UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The enemy of my enemy is not my friend : a theory of rebel alliance patterns in civil war Moncrief, Stephen Edward 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2012_fall_moncrief_stephen.pdf [ 297.79kB ]
Metadata
JSON: 24-1.0072956.json
JSON-LD: 24-1.0072956-ld.json
RDF/XML (Pretty): 24-1.0072956-rdf.xml
RDF/JSON: 24-1.0072956-rdf.json
Turtle: 24-1.0072956-turtle.txt
N-Triples: 24-1.0072956-rdf-ntriples.txt
Original Record: 24-1.0072956-source.json
Full Text
24-1.0072956-fulltext.txt
Citation
24-1.0072956.ris

Full Text

         i THE ENEMY OF MY ENEMY IS NOT MY FRIEND:  A THEORY OF REBEL ALLIANCE PATTERNS IN CIVIL WAR    by  Stephen Edward Moncrief  B.A., Western Washington University, 2010    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF ARTS  in  The Faculty of Graduate Studies  (Political Science)      UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)   July 2012   © Stephen Edward Moncrief, 2012          ii Abstract   Civil wars are rarely two-player affairs. Indeed, civil wars often feature several distinct rebel organizations contesting an incumbent’s territorial control, and while it would seem efficient for these rebel groups to ally with one another against the incumbent, the opposite occurs with surprising frequency: distinct rebel groups regularly fight one another even as they fight the same incumbent.  I offer a simple theory that aims to explain why insurgent groups fighting the same incumbent will ally in some instances, but not in others. I argue that when an incumbent boasts military capability sufficient to credibly threaten the elimination of the opposition, rebel groups will be more likely to ally with each other in order to avoid destruction. However, when rebel groups do not fear elimination, they are less likely to ally and more likely to fight amongst themselves, even as they continue their campaigns against the incumbent.  There are two reasons that these groups will fight each other: (1) in order to decrease the number of potential bargaining partners for the incumbent or a key sector of the civilian population, and (2) to avoid being disadvantaged when it comes time to divide valuable war spoils, especially when those spoils are won by supplanting the incumbent. I demonstrate the empirical plausibility of this theory with three well- documented cases, and conclude with suggestions for future research on the topic of internecine targeting between rebel groups.      iii Table of Contents  Abstract…………………………………………………………………...……………….ii Table of Contents……………………………………………………………………..…..iii Acknowledgements………………………………………………………….....................iv 1 Introduction 1.1 The Puzzle……………………………………………………………………..1 1.2 Beyond Two-Player Models of Civil War…………………………………….3 2 A Theory of Incumbent Capability and Internecine Targeting     2.1. Rebellion is Risky Business…………………………………………………..7  2.2. The Argument…………..…………………………………………………….8 3   The Second United Front in China (1936-1941)  3.1. Japan as the Incumbent in Eastern China…………………………….……..13  3.2. The Xi’an Incident…………………………………………………………..17  3.3. Alterations to Relative Capabilities…………………………………...…….21 4   The LTTE in Sri Lanka (1983-2009)  4.1. The Movement for a Tamil Homeland……………………………………...24  4.2. Why Tigers Don’t Need Allies……………………………………………...26  4.3. Significances, Empirical and Theoretical……………………………...……31 5    The Angolan War for Independence (1961-1974)  5.1. The MPLA, FNLA, and UNITA...……………………………………….....33 5.2. Internecine Targeting Within the Angolan Resistance...................................36  5.3. The Decolonized State as a Super-valuable Good…………………………..38 6  Conclusion……………………………………………………………………….43  Bibliography……………………………………………………………………..47        iv Acknowledgements First, I would like to express my gratitude for the opportunity to work in an environment as supportive as the UBC political science department. More specifically, I wish to gratefully acknowledge the guidance of Dr. Arjun Chowdhury, whose fine insights prompted me to sharpen my writing and thinking, and who also suggested a much more compelling title for this project. I wish also to thank Dr. Brian Job, for coupling high standards with patience. Second, I thank my parents, for their limitless wisdom.                                     1 1 Introduction 1.1. The Puzzle  Researchers have only recently started unpacking the colorful patterns of alliance and defection that characterize rebel strategy in civil wars. Tracing these patterns is no light exercise: side switching and fractionalization are common among insurgent groups.1 However, analytical challenges make the task of clarifying rebel alliance patterns no less urgent. Rarely are civil wars simple competitions between one state and one monolithic insurgency, and this renders two-player models of conflict inappropriate for the in-depth study of alliance and defection patterns among violent non-state actors.2 Indeed, civil wars often feature two or more distinct rebel groups violently targeting an incumbent, and these groups may claim to represent rival communities, ethnicities, or ideologies. It is not unusual for rebel organizations to violently clash with each other even while they target the same incumbent, and this is no trifling point: by under-theorizing relationships between rebel groups, students of intrastate conflict may be missing important variables that could help explain the intensity, duration, or even outcome of a civil war. Hironaka, for example, points to the fracturing of opposition groups as a possible factor in the inordinately long lifespans of some civil wars.3  If several distinct insurgencies target the same incumbent, it would seem that the most efficient wartime strategy would be for these groups to bracket their mutual                                                         1 Bakke, Kristin, Kathleen Cunningham and Lee Seymour, “Shirts Today, Skins Tomorrow: Dual Contests and the Effects of Fragmentation in Self-Determination Disputes,” The Journal of Conflict Resolution 56, no. 1 (2012): 67-93. 2 Jesse Driscoll, “Commitment Problems or Bidding Wars? Rebel Fragmentation as Peace Building,” Journal of Conflict Resolution 56, no. 1 (2012): 122. 3 Ann Hironaka, Neverending Wars: The International Community, Weak States, and the Perpetuation of Civil War (Cambridge, MA: Harvard University Press, 2005), 78-9. 1   2 suspicions and create a united front, or at least an understanding akin to a non-aggression pact until the incumbent is defeated. As shown below, there are indeed conditions under which this occurs. However, in many instances, the exact opposite takes place: distinct insurgent groups fight one another while they fight the same incumbent. Such violent internecine targeting occurs even between insurgent groups drawn from the same ethnic community, and between groups that claim similar policy goals. Alternatively, ethnically and ideologically differentiated rebel groups sometimes form alliances and fight the state along united fronts. If ideological difference and ethnic hostility are not reliable predictors of violence between rebel groups, what explains the decision to ally versus the decision to fight?  Below, I propose a simple theory of internecine targeting, centered on one straightforward but analytically valuable independent variable: the probability of an incumbent’s decisive victory, which becomes clearer as the war progresses and information on relative capabilities is provided to all sides. This is not to say that the sides necessarily disagree on the balance of power, as insurgencies are often small and poorly funded in comparison to the incumbent militaries they fight. Rather, crucial information is provided when the incumbent demonstrates its ability to effectively execute counterinsurgency techniques that threaten to eliminate the rebel groups. Under these circumstances, the rebel groups should be more likely to ally in order to avoid destruction. If, however, at least one rebel group believes that it can survive and impose costs on the incumbent sufficient to effect a victory or a desirable negotiated settlement, this group will be more likely to initiate violent internecine targeting against fellow rebel organizations. This is because the conditions of war are conducive to the development of   3 a series of commitment problems between rebel groups, making internecine targeting rational, if inefficient.  The balance of the paper is in four parts. I begin by reviewing the emergent literature on rebel group fractionalization, and present a theory of internecine targeting centered on incumbent capability and commitment problems between rebel organizations. After spelling out the logic of the theory, I present three illustrative cases that span time and geography: the united front between the Chinese Communist Party (CCP) and the Guomindang contra the Japanese incumbent in eastern China (1937-41), the Liberation Tigers of Tamil Eelam’s (LTTE) fratricidal relationship with other Tamil independence organizations (1983-2009), and the Angolan war for independence from the Portuguese colonial state (1961-1974). These cases are selected for their heuristic value, and are not intended as definitive tests of my hypotheses; I employ them to demonstrate the empirical plausibility of the theory. I conclude by suggesting additional cases for future research, including a few difficult tests that may assist in refining the theory.  1.2. Beyond Two-Player Models of Civil War  The call to disaggregate the study of intrastate war is fairly new, so it should not be surprising that presently, our knowledge of alliance and defection patterns among violent non-state actors in civil wars is neither unified nor systematic. In-depth ethnographic studies of rebel groups have yielded rich results, and made important contributions to our understanding of rebel organization, ethnic cohesion, and individual   4 motivations for fighting.4 It has, however, been difficult to put this case study knowledge in meaningful dialogue with quantitative work on the topic, because although large-N studies have identified the characteristics of especially long-lasting rebellions, such perspectives focus on modeling commitment problems between states and allegedly monolithic insurgencies, but not on commitment problems between rebel groups.5  I focus specifically on armed non-state organizations that, using the techniques of insurgency, seek initially to either capture the mantle of the state, or to secure a sub-state region’s full autonomy or secession. These rebel groups may have originated as factions within broader nationalist or separatist movements, but I do not select groups based on their historical origins. I instead select groups based on their organizational independence from one another. The theory seeks to explain the behavior only of insurgencies that are clearly differentiated by leadership and member composition; I am not looking to apply the logic of internecine targeting to different factions of the same umbrella group.  Bakke, Cunningham, and Seymour find that the number of identifiable factions within an initial self-determination movement is correlated with the movement’s propensity to split into several independent organizations, some of which may embrace violence in order to achieve dominance within the movement.6 These authors take seriously power disparities between opposition groups, and generate a sensible list of                                                          4 For an in-depth treatment of these topics, see Jeremy Weinstein, Inside Rebellion: The Politics of Insurgent Violence (New York: Cambridge University Press, 2007). 5 For examples, see James Fearon, “Why Do Some Civil Wars Last So Much Longer Than Others?” Journal of Peace Research 41, no. 3 (2004) 275-301; David Cunningham, Kristian Gleditsch and Idean Salehyan, “It Takes Two: Dyadic Analysis of Civil War Duration and Outcome,” Journal of Conflict Resolution 53, no. 4 (2009): 570-597. 6 Kristin Bakke, Kathleen Cunningham and Lee Seymour, “A Plague of Initials: Fragmentation, Cohesion, and Infighting in Civil Wars,” Perspectives on Politics 10, no.2 (2012): 265-283.   5 hypotheses regarding how these disparities will impact relations among the self- determination groups. The present study shifts its focus slightly to concentrate on relations between distinct insurgencies that have already formed and differentiated themselves. It distinguishes itself further by drawing on rationalist inquiries into the outbreak and termination of war in order to explain alliances and internecine violence among rebel organizations. As shown below, thinking in terms of commitment problems between insurgent groups may help to refine the conclusions reached in the emergent works on internecine targeting.  