Abstract
Multinational military exercises are among the most notable demonstrations of military cooperation and intent. On average, one is initiated every 8.9 days. But it has often been argued that joint military exercises (JMEs) increase the risk of war. Using a relational contracting approach, we claim that formal military alliances mediate the effect of JMEs. Exercises and alliances serve complementary functions: The former allows targeted responses to military provocations by adversaries, while the latter provides institutional constraints on partners and establishes a partnership's overall strategic limitations. In combination, alliances dampen the conflict escalation effects of exercises, deterring adversaries while simultaneously restraining partners. We test this theory using a two-stage model on directed dyadic data of JMEs from 1973 through 2003. We find that JMEs in general do not escalate conflict, and that JMEs conducted with allies in particular reduce the probability of conflict escalation.
