A two-stage Asymmetric Escalation Game is developed to explore the connection between stage credibility and deterrence stability. There are two players in the model: Challenger and Defender. Challenger may initiate or not. If Challenger initiates, Defender may do nothing, respond in kind, or escalate; Challenger may then escalate or counterescalate, and so on. Each player is uncertain about the other's intentions at the final stage of the game. Escalation represents a choice that both players believe is qualitatively different from other available responses. Thus the model applies to any situation in which Defender may respond by crossing a threshold, thereby inducing a (psychologically) distinct level of conflict.
The Perfect Bayesian Equilibria are identified and interpreted, and inferences are drawn about the viability of limited war options and various competing flexible response deployment policies. In general, the model reveals that substrategic deployments add little to overall deterrence stability. Under certain relatively rare conditions, a policy called no-first-use in the super power context offers Defender advantages that might conceivably warrant the deployment stance associated with it. But a war fighting deployment never benefits Defender. Within the confines of the model, therefore, limited or substrategic wars are possible but unlikely.