Of course, the role of commitment problems in elongating civil wars is well documented. Because the losing side is typically expected to demobilize and return to daily life in close proximity with former enemies, lasting peace settlements can be elusive, especially if one or both sides are judged responsible for atrocities committed during wartime, or if the war’s master cleavage falls along easily identifiable ethnic lines.7 It appears, however, that the field is only now beginning to take seriously the possibility of commitment problems among violent non-state actors in civil wars, and the present study is in keeping with this trend.  Driscoll, for example, uses the end of the Tajik civil war to show that in a game of post-war state building, “all warlords- ‘insurgent’ and ‘incumbent’ alike- must contend with the possibility that they will be marginalized in a future coalition.”8 In Driscoll’s model, the state represents a network of valuable resources for violence entrepreneurs, so capturing even a sliver of the post-war state can greatly enhance a warlord’s ability to                                                          7 Shanna Kirschner, “Knowing Your Enemy: Information and Commitment Problems in Civil Wars,” The Journal of Conflict Resolution 54, no. 5  (2010): 745-770. 8 Driscoll, “Commitment Problems or Bidding Wars?” 123.   6 consolidate power and secure capital. I borrow from this logic and suggest that rebel groups tend to suffer from a similar fear of the future that arises before the game of state- building begins, generating commitment problems that prompt rebel groups to abjure united fronts and target one another, even as they fight the same incumbent.  In another recent piece, Staniland takes internecine targeting between insurgent groups as a given, and treats defection as the dependent variable of interest.9 Accounting for the origin of internecine targeting, however, shows that the relationship between defection and rebel-on-rebel violence may be more complex: under certain circumstances, group A ’s fear of group B’s eventual defection creates an ex ante incentive for A  to eliminate or substantially weaken B, which in turn encourages B to defect, regardless of whether B would have done so before. The relationship between the possibility of defection and internecine targeting is thus more interactive than we might expect.  In short, there are several moving pieces floating freely between the recent works on internecine targeting between insurgent groups, but there is no systematic theory of what causes some rebel groups to ally, and others to fight each other. As a modest step toward addressing this issue, I seek to assemble these pieces into a coherent, testable explanation of an under-theorized phenomenon.                                                             9 Paul Staniland, “Between a Rock and a Hard Place: Insurgent Fratricide, Ethnic Defection, and the Rise of Pro-State Paramilitaries,” Journal of Conflict Resolution 56, no.1 (2012): 21.    7 2    A Theory of Incumbent Capability and Internecine Targeting 2.1. Rebellion is Risky Business  I begin with two simple propositions about the nature of civil war and rebel decision-making, followed by a straightforward series of arguments laying out the conditions under which independent rebel groups will ally against a mutually targeted incumbent, the conditions under which they will fight each other while fighting the same incumbent, and the rational calculations behind each decision.  First, the decision to participate in an insurgency is a risky one, and the costs of losing are significant. Beyond the obvious risks of being killed or injured, an insurgent risks serious reprisals if the war is concluded unfavorably. Losers may be imprisoned or ostracized on account of their participation, and even if individual insurgents are not held personally accountable, the winner and its allies may seek retribution against the families, communities, or ethnic groups associated with a losing insurgency. These reprisals could be violent, or they could take the form of long-term exclusion from the spoils of winner- controlled enterprises and resources. In short, rebellion is risky business, so insurgent organizations have cause to be discerning in their choice of alliance partners.  Second, potential allies are also potential defectors. Once an insurgent organization begins targeting the incumbent, it must be selective about what other organizations it works with. Rebel groups can vary significantly in their resolve and motivations, so it will not do to share information or materiel with another organization if unconvinced of its reliability. This speaks to the difficulty that insurgent groups are likely to encounter in signaling and discerning long-term intentions. If group A  draws its membership from a particular ethnicity or community, and group B draws from a pool of   8 A ’s traditional social rivals, the two organizations are likely to be suspicious of one another, even if they find a common adversary in the state. Alternatively, some groups enter into the conflict late, which could be taken as a signal of their proclivity to free ride on groups that had been fighting from the beginning.10  Also, while some rebel groups may legitimately intend to secure political goods for a valued constituency by violently imposing costs on the state, others may be predatory criminal bands disingenuously espousing political goals, and some may move back and forth along this spectrum over time. The costs of choosing to ally with the wrong group could be significant: an alliance with a partner lacking in resolve leaves everyone more vulnerable to defection and exposure, and an alliance with a particularly disreputable band could plausibly undercut legitimacy among civilian populations.11 On average, then, insurgent groups should be discerning when forming alliances with other independent rebel organizations. Having established the context within which decisions about alliances must be made, I isolate the conditions under which distinct insurgent groups will partner, and those under which they will fight.  2.2. The Argument (1) First, I consider incumbent capability, and whether it is sufficient to eliminate unallied rebel groups. When the incumbent C does not appear to constitute an overwhelming coercive presence, opposition groups A  and B are less likely to ally. However, when the incumbent C appears capable of defeating A  and B individually, and                                                          10 On the reputational advantages to being early participants in rebellion, see Weinstein, Inside Rebellion, 51-2. 11 On the consequences of losing civilian support, see Weinstein, Inside Rebellion, 203-8.   9 both opposition groups agree on this assessment of relative capabilities but choose to keep fighting (possibly because the costs of surrendering are judged to be unacceptable), then A and B will ally until conditions weaken C so as to make the alliance unnecessary for the survival of A  and B. This is an intuitive but necessary first step in illustrating the logic of targeting between opposition groups. (2) Second, I allow for incumbent capabilities that are insufficient to credibly threaten the elimination of unallied rebel groups, but not so inferior that the state risks losing the war in the short term. Under these circumstances, A  and B will have fewer incentives to ally, but why should one target the other? It is here that the first important commitment problem between the groups arises. Fearon’s conception of a commitment problem between two states that precludes a mutually preferred, peaceful bargain is similar to the situation in which A and B find themselves.12 For example, even if A  and B prefer not to fight each other, no mechanism exists by which the terms of an alliance can be consistently enforced, and conditions are such that one may have an incentive to defect to the incumbent at a later time. The fear of the future is well justified. Powell shows that in competitive environments where the balance of power can shift quickly and unexpectedly, complete information on relative capabilities in the present is not sufficient to prevent bargaining failure, and the conditions of civil war are no different: sudden shifts in the balance of power are quite likely in intrastate war, given the propensity of alliances to change, for external support to suddenly bolster one side, or for one group to corner the market of civilian support.13 In situations where C is clearly the superior coercive entity,                                                          12 James Fearon, “Rationalist Explanations for War,” International Organization 49, no.3 (1995): 379-414. 13 On complete information in competitive environments, see Robert Powell, “The   10 the alliance between A  and B is largely self-enforcing because without it, A  and B believe that they will be eliminated. Absent that belief, relations between A  and B suffer a commitment problem: one’s defection could severely disadvantage the group that did not defect first, and agreements to abide by the terms of an alliance are simply not credible.  For A  to target B in this scenario makes sense because if B cannot commit to an alliance as the war goes on and conditions change, it is in A ’s interest to weaken B, making B a less attractive partner for C in the future. This not only makes the potential alliance between B and C a less powerful combined force, it also makes A  the more viable bargaining partner for the incumbent. The same dynamic is at play among opposition groups that put a high premium on support from the same civilian population. These groups may find themselves competing for the support of a specific population that can provide food, money, shelter, or any number of other desirable goods.14 Indeed, the fewer viable alternatives behind which a domestic population can throw its support, the better off the surviving opposition group. This logic coalesces into a broader strategy of limiting the number of bargaining options for the incumbent and/or some critical sector of the civilian population through internecine targeting, even while the war against the incumbent continues. (3) Third, I allow not only for incumbent capabilities that are insufficient to credibly threaten the elimination of the unallied rebel groups, but I also account for a commitment                                                                                                                                                                       Inefficient Use of Power: Costly Conflict with Complete Information,” American Political Science Review 98, no. 2 (2004): 232-7. 14 Jannie Lilja, “Trapping Constituents or Winning Hearts and Minds? Rebel Strategies to Attain Constituent Support in Sri Lanka,” Terrorism and Political V iolence 21, no.1 (2009): 308; see also S. Koc Menard, “Fragmented Sovereignty: Why Sendero Luminoso Consolidated in Some Regions of Peru but Not in Others,” Studies in Conflict and Terrorism 30, no. 2 (2007): 176.    11 problem that may arise directly from the groups’ ultimate goal. More specifically, I allow for a commitment problem stemming from a disagreement over how to divide a good that can be captured if the insurgency is ultimately successful. This good is vested with such high value that it could radically alter the relative bargaining power of each opposition group later, if not equitably distributed. Here, I borrow from Reiter’s work on the nature of a good that comes in super-valuable increments. Reiter shows that some goods over which actors fight come in increments that are not linear in their value distribution, but take the form of a step function, with certain increments being far more valuable than others. For example, some pieces of territory within a state are richer in natural resources or of greater military value than other pieces, so it is difficult to divide such territory equitably among different actors.15  I borrow also from Powell’s treatment of the conflict that may arise if different domestic actors cannot commit to the equitable division of resources that come with domestic control, which tends to occur if an inequitable distribution could implicate each group’s future bargaining power.16 In these cases, several actors aim to control a good or set of goods they believe to be super-valuable (whether the goods are actually super- valuable is of secondary importance). Even if an agreement on the good’s distribution can be reached, it may be difficult for each group to credibly commit to abide by that agreement; incentives to renege are obvious.  Rebel groups often fight the incumbent over goods such as control of the state and its territory, or over the right to secede. As shown below, groups can set out with either                                                          15 Dan Reiter, How Wars End (Princeton, NJ: Princeton University Press, 2009), 44. 