Diaspora Support in Intrastate Conflict

Transcription

1 Diaspora Support in Intrastate Conflict Lucia Bird April 25, 2016 Abstract Militant groups engaged in civil conflict may be incentivized to seek outside support. While past work has considered the benefits of state sponsorship, I argue that transnational nonstate actors, specifically foreign diasporas, can influence the dynamics of the intrastate conflict and its termination. If diasporas are willing and able to support civil war actors, they can increase the capacity of either the militants or the government in their homeland, as well as contribute to peace building. The government in the diaspora s host state determines whether to allow diasporas to send aid to recipients in the homeland, improving their probability of success, or to block support due to the international costs associated with intervention in another state s domestic affairs. I argue that the delivery of such aid to militant groups, the conflict actors of interest in this paper, increases their capacity to militarily compete with or force the home state into negotiations. I provide a three player sequential game between the militant group, the host state of the diaspora (external government), and the militants home state (internal government). Preliminary conclusions from the model suggest that, when the external government s ability to punish the militants is small to moderate, the probability the militants seek diaspora aid increases as the affinity between the militants and the external government rises. 1 1 Please note that this is a preliminary draft of my prospectus. 1

2 Introduction Most states publicly commit to counter violent nonstate actors domestically and abroad; however, these states inaction regarding potential benefactors, or passive support, may nonetheless yield positive externalities for these militant groups. For example, a major tenet of American foreign policy is counter-terrorism; yet, the United States (U.S.) allowed Irish American diaspora members to support the Provisional Irish Republican Army materially and financially throughout the Troubles of the 1970s. The Irish diaspora in the U.S. sent personal financial contributions through the Irish Northern Aid Committee (NORAID) that were instrumental in purchasing weapons and supporting the dependents of imprisoned IRA members (Byman, 2005b). Beyond financial support, Irish Americans were able to send useful weapons due to the relative availability of arms in the U.S., to provide refuge for IRA operatives trying to escape legal repercussions of violent activity, and to lobby the U.S. government to pressure the United Kingdom into concessions concerning Northern Ireland (Byman, 2005b). While diaspora intervention initially proved helpful to the PIRA, the Irish American diaspora eventually became disillusioned with the violence of the conflict. As a consequence of this growing disillusionment and the condemnation of PIRA activities by Ireland, the Irish American diaspora lobbied the U.S. government to advocate a political resolution to the conflict (Cochrane, 2007). The evolving interactions between the militant group (the PIRA), the Irish American diaspora, and the U.S. government, and its impact on the dynamics of the conflict and its termination, exemplify the political phenomena motivating this paper. While conventional wisdom suggests that seeking outside sponsorship from transnational diasporas benefits needy or ambitious militant groups, this decision requires militants to additionally be somewhat risk-acceptant. As demonstrated by the experience of the PIRA and the Irish American diaspora, these transnational nonstate actors can either augment or undermine the probability that violent nonstate actors achieve military victory in civil conflict in the homeland. Given that the militants may not have perfect knowledge of the 2

3 interests or resources of diasporas prior to soliciting their participation, the militants may unintentionally sabotage their own agenda. Moreover, diasporas do not act in a vacuum; their actions are influenced by the policy preferences of the governments of the host states in which they reside. Should such governments view the militants favorably, they are likely to permit diaspora aid to militants. Conversely, if the diaspora s host state disapproves of the militants, it is likely to block diaspora aid and possibly punish the militants further by enhanced monitoring or intelligence sharing with the government in the homeland. Riskaverse militants may be able to more securely fund themselves autonomously or obtain financing from domestic sources (albeit on a potentially smaller scale). Some groups still seek support from transnational diasporas, even given the uncertainty associated with this choice, and I explore the conditions under which militants seek such aid. In the following paper, I present a model that yields some preliminary hypotheses regarding when militants are incentivized to solicit foreign diasporas for resources. Specifically, my model suggests that the degree of affinity between the government of the state in which the diaspora resides and the militants is directly related to the probability with which the militants are motivated to seek transnational support. Due to the potentially grave consequences related to provoking hostile host governments, this model additionally suggests that militants will take the risk of going abroad for diaspora support only if the external government s punishment capacity is fairly low. Ultimately, I plan to empirically test this argument using a logistic regression model on the universe of cases identified by the Non-State Actor Data (Cunningham, Gleditsch and Salehyan, 2013). Relevant Literature Much recent international relations scholarship has been devoted to studying the causes and consequences of state support of civil war actors (for example, Byman 2013, Salehyan 2007, and Carter 2012). However, the scholarship on nonstate support of militant groups is less 3

