A decomposition of the variance of international trade flows

To what extent do our theoretical models explain differences in bilateral trade flows? To answer this question, I estimate a model of international trade that decomposes bilateral trade flows into five economically meaningful components: importer's expenditure and price index, exporter's income, ad valorem bilateral trade costs and multilateral resistance terms (MRT). I then quantify the separate contributions of each component in explaining the variance of bilateral trade flows. The main contribution is to implement a decomposition method that (always) yields positive estimated shares and properly allocates the variance of trade across each component. The empirical results suggest that the measures of country size, alone, explain about 50 percent of the variance of trade flows, the measures of trade costs explain about 30 percent and that, together, the MRT explain about 20 percent of the variance. This last result is important because the MRT capture the general equilibrium adjustments specific to the theoretical model.