The logic of secret alliances

Abstract Alliances are typically understood as agreements intended to deter aggression from enemy states. By signaling an ally's commitment to a protégé state, a shared enemy may be deterred from attacking. In light of this signaling logic, secret alliances are puzzling. Because they are not observed, secret alliances by definition cannot achieve deterrence through signaling. So why do states enter into these secret agreements? We argue that alliances may also act as a signal of a state's intentions, potentially undermining deterrence. If a new alliance signals that the members' interests are unaligned with those of a shared enemy, the enemy may be provoked into war. As a consequence, states sometimes want to keep alliance agreements secret. To develop this argument, we analyze a formal model in which states may enter into secret or public alliance agreements. We find that the possibility of secret alliances has general effects on deterrence—possibly undermining the deterrent value of concurrent public alliances.