4 developed. Nonstate supporters might include diaspora groups, refugees or other displaced persons, nongovernmental organizations, transnational crime networks, and religious leaders (Byman et al., 2001a; Asal, Pate and Wilkenfield, 2008). Since Collier and Hoeffler (2004) identified the positive and statistically significant relationship between the size of diasporas and likelihood of recurring intrastate conflict, scholars have considered the role that such diaspora groups might have in politics in the homeland and host state (for example, Saideman 2012, Smith and Stares 2007, and Sheffer 2014). Since this study will continue in a similar vein by identifying those factors that motivate militants to seek support from external nonstate actors, I will review research relevant to the subject of diaspora involvement in civil conflicts in the homeland in this section. To execute attacks that undermine the state and maintain organizational capacity, militant groups require financial resources. On the operational side, such funds help militants promote the group s ideology via propaganda, pay salaries for militants and their dependents, finance travel and communications costs, train new militants, and provide logistical support (Force, 2008). In terms of sustaining the organization, militant groups may also allocate welfare goods and administer mass media outlets (like television channels and websites) (Force, 2008). In short, contrary to the conventional thoughts of policymakers, carrying out violent attacks and sustaining the organization is not cheap, and so militant groups must secure substantial financing to survive (Prober, 2005). Militants can either fund themselves autonomously, through legitimate or illicit means, or seek outside resources (both domestic and transnational) to meet their financial needs. Militant groups can self-finance when operatives are engaged in legitimate businesses or criminal activities if the funds can be effectively diverted from making profit to war. So, militants can draw on profits from legitimate businesses more easily in certain sectors, such as those lacking formal qualifications or significant startup investment, when authorities ability to ascertain the reported to actual sales ratio is difficult (Force, 2008). Additionally, militants may engage in criminal activities to support their conflict goals, especially as their 4

5 hierarchical structures become flattened (Dishman, 2005). Levitt identifies several illicit fund-raising activities, including prostitution, forgery, drug and arms trafficking, in which militants engage (2002). For example, the southern Lebanese militant group Hezbollah acquires funding to support Israeli Arabs intelligence efforts with opium profits from poppy fields in the Bekaa Valley (Kraft, 2000). Alternative forms of illegitimate funding involve fraud using stolen credit cards and checkbooks (Force, 2008). Finally, some militant groups can leverage primary commodities in civil conflict, which Collier and Hoeffler hypothesize may provide opportunities for extortion by the groups or be correlated with poor governance by the state (2004). The ability of rebels to take advantage of primary commodities, such as opium, diamonds, or coca, may elongate the war because multiple civil war actors can earn outside income from the contraband, regardless of fighting (Fearon, 2004). For example, Levitt identifies militant Islamist groups that profit from exchanges on the gold and diamond black markets (Levitt, 2002; Farah, 2001). Conversely, if militant groups cannot sustain themselves independently, they may turn to external financing from domestic and foreign, as well as state and nonstate, alternatives. Beyond personal profits and savings, militants can obtain money from other domestic sources, such as sympathetic relatives and friends (Force, 2008). Foreign states may be incentivized to aid militants in order to undermine rivals (Saideman, 2002; Maoz and SanAkca, 2012), to pursue geostrategic interests like regime change (Heraclides, 1990), or to support similarly oriented groups (Byman et al., 2001a). Such support may be beneficial militarily, involving weapons provision or intelligence sharing (Jenkins, 1986), or non-militarily, like providing ideological direction (Byman, 2005a). Furthermore, states may inadvertently (or neglectfully) aid militant groups by allowing them safe havens in their territory. Even though states may face some costs for hosting violent substate actors, this also enables militants sponsors to obtain some bargaining power relative to the host state that is valuable to the state providing the safe haven (Bapat, 2007). Militant groups frequently take advantage of such external sanctuaries to avoid repression by the opposing state and to capitalize on 5

6 the ability to organize for military action (Salehyan, 2007). Similarly, even states that do not deliberately support or allow sanctuary for militants may permit outflows of aid from domestic actors to militants abroad in a form of passive sponsorship (Byman, 2005a). Militants may also obtain financing from outside nonstate sources, such as charities, humanitarian agencies, and other front companies (Levitt, 2002). This funding may be voluntary, such as donations collected for widows and orphans from diasporas that intentionally or fraudulently are used in militant activities, or forced by threatening retribution against the diaspora members abroad or their family and friends at home if funding is refused (Force, 2008). The relationship between external actors and militants groups may be more nuanced than previous authors have posited. While some scholarship has worked to theorize the triangular relationship present between diasporas incentivized to become involved in homeland politics, the host state in which they reside, and the civil conflict actors (Sheffer, 2003), this paper will more comprehensively assess and offer novel predictions related to this relationship. Employing a domestic politics approach, Saideman argues that policymakers foreign policy decisions are determined by their constituents interests abroad (2012). Building on selectorate theory (Bueno De Mesquita et al., 2005), Saideman asserts that politically vulnerable leaders endeavor to keep constituents in their winning coalition, and therefore are incentivized to send aid to civil actors with whom their constituents share an ethnic identity as a demonstration of credible commitment to their constituents (2012). Similarly, Adamson asserts that diasporas with kin groups involved in civil wars in the homeland can lobby for the militants in order to accrue sympathy and political support from the host state (2013). While these scholars provide useful initial frameworks through which to consider this question, some space for improvement exists. For example, current scholarships fails to adequately address the characteristics of the relationship between the militants and the host state, which is mediated by the diaspora, that might influence militant groups choices regarding support acquisition. Additionally, further theorizing on the foreign policy impact 6