16 Robert Powell, “War as a Commitment Problem,” International Organization 60, no.1 (2006): 189.    12 the intention of capturing the state (as in the Chinese and Angolan cases) or securing control over an autonomous region (as in the case of the Sri Lankan civil war), but control over the administrative apparatuses of a territory is super-valuable in that each group’s relative bargaining power can be substantially altered based on the distribution of that control. Capturing the mantle of the state in an international community that prefers not to conduct business with sub-state actors has major advantages, and if a power- sharing arrangement cannot be agreed upon following the incumbent’s defeat, the opposition groups will fight each other before defeating the incumbent.17  In brief summary, the conditions I propose take the following form: (1) When the incumbent can credibly threaten to eliminate the individual rebel groups, the groups are more likely to ally. When the incumbent cannot credibly threaten to eliminate the rebels, the rebels are less likely to ally. (2)  When the incumbent cannot credibly threaten to eliminate the rebels, one rebel group may actively target another in order to decrease the number of potential bargaining partners for the incumbent or a key sector of the civilian population. (3)  Additionally, when the incumbent cannot credibly threaten to eliminate the rebels and at least some of the rebel groups seek control over a good that may come in super- valuable increments, the groups will fail to ally, and will actively target each other to avoid being disadvantaged when it comes time to divide the good.                                                           17 On a state’s preference for similarly constituted bargaining partners, see Hironaka, Nevernding Wars; Jeffrey Herbst, States and Power in Africa: Comparative Lessons in Authority and Control (Princeton, NJ: Princeton University Press, 2000); Robert Jackson and Carl Rosberg, “Why Africa's Weak States Persist: The Empirical and the Juridical in Statehood,” World Politics 35, no.1 (1982): 1-24. On the lucrative potential of capturing the state, see William Reno, Warlord Politics and African States (Boulder, CO: Lynn- Reiner, 1998).    13 3     The Second United Front in China (1936-1941) 3.1.  Japan as the Incumbent in Eastern China  Before proceeding, two historiographical notes are in order. First, space limitations restrict the detail in which I can depict the internal politics of the CCP and the Guomindang in the years preceding the Second Sino-Japanese War, and this is no trifling point: like any political party cum insurgency, these were not monolithic organizations, and internal disagreements over policy direction were not uncommon. However, detailed biographies of these organizations and their elite leadership circles are available elsewhere, and the modest historical sketch that follows is intended as an illustration of the effect that overwhelming incumbent capability may have on insurgent alliance strategy, not as a holistic summary of the Second Sino-Japanese War, nor as a detailed description of its main belligerents. Second, although the Japanese occupation of northeastern China and its clashes with the CCP and Guomindang may not have constituted a civil war in the conventional sense, the wartime conditions in China from 1937-45 mirror the common circumstances of civil war in an analytically useful way, and render this episode in Chinese military history appropriate for comparative study.18 A reading of the Second Sino-Japanese War as a reflection of the power dynamics common to contemporary intrastate war does not require a radical reinterpretation of the conflict. As I show below, the comparison is both fitting and straightforward.  If, in 1937, one had to select from among all the purveyors of organized violence in northeastern China and designate that actor as the “incumbent”, Japan would have                                                          18 The standard understanding of civil war posits an internationally recognized sovereign engaging an armed, non-state challenger in a military contest that produces a total of 1000 casualties and produces an average 100 deaths per year (see Fearon, “Why Do Some Civil Wars Last So Much Longer Than Others?” 287).   14 been the most appropriate choice. Indeed, Japan had already established itself as the mainland’s eminent coercive entity, materially superior to the Guomindang, the CCP, and a colorful smattering of regional warlords, bandits, and private armies by virtually every sensible indicator. As early as 1928, the Japanese military boasted two aircraft carriers and 39 submarines, and less than nine years later, it claimed 968 fighter planes to the Guomindang’s 37.19 Japan’s coercive advantage was not expected to wane quickly: by 1937, the Japanese military’s artillery, machine gun, and tank counts far exceeded the Guomindang’s and the CCP’s, leading an aide to US General Joseph Stilwell to remark that ‘“...in tanks and ammunition supply Japanese strength was overwhelming; in mobility the Chinese were not even comparable.’”20 Although Chiang Kai-Shek remained the formally recognized head of the Chinese state in 1937, his party clearly could not claim control over the country’s eastern seaboard, and was simply overmatched by the Japanese incumbent, making the second united front between the Guomindang and CCP akin to cooperation between insurgent groups looking to wear down an incumbent through strategies of attrition, since neither was prepared to fight a conventional war.  Indeed, in the year preceding the formation of the second united front, the CCP was scattered and exhausted from the Long March of 1934-35. Only one-tenth of the original insurgent band that left Jiangxi Province with Mao Zedong ended the march with him, and throughout their search for a new base of operations, CCP troops were dogged                                                          19 Ian Buruma, Inventing Japan: 1853-1964, (New York: Modern Library, 2003), 84. Also speaking to relative capabilities is Jonathan Spence, The Search for Modern China 2nd ed., (New York: W.W. Norton, 1999), 443. 20 Quoted in Andrew Bingham Kennedy, “Can the Weak Defeat the Strong? Mao’s Evolving Approach to Asymmetric Warfare in Yan’an,” The China Quarterly 196 (2008): 887.   15 by the Guomindang, who inflicted heavy casualties on Mao’s forces.21 In 1936, the CCP was clearly in no condition to mount a serious threat to the Japanese military.  But the Japanese presence in mainland China during this period was not merely that of an occupying army, as the example of Manchukuo makes clear. Although Japanese-administered polities like Manchukuo suffered from an awkward political positioning, recognized neither as independent states nor as colonial holdings, they were not mere legal fictions, and the Japanese “quasi-states” in China’s east boasted fragmented but tangible empirical presences.22 The Japanese military, working in tandem with private enterprise, consolidated control over essential eastern rail lines throughout the 1930s, lending Japan considerable leverage in controlling the region’s trade and transportation, and effectively forcing the Guomindang from their urban strongholds in the east and exiling them to the southwest.23 The Japanese vision of its own incumbency was long-term: the South Manchuria Railway Company, a Japanese holding, was a focal point in Tokyo’s plan for a transportation network that would aid not only in the immediate war effort, but also grow into an arterial infrastructure sustaining puppet states like Manchukuo well into the future.24                                                          21 Patricia Buckley Ebrey, James B. Palais & Anne Walthall, Modern East Asia From 1600, (Boston, MA: Houghton Mifflin, 2006): 511-14. 22 See Thomas Dubois, “Inauthentic Sovereignty: Law and Legal Institutions in Manchukuo,” The Journal of Asian Studies 69, no.3 (2010): 759; Bruce Elleman and Stephen Kotkin, Manchurian Railways and the Opening of China: An International History, (Armonk, NY: M.E. Sharpe, 2009), 51. 23 Ebrey, Palais and Walthall, Modern East Asia from 1600, 515; Suzanne Pepper, “The KMT-CCP Conflict,” in The Cambridge History of China Vol. 13, Part 2 (Cambridge University Press: 1986), 738. Cambridge Histories Online, DOI: 10.1017/CHOL 9780521243384.014. 24 Elleman and Kotkin, Manchurian Railways and the Opening of China, 51; Li Narangoa, 2004, “Japanese Geopolitics and the Mongol Lands, 1915-1945,” European Journal of Asian Studies 3, no.1 (2004): 66-7.   16  Further, the legal codes and strictures officially promulgated by Manchukuo’s Japanese-appointed jurists in 1937 were five years in the making, and included an entire subfield dedicated to investment and banking in Manchukuo. The Japanese state also looked to resettle China's northeast with ethnic Japanese: in 1936, Tokyo and various allied firms looked to solicit commitments from more than one million Japanese to resettle in China’s northeast, which already had a Japanese population that grew from 68,000 in 1909 to 219,000 in 1930.25 These efforts were complemented with a system of Japanese-administered public education throughout the territory, which took as its central aim a “geopolitical socialization” that recast Manchukuo as a legally and politically legitimate territory with a subjugated identity amenable to Japanization.26  It would seem, then, that the Japanese occupation of eastern China during this period closely resembles common conditions of contemporary civil war in its more conventionally understood form: the incumbent has a strong empirical presence in a limited area, but that presence fragments as one moves further into the hinterland, where armed opposition groups contest the incumbent’s monopoly on violence. As with Japan’s resettlement campaigns, the incumbent may also attempt to resettle tracts of strategically desirable land with members of a favored ethnic group, looking to assimilate the indigenous populations and/or consolidate control over critical natural resources.27 How, then, did Japan’s coercive superiority and its bid for incumbency on the eastern mainland                                                          25 Thomas Dubois, "Inauthentic Sovereignty: Law and Legal Institutions in Manchukuo," The Journal of Asian Studies 69, no.3 (2010): 756-60. See also Narangoa, “Japanese Geopolitics and the Mongol Lands,” 50. 26 Narangoa, “Japanese Geopolitics and the Mongol Lands,” 55-6. 27 Fearon, “Why Do Some Civil Wars Last So Much Longer Than Others?” 283. See also Buhaug, Halvard Scott Gates & Paivi Lujalam, “Geography, Rebel Capability, and the Duration of Civil Conflict,” Journal of Conflict Resolution 53, no.4 (2009): 546.   17 spur the CCP and Guomindang to ally in 1936, and what is the theoretical significance of this alliance?  3.2. The Xi’an Incident  Negotiations are always more interesting when a nicely executed kidnapping provides the impetus for bargaining, and the Xi’an Incident was no exception. On December 12, 1936, Chiang travelled to Xi’an City in north central Sha’anxi Province, and met with Zhang Xueliang, an allied warlord charged with coordinating a major anti- communist offensive in the region. Upon Chiang’s arrival in Xi’an, Zhang detained the Guomindang leader, and refused to release him until he agreed to form an anti-Japanese united front with the CCP. Zhang’s reasoning was that Japan presented a far greater military threat to the Guomindang (and, by extension, to the territorial integrity of China) than did the communists, with whom Zhang’s military had already come to a de-facto ceasefire.28 This was to be the second such united front between the CCP and the Guomindang: in 1926, the two had allied for the Northern Expedition, a brief offensive designed to rid China of the warlords that had become particularly powerful political figures in the post-Qing landscape. This first alliance was short-lived: in April of 1927, after the Guomindang retook Shanghai, Chiang’s troops reneged on the deal and began attacking communists, driving the party from its urban power bases and deep into rural                                                          28 So Wai Chor, “The Making of the Guomindang's Japan Policy, 1932-1937: The Roles of Chiang Kai-Shek and Wang Jingwei,” Modern China 29, no.2 (2002): 242-3. See also Ebrey, Plais and Walthall, Modern East Asia from 1600, 515, and Ji Yiqiao, “The Xi’an Incident: The Reminiscences of Mr. Ji Yiqiao,” Chinese Studies in History 22 no. 3 (1986): 106.   