to be revealed the beneficiary of such funds were engaged in human rights abuses or operating outside of humanitarian norms.59
Militias present a useful prospect for states inclined to engage in ruthless counterinsurgency practices whilst blurring the lines of accountability to an international audience. Though the outsourcing hypothesis may apply to all forms of militia, instrumentalism would hold that rival ethnic groups may be the best option for state elites wishing to devolve violence to irregular groups due to the relative ease they may be able to mobilise such ethnic groups against ethnically-motivated insurgencies. Furthermore, in neopatrimonialist political societies state officials may be able to exercise some level of informal control over militias organised within their ethnically homogenous patrimonial network, allowing some for some degree of direction and coordination with militias without explicit institutional accountability.
57
Peterson, Understanding Ethnic Violence: Fear, Hatred and Resentment in Twentieth-Century Eastern Europe (Cambridge University Press, 2012).
58
Campbell & Brenner, Death Squads in Global Perspective: Murder with Deniability, (Springer, 2002): 6. 5959
Carey, Colaresi & Mitchell, ‘Governments, Informal Links to Militias and Accountability’, Journal of Conflict Resolution, 59:5 (2015).
16 5. Methodology
This research project explores the relationship dynamics between insurgent groups, aggrieved populations and counterinsurgency forces, primarily by investigating how these dynamics influence conflict intensity. More specifically, the project seeks to understand how the demographic composition of certain counterinsurgency forces effect these dynamics through quantitative methods applied to an original dataset. Entry into the dataset requires two qualifying components: ‘militia’ and ‘counterinsurgent’. Firstly, to qualify as a militia the group must be armed, demonstrate some organisational structure, and enjoy some degree of autonomy from the state relative to regular state security forces. This rather broad classification diverges from the ‘non-state armed group’ label and seeks to accommodate to the high degree of variation between militia in their levels of integration with the state. As Krause & Milliken highlight, assigning the label of ‘non-state armed group’ “represents a barrier to utterly understanding the role and behaviour of armed groups that are also deeply entangled with state power and state agents in complex ways”.60 The criterion to qualify for the ‘counterinsurgent’
component of the militia group demands that (A) their core goal is the destruction of the insurgent organisation (qualifying the militia as counterinsurgents and not party-political driven death squads or other insurgent groups) and (B) they are recruited from geographically local communities to the insurgents constituent population. Lastly, this research only considers militia that amount to a strength of over 999 members.
This research approaches militias through the paradigm of the three ‘identity’ types explored in the preceding section:
1) Co-ethnic/identity: Applicable when the conflict is largely drawn across lines of identity, and the insurgent group exclusively or largely recruits from one distinct ethnic or religious group or coalition of aligned identity groups. This militia draws extensively on recruits from the insurgent group’s constituent population.
2) Non-ethnic/identity: In cases where the conflict is not drawn along lines of identity or driven by the grievances of a specific ethnic or religious groups, local non-ethnic militia may operate as irregular forces mobilised from local communities in the insurgent’s area of operation.
60 Krause & Milliken, ‘Introduction: The Challenge of Non-State Armed Groups’, Contemporary Security Policy, 30:2 (2009), 202.
17
3) Rival ethnic/identity: Applicable when the conflict is drawn largely across lines of identity and the insurgent group exclusively or largely recruits from one distinct ethnic or religious group or coalition of ethnic/religious groups that excludes the insurgent constituent group. The counterinsurgent force in question is composed of an especially conscientious group in the eyes of the insurgent group’s constituent population, and may be responsible for the grievances generating the onset or continuation of the insurgency.
In analysing the saliency of ethnic identities in conflict, this research is largely aligned with the Bormann, Cederman & Vogt’s work that attempts to distance itself from “a formal, encompassing approach to ethnicity that draws no distinction between its key dimensions, such as language, religion, caste, or race”.61 Properly identifying intragroup fissures along
ethnolinguistic, religious and racial lines can better inform fine-grained analysis of identity groups rather than treating each identity as a strictly defined unitary structure. This paper uses a group of datasets under the umbrella of the ‘Ethnic Power Relations (EPR) Dataset Family’ produced by research staff & ETH Zurich’s Centre for International Studies.62
Assessing the role of co-ethnic or rival-ethnic militia first requires some comprehension of the saliency of ‘ethnicity’ in the generation of grievances of the aggrieved population. The ACD2EPR provides a strong platform to identify ethnically or religiously driven insurgency movements as it codes the rebel groups position on making an exclusive claim to fight on behalf of an identity group, recruit from a distinct identity group and whether the group receives support from at least 50% of their constituent ethnic or religious group.63
Rival ethnic groups (to the insurgent’s constituent community) are identified through two methods. Firstly, the insurgent group’s population may have been previously excluded from power. In this case the EPR-CORE dataset provides detailed information regarding 800 ethnic group’s access to power structures across a spectrum of exclusion ranging from monopolies on power to discriminatory experiences.64This data gives an indication of an identity group’s
access to power structures but a second method must also be used to contextualise this information. Qualitative research is carried out across a variety of academic sources to consider
61
Bormann, Cederman & Vogt, ‘Language, Religion & Ethnic Civil War’, Journal of Conflict Resolution, 61:4 (2017), 744.
