Interregional Trade in Russia: Gravity Approach

The paper analyzes interregional trade in Russia using gravity models. The model estimates the trade elasticity with respect to the size of exporting and importing regions and the distance between them. In addition, the impact on trade of additional factors, such as the common border of trading regions, the presence or absence of railroads, land or sea borders with other countries, is studied. Special attention is given to the issue of measuring distances between regions. The influence of the method of calculating the distance matrix (from the simplest orthodromic to the proposed weighted matrix of the shortest road and rail distances) on the coefficients of the models is studied. The all-Russian estimates of trade elasticities by the size of the exporting and importing region, equal to 1.15 and 1.05, showed high accuracy and robustness to the set of factors included in the model, the observation period, and the distance matrix. Both values were greater than one, which is significantly higher than typical estimates for international trade. This suggests that large and wealthy regions in Russia trade more, further increasing their welfare, while small and depressed regions are unable to escape the poverty trap, further increasing the current high level of regional heterogeneity. Distance is also very important in Russia (the elasticity of trade with respect to distance is –1.15, which is much higher than the world average, but still lower than the previous estimates for Siberia and the Russian Far East). This indicates insufficient transport infrastructure, higher costs of information search, transactions, contract execution, and other difficulties associated with long-distance trade. The absence of railroads in a region reduces its trade by about one-third, while neighboring regions increase the quantity of goods transported between them by about 75%. An external land or sea border facilitates domestic imports, some of which are re-exported abroad and some are consumed with the money earned from exports. At the same time, domestic exports from border regions, which cannot compete with external exports, are reduced. The method of calculating the distance matrix has a significant effect on the elasticity of trade with respect to distance, and to a limited extent on other coefficients of the model. In this case, it is recommended to use the weighted matrix proposed in this paper, which uses road distances for nearby regions and rail distances for distant regi