Detecting Audience Costs in International Disputes

Selection effects in crisis bargaining make it difficult to directly measure audience costs because state leaders have an incentive to avoid incurring audience costs. We overcome this inferential problem of selection bias by using a structural statistical model. This approach allows us to estimate the size of audience costs, both incurred and not incurred, in international crises. We show that although audience costs exist for state leaders of various regime types, democratic leaders face larger audience costs than nondemocratic leaders do. Audience costs can be so large that war might be preferable to concessions, especially for leaders of highly democratic states. Audience costs also increase a state's bargaining leverage in crises because the target state is more likely to acquiesce if the challenge carries larger audience costs. We also find evidence that audience costs generate selection effects.