7 of multiple salient ethnic groups with distinct demands living in one state is needed. Diaspora groups should not be conceptualized as a uniform bloc due to intra-group variation, but they share characteristics that make them unique actors in international politics. While scholars debate the definition most appropriate for these increasingly relevant transnational actors, I will briefly outline some broad areas of agreement. Smith defines diasporas as groups that are established in foreign countries (aside from the recognized homelands from which they or their ancestors departed) (2007). Cohen identifies the collective memory or mythology of the idealized homeland and feelings of distinction from the mainstream society as fundamental traits that diasporas share (1997). Also, diasporas simultaneously endeavor to integrate into their host states while still maintaining strong connections with the homeland and other diasporas abroad (Sheffer, 2003; Keles, 2015). Such linkages foster long-distance nationalism, which unites communities abroad around common identities, traditions, and practices in an effort to inspire participation in homeland politics (Schiller, 2009). Finally, diasporas are unique actors in civil wars because they can support armed conflict in their homelands while avoiding personal experience of the violence (Adamson, 2013). This may change diaspora groups preferences regarding the outcome of the civil conflict. For example, diasporas may prefer continued fighting if they believe conflict will ultimately result in favorable concessions. Of course, the possibility of retributive violence against relatives and friends of the diaspora members casts doubt on this claim. While the conditions under which diasporas identify with their homelands occupy much of the literature, questions remain regarding what characteristics of diasporas and their host states incentivize some militant groups (but not others) to seek aid from abroad, as well as why these diasporas are not as interested in complete integration into their host state. Diasporas can affect the civil war in the homeland by providing direct aid, such as weapons, to either conflict actor. Moreover, diasporas may more subtly influence conflict dynamics by lobbying the host state or other international actors. Diasporas, typically wealthy compared to coethnics in the homeland, may support militants financially via donations; membership 7

8 dues for social, religious, or political associations; or contributions to legitimate or fraudulent foundations that ultimately enable violence (Byman et al., 2001b; Hess, 2007). For example, Elu and Price find that remittances contribute to the incidence of terrorism in Sub-Saharan Africa (2012). Additionally, diasporas may provide other types of direct support, such as fighters, intelligence reports, and arms (Sheffer, 2003). Hegghammer finds that recruits from diaspora communities more often fight abroad because the foreign conflict is perceived as more legitimate than attacks executed in the host state (2013). While this is an interesting normative argument, the analysis applies only to recent Islamic militants and may not generalize across time and space. Some scholars argue that diasporas influence on the conflict is insignificant relative to state sponsors impact (Byman, 2013), but I argue that militants relationship to potential state supporters is mediated by diasporas abroad. Diasporas may influence the prospects of militants in their homeland by lobbying policymakers in their host state, much like a transnational advocacy network makes efforts to pressure states into taking political, and sometimes military, actions (Keck and Sikkink, 1998). These mobilized diaspora groups can act as ethnic lobbies, especially in advanced, industrialized democracies, to raise international sympathy for the plight of militants in the homeland using mass and social media (Lia, 2007). For example, Van Bruinessen hypothesizes that exile...transformed Kurdistan from a vaguely defined geographic entity into a political ideal by engaging educated Kurdish diaspora members abroad in homeland politics and providing political opportunities to influence governments in host states, like Belgium, Germany, France, and the U.S. (2007). The diaspora s capacity to influence foreign policymaking by the host government is directly related to the degree of congruence between the diaspora s and the host state s objectives, in addition to the prevalence of the diaspora abroad (Shain, 1995, 1999). While some scholars have theorized the relationship between diaspora support and its benefits for militant groups, work remains to be done to determine under what conditions militants make the risky choice to seek aid from diaspora actors abroad. The militants are taking a risk by 8

