Threats to use force: Costly signals and bargaining in international crises

The dissertation examines the use, interpretation, and effects of military threats employed in international disputes. Drawing on recent developments in game theory, I argue that crisis bargaining occurs because state leaders typically have private information about their willingness to use force rather than accept possible negotiated settlements, and because they can have incentives to misrepresent this private information. These incentives imply that quiet diplomatic conversations and other costless signals often will not allow states to learn about an opponent's resolve on the issues or to credibly reveal one's own.;Instead, to enable learning about the true "balance of resolve" crisis signals must be costly in a way that depends on a state's private information: Actions such as troop mobilizations and public statements by leaders reveal information by being more costly for a state with low resolve than for one with high resolve. The costs that render crisis threats informative come in two major forms: audience costs, which arise when a domestic political audience may punish a leader for backing down after having made a "show of force"; and costs linked to risk of preemptive attack if a state delays making concessions. Crises are analyzed as sequences of such costly signals, during which states learn about the balance of resolve in a necessarily "noisy" and dangerous manner--the costs and risks of crises prove to be the price of learning.;Formalized in a game theoretic model, the theory yields specific hypotheses about the effects of threats, which are evaluated using Huth and Russett's (1988) data set on 58 international crises since 1885. Whereas existing rationalist arguments predict a threat will succeed when the state is favored by the balance of capabilities or interests, costly signaling theory predicts that how the balance matters depends on when information about it is revealed. I find that pre-crisis information--such as alliances, geography, nuclear status, crude measures of military capability, and the past history of an adversarial relationship--appears to have a different impact or works via a different causal logic than information revealed during a crisis by the states' bargaining behavior.