This paper compares various estimation methods often used in the estimation of gravity models of international trade. The authors first discuss different structural and consistent estimation techniques, their underlying assumptions and their impact on estimated coefficients. They then estimate the gravity model for global bilateral trade flows using various empirical methodologies. They focus on a comparison of the distance and border effects across estimation techniques. For the border effects they take into account adjacency effects as well as the distinction between intra-regional and inter-regional trade. Their findings confirm the significantly negative distance and the significantly positive adjacency effect across estimation methods, although the size of the effects varies substantially across methods. The border effects by global regions are much more sensitive - both in size and direction - to the applied estimation method. Although all estimation methods have their own merits and drawbacks, the authors provide some guidelines for future empirical research based on their findings.
gravity model empirical estimation distance effect border effect