Journals Proceedings

Abstract

The paper presents the use of the augmented gravity model for panel data. Silva and Tenreyro (2006) showed that the formerly used Hausman-Taylor estimator fails to provide consistent estimates in this class of models. They propose the use of Poisson pseudo maximum likelihood (PPML) approach, that consists in solving a set of first order condition equations that stem from the Poisson regression model but are applied to attain estimates in the log-linear regression. We expand their results by including random country effects in the model, which allows for the more efficient use of country-level panel data. We use the above described model to estimate the effects of a couple of factors on the value of exports of the group of Visegrad countries (V-4, Visegrad four). These include the EU accession and the recent economic crisis, as well as other countries’ characteristics typically included in the gravity models. We allow for structural breaks and check whether the moment of joining the EU could be considered as one. The results assessed for the group of the EU countries trading with 234 countries of the world in the 1999- 2013 period demonstrate the beneficial effects of the EU membership as well as the existence of structural break at the moment of the EU enlargement. The robustness checks confirm validity of the proposed estimation approach