Gravity Equation Analysis in the Context of International Trade: Model Specification Implications in the Case of the European Union

Abstract

We examine model specification implications pursuant to gravity equation analysis in assessing the validity of new trade theory (NTT) models. The testing is conducted on twenty-three EU members, consisting of eight postsocialist transitional countries and fifteen developed countries, for the period 1994-2004. Our model specification borrows from Baltagi et al. (2003), employing fixed effects in the panel estimation. In accordance with them, we find that the appropriate model is specified as generally as possible. Divergent from them, we find that estimation results do not strongly depend on model specification. Yet, in analyzing trade flows for disaggregated samples, considerable sensitivity to specification is observed and an enigma encountered per findings with respect to trade behavior that support neither NTT nor the theory of comparative advantage. Estimation results are found not to depend on whether gross domestic product data are measured in nominal or purchasing power parity terms.