Comparison of Predictive Power of Gravity and Structurial Models of Trade Flows

Predicting international trade flows is an important task of the modern economy. Great progress in this direction was achieved by using a gravity method based on a simple observed dependence of the magnitude of trade flows on the volumes of GDP of the interacting countries and the distance between them. The downside of this model is that, without having sufficient micro-bases, this model is doomed to face high uncertainty in predicting trade in completely new situations. This, in turn, leads to searching for new models and to developing of known ones. Nevertheless, to date there is no universally recognized structural model that satisfactorily describes international trade. This explains the relevance of this paper. In this paper, we derive a structural model of international trade for a large number of countries on the basis of the Krugman model and prove the existence of an equilibrium of this model. After it was chosen the ten largest countries for export, for which simulations were carried out. Predictions of trade flows were built using data from open sources for 2010. In order to understand how "well" the constructed model shows itself, the predicted trade flows were compared with the real ones, as well as with the predictions of a simple pairwise gravitational model. The gravity model was calibrated on data for 2000-2009. The conclusion of the simulation is that, judging by the nature of the deviations of Krugman's model from reality, they are explained, first of all, by the insufficient number of countries included in the analysis. The second notable reason for the deviation is that the model takes into account only geographic but not cultural distances between countries in "trade costs", so that the gravity model has shown itself to be much better. The work revealed the conditions for the existence of solutions and directions for improving the well-known theoretical model of international trade, in order to bring its predictions closer to reality.