62
Data available at: https://icr.ethz.ch/data/epr/. 63
Vogt et. al., ‘Integrating Data on Ethnicity, Geography, and Conflict: The Ethnic Power Relations Data Set Family’, Journal of Conflict Resolution, 59:7 (2015).
18
the severity of inter-group tension and other factors, like the presence of anti-group rhetoric and behaviour in the insurgent group’s discourse, are also considered.
Once the militia’s locality to the insurgent group has been identified, their presence is measured by estimating the group’s manpower relative to other forces comprising the COIN regime as a percentage. This measurement of component size is arguably better equipped to describe the prominence of militia in an asymmetrical conflict than research that codes only the presence of militia groups under equal coding, regardless of size. Aliyev’s research, for example, offers only a binary dummy variable for the presence of militia rather than an approximation of how large this force was relative to other counterinsurgency forces.65
This research measures conflict intensity by the number of deaths per 10,000 members of the conflict-affected population and is reliant on the Uppsala Conflict Data Program (UCDP)66
to code deaths related to the insurgent group occurring in the conflict area. Analysing this dependent variable against patterns of militia deployment offer to paint a clearer picture of the consequences or benefits of counterinsurgent irregulars.
5.1. The Dataset: Case Selection & Data Gathering
Cases of counterinsurgency campaigns were selected by scrutinising the UCDP/PIRO Armed Conflict Dataset (Version 20.1)67
, to identify conflicts that could be considered ‘intrastate’ that began between 1991-2015. This period was chosen as this research aims to detach itself from grouping contemporary cases of counterinsurgency with the distinct experiences of colonial or Cold War proxy battles of the late 20th century. From here, conflicts had to meet several criteria
to be considered in this research. The conflict must have been fought between belligerents of asymmetric strength and fit the general description of an insurgent group to exclude, for example, coups. Furthermore, the conflict must have produced over 999 deaths over the course of its lifespan and survive at least 20 conflict months to meet the ‘prolonged’ definition of insurgency. After these features were confirmed, conflicts were investigated via sources of scholarly work and journalistic reporting to reveal whether counterinsurgent militia groups (that reached a size of over 1,000 militants) were present at any stage of the fighting. From here
65 See Aliyev & Souleimanov, ‘Ethnicity and Conflict Severity: Accounting for the effect of Co-ethnic and Non-Ethnic Militias on Battlefield Lethality’, Third World Quarterly, 40:3 (2019).
66
Unless explicitly mentioned otherwise, any subsequent mention of the UCDP in this paper will be citing the UCDP Georeferenced Event Dataset (GED, V.20.1) by Pettersson & Öberg, ‘Organized violence, 1989-2019’, Journal of Peace Research, 57:4, (2020) and Sundberg, Ralph and Erik Melander, ‘Introducing the UCDP Georeferenced Event Dataset’. Journal of Peace Research, 50:4 (2013).
67
Pettersson & Öberg, ‘Organized violence, 1989-2019’, Journal of Peace Research, 57:4, (2020). & Gleditsch et al.,‘Armed Conflict 1946-2001: A New Dataset’, Journal of Peace Research, 39:4 (2002).
19
more in-depth research could be commenced to code the aforementioned variables. Research into the size of various regular and irregular components of the counterinsurgency regime was undertaken with the assistance of a broad range of datasets, NGO reports and security Yearbooks, whilst empirical case studies were useful for contextualising the organisation and activities of non-sate armed groups.
5.2. Empirical Strategy
The subsequent of this paper rely on a mixed methods approach. The first section of the findings is dedicated to quantitative data analysis that provide an important indication of the relationship between C-EM and conflict intensity. The latter is a brief exploration of the processes of ethnic defection in two of the cases used in this dataset.
