The gravity model revisited: An evaluation of the predictive ability of the gravity model.

Abstract: The end of the communist era in central and Eastern Europe raised many questions about which new patterns of trade would emerge when these countries would liberalize and open up for trade. To answer this question, among other methods, the gravity model of international trade was used to make predictions about future trade flows. In this thesis we evaluate the predictive ability of the gravity model by comparing the trade flows predicted in the end of the 1980s with the actual trade flows referring to the same time. We find that the gravity model gives very poor results in making predictions about future trade flows. There are several reasons for this; first, the model is specific considering both the countries and time for which it was being calibrated. Second, the method for using the gravity model is ad-hoc, based on conventions and has not undergone a thorough theoretical discussion and in connection with this the proxy variables used by the model are poorly examined. Third, the model is not specified to explain trade re-orientation, yet it is for this purpose it has been used. From this we conclude that before the gravity model is used again for explaining changes in trade patterns it needs a thorough theoretical discussion aimed at correcting these problems. Otherwise, we suggest that the model should not be used at all for this purpose.