18 Jiangxi Province, where Mao Zedong would later begin the Long March.29  Zhou Enlai, a central ally of Mao Zedong and a leading diplomatic light in the CCP, offered to “negotiate” Chiang’s freedom from Zhang in exchange for a preliminary commitment to an anti-Japanese alliance. Acquiescing to Zhou’s terms, Chiang was released on December 25th without further incident.30 In August of 1937, the second united front was formally created, and although it saw little active coordination between the CCP and Guomindang militaries, the two organizations were able to agree at least to a ceasefire while they each turned their attention to the Japanese incumbent.31 This is not to say, however, that the kidnapping was causally prior to the power disparity between incumbent and challengers: if Chiang and key elements of the Guomindang leadership did not agree on the probability of being eliminated by the Japanese incumbent, then Chiang would have abrogated the second united front immediately after he was released. Instead, the partnership lasted for approximately four years.  The awkwardness of this second partnership between the CCP and the Guomindang can hardly be overstated. Clearly, Chiang’s agreement to an anti-Japanese alliance was not easily won: as early as 1931, both Chiang and an important factional ally, Wang Jingwei, remained committed to wiping out the CCP before turning the Guomindang’s military focus on Japan.32 Wang’s distrust of the CCP was especially profound: he described allying with the communists as ‘“drinking poison to quench thirst”’, and he eventually defected to Japan after Chiang assented to the second united                                                          29 Charles Dobbs, “The Xi'an Incident,” in Encyclopedia of Modern Asia, Vol 6, ed. Karen Christensen and David Levison (New York: Scribner's, 2002), 364-5. 30 Youwei Xu and Philip Billingsley, “Behind the Scenes of the Xi'an Incident: The Case of the Lixingshe,” The China Quarterly 154 (1998): 284. 31 Kennedy, “Can the Weak Defeat the Strong?” 887. 32 Chor, “The Making of the Guomindang’s Japan Policy,” 213.   19 front.33 Citing surprise attacks near Shanghai and the Great Wall in the early 1930s, Wang argued that the CCP could not be trusted to abide by the terms of a ceasefire, and he feared that the communists would merely use the respite to improve their capabilities vis-a-vis the Guomindang and attack later.34 This general policy of targeting the CCP before focusing on Japan was maintained until 1935, when it became clear to Chiang that the Soviet Union, from whom he was seeking assistance against the encroaching Japanese military, would not support the Guomindang independent of the CCP. In response, Chiang sent envoys to CCP representatives in Guangzhou, Shanghai, and Moscow to signal his willingness to open talks with the communists, but his conditions included the non-negotiable demobilization of the Red Army, and its reorganization under the operational command of the Guomindang.35 Such a proposition was unlikely to have been accepted by the CCP, even if the Guomindang had not also been executing an offensive against communists in Shaanxi province at the time the offer was made, while also confining other CCP troops to their base in Ya’nan.36  Mao’s response to Chiang’s preconditions underscored the commitment problem that bedeviled CCP-Guomindang relations, and they mirrored Wang Jingwei’s fears regarding the communists. Mao expressed to Zhou Enlai a concern that Chiang would only use a ceasefire to rearm and attack the CCP later.37 Many other CCP elites were also reticent about the possibility of cooperating with the Guomindang. In 1935, Wang Ming, a senior CCP member, declared at the Seventh Congress of the Comintern that the threats                                                          33 Ibid, 246. 34 Ibid, 243. 35 Ibid, 242. 36 Ibid, 242; Ebrey, Palais, and Walthall, Modern East Asia from 1600, 515. 37 Kennedy, “Can the Weak Defeat the Strong?” 895.   20 posed by the Japanese military and the Guomindang were intrinsically connected, and that Chiang, who had already ‘“committed countless traitorous and cruel crimes”’ during his tenure as head of the Guomindang, was liable to ally with the Japanese and sell out the other Chinese resistance groups.38 Wang Ming’s accusations were not without some merit, and his instincts were basically correct: immediately prior to the Xi’an Incident, Chiang was indeed weighing the benefits of bargaining with the Japanese against those of seeking an alliance with Moscow.39  In many ways, the second united front was a surprising outcome. Neither the CCP nor the Guomindang had any reason to trust that the other would abide by the terms of a ceasefire or an alliance for very long, given Chiang’s abrogation of the first alliance and the grim outcome of the Northern Expedition. Further, the elite circles of both organizations, regardless of their factional persuasions, appear to have held each other in considerable contempt on the eve of the Second Sino-Japanese War. However, in late December of 1936, Chiang was coaxed into an anti-Japanese united front with the substantially weaker CCP, and by August 1937, the Red Army had been reconstituted as the Eighth Route Army, although the CCP was not made to relinquish administrative control over these troops.40  In short, the overwhelming coercive capacities of the Japanese incumbent relative to any Chinese contender positioned the CCP and the Guomindang to overcome their commitment problem for long enough to form an anti-Japanese united front. Following its formal creation, the Guomindang suspended its campaigns against the CCP, and the                                                          38 Michael Sheng, “Mao, Stalin, and the Formation of the Anti-Japanese United Front: 1935-37," The China Quarterly 129 (1992): 153-4. 39 Chor, “The Making of the Guomindang’s Japan Policy,” 241-2. 40 Kennedy, “Can the Weak Defeat the Strong?” 887.   21 two groups focused their attention on weakening the Japanese military, or at least focused less of their attention on fighting each other. This is not to say that the alliance by itself would have been sufficient to defeat the incumbent. Chiang’s forces suffered over six times the number of casualties they inflicted on Japanese troops near Hangzhou Bay in the autumn of 1937. 41  In the spring of 1938, the Guomindang demonstrated an underwhelming acumen for strategy, blowing up a set of dikes on the Yellow River in an attempt to slow the progress of Japanese troops into western China, and succeeding in drowning 4,000 villages while slowing the Japanese advance for only three months.42 Had the US not entered the war against Japan, the second united front very well could have lost the war. For the period that they were united, however, the CCP and Guomindang survived, and this largely fits within the contours of our first set of expectations, and even if the second united front would not have enabled the CCP and Guomindang to dispatch the Japanese incumbent, it probably enabled them to survive for longer than they might have otherwise.  3.3. Alterations to Rebels’ Relative Capabilities   The fracturing of the second united front in 1941 points to the importance of changes in relative capabilities that can impact the calculations of insurgent elites and bring the commitment problem back into focus. In 1940, the Roosevelt administration shipped 100 new fighter planes to Chiang Kai-Shek’s forces in Chongqing, but denied any aid to the CCP, leaving the Guomindang in a substantially better position vis-a-vis their united front partner in terms of both materiel and morale, further improving the                                                          41 Spence, The Search for Modern China, 423. 42 Ibid, 424-5.   22 Guomindang’s coercive capabilities relative to the Japanese.43 Indeed, by August of 1946, the US had supplied $900 million worth of military aid to the Guomindang, almost entirely to the exclusion of aid to the CCP.44 In light of each side’s preexisting fears that the other would defect from the second united front should the opportunity present itself, it should not be particularly surprising that after securing significant foreign support, the Guomindang abrogated the terms of the alliance and (again) attacked a CCP flank during the Fourth Army Incident of 1941.45  The formal entry of the United States into the war against Japan was another major exogenous change that altered the relative power of the incumbent and the armed opposition groups, since Japan would from then on have to fight on more fronts.46 The first set of conditions detailed in the model above no longer held: with a weaker Japanese incumbent, Guomindang elites no longer believed it necessary to ally with the CCP, and the Guomindang defected from the alliance. The two organizations went back to war with one another to resolve the question of which group would control the apparatuses of the state after supplanting the incumbent. Such control should be conceptualized as a good available in super-valuable increments, and this outcome aligns with the third set of expectations in the theory.  Indeed, Pepper documents the numerous failed attempts to form a CCP- Guomindang coalition government in 1945, and this speaks to the difficulty of dividing a super-valuable good that could significantly alter future bargaining power: when Chiang                                                          43 Ibid, 443. 44 Suzanne Pepper, “The KMT-CCP Conflict, 1945-1949,” 736. 45 Spence, The Search for Modern China, 443; Ebrey, Palais and Walthall, Modern East Asia from 1600, 516. 46 Spence, The Search for Modern China, 444.   23 argued that the remaining CCP strongholds in the countryside should be brought into the fold under a unified central government, the communists balked and negotiations stalled.47 Additionally, even as the two sides sought a power-sharing arrangement, CCP and Guomindang troops jostled with each other to establish an empirical presence in the strategically desirable northeast, which until 1945 had been administered by the Japanese incumbent. When General George Marshall was charged with overseeing the implementation of a ceasefire and the transition to a coalition government on the mainland, he found it impossible to assure the CCP of American impartiality, and as arms continued to flow from the US to the Guomindang, negotiations fell apart, and fighting between the CCP and Guomindang would not end again until 1949.48  In sum, what can we glean from this case? As per the theory, when insurgents believe that the incumbent boasts coercive capabilities sufficient to eliminate its armed non-state competitors, we can expect those vulnerable insurgent groups to ally, even if mutual suspicion textures their relationship. However, the alliance structure may fragment if one or both of the rebel groups anticipate a positive change in their capabilities relative to the incumbent. This may occur if at least one of the groups receives or anticipate receiving external material support that is not conditioned on the formation or continuation of an alliance between itself and other rebel groups, as in the Chinese case. If such alterations occur, at least one is no longer as dependent on the other for survival. The specter of the commitment problem comes back into play, prompting internecine targeting.                                                           47 Pepper, “The KMT-CCP Conflict,” 724. 48 Ibid, 729-30.   24 4   The LTTE in Sri Lanka (1983-2009) 4.1. The Movement for a Tamil Homeland  The Liberation Tigers of Tamil Eelam, and many of the other Tamil liberation groups with which the LTTE competed throughout the 1980s, were different from the other insurgencies in the present study because they never sought to supplant the incumbent. However, most of these groups at least nominally sought an autonomous Tamil homeland secure from Sinhalese political domination, although the LTTE was easily the most reluctant to seek a diplomatic solution to Tamil grievances.49 The secessionist nature of the conflict, however, did not make the commitment problems that developed between these liberation groups any less severe. The Tamil homeland (whether realized as an autonomous region within Sri Lanka or as an entirely separate state) would have required administrative and political control by a limited number of Tamil elites. Thus, control of an autonomous homeland should have represented a good available in super-valuable increments, the inequitable division of which could appreciably impact future bargaining between Tamil liberation organizations and political parties, severely disadvantaging some while disproportionately benefiting others. It should not, therefore, be surprising that internecine targeting came to characterize the early phases of the Sri Lankan civil war.  However, there is a second issue at play in this case. The incentives for internecine targeting may have been especially strong in the early phases of the Sri Lankan civil war because all the Tamil insurgencies espoused overlapping sets of goals.                                                          