9 asking for diaspora aid since they could be ignored by diasporas abroad or even undermined by diasporas interested in supporting the homeland government or immediately resolving the civil conflict (Bercovitch, 2007). External state and nonstate actors can be beneficial or detrimental to the militants. Scholars debate the impact of outside support, even that which exclusively increases militant capacity, on overall violence levels. While Greene (1990) suggests outsider intervention on behalf of the rebels increases the incidence of violence, Kalyvas (1999) argues that the probability of rebel violence decreases with foreign support because civilian defection to the state is less costly. Also, scholars have neglected to theorize the role of diasporas in conflict resolution. Smith asserts that almost all diaspora groups want peace in their homelands on their terms, which may involve arming militants or, instead, building state capacity to oppose dissident militants in the homeland (2007). Diasporas can be instrumental in the peace process since they can promote dialogue, mediation, and problem solving between civil war actors (Bercovitch, 2007). For example, Tamil diaspora groups sponsored the Liberation Tigers of Tamil Eelam (LTTE) but disliked the group s post-9/11 association with terrorism; consequently, diaspora members urged local LTTE branches to eschew violence and seek a diplomatic resolution (Fair, 2007). Additionally, some scholars argue that diaspora aid is costless since diasporas are generally reliable and refrain from trying to control militants agenda (Byman et al., 2001a), but there are some consequences to the militants that are associated with accepting such support. For example, militants may still face a loss of agenda control if diasporas or host states have distinct objectives in the homeland (Horowitz, 1985). Weinstein suggests another potential pitfall of external support: when foreign aid is available, militants may depend less on the population and eventually alienate and possibly abuse the civilians supposedly under their jurisdiction (2006). While scholars originally posited many of these consequences for militants with states (not diasporas) as the foreign sponsors, I argue that the same mechanisms are at work regardless of the state or nonstate status of the aid provider. 9

10 A Theory of Diaspora Support In this section, I present a model of a militant group s request for diaspora aid, which is mediated by the government in the diaspora s host state, in response to an ongoing conflict between the militants and the government of the diaspora s homeland. The players in this game consist of the militants in the homeland, the external government of the diaspora s host state, and the internal government in the homeland. Before the game starts, the militants and internal government are engaged in a civil conflict, the outcome of which interests the diaspora and its external government. Militants Transnational Domestic Internal Government Resist Concede External Government Deny Allow p-c M 1-p-C IG Internal Government θp-c M 1-θp-C IG -βc EG Resist Concede βp-c M 1-βp-C IG βpλ-k EG 1 -K IG βλ-k EG Figure 1: Diaspora Support of Militants Game Figure 1 describes the extensive form representation of the game. The game begins with the militants choice to seek aid from transnational diasporas or keep the conflict, and its financing, internal. If the conflict remains domestic, the militants opt to self-fund or only seek support from actors within the homeland, such as friends and family members. 10

11 If the militants decide to keep the civil war within the boundaries of the homeland by autonomously financing the conflict or requesting support from the domestic population, the game ends with the homeland government (henceforth, referred to as the internal government) electing to resist the militants demands by continuing the military conflict or conceding to negotiations. If the internal government decides to fight, the payoffs to the militants, internal government, and host state government (henceforth, the external government) consist of (p-c M ; 1-p-C IG ; 0). The militants payoff is a function of their probability p, with p [0,1], of winning less the cost C M, with C M [0,1], of fighting the internal government. The internal government s payoff, 1-p-C IG, is given by the utility of winning less the probability the militants win and the cost of fighting the civil war. Because the militant group stays domestic, the external government does not become involved and thus receives nothing from the interactions between civil war actors. If the internal government concedes, the payoffs to the militants, external government, and internal government are (1; 0; 0). I assume that negotiations will lead to a change to the status quo that will benefit the militants, so I normalize that outcome to 1. The internal and external governments both receive 0 since the internal government is forced to negotiate with the opposition and the external government does not play on this branch of the game. If the militants seek aid from actors abroad, the external government makes the next move. The external government may prohibit or allow resources to flow from the diaspora to the militants. By denying diaspora aid, the external government suffers some costs in the form of technical, financial, or human resources that must be diverted from normal public spending. This prohibition of diaspora aid is painful to the militants because of the constraints their transnational operatives may face due to increased scrutiny by the external government and dangers associated with increased exposure of the militant group to the international community, which may make repression of the militants by the internal government easier. If the external government denies aid from the diaspora group in the host state, the 11

