Abstract: One way nuclear agreements might keep signatories from proliferating is by constraining nuclear capacity. Theoretical work on nonproliferation often points to such constraints as an important driver of nonproliferation success. Some have argued that, absent sufficient constraint, states with the desire and capability to proliferate will do so. Faced with more costly routes to a weapon, states subject to technological constraint may abide by the terms of the deal. This perspective poses an important empirical question: do nonproliferation agreements result in significant technological constraint in practice? This paper evaluates the empirical prevalence of constraints arising from nonproliferation deals. Doing so requires (1) providing an appropriate measure of nuclear proficiency and (2) developing an estimate of the counterfactual, no-agreement capacity of states that received such agreements. This study addresses both of these points. First, new data is gathered to estimate proficiency, improving upon existing measures in the literature. Second, the generalized synthetic control method is applied to estimate counterfactual proficiency levels for the recipients of agreements. With this approach, the constraining effects of deals the United States implemented with Japan, South Korea, and Taiwan and the Declaration of Iguacu between Brazil and Argentina is evaluated. The findings indicate that the constraining effect of these nonproliferation agreements is minimal.