Initially, the dataset is used to test for correlation between conflict intensity and militia to compare the efficiency of co-ethnic, non-ethnic and rival-ethnic militia in tackling insurgencies. If C-EM are indeed the most efficient form of counterinsurgent militia, we should expect significant negative relationships, stronger than that of other militia types, signalling that as C-EM size increases; conflict intensity should decrease.68
A simple correlation test alone, however, can be misleading as it does not account for the role other variables could potentially come in to play and effect the dependent variable, in this case conflict intensity. Therefore, this paper employs a its primary quantitative test, multiple linear regression analysis, to clarify the statistical significance the militia variable has in predicting the variance of battle deaths whilst controlling for other variables, a full list of which can be found alongside the dataset in the Appendices.
Should militia play a statistically significant role in affecting the variance in conflict intensity we should expect to find C-EM size playing a statistically significant role in predicting conflict intensity in our regression model. Especially relevant for this paper would be to note how the statistical relationships vary depending on the type of ethnic or non-militia.
68 This paper assumes a very weak relationship when r=0.00-0.19, weak when r=0.20-0.39, moderate when r=0.40-0.59, strong when r=0.60-0.79 and very strong when r=0.80-1.0.
20 5.3. Limitations
There are two core weakness to relying on a quantitative research to address research problems investigating non-state armed groups. The first issue is of data availability. One recent report notes that “militia are among one of the most inaccessible groups for qualitative research”, let alone quantitative methods.69 This presents an issue to gathering accurate data, especially of
militia group’s manpower, on a monthly basis across conflicts than span years. The informal nature of these groups means exact records are rarely, if ever, kept on membership. Yet any attempt at estimating such figures could also be extremely misleading. Some information on militia sizes is available on an annual basis and then projected onto the months of that year, and there is the rare occasion where monthly updates or key months in recruitment cycles are identified. In sum, the lack of availability of data, though it has improved recently with the establishment of new key datasets, is a key obstacle for this research.
The second issue of this quantitative is of misdirecting an emphasis on ‘correlation’ without carefully considering causation. Therefore, rather than completely relying on quantitative methods, this research offers a combined quantitative/qualitative approach to the research problem. The calculations applied to the data offer a firm foothold to signal overall trends but do little to temper the issues equating ‘correlation’ with ‘causation’.
For the purpose of a between-case comparison, this study selects two cases of C-EM deployment based on their performance in a Pearson correlation tests. The case that holds the strongest statistically significant correlation between C-EM deployment and conflict intensity and a case where the relationship between the two variables is insignificant and weak. Qualitative analysis can then be applied to each case to distinguish what circumstances drove the relative success and failure of each case in decreasing conflict intensity. A mixed methods approach lends itself to better exploring the complex dynamics of insurgency and supposed ‘non-state’ violence that aligns statistical conclusions with relevant investigative context. The result of this methodology is, hopefully, the provision of answers to research questions that are rich in evidence and not overly dependent on dense statistical conjecture that can overlook the intricate processes of civil war.
69
21
Standard Errors in parenthesis. Significance Levels: *p < 0.05, **p < 0.01, ***p < 0.001. Calculated in IBM SPSS. Full computation output for each model available in appendix. Table I., Multiple Regression Models – Conflict Deaths
Model 1 (Co-ethnic Militia) Model 2 (Rival-ethnic Militia) Model 3 (Non-Ethnic Militia) B β B β B β
Co-Ethnic Militia Strength -0.254*** -0.254*** -0.119* -0.104*
(0.043) (0.060)
Rival-Ethnic Militia Strength 0.037 0.028 0.124* 0.124*
(0.054) (0.051)
Non-Ethnic Militia Strength -0.138 -0.138*
(0.051)
Incumbent State Strength 0.080* 0.080* 0.167** 0.167** 0.243 0.243***
(0.040) (0.053) (0.048)
Expeditionary Force Strength -0.038 -0.024 0.457** 0.137** -0.119 -0.090*
(0.059) (0.144) (0.057) CivAction 0.003 0.002 -0.114 -0.80 0.196 0.172* (0.065) (0.085) (0.081) CivTarget 0.281*** 0.170*** 0.328*** 0.262*** 0.183 0.056 (0.076) (0.064) (0.122) CivSecure -0.077 -0.061 0.143 0.083 -0.231 -0.220** (0.052) (0.092) (0.067) Peacekeepers -0.156 -0.061 -0.005 -0.002 0.048 0.014 (0.106) (0.101) (0.178)