12 payoffs to the militants, internal government, and external government are (θp-c M ; 1-θp- C IG ; -βc EG ). If the militants were to exclusively seek aid when diasporas favor the militants agenda, the outcome associated with external government denial (θp) would yield the worst payoff for the militants. This is because the probability of militant victory p falls to θp. The parameter θ, with θ [0,1], indicates the external government s ability to constrain the militants by denying diaspora aid, more stringently monitoring the diaspora s transnational activities, or sharing intelligence on the militants with the internal government. However, I assume that militants make the request for diaspora aid under risk because the level and nature of diaspora aid β, with β [0,2], is unknown. When β is less than 1, receiving aid from the diaspora decreases the militants chance of success, possibly more than when aid is denied by the external government. The internal government receives 1-θp-C IG, which represents the payoff of winning (which I normalize to 1) less the reduced probability that militants win and the cost of fighting the civil war C IG, with C IG [0,1]. The external government s payoff reflects the costs required to halt diaspora support, C EG, with C EG [0,1], weighted by β. In this case, β represents the level of diaspora support or political influence in the host state. More populous, wealthy, or politically active diaspora groups will be more beneficial to the civil war actor they choose to back, as well as more costly to external governments that deny diaspora aid. Depending on the regime type of the external government and the nature of the coalition necessary for the leader to stay in power, the prohibition of diaspora support can be both politically and logistically expensive (Saideman, 2012). Conversely, the external government can permit, and perhaps even supplement, the aid package being sent to the militants in the homeland at this decision node. Such aid will benefit the militants capacity in conflict or bargaining leverage if negotiations ensue. If the external government does allow support to flow from the host state to the militants, the internal government makes the final decision of whether to keep fighting the militants or acquiesce to negotiations. The internal government may be willing to face an externally 12

13 supported militant group, though the war is likely to be costlier and longer in duration due to the increased capacity of the militants. Alternatively, the internal government may agree to negotiations if externally supported militants pose a sufficient threat to the sovereignty of the state. With outside aid, the militants become more legitimate rivals in conflict, which may force the internal government to bargain. If the external government permits diaspora aid to flow to militants, the internal government may choose to negotiate or to continue resisting the now externally supported militants. If the internal government agrees to any negotiated settlement, the payoffs to the militants, internal government, and external government are (1; -K IG ; βλ-k EG ). I again assume that the militants receive a payoff of 1 because they accrue benefits from the change to the status quo. 2 The internal government, at this outcome, incurs a cost -K IG, with K IG [0,1]. The parameter K IG substantively refers to the increased cost to the internal government of dealing with an externally supported group due to the internationalization of the civil conflict, the likely improved bargaining position militants have due to the incoming flow of resources from the foreign diaspora, and the perceived legitimation of militants cause due to the tacit support from the external government. The external government s payoff refers to the benefit it receives from allowing diaspora aid to the militants less the cost of permitting diaspora aid. The benefits to the external government are composed of β and λ, with λ [0,1]. The parameter λ describes the degree of affinity between the militant group and the external government. The parameter λ substantively refers to how much affinity, in the form of ideological similarity, strategic objectives, or historical interactions, exists between these two actors. I assume that increasing affinity means that the external government is more likely to be sympathetic to the militants plight and the diaspora s lobbying efforts. The cost of permitting diaspora aid K EG, with K EG [0,1], refers to the external government s 2 Some scholars, such as Bapat 2007, note that militants receiving state support may be forced to pay a tax of sorts to their state sponsors in exchange for help in the civil war. While this assumption is valid for state sponsors, the primary outside supporters in this model are diaspora groups. The diaspora, though they can influence civil conflicts, are not likely to provide as much or as varied support, nor risk as much as state sponsors. Therefore, I assume the diaspora do not require a tax from the militants in this game. 13

14 potential to receive sanctions or other international reputation costs due to at least implicitly helping the militants (Byman et al., 2001a). However, the internal government may decide to continue the fight against the externally supported militants. If the game concludes with ongoing conflict, the payoffs to the militants, internal government, and external government are (βp-c M ; 1-βp-C IG ; βpλ-k EG ). The militants probability of winning is weighted by β less the cost of fighting the civil war. If I assume that the militants seek transnational support only when the diaspora seem sympathetic to the cause, the parameter β should increase the militants chance of winning. If, however, the militants gamble for financial resources does not pay off because the diaspora are actually in favor of peace building or victory by the internal government, the β level of support actually undermines the militants ability to overcome the internal government. 3 The internal government receives the payoff of winning less the probability that militants win (weighted by β) and the cost of resistance. The external government s payoff here is the probability of the militants win, influenced by the level of diaspora support β and affinity between the militants the external government λ, less the cost of permitting the aid to the militants. In this case, I assume that the external government s payoff is influenced by the outcome of the conflict, which is why the probability of militant victory p is included. Equilibrium Solutions I solve this sequential game using the subgame perfect Nash Equilibrium solution concept. The formal solution in its entirety is presented in the Appendix. I describe the equilibria as a function of β and discuss them based on actions taken by the internal and external governments at low, median, and high levels of β. I first find the equilibria under complete information, in which I assume that all the players are informed on the level of β, or support provided by diaspora. Substantively, this means that each player knows both the amount of resources the diaspora is willing to send and the civil war actor to whom the aid will be 3 In this scenario, β<1 14

