A Double-Edged Sword: Nuclear Deterrence and Nuclear Caution

Most empirical studies of the effects of nuclear weapons only consider whether a state at a given time has nuclear arms or not, but there are strong reasons to think that the effects of a state’s nuclear arsenal are conditioned by a range of other variables including how nuclear weapons are managed. I argue that the effect of nuclear weapons in non-existential disputes is determined by what I call nuclear risk: the overall likelihood that these weapons may be used without authorization by political leaders. Using a formal model, I hypothesize that higher nuclear risk leads to greater deterrent power, but it also makes the leaders of the nuclear state more cautious if their deterrence fails. I test these two hypotheses using the Correlates of War data. I first measure nuclear risk using a simple index, and then propose a novel Bayesian technique for imputing risk. Both hypotheses are borne out by empirical results. That is, nuclear-armed states with riskier postures are more likely to get concessions from their opponents, but they are also less likely to reciprocate a dispute when they are targeted by other states.