49 Deepa Ollapaly, The Politics of Extremism in South Asia (New York: Cambridge University Press, 2008); Jannie Lilja, “Ripening Within? Strategies Used by Rebel Negotiators to End Ethnic War,” Negotiation Journal 27, no.3 (2011): 320.    25 While these commonalities may have had some intuitive appeal for a united front among the Tamil liberation groups, they also represented a domain of competition: these organizations were vying for the support of Tamil civilians, or at least for unrestricted access to the human resources and financial capital this pool of civilians represented. So, how is it that insurgent organizations drawn from the same ethnic bloc might come to view themselves as competitors rather than allies?  There are two possible answers to this question: (1) it could be that an insurgent group A  fears that B will defect to the incumbent, along with B’s base of civilian support. This base would then represent a pool of resources to be mobilized against A  sometime later in the war, so internecine targeting becomes a perfectly rational strategy to be initiated by A . Indeed, “control over civilian populations is a central goal of both insurgents and states because the distribution of civilian loyalty potentially shapes war outcomes.” 50 At the very least, an insurgency’s ability to coax or extort civilian support will probably enable it to survive longer, and elite Tiger rhetoric often reflected this point, laced as it was with references to the traitorous proclivities of other Tamil liberation groups that were more amenable to bargaining with Colombo. Douglas Devananda, a former Tamil insurgent who defected to the state in 1987 and lost an eye in a highly personalized LTTE reprisal, put the point succinctly: ‘“we’re seen as traitors.”’51  However, it is also plausible that (2) A  is unconcerned by the possibility of B’s defection, and is simply interested in acquiring more civilians for A ’s predatory activities.                                                          50 Reed Wood, “Rebel Capability and Strategic Violence Against Civilians,” Journal of Peace Research 47, no.5 (2010): 602. 51 Quoted in Ian MacKinnon, “The Man Behind the Tamil Tiger's New Offensive; ‘You Will Have to Die Someday,”’ Newsweek, May 2000, 29.    26 This would only be the case, however, if A  is (or believes itself to be) sufficiently viable contra the incumbent regardless of whether or not B defects. This possibility suggests an important dimension of sequencing in the theory, and speaks to the role of organizational learning based on battle outcomes. In the early phases of the conflict, A  may fight the incumbent but not turn its weapons on B, unsure of whether it will need B later. As the course of the war reveals more information, and it becomes clear that A  can probably survive by itself, there is no reason for A  not to initiate internecine targeting, because under such circumstances, B can represent only three things to A : a defection risk, an ex post commitment problem over the equitable division of supervaluable war spoils, or an obstruction to the acquisition of more civilians for A ’s predatory activities.  4.2. Why Tigers Don’t Need Allies  Armed Tamil organizations coalesced throughout the 1970s and peaked in number at about 35 by the early 1980s, but eventually dwindled back down to the Eelam People’s Revolutionary Liberation Front (EPRLF), the Eelam Revolutionary Organization of Students (EROS), the People’s Liberation Organization of Tamil Eelam (PLOTE), the Tamil Eelam Liberation Organization (TELO), and the LTTE, which had formally come into being in 1976.52 After the “Black July” riots of 1983, which saw two to three thousand civilians massacred and one hundred fifty thousand displaced by anti- Tamil violence, Tamil militant groups began to campaign against the Sri Lankan state in                                                          52 Lilja, “Trapping Constituents or Winning Hearts and Minds?” 312; Olapally, The Politics of Extremism in South Asia, 146; Alice Stack-O’Connor, “Lions, Tigers, and Freedom Birds: How and Why the Liberation Tigers of Tamil Eelam Employs Women,” Terrorism and Political V iolence 19, no.1 (2007): 44.   27 earnest.53  It seems that these insurgent organizations would have had every reason to ally: “The stated agenda of all these groups” ... “contained nearly indistinguishable objectives on Tamil Eelam, and there was even a significant effort to put forward a united front when dealing with the Sri Lankan government.”54 However, the LTTE avoided serious, sustained involvement with any kind of pan-Tamil united front, and by the mid-1980s, it was actively targeting the other insurgent organizations operating within Sri Lanka. Mutual suspicion naturally came to characterize relations between the LTTE and its rivals. As a former member of PLOTE recalls, the LTTE assassinated TELO head Sri Sabaratinam and murdered nearly 150 other TELO members in Jaffna in 1986.55 While discussing the internal discipline structure of the PLOTE, another member recalls that anyone in the ranks expressing a desire to quit was accused of being an LTTE or EPRLF informant, and was subjected to torture.56 By 1988, the LTTE had killed approximately 600 members of other Tamil militant groups.57 By July 1989, the LTTE had largely succeeded in its campaign, and “no viable alternative to the LTTE has existed since then.”58                                                          53 Gamini Samaranayake, “Political Terrorism of the Liberation Tigers of Tamil Eelam (LTTE) in Sri Lanka,” South Asia: Journal of South Asian Studies 30, no.1 (2007): 173. See also Sumantra Bose, Contested Lands: Israel-Palestine, Kashmir, Bosnia, Cyprus, and Sri Lanka (Cambridge, MA: Harvard University Press, 2007), 28. 54 Olapally, The Politics of Extremism in South Asia, 160. 55 Chitra Sivakumar, Transformation in the Sri Lankan Tamil Militant Discourse: Loss of the Tamil Self, V iolence, and the Hermeneutic of Recovery (Madras: University of Madras, 2001),10. 56 Ibid, 25. 57 Neil Devotta, “The Liberation Tigers of Tamil Eelam and the Lost Quest for Separatism in Sri Lanka,” Asian Survey 49, no.6 (2009): 1029. See also Lilja, “Trapping Constituents or Winning Hearts and Minds?” 314. 58 Olapally, The Politics of Extremism in South Asia, 160.   28  Staniland expresses surprise that “despite successes against the Sri Lankan state during this period, in 1986 and 1987 the LTTE turned its guns on its rivals within the Tamil militant movement.” 59  Certainly, it seems counterintuitive that a reasonably successful wartime arrangement would suddenly be abrogated by one of the insurgencies, and the LTTE’s attacks on other Tamil militants certainly shocked its civilian constituents. 60  However, if the conditions proposed above accurately convey the incentive structures of insurgent leadership circles, this outcome should not be surprising. If the incumbent proves incapable achieving an absolute victory over the varied insurgent groups, and it seems that the probability of surviving against the incumbent is reasonably high for at least one of the insurgencies, then a commitment problem should come into play, making internecine targeting more likely. The targeting would be especially likely if one of the insurgent groups has substantially more military acumen than the others, as was the case with the LTTE.61  On the other hand, if the probability of being eliminated by the incumbent is high for all insurgent groups, and none of the insurgencies trust the incumbent to abide by the terms of a ceasefire, then the insurgent groups should look to form an alliance. One might ask why it is that, if commitment problems color the relationship between A and B as well as between A  and the incumbent, why A  would opt to ally with B rather than the state, if it trusts neither. The answer is that if the incumbent is the overwhelming coercive presence, then even if B breaks the terms of an alliance with A , the consequences for A will be less dire: the CCP managed to survive despite the Guomindang’s abrogation of                                                          59 Staniland, “Between a Rock and a Hardplace,” 32. 60 Lilja, “Trapping Constituents or Winning Hearts and Minds?” 314. 61 Ibid, 311. See also Lilja, “Ripening Within?” 320.   29 the second united front, whereas if it had allied with the Japanese incumbent, it would probably have been less likely to survive the incumbent’s abrogation of the deal. Although the LTTE had always been significantly outgunned by the Sri Lankan military, it was never in danger of being completely wiped out until much later in the conflict, and their guerilla capabilities enabled the Tigers to inflict appreciable damage on all three branches of the incumbent military throughout the conflict, which no other Tamil insurgency could do.62  After the first few years of fighting, the Tigers would have learned that they could survive without establishing a united front with other Tamil insurgent groups. This might go some way in explaining why most other Tamil rebels eventually defected to Colombo, whereas the Tigers remained committed to violence and strategies of costly signaling. LTTE head Villupillai Prabhakaren ordered his subordinates to shoot him if he defected from the Tigers or at any point compromised on the issue of secession, and this order remained active even when Prabhakaren was on his deathbed watching Clint Eastwood films.63 The LTTE was also notable for its members’ use of cyanide capsules, which were to be swallowed in case of capture.64 Well into the war’s third decade, the LTTE continued to send unmistakable signals of resolve to Colombo: from 1980 to 2003, the LTTE carried out more suicide attacks than any other terrorist group or insurgent organization in the world, one of the most spectacular being carried out by a pregnant                                                          62 P. Sahadevan, “On Not Becoming a Democrat: the LTTE’s Commitment to Armed Struggle,” International Studies 32, no.3 (1995): 272-3. 63 Economist, “Obituary: Villupillai Prabhakaren,” May 2009; Devotta, “The Liberation Tigers of Tamil Eelam and the Lost Quest for Separatism in Sri Lanka,” 1030. 64 Ibid, 1029.   30 woman in front of Colombo’s army headquarters in 2006.65 The point is clear: the LTTE proved its acumen fairly early in the war and did not trust the resolve of other groups, casting them as potential defectors and thus, appropriate targets for internecine violence.  The LTTE’s monopoly on the Tamil resistance movement paid off nicely: the Tamil diaspora, concentrated in India and various western states like Canada, was a critical source of funding for the Tigers, and the ability to monopolize the movement for a sovereign Tamil homeland meant that any member of the diaspora willing to donate to an armed resistance group had only the LTTE to support.66 The Tigers were, through methods of persuasion and extortion, able to garner $50 million in donations from the Tamil diaspora well into the mid-2000s, while Tamil communities in the northern areas of the island, dominated as they were by the LTTE, provided a sizable pool of voluntary and forced recruits.67 Forcible conscription, extortionate levels of taxation, and civilian abuse eventually came to characterize LTTE relations with many Tamil civilians, indicating the organization’s metamorphosis into something more akin to a predatory mafia than a liberation group.68 From 2002 to 2006, the Tigers also executed campaigns against various splinter groups, further consolidating the LTTE’s claim to the Tamil population, which it would have no incentive to share with other Tamil insurgent groups                                                          65 Bose, Contested Lands, 10. 66 Devotta, “The Liberation Tigers of Tamil Eelam and the Lost Quest for Separatism in Sri Lanka,” 1023. 67 Kyle Beardsley and Brian McQuinn, “Rebel Groups as Predatory Organizations: The Political Effects of the 2004 Tsunami in Indonesia and Sri Lanka,” Journal of Conflict Resolution 53, no.4 (2009): 633-4. 68 Kristine Hoglund, “Violence and the Peace Process in Sri Lanka,” Civil Wars 7, no. 2 (2005): 161; Devotta, “The Liberation Tigers of Tamil Eelam and the Lost Quest for Separatism in Sri Lanka,” 1032.   31 if it could survive alone.69 Pursuant to this point, Staniland considers the instance of Colonel Karuna, who led a splinter group that defected to the state late in the war. Karuna controlled an independently viable flank of soldiers, and when summoned by Prabhakaren in 2004 to face corruption charges and possible execution, Karuna’s faction went to war with the rest of the LTTE.70 It makes sense, then that if Prabhakaren feared Karuna’s defection and his group’s independent viability, he would initiate tactics to either decapitate the faction or bring it to heel.  4.3. Significances, Empirical and Theoretical  What does the LTTE’s strategy of internecine targeting illustrate? In this instance, a rebel group targeted and severely weakened the other rebel organizations drawn from the same ethnic bloc and espousing most of the same goals. However, the LTTE’s strategy was perfectly rational, if seemingly counterintuitive. The strategy decreased the number of alternatives behind which the Tamil population could throw its support, or, put differently, it lowered the number of armed groups that could extort recruits and finances from the civilian population. Indeed, the campaign against rival insurgencies saw the LTTE “eliminating the leaders of other groups (both moderate and militant ones) and forcing their supporters to either join its ranks or leave the areas under LTTE control.” 