15 delivered. I then find the equilibrium under incomplete information, in which the militants do not know the level of β. I derive a testable hypothesis from this model concerning the explanatory variables relevant to determining when a militant group is more likely to seek diaspora support. Complete Information Solution Case 1 I first consider the condition under which the internal government, at the final node, decides to resist. This occurs when β < β, with β = 1 C IG+K IG. I then consider the choices available p to the external government: to permit or to prohibit diaspora aid. The external government anticipates the internal government s decision to resist, and so denies when β < β, with β = K EG pλ+c EG. Under this set of conditions, the militants have a dominant strategy to keep the conflict and financing efforts domestic. Militants should remain domestic because, if they go external, their payoff θp-c M is always lower than the utility they would get by staying domestic, regardless of whether the internal government resists or concedes. Lemma 1. Militants choice to keep the conflict domestic is strictly dominant when β < β < β, 4 meaning that the external government denies and the internal government resists. 5 Case 2 I next consider the conditions under which the internal government concedes to the militants demands at the final node. The internal government makes this decision when β > β (in fact, β takes on the highest level possible under this case). The external government has a dominant strategy to allow any level of diaspora support to flow to the homeland if the internal government concedes to the militants. This is because the utility of allowing support when the internal government concedes, βλ-k EG, is positive and always greater 4 β < β based on the assumption that the internal government faces higher restrictions than the external government on actions related to diaspora support of militants. 5 See proof in Appendix. 15

16 than -βc EG, which is the payoff the external government receives from denying support. If the militants seek domestic support exclusively, the internal government has a dominant strategy to resist because 1-p-C IG is greater than 0, which is the utility for the internal government of conceding. Therefore, if the external government allows the diaspora to send aid, the militants have an optimal strategy to seek transnational support. The militants should request diaspora aid because their payoff of going transnational, given that the internal government concedes, is 1. This is greater than the militants payoff of remaining domestic p-c M, given the internal government s dominant strategy to resist if the militants remain domestic, and is the highest payoff available to the militants in the game. Lemma 2. Militants choice to request transnational aid from diasporas is optimal when β > β > β, that is, when the external government allows and the internal government concedes. 6 Case 3 The third case represents the most substantively interesting set of conditions. Under these conditions, the internal government decides to fight the militants (β < β), yet the external government chooses to allow diaspora aid (β > β ). Because the internal government faces higher constraints than the external government regarding decisions on the militants, I assume that β < β. I make this assumption because the militants challenge the sovereignty of the internal government, meaning the internal government is more constrained in its policy choices than the external government, which does not face such a threat. Since the external government s threshold for choices related to diaspora support is lower than the internal government s, the external government has more flexibility in dealing with diaspora support of militants. The militant group anticipates the decisions of the internal and external governments. 6 See proof in Appendix. 16

17 The militants then decide to go transnational if the utility associated with seeking diaspora support, βp-c M, is greater than the payoff received by keeping the conflict domestic, p-c M. For the militants to receive p-c M by keeping the conflict and resource acquisition efforts domestic, the internal government must have an optimal strategy to resist. This occurs under certain conditions. The internal government must optimally prefer to resist rather than concede if the militants remain domestic, and this occurs when the payoff of continuing to fight, 1-p-C IG, is greater than the payoff of conceding, 0. This implies that 1-p is greater than C IG. Given that β = 1 C IG+K IG p and 1-p is greater than C IG, β must exceed 1. I assume that β is less than β and β is greater than 1, so the relationship between β and 1 influences the conditions under which the militants decide to seek external support from diasporas or remain domestic. Lemma 3. Militants choice to request transnational aid from diasporas is optimal when max{β, 1} < β < β. Lemma 4. Militants decide to keep the conflict and the associated pursuit of resources domestic when β < β < 1. Incomplete Information Solution In the incomplete information equilibrium, I again assume that the internal government is more constrained in its dealing with the militants than the external government, so β < β. The militants do not know the level of β, meaning they lack information on both the degree of support provided by the diaspora and whether aid will be sent to the internal government or the militants (though the external and internal governments are aware of the level of β). Under incomplete information, my prior belief is that β is uniformly distributed between 0 and 2. To find the total expected utility of seeking transnational aid, I associate the probability of β falling into the intervals defined by my cutoff points of β (β and β) with the payoffs militants receive (as illustrated by Figure 2). Because the militants are unsure of the level of β, I cannot use the β that is included in the payoffs. Instead I integrate to find 17

