Everything You Always Wanted to Know About EU Membership Trade Effects but Were Afraid to Ask

This paper studies the role of dataset choices for trade policy estimates in structural gravity models. Using twelve publicly available data sources and applying alternative and commonly used dataset restrictions, we obtain 586 estimates for the EU membership trade effects from a standard structural gravity model specification. These estimates are used in a meta-regression analysis to shed light on potential sources for heterogeneity in the obtained EU membership trade effects. The meta study reveals a crucial role of domestic trade flows for the effect size of the EU membership trade effect estimate. The estimated EU trade effect is on average 20 percentage points larger in datasets that include domestic trade flows. The effect size of the EU membership trade effect estimates also vary with the time- and country coverage, across time interval and consecutive year panel data, alternative product classifications, the level of sectoral disaggregation in the data and the use of imports as alternative measure for trade flow. Alternative estimation packages do not significantly alter the effect size of our estimates for the EU membership trade effects. The paper concludes with some recommendations on data source selection and dataset restrictions.