Extended deterrence and alliance formation

Inductive approaches have little conclusive to say about whether alliances make war more or less likely. Moreover, the empirical associations are ambiguous. Modeling alliance formation in an incomplete information setting reveals that alliances have predictable effects on the behavior of nations. The primary effect of alliances is to increase the probability that a nation will intervene in an ongoing war. Since an alliance provides a signal that a nation is likely to intervene, alliances affect the behavior of other states. Specifically, the formation of a defensive alliance deters potential aggressors from attacking, and encourages those nations that are attacked to resist. Although an alliance helps deter enemies and helps reassure friends, success is not guaranteed; deterrence failures occur. Although alliances have a predictable effect on the behavior of nations, the aggregate relationship between alliance formation and the occurrence of war is ambiguous. As well as providing an explanation of much of the empirical evidence, this model generates new testable hypotheses.