18 the expected utility of β for militants when the internal government resists and the external government allows (β < β < β). I use this expected value to construct the total utility for militants of going abroad for aid. Nature β < β β β, β) β > β Militants Militants Militants transnational transnational transnational domestic domestic domestic θp-c M p-c M βp-c M p-c M 1 p-c M Figure 2: Payoffs for Militants on Distribution of β This incomplete information situation is interesting because the militants are uncertain whether it is better to fund themselves with domestic sources exclusively or seek aid from foreign diasporas. Seeking transnational aid is optimal when their expected utility from going abroad for diaspora support, β 2 (θp - C M) + ( β 2 β 2 )p 4 - β β 2 C M + 2 β, exceeds the 2 utility of remaining domestic, p-c M. This occurs when β p(2θ - β ) - β(2- βp+2c M ) + 4(1-p + C M ) > 0. Using comparative statics, I find some relationships between parameters from the model and my phenomenon of interest, the probability with which militants seek transnational support from diasporas. Specifically, I identify a direct relationship between λ, the parameter representing the affinity between the militants and the external government, and the militants expected utility of seeking support from transnational diasporas. The substantive implication of this is that militants are more likely to solicit diaspora support as the affinity between militants and the external government increases. This relationship holds under the condition that θ, the ability of the external government to punish the militants, is sufficiently small (θ < K EG (pλ+c EG ) ). 18

19 Theoretical Implications and Discussion This model provides insights into the conditions under which militants involved in intrastate conflict are willing to take the risk of seeking aid from transnational diaspora actors. By provoking diaspora s participation in the conflict in the homeland, the militant group may gain significant advantages in capacity to fight against or leverage in bargaining with the internal government. However, there is a level of risk associated with this decision in that the diaspora may, in reality, not align with the militants in terms of objective or strategies and instead prefer immediate conflict resolution or even support the internal government. This risk is captured in the model with the parameter β, which varies between 0 and 2, meaning that its effect on the probability p of militant victory may be helpful or harmful. For example, the Colombian diaspora generally lacks any unilateral preference on a civil war actor in the homeland and is reluctant to engage in the conflict in the homeland; however, diaspora members in the U.S. overwhelmingly supported a staunch anti-guerrilla presidential candidate in the 2002 Colombian elections (Bouvier, 2007). So, while Colombian diaspora members do not participate in the substate conflict beyond voting for candidates that strongly oppose guerrilla activity, the diaspora s preferences for stability and against guerrilla violence suggest that more direct diaspora participation would undermine Colombian militant groups. Even given the risks associated with seeking transnational aid, militant groups do make the decision to involve foreign diasporas, and, by association, their external governments. The choice to turn to transnational options becomes more likely under certain conditions, derived (as previously mentioned) from the comparative statics of the model. As the affinity (λ) between the militants in the homeland and the external government increases, the militants payoff for seeking support abroad increases. Substantively, this implies that militants are more inclined to involve diaspora groups located in host states with (external, in the terms of the model) governments that are sympathetic to the militants agenda. This is intuitive because the militants are only willing to accept the risk of seeking transnational 19

20 supporters, in the form of foreign diasporas, when the external government is sufficiently similar to the militants such that it is likely to allow the diaspora aid to flow to the homeland. For instance, the American Cuban diaspora, prominent due to wealth and electoral relevance, successfully aligned its own anti-castro interests with the anti-communism agenda of the United States after the fall of the U.S.-supported Batista regime (Grugel and Kippin, 2007). Opposition groups to Castro in Cuba benefited from the diaspora s intervention on its behalf as the U.S. allowed the diaspora to direct funding allocations, television and radio broadcasts, as well as democratization efforts (Vanderbush, 2009). Additionally, the probability that militant groups would attempt to acquire aid from transnational diasporas is affected by the degree to which the external government can inflict cost on the militants or their associated diasporas that otherwise might have been supportive (represented in the game as θ). From the comparative statics of the game, the degree of punishment that militants expect from the external government must be low to moderate if they are still incentivized to risk requesting aid from foreign sources. For the militants, this penalty represents a cost, such as increased legal prosecution of transnational operatives or disrupting flows of financial or material resources, that will ultimately weaken military capacity or bargaining power. This is costly because, had the militants instead chosen to keep their civil conflict and extraction of resources to support violent activity domestic, their probability of victory would have been higher. For example, several former leaders of an American Muslim charity, the Holy Land Foundation, were convicted of sending more than $12 million to Hamas, the Palestinian terrorist organization and political leadership in the Gaza Strip (FBI, N.d.). Under the pretense of collecting charitable donations for humanitarian relief in the Levant, the Holy Land Foundation leaders knowingly sent financial contributions that were instrumental to Hamas violent activities, such as suicide bombing (Levitt, 2004). While the loss of funds from the Holy Land Foundation likely somewhat depleted Hamas funds, the militant group still operates and launches violent attacks from Gaza. So, the external government s ability to punish Hamas, in this case, is unlikely to 20

