Dear all,

I am about to use a gravity model in order to identify the ex post trade effects of EU Membership on bilateral export trade flows.

My question is about the application of the gravity model on my panel data (120 countries, 1950-2015) using Fixed Effects in EViews (10). Theory and several articles suggest including importer-year (it), exporter-year (jt) and time (t) fixed effects to control for the Multilateral Resistance Terms by Anderson and Van Wincoop (2004). Unfortunately, this leaves me with two questions when it comes to the practical application.

(1) Due to the structure of my data, I have bilateral trade of country-pairs (ij) for each year. This is in line with recent studies using the gravity approach for the EU Membership Effect on trade. In order to reproduce their results I want to follow their application. But they seem to depart from theory by only including country-pair (ij) and time fixed effects and not im- and exporter-year fixed effects! Is this a violation of the theory-consistent estimation? (e.g. the WTO Practical Guide (2012:108) (Annex) recommend to generate these fixed effects to each country and year (it and jt). This would imply to include 2nT dummies in the estimation!)

(2) I am using EViews for the estimation and I am not sure whether using fixed cross-section and time effects in the panel options do correctly account for (a) country-pair or (b) im- and exporter-year fixed effects?

I really appreciate your answers in advance and will provide further information if necessary.

More Marvin Sachs's questions See All
Similar questions and discussions