Abstract
Direct nuclear deterrence should provide significant stability considering the defender's capability to inflict enormous costs on the challenger and its high level of resolve to protect its territory and people. In the past, however, nuclear powers' deterrence strategy to prevent adversaries from escalating a crisis has failed. We argue that to manage deterrence failure crises, nuclear weapons would not matter much until the crisis actors overcome the inherent political difficulties in following through on the threat of nuclear punishment. If a defender passes this bar, then it would escalate the crisis after the deterrence failure. Conversely, if the challenger possesses high political feasibility to follow through on a threat of nuclear use, then the crisis will be terminated less violently as most defenders will view nuclear escalation as more likely and decide to compromise. We identify a total of 156 cases of direct deterrence failure after 1947 from the International Crisis Behavior Data data set. Then, using ordered logistic regression analysis, we find evidence for this feasibility of punishment hypothesis.