71 Given that the support of a civilian population does not easily lend itself to equitable division, it makes sense that the LTTE and other Tamil militant organizations would                                                          69 Beardsley and McQuinn, “Rebel Groups as Predatory Organizations,” 634-5; Hoglund, “Violence and the Peace Process in Sri Lanka,” 164. 70 Staniland, “Between a Rock and a Hard Place,” 34-5. 71 M. Mayilvaganan, “Is it Endgame for the LTTE?” Strategic Analysis 33, no.1 (2009): 25.   32 target one another to decrease the number of ways in which that support might be split. This strategy later placed the LTTE in a desirable position: by the mid-1990s, the LTTE was in control of one-fourth of the state’s territory, and boasted a 20,000-strong military force.72 Further, the ceasefire that (briefly) quieted Sri Lanka’s civil war in 2002 was a bargain brokered between Colombo and only the LTTE, because it was the only serious domestic threat that was left.73  This brief historical rendering makes no claims as to discerning the LTTE's “real” motivations. The greed/grievance dichotomy is an overly restrictive one, and while many Tigers were surely committed to the overarching goal of Tamil liberation (hence a pregnant suicide bomber), others were probably motivated more as war entrepreneurs seeking opportunities for profit and predation.74 Korf is on point in arguing that seeking justice and seeking profit were inseparable motivations for the LTTE, and whatever an insurgency's “real” goals may be initially, they may morph over time.75 A  and B may seek to secure political goods for a valued constituency by seceding or capturing the state, in which case disagreements over the ex post division of a super-valuable good is likely to prompt internecine targeting, absent overwhelming incumbent capabilities. A and B may also seek to profit from the development of a war economy, in which case disagreements over how to divide goods such as civilian support are likely to develop between the insurgencies.                                                          72 Devotta, “The Liberation Tigers of Tamil Eelam and the Lost Quest for Separatism in Sri Lanka,” 1023. 73 Kristin Hoglund and Isak Svensson, “Mediating Between Tigers and Lions: Norwegian Peace Diplomacy in Sri Lanka's Civil War,” Contemporary South Asia 17, no. 2 (2009): 175. 74 Beardsley and McQuinn, “Rebel Groups as Predatory Organizations.” 75 Benedikt Korf, “Functions of Violence Revisited: Greed, Pride, and Grievance in Sri Lanka's Civil War,” Progress in Development Studies 6, no.2 (2006): 118.   33 5    The Angolan War for Independence (1961-1974) 5.1. The MPLA, FNLA, and UNITA  Although the Portuguese colonial incumbent was not dispatched from Angola until 1974, indigenous Angolan insurgencies had been coalescing in earnest since the late 1950s.76 However, only three distinct insurgencies developed into appreciable military presences during the period, and it is with these organizations that the present study is principally concerned. Unlike the Chinese and Sri Lankan insurgencies discussed above, the Angolan groups had ethnically differentiated support bases. The Movimento Popular de Libertacao de Angola (MPLA) formed in 1956, drew primarily from the Mbundu intelligentsia of north-central Angola, and was led by Augustinho Neto.77 The Frente Nacional de Libertacao de Angola (FNLA) formed in 1962, drew from the Bakongo ethnicity in the north, and was headed by Holden Roberto, a brother-in-law to Zairean president Mobutu Sese Seko.78 The Uniao Nacional para a Independencia Total de Angola (UNITA) formed in 1966 as a breakaway from the FNLA, played mostly to the Ngangela, Chokwe, and Ovimbundu peoples of the south, and was led by Dr. Jonas Savimbi (whose PhD was, indeed, in political science).79 Two of these organizations also established their early bases in exile: the MPLA worked out of Congo-Brazzaville in 1961, and the FNLA conducted its early operations from Mobutu’s Zaire.80                                                          76 Inge Tvedten, Angola: Struggle for Peace and Reconciliation (Boulder, CO: Westview Press, 1997). See also Martin James III, Political History of the Civil War in Angola, 1974-1990 (New Brunswick, NJ: Transaction Publishers, 1992). 77 Steven F. Jackson, “China's Third World Foreign Policy: The Case of Angola and Mozambique, 1961-93,” The China Quarterly 142 (1995): 395. 78 “The Legacy of Oil and Geopolitics,” New African, November 2009. 79 Gordon McCormick and Lindsay Fritz, “The Logic of Warlord Politics,” Third World Quarterly 30, no.1 (2009): 85. 80 Inge Brinkman,”Language, Names and War: The Case of Angola,” African Studies   34  Although all three groups were targeting the Portuguese incumbent by 1967, they consistently articulated themselves as alternatives to each other throughout the war for independence.81 Rather than developing an enduring alliance or non-aggression pact, these three rebel groups instead “...battled each other more than the Portuguese, with MPLA rejecting attempts to form even a common front against enemy forces.”82 Why would these groups actively fight each other as they fought the Portuguese incumbent? There are several explanations boasting some intuitive appeal, but none of them hold up to scrutiny. Lockyer, for example, lists “contending political ideologies, their leaders’ egos and the fact that they each recruited from different ethnic populations” as the central variables explaining the internecine targeting between the MPLA, FNLA, and UNITA.83 I briefly consider these in order.  The ideological explanation is unsatisfactory on two levels. First, it infers political motivations from proclaimed ideologies, but this may not be valid: “the FNLA had little real ideology to speak of, and no regular organization beyond what Holden Roberto could personally control.”84 Even less charitably, “the FNLA, as the CIA would later admit, was nothing more than a motley crew of ragtag dogs of war.” 85 Second, even if one were to grant that the FNLA took something approaching an anti-communist line, the ideological explanation would have some difficulty accounting for the formation of a brief alliance between Savimbi’s nominally Maoist insurgency and Roberto’s FNLA in                                                                                                                                                                       Review 47, no.3 (2004): 144-5. 81 Tvedten, Angola: Struggle for Peace and Reconciliation, 30-2. 82 James, Political History of the Civil War in Angola, 1974-1990, 45. 83 Adam Lockyer, “Foreign Intervention and Warfare in Civil Wars,” Review of International Studies 37 (2011): 2348. 84 Jackson, “China’s Third World Foreign Policy,” 395. 85 “The Legacy of Oil and Geopolitics,” New African, 45.   35 1974. Further, the MPLA’s communist leanings should have aligned more closely with UNITA’s Maoist ideology than with that of the FNLA, but Neto and Savimbi never allied against Roberto.  Proceeding to the second of Lockyer’s explanations, it would seem that the size of an insurgent leader’s ego would be a difficult variable to operationalize, but even if it could be convincingly measured, the problem of egotism in an insurgency’s elite leadership circle would not account for the variation in insurgent alliance patterns discussed in the present study. It seems unlikely that the egos of Neto and Roberto, who fought, would have significantly overshadowed those of Chiang and Mao, who allied. Further, it would be difficult to convincingly demonstrate that an insurgent leader’s strategic decisions were based on egotism rather than privately held information. Individually focused, non-falsifiable psychological explanations do not seem to have the same analytical potential as a rationalist explanation for internecine targeting.  Lastly, the explanation centered on ethnic hostility has significant shortcomings. Although it is certainly true that the Angolan insurgencies played to different ethnic blocs, ethnic difference is only a descriptor of the lines along which the insurgencies differentiated themselves, not a causal motivation for violence between them. In general, Kalyvas finds “considerable heterogeneity and fluidity in the behavioral expressions of ethnic identities within civil war.”86 Indeed, the FNLA and UNITA would cooperate against the MPLA in 1974, even though the FNLA and UNITA were formed from different ethnic blocs.87 Further, the Sri Lankan case shows that insurgencies drawn from                                                          86 Stathis Kalyvas, “Ethnic Defection in Civil War,” Comparative Political Studies 41, no.8 (2008): 1045. 87 See Tvedten, Angola: Struggle for Peace and Reconciliation.   36 the same ethnic group will target one another under certain material conditions, making the rationalist explanation proposed here more satisfying.  5.2. Internecine Targeting Within the Angolan Resistance  In examining the choice of insurgent elites to fight each other as well as the Portuguese incumbent, the conditions isolated in the theory again reveal themselves: an incumbent without overwhelming coercive capabilities was coupled with an inability among the insurgent groups to commit ex ante to the equitable distribution of super- valuable war spoils. Not surprisingly, the three insurgent groups battled each other throughout much of the Angolan struggle for independence.  When small-scale guerrilla attacks began in 1961, the 6,500-strong Portuguese colonial military demonstrated little acumen for counterinsurgency, and although incumbent forces would swell to 70,000 by 1970, the Portuguese consistently failed to make the long-term infrastructural developments that might have won over essential rural populations during the war.88 Although the incumbent military eventually improved its counterinsurgency capabilities and boasted considerable numerical and materiel advantages over the MPLA, FNLA, and UNITA, the three insurgencies were never in serious danger of being completely eliminated by the incumbent, and the costs to the regime in Lisbon were such that “the April 1974 coup should not have come as a surprise.”89 Indeed, a US National Security Memorandum from the period describes the war in Angola as a costly stalemate: ‘“the rebels cannot oust the Portuguese and the                                                          88 Stephen Weigert, Angola: A Modern Military History, 1961-2002, (New York: Palgrave Macmillan, 2011), 43-50. 89 Ibid., 50.   37 Portuguese can contain but not eliminate the rebels.”’ 90  Weigert documents in detail the internecine targeting between the three Angolan insurgencies, and in reviewing this dimension of the Angolan war for independence, the ubiquity of the phenomenon comes into focus: the MPLA and FNLA were targeting one another’s columns as early as 1963 and continued to battle sporadically until 1974, while the MPLA began ambushing UNITA flanks almost as soon as they appeared in eastern Angola in 1966. 91  Internecine battles between all three organizations became commonplace: in 1971, Savimbi formally declared that he was at war with the other two insurgencies. Each organization would also come to view the others as defection risks: accusations of collusion with the colonial incumbent were commonly leveled by each insurgency throughout the war, and the accusations were probably not inaccurate.92 In keeping with the predictions of the model, the Portuguese incumbent proved incapable of implementing an effective counterinsurgency strategy, and since at least one of the Angolan insurgencies would not have perceived a serious risk of being eliminated in the early phase of the war, it should not be surprising that these organizations did not ally. Why, however, would they have actively targeted each other, especially since they would not have been competing for the same civilian support base, as were the LTTE and other Tamil insurgencies?                                                            90 Quoted in James, Political History of the Civil War in Angola, 1974-1990, 41. 91 Weigert, Angola: A Modern Military History, 53-5. 