Abstract:
We combine the latest difference-in-differences estimators for treatments with exit and structural gravity literature to evaluate the effects of sanctions on trade, when they are in place and when they are lifted. Our analysis shows that sanctions reduce trade between senders and targets by 58%, with estimates from our preferred model 50% larger than those from traditional two-way fixed effects (TWFE) models. A bias decomposition highlights arbitrary weighting and contamination bias in TWFE estimates. Sensitivity checks confirm the robustness of our findings, emphasizing the relevance of these methods for gravity estimations, including trade, migration, foreign investment, and other bilateral flows.