We are doing science for policy
The Joint Research Centre (JRC) is the European Commission's science and knowledge service which employs scientists to carry out research in order to provide independent scientific advice and support to EU policy.
This paper undertakes the simultaneous estimation of import elasticities of substitution (trade elasticities) within European Union (EU) regions, differentiating between imports from regions belonging to the same country (national or interregional trade) and regions belonging to other EU countries (international trade within the EU). We use a nested CES utility structure to derive the corresponding trade gravity equations and estimate them by way of Poisson pseudo-maximum likelihood regression. As the EU is a single market, the usual approach followed in the international trade literature that relies on changes in bilateral tariffs cannot be used to identify the trade elasticities. To address this issue, a very detailed definition and calculation of the ad valorem specification of transport costs is performed. The methodology takes into account the transport engineering and logistic characteristics of road freight transportation, which allows us to obtain a reliable measure of the generalized transport costs between regions. Trade elasticities are calculated at several levels of industrial aggregation, including individual sectors at 2-digit CPA classification, and their higher-level categories corresponding to agriculture, mining, and manufacturing. Results show that the trade elasticity increases the closer are the trading partners; i.e., national vs. foreign elasticities, thereby providing the first evidence of this widely presumed hypothesis. National trade elasticities are broadly double the value of their foreign counterparts. We also find that trade elasticities substantially decrease as commodities are considered at a higher level of aggregation. Our calculated trade elasticities can be adopted in a wide array of models of international trade, or spatial economic models such as Regional Computable General Equilibrium models (e.g. the RHOMOLO model), improving the results obtained from simulations aimed at policy analysis.