92 Edward George, The Cuban Intervention in Angola, 1965-1991: From Che Guevara to Cuito Cuanavale (New York: Frank Cass, 2005), 52; Weigert, Angola: A Modern Military History, 54-5.   38 5.3. The Decolonized State as a Super-valuable Good  During the war for independence, it does not appear that the three insurgent organizations had clear ideas regarding how Angola might be administered once the Portuguese withdrew, and this is an important point because territory in Angola is not evenly distributed in terms of its value. The Cabinda region housed substantial oil deposits, Lunda Norte produced considerable diamond wealth during the 1960s, and Huila province contained substantial beds of iron ore.93 Control of the state following the withdrawal of the Portuguese would, then, be a highly desirable good because it could determine the allocation and management of essential resources and pieces of territory, which could in turn substantially alter the future bargaining power of each group if distributed inequitably. Indeed, control over the decolonized Angolan state should have represented control over the distribution of rents from Angola’s spectacular natural resources, and conceptualizing the decolonized state in this way would not have been wrong: LeBillon shows that the oil sector has come to be “the key resource of the Angolan economy and the government” ... “and as such is inaccessible to most Angolans except via revenues directed to public expenditure.” 94 Indeed, the state monopoly over oil wealth, realized through SONANGOL, enabled the MPLA to buy the loyalty of the state’s military for several decades after it won control of the capital in 1975, and import $5 billion worth of arms in the 1990s alone.95  The point is not merely retrospective: by the time the Portuguese incumbent was removed, it had become clear that all three of the insurgent groups in Angola sought to                                                          93 James, Political History of the Civil War in Angola, 21-2. 94 Phillippe LeBillon, “Angola's Political Economy of War: The Role of Oil and Diamonds, 1975-2000,” African Affairs 100, no. 398 (2001): 61-3. 95 Ibid, 64.   39 control a good they conceptualized as supervaluable: “After forging the common platform in Mombasa, the movements had engaged in tactical maneuvers with one simple objective: taking over power unilaterally in November 1975.” 96 Indeed, “in Angola three parties were all competing for formal recognition and power first within Angola, secondly in the OAU, and finally around the world.” 97 The Alvor Peace Accords, convened by the Organization for African Unity to establish a power-sharing agreement between the three groups in 1974, collapsed because none of the three could agree on an equitable distribution of the shares of state power, and the groups subsequently went back to war, this time with the MPLA as the incumbent and the FNLA and UNITA once again in roles as the armed opposition. Even if they could have agreed on how the power was to be distributed, it is not at all clear that each would have been able to credibly commit to respect the division after years of fighting one another, nor is it clear by what serious, unbiased mechanism the terms of power-sharing would have been enforced. The distrust among the different groups at the Alvor Accords was clear. One delegate from the MPLA noted: ‘“There is no unity. The only unity is a common text brought from Mombasa to discuss with the Portuguese.”’ 98  The theory’s analytical value in explaining the Angolan case does not end here, however. Throughout 1975, the MPLA assumed the role of incumbent, consolidating control in Luanda and in several provincial capitals to boot.99 The MPLA’s rapid military ascension during this period demonstrates the value of controlling the mantle of the state:                                                          96 Gilbert Khadiagala, “Negotiating Angola's Independence Transition: The Alvor Accords,” International Negotiation 10, no. 2 (2005): 302. 97 Jackson, “China’s Third World Foreign Policy,” 404. 98 Quoted in Khadiagala, “Negotiating Angola’s Independence Transition,” 303. 99 Weigert, Angola: A Modern Military History, 59-60.   40 its “increasingly complex network of international allies” provided the MPLA with a colorful array of new weapons, and $200 million in military aid from Moscow between late 1975 and mid 1976 alone.100 This was not, of course, a wholly new development: in the early 1970s, the MPLA received military support from the Soviet Union that ran the gamut from Sam-7 anti-aircraft missiles to a PT-76 amphibious tank.101 However, the payoff to being a sovereign as opposed to a rebel is considerable: Henry Kissinger sought to ensure that US support for the FLNA and UNITA remained modest and wholly covert, ‘“referring to legal obstacles to military assistance to insurgents.”’ 102  It seems that, by 1975, UNITA and the FNLA came to see the MPLA as a significant threat as it asserted control over the capital. Given that Neto was backed with military aid and advising from the USSR and Cuba that dwarfed the support given to Roberto and Savimbi by Pretoria and Washington, the FNLA and UNITA would have realized that they would not be facing a counterinsurgency strategy as ineffective as that implemented by Portugal. Neto now not only had the FNLA and UNITA significantly outgunned, but the MPLA had several years’ experience fighting both.103 As such, the new incumbent would have known not only the opposition’s capabilities but also the anatomy of an Angolan insurgency. Indeed, Soviet military aid “enabled the MPLA to field thousands of well-equipped soldiers, encouraging its leadership to exterminate its                                                          100 Ibid, 60-2. 101 James, Political History of the Civil War in Angola, 54. 102 Quoted in Stephen Weissman, “CIA Covert Action in Zaire and Angola: Patterns and Consequences,” Political Science Quarterly 94, no.2 (1979): 283. 103 On the differences between access to external support between the three groups, see Weigert, Angola: A Modern Military History, 59-60; Weissman, “CIA Covert Action in Zaire and Angola,” 284.   41 rivals rather than negotiate with them.” 104  In 1974, the MPLA and FNLA had roughly similar capabilities, but one year later, the distribution of those capabilities had been dramatically altered.105 As the theory would predict, the FNLA and UNITA began cooperating (with assistance from the South African Defense Forces and the CIA) to launch campaigns against Neto and the MPLA government in April 1975, although by 1976, the militarily superior MPLA had sent the FNLA/UNITA alliance back to the hinterlands, where they would lose access to military aid from patrons that saw no utility in further funding the losing side.106 Indeed, in early 1976, observers believed that having defeated the FNLA, the MPLA “would soon destroy UNITA’s forces,” though this of course did not come to fruition until several decades later.107  What does the Angolan case tell us? The protracted conflict between the MPLA, FNLA, and UNITA appears to be a good illustration of the third set of conditions proposed in the theory. The Portuguese incumbent was incapable of convincingly defeating the opposition groups, and the groups could not agree on how to divide ownership of the super-valuable decolonized state. As a result, not only did they fail to ally, they actively targeted each other to avoid being disadvantaged when it came time to divide the good, and quite predictably, the attempt to divide the good ended in failure. When the MPLA consolidated its control in Luanda and assumed the incumbent role, the                                                          104 George, The Cuban Intervention in Angola, 60. 105 Lockyer, “Foreign Intervention and Warfare in Civil Wars,” 2352-4. 106 George, The Cuban Intervention in Angola, 60; Lockyer, “Foreign Intervention and Warfare in Civil Wars,” 2349-60; Weigert, Angola: A Modern Military History, 60-4. 107 Linda Heywood, “UNITA and Ethnic Nationalism in Angola,” The Journal of Modern African Studies 27, no.1 (1989): 58.    42 latter two allied to avoid destruction as information regarding the MPLA’s rapidly improving capabilities became clear to all parties. It would seem, then, that if insurgencies A  and B perceive that the incumbent’s capabilities are improving, and neither of the insurgent leadership circles have any reason to believe that such a trajectory will change in the near future, they are more likely to ally in order to avoid elimination.                     43 6   Conclusion  Bakke, Cunningham and Seymour consider in detail patterns of fragmentation and their impact on relative power between factions in broad-based nationalist movements, providing helpful insights on whether and how infighting within an opposition movement will proceed. Centralization, cohesion, and a high degree of institutionalization are all predicted to lower the probability of infighting, and while many of these predictions are wholly reasonable, they are not specific to rebel groups that fight one another while targeting the same incumbent, and they do not focus on commitment problems between rebel groups as explanations for this type of wartime dynamic.108 By specifying perceived incumbent capability as the key independent variable determining whether two or more rebel groups will ally during wartime, and then proposing the commitment problem as an explanation for why unallied groups will fight before dispatching the incumbent or expropriating a desired concession, the present study makes a unique contribution to our knowledge of internecine targeting among rebels during civil war.  While the cases detailed above render it empirically plausible, the theory is not invulnerable to falsification. A case suggesting the absence of a commitment problem between independently viable rebel groups would undermine the model. For example, if two or more rebel groups do not engage in internecine targeting while fighting the same weak incumbent, and do not fight one another following their defeat of that incumbent, there may have been no operative commitment problem in that case, or it may have been mitigated by some exogenous mechanism. One might plausibly argue, for example, that the case of FRELIMO in Mozambique does not comport with the theory’s expectations.                                                          108 For a list of central variables and hypotheses, see Bakke, Cunningham and Seymour, “A Plague of Intials,” 279.   44  Indeed, FRELIMO was an amalgamation of several distinct liberation groups that cooperated to oust the Portuguese colonial incumbent, and although defections and infighting occurred in 1969 following the death of FRELIMO head Eduardo Mondlane, the remnants of the united front soldiered on until FRELIMO assumed the mantle of the state in 1975.109 At independence, the FRELIMO coalition certainly projected a united image.110 However, after FRELIMO assumed control of the Mozambican state, struggles over how to distribute land and administrative power exacerbated preexisting regional tensions, and FRELIMO began targeting the rival independence groups that had defected from the united front, some of whom joined the nascent RENAMO insurgency the following year.111  The Sikh insurgency in Punjab might also, on face, appear to be a difficult case for the theory to explain. By the early 1990s, the movement for an independent Sikh state boasted seven distinct insurgent groups, who fought one another more frequently than they fought the incumbent.112 Indeed, the Indian state claimed conventional military superiority by a wide margin, and so it may seem surprising that the Sikh rebels engaged                                                          109 Jeremy Weinstein and Laudemiro Francisco, “The Civil War in Mozambique: The Balance Between Internal and External Influences,” in Understanding Civil War: Evidence and Analysis, Vol.1: A frica, ed. Paul Collier and Nicholas Sambanis (Herndon, VA: World Bank Publications, 2004), 161-3. 110 Alice Dinerman, Revolution, Counter-Revolution and Revisionism in Post- Colonial A frica: The Case of Mozambique, 1975-1994 (New York: Routeledge Studies in Modern History, 2006), 49. 111 Weinstein and Francisco, “The Civil War in Mozambique,” 164; Weinstein, Inside Rebellion, 73-6. 112 Jugdep Chima, “Controlling the Sunni Insurgency in Iraq: ‘Political’ and ‘Military’ Strategies from Successful Counterinsurgency in Punjab-India,” Small Wars & Insurgencies 18, no.4 (2007): 620.    