Zusammenfassung:
Most empirical estimates of the trade elasticity exploit tariff variation generated by Preferential Trade Agreements (PTAs). Because firms only partly utilize preferential tariffs, standard regressions identify an eligible-tariff elasticity rather than the structural elasticity with respect to tariffs actually paid that is required for quantitative trade models. I model PTA tariff cuts as a treatment with partial compliance and estimate the effective-tariff elasticity using an instrumental variable approach that explicitly accounts for preference utilization. The effective-tariff elasticity lies between 10 and 12, roughly three times larger than eligible-tariff estimates around 3.5. I show that the eligible-tariff elasticity varies with the tariff preference utilization of the estimation sample and is therefore not portable across policy environments, while the effective-tariff elasticity identifies a policy-invariant elasticity that can be used for quantitative modelling.