21 prevent the group from future requests for aid from transnational diasporas. Thus, I am able to develop the following hypothesis based on my model: H1 : As the affinity between the militants and the external government rises, militants are increasingly likely to seek aid from transnational diaspora groups when the ability of the external government to punish the militants is small to moderate. Research Design In this section, I will discuss future steps for this project in terms of empirically testing the hypothesis derived from the formal model. To appropriately test the proposed hypothesis, I will gather data from several sources to operationalize the dependent and key independent variables. I will use the Non-State Actor (NSA) Data, which augments the Uppsala Armed Conflict Data, to identify my universe of cases (Cunningham, Gleditsch and Salehyan, 2009). This dataset includes information regarding military capacity and political characteristics of over 500 violent nonstate actors, as well as details on the nature of the termination of the conflict. I will employ the NSA dataset primarily to acquire information on the dependent variable of interest in this analysis, which is whether militants seek aid from diasporas abroad. To gather data on the key independent variables, I will use a number of datasets, such as the UCDP External Support Data (Högbladh, Pettersson and Themnér, 2011) and the Klein et al. rivalry dataset (Klein, Goertz and Diehl, 2006). The unit of analysis in my dataset will be conflict dyad-period, where the dyad consists of the militant group and the internal government (to use the language of my game). To identify the external government, I associate the diasporas, which are likely solicited for aid by the militants, with their host state. The militants in the homeland may potentially be able to draw upon the support of more than one diaspora. This complicates the link between the diaspora and the external government. If multiple diasporas abroad might be sympathetic to and willing to help the militants, I will code the country with the largest share 21

22 of diaspora members as the external government. I will use the World Bank s Global Bilateral Migration Database, which includes estimates of bilateral immigration from by decade, to approximate diaspora sizes (Ozden et al., 2011). Because of the limitations on diaspora data, I drop cases of militant groups whose conflicts with the internal government fall outside of the specified range of the Global Bilateral Migration Database. The dependent variable of interest for this analysis is the choice by militant groups in the homeland to seek aid from foreign diasporas. Though likely unobservable, the decision by militants to go abroad for support can be identified by using several relevant variables from the NSA Data. First, NSA includes two variables that demonstrate whether transnational nonstate actors provide aid to states in a combative or non-combative manner. A third useful variable to approximate the decision to seek transnational aid is militants presence in foreign states. At these locations abroad, groups operatives might be able to request financial, material, or human resources from targeted diasporas. In the original dataset, these variables are ordinal (with three levels of support possible). I will recode these measures as binary, with a 1 if the militants receive any degree of support in either category. Furthermore, I will create an additive index of transnational nonstate support with seven potential levels by aggregating these three variables. Statistical analysis with an ordinal variable will be helpful in specifying the conditions under which militants receive increasing levels of transnational nonstate support instead of just existence or lack of support. According to my hypothesis, the key independent variable is the affinity between the external government and the militants in the homeland (λ in the model). My theoretical framework suggests that increasing levels of affinity imply that the external government will be more favorable to allowing the diaspora to send aid to the militants. I will measure this affinity as a continuous variable that takes into account multiple facets of the relationship between the militants and the diaspora s host government. 7 I will consider the ideological similarity between the militants and the external government by reviewing the group s polit- 7 The details of the values that these indicators will take on remain to be specified. 22

23 ical objectives (included in the NSA case descriptions) and the degree to which they parallel national policy goals of the state. This affinity variable will also indicate whether the militants have benefited from interactions with the external government during their campaign, as demonstrated by evidence in the UCDP External Support (Disaggregated/Supporter) Dataset (Högbladh, Pettersson and Themnér, 2011). Additionally, I will consider the relationship that exists between the internal and external governments, which is likely to affect how much affinity the external government and militants share. The affinity between the external government and the militants should be higher when a rivalry exists between the external and internal governments, which I will identify using the Klein et al. data on rivalry (Klein, Goertz and Diehl, 2006). Furthermore, the degree of trade interdependence between the host state and homeland should demonstrate the geostrategic importance of this relationship to the external government. Low volumes of trade between the two states should be correlated with higher affinity between the external government and militant group. Therefore, I will aggregate these various indicators to operationalize the affinity parameter λ. Based on my theory, the effect of the affinity between the militant group and the external government on the probability that militants seek transnational aid is conditional on the capacity of the external government to punish the militants (the parameter θ from my game). I will operationalize this ability to punish the militants as the capacity of the diaspora s host state. To measure state capacity, I will employ the the relative political extraction variable from the Relative Political Capacity Dataset. This is more useful than national GDP, which is commonly used to measure state capacity, because it more specifically shows the degree to which governments can apportion national output to achieve policy goals. Moreover, two versions of this variable, one for developing and the other for developed countries, exist. In this way, I can more appropriately identify capacity potential depending on development level of the host state. Finally, I will include some control variables on which to condition my explanatory variables of interest. First, I will include the population of the host state and the size of the 23