Explaining Variation in the Type of Support Foreign Sponsors Provide to Warring Factions in Civil War

这项研究旨在分析防务条约对国家间危机的开始和升级的影响。对防御性联盟作用的主流解释是由扩展威慑理论提供的：第三方捍卫者威胁使用武力来保护其他国家、他们的领土和他们的价值观免受攻击(Huth 1988)。超级大国在冷战期间对其在欧洲和世界各地的盟友的安全保障就是一个很好的例子。然而，扩展威慑范式显示出理论和经验上的局限性。防务条约在预防争端方面的有效性可能会因安全困境和道德风险考虑而受到阻碍。此外，对联盟威慑作用的研究一般对国家间危机、升级和威慑的复杂性知之甚少(Johnson and Leeds 2011；Benson 2012)。考虑到国家间危机的复杂性，本文试图系统化我们对防御条约在危机中的作用的理解。虽然威慑的目的可以是防止争端，但也可以为危机管理部署成功的威慑战略，防止其达到战争的水平。当危机开始时，发起者仍然可以在没有达到武力水平的情况下进行危机谈判；在这种情况下发出的威慑威胁明确规定了不应越过的红线，以维持和平。文献将上文所述的前一种威慑称为“一般威慑”，将后者称为“立即威慑”(Fearon 2002)。历史上记录的最具标志性的威慑案例是两个潜在的交战国在一场危机中解决争端而不诉诸武力的例子，比如古巴导弹危机。我研究的初步假设是，在两个对立的国家中，鉴于防御性承诺的声誉成本与危机爆发之间缺乏明确的联系，无论是门徒还是反对者，都可能被防御性联盟的存在所感动，从而引发争端。然而，一旦危机开始，针对联盟的战争预期成本就会增加，从而降低对手发起暴力挑战的可能性。即使盟友的干预不确定，对手发起暴力争端的理性动机也比没有联盟的情况下要少。与进攻性条约或中立性条约相反，防御性联盟不仅不是为了鼓励大胆的行为。此外，值得期待的是，后卫会阻止门将S的放肆。捍卫者不太可能愿意支持门徒强行改变现状，因为随之而来的冲突可能会退化为成本更高的更广泛的冲突。例如，在冷战期间，这两个超级大国担心危机可能演变为核对抗，并试图约束他们的伙伴。因此，在这一新的理论范式中，防御条约应该有助于危机的爆发(低总体威慑力)，但对升级(高即时威慑力)具有抑制作用。我的目标是用混合方法检验这些假设。首先，我将对潜在挑战者和潜在目标之间的争端开始和升级进行量化分析。对记录的危机进行大N分析，似乎是检验有关联盟总体威慑效果的一般说法的最合适方法。为了证实这一分析，我打算分析几个案例。很容易强调统计方法的不可避免的局限性：在这一点上，就像在其他将现实简化为本质的模型中一样，“演绎能力通常是以牺牲历史准确性为代价的”(Achen和Sndal，1989)。虽然历史案例的现实并不一定反驳整个逻辑。

The research aims to analyse the impact of defence pacts on the onset and escalation of inter-state crises. The mainstream interpretation of the role of defensive alliance is provided by extended deterrence theory: a third-party defender threatens the use of force to protect other countries, their territories, and their values from attack (Huth 1988). The superpowers' security guarantees to their allies in Europe and around the world during the Cold War are a case in point. However, the extended deterrence paradigm displays theoretical and empirical limitations. The effectiveness of defence pacts in preventing disputes can be hindered by both security dilemma and moral hazard considerations. Additionally, studies of the deterrent effect of alliances generally share little awareness of the complexities of inter-state crises, escalation, and deterrence (Johnson and Leeds 2011; Benson 2012).This thesis attempts to systematize our understanding of the role of defence pacts in crisis taking into account the complexities of inter-state crisis. While deterrence could be aimed at the prevention of disputes, a successful deterrence strategy may also be deployed for crisis management, preventing it from reaching the level of war. When a crisis begins, the initiator can still engage in crisis bargaining without reaching the level of force; deterrent threats issued in that context clearly state the red lines that should not be passed for peace to be maintained. The literature calls the former type of deterrence described above "general deterrence", and the latter, "immediate deterrence" (Fearon 2002). The most iconic cases of deterrence recorded in history are examples in which two potential belligerents resolved a dispute within a crisis without resorting to force, like the Cuban Missile Crisis. The preliminary hypotheses of my research are that, in a dyad of rival states, given the absence of a clear link between reputation costs of a defensive commitment and crisis onset, either the protégé or the opponent could be moved by the presence of a defensive alliance to initiate a dispute. Once a crisis has started, however, the expected costs of war against the alliance increase, and therefore lower the likelihood of an opponent issuing a violent challenge. Even if the ally's intervention is not certain, an opponent has fewer rational incentives to start a violent dispute than in the case of no alliance. Not only is the defensive alliance not meant to encourage emboldening behaviour, as opposed to an offensive treaty or neutrality pact. Additionally, it is worth expecting that the defender will prevent the protégé's emboldenment. The defender is unlikely to be willing to support a forceful change to the status quo by the protégé, as the ensuing conflict could degenerate into a broader conflict with higher costs. During the Cold War, for example, the two superpowers feared that crises could turn into nuclear confrontations and tried to restrain their partners. As a result, in this new theoretical paradigm, defence pacts should be conducive to crisis onset (low general deterrent power) but have a restraining effect on escalation (high immediate deterrent power).I aim to test these hypotheses with a mixed method approach. First, I will carry out a quantitative analysis of dispute onset and escalation between a potential challenger and a potential target. A large-N analysis of recorded crises appears to be the most appropriate method to test general claims about the overall deterrent effect of alliances. To corroborate this analysis, I aim to analyse several case studies. It is easy to highlight the inevitable limitations of the statistical methodology: in this, as in other models that reduce reality to the essentials, "deductive power is usually purchased at the cost of historical accuracy" (Achen and Snidal 1989). While the reality of historical cases does not necessarily disprove the entire logic...