Relative Insurgent Strength 0.241*** 0.241*** 0.191** 0.191** 0.046 0.046
(0.043) (0.058) (0.050)
State Authority Over Terrain -0.064 0.064 0.020 0.019 -0.324 -0.324***
(0.041) (0.052) (0.053) Governance 0.190*** 0.190*** 0.225*** 0.255*** 0.092 0.092* (0.042) (0.052) (0.044) Polity2 -0.072 -0.059 -0.229** -0.175** -0.051 -0.044 (0.053) (0.073) (0.049) Peace Agreement 0.039 0.012 -0.101 -0.042 -0.339 -0.07 (0.126) (0.106) (0.187) Political Terror 0.132* 0.080* 0.213*** 0.174*** 0.139 0.139** (0.064) (0.057) (0.049)
Constituent Group Power Access 0.096* 0.084* 0.088 0.079 (0.043) (0.066) Constant 0.025 0.026 285.540 (0.035) (0.049) (13.267) Total Observations 868 630 599
22
6. Data Results & Analysis: Multiple Regression Analysis
Correlation tests alone are unable to account for other variables that may influence conflict intensity. By utilising multiple linear regression models, it was possible to assess the contribution of each independent variable in predicting variance of conflict intensity. Table I shows the variable coefficients of three multiple linear regression models. Each model is generated using the all conflict months from a population of cases where co-ethnic militia (model 1), rival-ethnic militia (model 2), and non-ethnic militia (model 3) were respectively deployed. Standardised z-scores for all variables (excluding dummy variables) were calculated discretely for each conflict before being entered into the regression analysis software to ensure cross-case comparisons were relating standardised values. Table I shows both the raw (B) and standardised (β) coefficients of each variable. The standardised coefficient is of importance to this paper, as it provides us with values we can use to compare the relative importance of each variable and its unique predictive contribution to the model.
Model 1 suggests we can discount our null hypothesis, as C-EM strength has a statistically significant impact our independent variable. In model 1, which was generated with the total population of conflict months from wars where C-EM were deployed at some stage, the co-ethnic deployment variable is negative and statistically significant (p < 0.001). This implies that a larger C-EM deployment is a strong predictor of decreased conflict violence. Furthermore, the standardised coefficient (β=-0.254) is the greatest value of all the variables. The model, therefore, suggests that the deployment and strength of co-ethnic irregular counterinsurgents is a greater predictor of conflict fatalities than the other statistically significant variables, including the COIN regime’s targeting of civilians, the strength of insurgent group relative to the total strength of the combined COIN regime, and the quality of governance provided by the counterinsurgency regime. In other words, the deployment of co-ethnic forces had a more significant predictive impact on conflict intensity over strategic approaches that have been critical in classical counterinsurgency literature.
Model 2 is generated using all conflict months from the population of cases where rival-ethnic militia are deployed. In this case, rival-ethnic militia do play a somewhat statistically
23
significant (p<0.05) role in predicting conflict intensity. Unlike C-EM in model 1 where the relationship between militia strength and conflict intensity is inversed, rival-ethnic militia holds a positive predictive relationship with conflict intensity, suggesting both in parallel. This being said, the role of rival-ethnic militia in the model is less important than several other variables. The strength of other elements of the counterinsurgency regime (expeditionary force strength & incumbent strength) both have higher β values, as does incumbent civilian targeting and various governance and political system variables. This suggests that the predictive importance of rival-ethnic regime size is lesser than the importance placed on chances in counterinsurgent strategy and political reform. It should also be noted that the co-ethnic forces included in conflicts grouped by rival-ethnic militia deployment remain a statistically significant negative predictor of conflict intensity, albeit at a lesser level of significance and importance than in model 1. Finally, model 3 is generated by a total of 599 observations over the population of cases where non-ethnic militia are, at some point, deployed. Non-ethnic militia hold a statistically significant negative relationship with conflict intensity, illustrating that a growth in non-ethnic militia deployment generally predicts lower levels of conflict intensity. Similarly, to rival-ethnic militia in model 2, however, several of the strategic controls are more statistically significant than the non-ethnic variable, with higher levels of importance. Changes to the security of civilians in the conflict zone, state authority over terrain, the instrumentalization of political terror and the size of the indigenous state force were all of greater importance to the predictive model.
