| The dissertation develops a deductively systematic and inductively successful theory of dispute settlement by encompassing the insight that, uncertain about the future, disputants do not make all the necessary decisions for the initiation, continuation, and termination of their conflict at one moment. For this purpose, I discuss how leaders of states in a dispute choose policies and update beliefs, model disputants' prewar and intrawar interplay into a two-person multi-stage bargaining game with asymmetric information, derive the perfect Bayesian equilibria of that game, discuss the implications of those equilibria at each level of analysis in international conflict, and test those implications on Jones, Bremer, and Singer's militarized interstate dispute data (1996) and Small and Singer's international war data (1992).; I develop the following theses about disputants. First, they have only limited information about each other's bargaining position at both domestic-political and international-structural levels. Second, they update prior beliefs as if they were Bayesian statisticians. Third, they choose current policies as if they were closed-loop decision makers. Fourth, they bargain for terms of settlement that maximize their respective expected payoffs rather than the likelihood of agreement.; I argue that, while uncertainty over relative bargaining power is a necessary condition for bargaining failure, states persist in dispute not only because of the presence of such uncertainty but also because they insist on negotiated terms that involve both some chances of settlement and some risks of disagreement and that any factor---structural, domestic, or individual---that affects such risky behavior at negotiation can be said to cause the variation in dispute outcomes. By applying comparative statics to the equilibria of the endogenous-war game, I assess the impacts of numerous variables at different analytical levels, demonstrate what biases or incorrect assessments inhere in the existing theories of dispute settlement, and fill the gaps between those theories. |