45 in internecine targeting.113 However, the probability that the Indian state would violently eliminate the insurgent groups was actually fairly low, simply because it had a different strategy. Chima shows that although the state indicated that it would never acquiesce to a Sikh secession, it would entertain greater autonomy for Punjab, and by the late 1980s, was essentially content to let the Sikh insurgencies burn out, which is largely what occurred: the insurgent organizations essentially devolved into criminal bands preying on civilian populations.114 It is possible, then, that the Punjab case might not fall within the purview of the theory: while the three insurgencies detailed in the previous section fought one another fearing alterations in future bargaining power, the Sikh organizations had a very low probability of creating an independent state, and thus of capturing a super- valuable good. This is not unlike the way that other criminal organizations will fight one another over turf, even when they have no serious chance of achieving a much larger political goal. A more detailed treatment of cases such as these, which do not at first appear to conform to the theory’s expectations, could provide valuable refinements to the theory, and further the field’s understanding of insurgent alliance patterns in civil war.  The internecine targeting described in the present study appears to be a puzzling wartime arrangement for rebel groups, and yet it is a fairly common phenomenon. Ideological disagreement and ethnic hostilities are not compelling as causal motivations for rebel on rebel violence, and do not accurately predict patterns of alliance and                                                          113 On the balance of power between the Indian state and Sikh insurgents, see Shale Horowitz and Deepti Sharma, “Democracies Fighting Insurgencies: Evidence from India,” Studies in Conflict & Terrorism 31, no.8 (2008): 758. 114 Chima, “Controlling the Sunni Insurgency in Iraq,” 619; Gupreet Bal and Paramjit Judge, “Terrorism and Rural Entrepreneurship in Punjab,” Journal of Entrepreneurship 10, no.2 (2001): 205.    46 defection. More helpful is a rationalist explanation centered on the development of commitment problems between insurgent organizations. Civil war is a risky enterprise for rebels, and the costs of allying with an unreliable partner can be very high. It is for this reason that the formation of a united front between rebel groups only becomes likely when the incumbent’s counterinsurgency capabilities are sufficient to credibly threaten the rebels’ violent elimination. When it appears that two or more distinct rebel groups will be able to survive contra the incumbent, two motivations for internecine targeting arise. First, civilian support as a means of improving standing relative to the incumbent constitutes a valuable prize, and one which insurgent groups should therefore be willing to contest. Second, groups with a decent probability of capturing the state or establishing a new state will struggle to credibly commit to the equitable division of such a super- valuable good, and the fear of being disadvantaged in future bargaining situations as a result of inequitable division will incentivize ex ante internecine targeting. Specifying a theory of internecine targeting constitutes a modest first step toward the explanation of this phenomenon, which can be refined with further process-tracing work, especially in cases seemingly inhospitable to the model.          47 Bibliography  Bakke, Kristin, Kathleen Cunningham and Lee Seymour. “Shirts Today, Skins Tomorrow: Dual Contests and the Effects of Fragmentation in Self-Determination Disputes.” The Journal of Conflict Resolution 56, no. 1 (2012): 67-93.  ------- “A Plague of Initials: Fragmentation, Cohesion, and Infighting in Civil Wars.”  Perspectives on Politics 10, no. 2 (2012): 265-283.  Bal, Gupreet and Paramjit Judge. “Terrorism and Rural Entrepreneurship in Punjab.”  Journal of Entrepreneurship 10, no. 2 (2001): 191-208.  Beardsley, Kyle and Brian McQuinn. “Rebel Groups as Predatory Organizations: The Political Effects of the 2004 Tsunami in Indonesia and Sri Lanka.” Journal of Conflict Resolution 53, no. 4 (2009): 624-645.  Bose, Sumantra. Contested Lands: Israel-Palestine, Kashmir, Bosnia, Cyprus, and Sri  Lanka. Cambridge, MA: Harvard University Press, 2007.  Brinkman, Inge. “Language, Names and War: The Case of Angola.” African Studies  Review 47, no. 3 (2004): 143-163.  Buhaug, Halvard and Jan Ketil Rod. “Local Determinants of African Civil Wars,            1970-2001.” Political Geography 25, no. 3 (2006): 315-335.  Buruma, Ian. Inventing Japan: 1853-1964. New York: Modern Library, 2003.  Chima, Jugdep. “Controlling the Sunni Insurgency in Iraq: ‘Political’ and ‘Military’ Strategies from Successful Counterinsurgency in Punjab-India.” Small Wars & Insurgencies 18, no. 4 (2007): 615-637.  Chor, So Wai. “The Making of the Guomindang’s Japan Policy, 1932-1937: The Roles of  Chiang Kai-Shek and Wang Jingwei.” Modern China 29, no. 2 (2002): 213-252.  Cunningham, David, Kristian Gleditsch and Idean Salehyan. “It Takes Two: A Dyadic Analysis of Civil War Duration and Outcome.” The Journal of Conflict Resolution 53, no. 4 (2009): 570-597.  Devotta, Neil. “The Liberation Tigers of Tamil Eelam and the Lost Quest for Separatism  in Sri Lanka.” Asian Survey 49, no. 6 (2009): 1021-1051.  Dinerman, Alice. Revolution, Counter-Revolution and Revisionism in Post-Colonial A frica: The Case of Mozambique, 1975-1994. New York: Routeledge Studies in Modern History, 2006.     48 Dobbs, Charles. “The Xi’an Incident.” In Encyclopedia of Modern Asia, Vol 6, ed. Karen  Christensen and David Levison, 174. New York: Scribner’s, 2002.  Driscoll, Jesse. “Commitment Problems or Bidding Wars? Rebel Fragmentation as  Peace Building.” Journal of Conflict Resolution 56, no. 1 (2012): 118-149.  Dubois, Thomas. “Inauthentic Sovereignty: Law and Legal Institutions in Manchukuo.”  The Journal of Asian Studies 69, no. 3 (2010): 749-770.  Ebrey, Patricia Buckley, James B. Palais & Anne Walthall. Modern East Asia from 1600.  Boston, MA: Houghton Mifflin, 2006.  Economist. “Obituary: Villupillai Prabhakaren,” May 23rd, 2009.  Elleman, Bruce and Stephen Kotkin. Manchurian Railways and the Opening of China:  An International History. Armonk, NY: M.E. Sharpe, 2009.  Fearon, James. “Rationalist Explanations for War.” International Organization 49, no. 3  (1995): 379-414.  ---------. “Why Do Some Civil Wars Last So Much Longer Than Others?” Journal  of Peace Research 41, no. 3 (2004): 275-301.  George, Edward. The Cuban Intervention in Angola, 1965-1991: From Che Guevara to  Cuito Cuanavale. New York: Frank Cass, 2005.  Herbst, Jeffrey. States and Power in Africa: Comparative Lessons in Authority and  Control. Princeton, NJ: Princeton University Press, 2000.  Heywood, Linda. “UNITA and Ethnic Nationalism in Angola.” The Journal of  Modern African Studies 27, no. 1 (1989): 47-66.  Hironaka, Ann. Neverending Wars: The International Community, Weak States, and the  Perpetuation of Civil War. Cambridge, MA: Harvard University Press, 2005.  Hoglund, Kristine. “Violence and the Peace Process in Sri Lanka.” Civil Wars 7, no. 2  (2005): 156-170.  Hoglund, Kristin and Isak Svensson. “Mediating Between Tigers and Lions: Norwegian Peace Diplomacy in Sri Lanka’s Civil War.” Contemporary South Asia 17, no. 2 (2009): 175-191.  Horowitz, Shale and Deepti Sharma.“Democracies Fighting Insurgencies: Evidence from  India.” Studies in Conflict & Terrorism 31, no. 8 (2008): 749-773.     49 Jackson, Robert and Carl Rosberg. “Why Africa’s Weak States Persist: The Empirical  and the Juridical in Statehood.” World Politics 35 (1982): 1-24.  Jackson, Steven F. “China’s Third World Foreign Policy: The Case of Angola and  Mozambique, 1961-93.” The China Quarterly 142 (1995): 388-422.  James, Martin. Political History of the Civil War in Angola, 1974-1990. New Brunswick,  NJ: Transaction Publishers, 1992.  Kalyvas, Stathis. “Ethnic Defection in Civil War.” Comparative Political Studies 41,  no. 8 (2008): 1043-1068.  Kennedy, Andrew Bingham. “Can the Weak Defeat the Strong? Mao’s Evolving  Approach to Asymmetric Warfare in Yan’an.” The China Quarterly 196 (2008):  884-899.  Khadiagala, Gilbert. “Negotiating Angola’s Independence Transition: The Alvor  Accords.” International Negotiation 10, no. 2 (2005): 293-309.  Kirschner, Shanna. “Knowing Your Enemy: Information and Commitment Problems in  Civil Wars.” Journal of Conflict Resolution 54, no. 5 (2010): 745-770.  Korf, Benedikt, “Functions of Violence Revisited: Greed, Pride, and Grievance in Sri  Lanka's Civil War.” Progress in Development Studies 6, no. 2 (2006): 109-122.  LeBillon, Phillippe. “Angola’s Political Economy of War: The Role of Oil and  Diamonds, 1975-2000.” African Affairs 100, no. 398 (2001): 55-80.  Lilja, Jannie. “Trapping Constituents or Winning Hearts and Minds? Rebel Strategies to Attain Constituent Support in Sri Lanka.” Terrorism and Political V iolence 2, no. 1 (2009): 306-326.  -------. “Ripening Within? Strategies Used by Rebel Negotiators to End Ethnic War.”  Negotiation Journal 27, no. 3 (2011): 311-342.  Lockyer, Adam. “Foreign Intervention and Warfare in Civil Wars.” Review of  International Studies 37 (2011): 2337-2364.  MacKinnon, Ian. “The Man Behind the Tamil Tiger’s New Offensive; ‘You Will Have to  Die Someday.”’ Newsweek, May 29th, 2000.  Mayilvaganan, M. “Is it Endgame for the LTTE?” Strategic Analysis 33, no. 1 (2009):  25-39.  McCormick, Gordon and Lindsay Fritz. “The Logic of Warlord Politics.” Third World  Quarterly 30, no. 1 (2009): 81-112.   50  Menard, S. Koc. “Fragmented Sovereignty: Why Sendero Luminoso Consolidated in Some Regions of Peru but Not in Others.” Studies in Conflict and Terrorism 30, (2007): 173-206.  Narangoa, Li. “Japanese Geopolitics and the Mongol Lands, 1915-1945.” European  Journal of Asian Studies 3, no. 1 (2004): 45-67.  New African, “The Legacy of Oil and Geopolitics,” November 1st, 2009.  Olapally, Deepa. The Politics of Extremism in South Asia. New York: Cambridge  University Press, 2008.  Pepper, Suzanne. “The KMT-CCP Conflict, 1945-1949.” In The Cambridge History of China Vol.13, Part 2: Republican China 1912-1949. Eds. John K. Fairbank and Albert Feurerwerker. Cambridge University Press, 1986. Cambridge Histories Online. DOI: 10.1017/CHOL9780521243384.014.  Powell, Robert. “The Inefficient Use of Power: Costly Conflict with Complete  Information.” American Political Science Review 98, no. 2 (2004): 231-241.  ---------. “War as a Commitment Problem.” International Organization 60, no. 1 (2006):  169-203.  Reiter, Dan. How Wars End. Princeton, NJ: Princeton University Press, 2009.  Reno, William. Warlord Politics and African States. Boulder, CO: Lynn-Reinner, 1988.  Sahadevan, P. “On Not Becoming a Democrat: the LTTE’s Commitment to Armed  Struggle.” International Studies 32, no. 3 (1995): 249-281.  Samaranayake, Gamini. “Political Terrorism of the Liberation Tigers of Tamil Eelam (LTTE) in Sri Lanka.” South Asia: Journal of South Asian Studies 30, no. 1 (2007): 171-183.  Sheng, Michael. “Mao, Stalin, and the Formation of the Anti-Japanese United  Front: 1935-37.” The China Quarterly 129, (1992): 149-170.  Sivakumar, Chitra. Transformation in the Sri Lankan Tamil Militant Discourse: Loss of the Tamil Self, V iolence, and the Hermeneutic of Recovery. Madras: University of Madras, 2011.  Spence, Jonathan D. The Search for Modern China, 2nd ed. New York: W.W. Norton,  1999.     51 Stack-O’Connor, Alice. “Lions, Tigers, and Freedom Birds: How and Why the  Liberation Tigers of Tamil Eelam Employs Women.” Terrorism and Political  V iolence 19, no. 1 (2007): 43-63.  Staniland, Paul. “Between a Rock and a Hard Place: Insurgent Fratricide, Ethnic  Defection, and the Rise of Pro-State Paramilitaries.” Journal of Conflict  Resolution 56, no. 1 (2012): 16-40.  Tvedten, Inge. Angola: Struggle for Peace and Reconciliation. Boulder, CO: Westview  Press, 1997.  Wayland, Sarah. “Ethnonationalist Networks and Transnational Opportunities: the Sri  Lankan Diaspora.” Review of International Studies 30, no. 3 (2004): 405-426.  Weigert, Stephen. Angola: A  Modern Military History, 1961-2002. New York: Palgrave  Macmillan, 2011.  Weinstein, Jeremy. 2007. Inside Rebellion: The Politics of Insurgent Violence. New  York: Cambridge University Press.  Weinstein, Jeremy and Laudemiro Francisco. “The Civil War in Mozambique: The  Balance Between Internal and External Influences.” In Understanding Civil War:  Evidence and Analysis, Vol.1: A frica, edited by Paul Collier and Nicholas  Sambanis. Herndon, VA: World Bank Publications, 2004.  Wiessman, Stephen. “CIA Covert Action in Zaire and Angola: Patterns and  Consequences.” Political Science Quarterly 94, no. 2 (1979): 263-286.  Wood, Reed. “Rebel Capability and Strategic Violence Against Civilians.” Journal of  Peace Research 47, no. 5 (2010): 601-614.  Xu, Youwei and Philip Billingsley. “Behind the Scenes of the Xi'an Incident: The  Case of the Lixingshe.” The China Quarterly 154, (1998): 283-307.  Yiqiao, Ji. “The Xi’an Incident: The Reminiscences of Mr. Ji Yiqiao.” Chinese  Studies in History 22, no. 3 (1986): 106-110.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072956/manifest

Comment

Related Items