6.1. Data Analysis: Theoretical Implications
This cross-case regression analysis of the impact of identity-based militia on conflict intensity has important consequences on our understanding of identity-based militia. The negative relationship between conflict intensity and C-EM is clearly demonstrated in the first two models and adds significant weight to existing theoretical arguments that imply C-EM can play an important role in the reduction of levels of violence. It firmly indicates that strengthening the prominence of co-ethnic irregulars in the counterinsurgency regime is key to decreasing conflict fatalities. Comparatively, increasing the prominence of a rival-ethnic militia in a counterinsurgency regime is likely to be detrimental to decreasing conflict intensity. Despite rival-ethnic militia’s similar levels of familiarity with conflict terrains they are either unwilling or unable to engage in operations that decrease the persistence threat of violence in intrastate conflict. Although non-ethnic militia play a significant role in decreasing conflict intensity,
24
their position in the model 3 is of lower importance and statistically significance than the influence of co-ethnic forces in model 1, often trumped by strategic changes correlating with a variation in conflict intensity. This indicates co-ethnic forces are the best equipped form of counterinsurgent irregular to limit violence in regions and populations impacted by an insurgent organisation.
This statistical analysis confirms has several implications on the validity of the theories explored in the prior chapters. Most notably is the ability of co-ethnic counterinsurgents to effectively reduce conflict intensity on a much more consistent basis than non-ethnic militia. Despite non-ethnic community militia representing the most local form of counterinsurgents to aggrieved populations (in non-ethnically organised insurgency campaigns), they were still, on average, outperformed by co-ethnic groups. One possible explanation for this is that the saliency of ethnicity in some societies ensures more robust local networks built around in-group patronage. This access to such comprehensive local networks ensures C-EM are granted greater access to the intelligence opportunities offered by operationalizing existing relations with the insurgent group’s constituent population. This could also explain why the Afghan case also experiences a significantly relevant negative relationship between conflict intensity and irregular manpower as a proportion of total counterinsurgent strength. Even though describing the Taleban as primarily a Pashtun ethnic movement is contentious, the saliency of tribal or ‘qawm’ structures and durability of kin networks in the south and southeast of the country certainly tangible.70 In other terms, these networks ensure irregular counterinsurgents can
exercise vigorous internal monitoring over the local population to a degree not accessible to militias organised around broader societal structures. Therefore, it should not merely be assumed that an insurgent organisation’s endeavours can be disrupted by the mobilisation of their neighbours. These deployments can only demonstrate their benefits upon the condition of a militia membership’s strong existing relations across the insurgent’s constituent population. Furthermore, the increase of conflict intensity generated by rival ethnic militia could indicate that the mobilisation of perceived ‘outsiders’ ultimately fails to offer the same benefits of network monitoring and struggle to exercise discriminatory targeting practices due to a lack of familiarity or relationship with aggrieved populations.
70
Glatzer, ‘The Pashtun Tribal System’, in G. Pfeffer & D.K. Behera (eds.) Concept of Tribal Society (Concept, 2002). and Giustozzi, ‘Afghanistan: Transition without End’, Crisis StatesWorking Paper, 2:40 (2008): 1.
25
Overall, this implies that it is a counterinsurgent’s familiarity of the human terrain and accessibility to local communities where identity is salient, rather than their experience with the local geographic topography of a conflict area, that is paramount for limiting conflict intensity. Although rugged terrain was constantly noted as a powerful tool for insurgent groups exercising guerrilla warfare in classical texts, it could be suggested that non-ethnic militia geographical ‘guides’ are more redundant in this dataset considering the modern technological advancements available to incumbents throughout these cases.
Perhaps surprisingly, model 1 goes well beyond highlighting the statistically significant role played by ethnic defector militias. The prominence of C-EM is actually of more importance to addressing conflict intensity than changes to other theoretically valued variables. The triad of variables that compose our controls for the variation in ‘Hearts and Minds’ strategies (CivAction, CivTarget, CivSecure) are largely irrelevant relative to the role co-ethnic irregular strength in accounting for a reduction in violence, with the exception of CivTarget which plays a statistically significant and important role in both models of ethnic militia. These findings suggest that greater attention should be paid to who exactly is fighting insurgents as much as what strategies counterinsurgents are employing.
These conclusions largely reinforce existing theoretical perspectives and authenticate prior statistical research on the effect of the presence C-EM on conflict deaths. What prior research fails to capture, however, is how the prominence of militia forces in each counterinsurgency regime affects battle deaths. This research shows that, across cases, militia size holds a statistically significant relationship with conflict intensity. Analysis of this data, therefore, offers an original conclusion that expands existing knowledge of C-EM deployment by suggesting that the more a counterinsurgency regime is comprised of co-ethnic local elements, the more likely conflict intensity will decrease on a much more consistent basis than other militia types.
7.1. Case Study: The Sons of Iraq
The Sons of Iraq were a militia organised 41 months into the lifecycle of Abu Musab al-Zaraqwi’s Tanzim Qaidat al-Jihad fi Bilad al-Rafidayn (TQJBR) organisation, what later became known as the Islamic State in Iraq. Although Zarqawi’s network was established well before its inroads into Iraq, it was the US-led coalition’s invasion that ensured his organisation could begin consolidating power in the country. Before this point al-Zarqawi’s network had
26
been something of a marauding party of Sunni militants, primarily composed of Jordanian and Palestinian Sunni Muslims, leaping from Jordan to Afghanistan, Afghanistan to Iran and eventually finding a habitable climate in the destabilised post-Saddam Iraq.71
The ISI was able to rapidly gain influence amongst Sunnis who felt alienated by the perceived empowerment of Shia & non-Arab minorities into the new Government of Iraq (GoI) configuration and the occupation of Iraq by a predominately Christian imperialist force.
The C-EM in question, the Sons of Iraq (SOI), is an official extension of an informal program local to al-Anbar province, suitably classified as the ‘Anbar Awakening’. By 2005, Sunni elites in the Anbar reion recognised the threat was not primarily the GoI or their American sponsors, but Zarqawi’s extremist Islamist alliance. The defection of Sunnis came as coalition forces became increasingly pessimistic about their chances of projecting influence into the Anbar region, leaving ISI as the uncontested source of authority in the province.72 The ISI, assured by
a deep conviction in the sanctity of their actions, went to great lengths to eliminate any threats to their fundamentalist vision. This included extensively targeting Sunni communities as ISI’s Fig.2
71
Lister, ‘Profiling the Islamic State’, Brookings Doha Analysis Paper, 13, (Brookings Institution, 2014): 6-8. 72
27
leadership sought to consolidate their proto-state in Iraq. The readiness of the insurgents to engage in a strategy of ruthless intimidation against the local Sunni population ensured they would become deeply unpopular in the eyes of many.73
Many of al-Anbar locals suspected it was this brutality that encouraged Sheikh Abdul Sattar Abu Risha to launch the Awakening concept in 2005, motivated by the murder of many members of his family by ISI operatives.74
There was also a large presence of foreign fighters within the hierarchy of the ISI, who, as al-Jabouri & Jensen describe, “did not know how to win the hearts and minds of the Iraqis”.75
Besides representing the strongest co-ethnic statistical relationships in our dataset, the Iraqi case highlights some important thematic scenarios that are replicated in other case studies. Figure 2 provides a telling picture of the influence of counterinsurgent C-EM formation. There is an observable decrease in violence throughout the period that the SOI program was in action. The average monthly conflict fatalities resultant of the ISI conflict in the period of C-EM deployment fell by approximately 37% (to 174.1 deaths per month) from the months prior to the SOI program’s launch (276.6 deaths per month). Similarly, conflict fatalities in the period following the militias disbanding rose by over 520% (1093.8 deaths per month). No other individual component of the counterinsurgency regime can match this fall in violence. Even if we compare the coalition’s peak deployment during the renowned troop ‘surge’ of 2007, the 6 months following the high-water mark of the surge was still 58% more deadly (211.17 deaths per month) than the co-ethnic’s peak deployment.
Although the seeds of the Awakening were planted by local tribal leaders, coalition commanders immediately recognised the potential of co-opting these forces into the counterinsurgency effort and played a key role in its extension across the Iraqi conflict area. Local US commanders initially looked the other way when Sunni militants attacked ISI strongholds, but the relationship between Anbar-based coalition officers and Sunni militias intensified swiftly as they proved their worth with vital intelligence and military victories against Islamist elements in the province.76 This relationship grew more and more
institutionalised as coalition commanders sought to solidify their relationship with local power-brokers; firstly as the ad-hoc formalisation of anti-ISI vigilantes through a major police recruitment push organised by American brigade staff stationed in Ramadi, the capitol of
al-73
Mapping Militant Organizations, The Islamic State (Stanford University, 2019). 74
Montgomery & McWilliams, Al-Anbar Awakening: Volume 2: Iraqi Perspectives (Marine Corps University Press, 2009): 24.
75
Al-Jabouri, & Jensen, 'The Iraqi and AQI Roles in the Sunni Awakening', PRISM, 2:1 (2010): 16. 76
