Is the Border Effect an Artifact of Geographic Aggregation?

The existence of a large border effect is considered as one of the main puzzles of international macroeconomics. Following Hillberry and Hummels (2008) insight, we are able to solve this puzzle by combining international shipments with intra-national shipments data characterised by a high geographic grid. At this fine grid, intra-national shipments are highly localised and dropping sharply with distance. The use of a small geographical unit of reference to measure intra-national bilateral trade flows allows to estimate correctly the negative impact of distance on shipments. When we use sector disaggregated export flows of 50 Spanish provinces in year 2005 exhaustively split into inter-provincial and inter-national flows, we find that Spanish provinces trade as much with other countries as they do with other  Spanish provinces, once we control for differences in size and proximity; that is, there is no border effect in Spain. The results are robust to changes in the geographic unit of reference and to the level of industry aggregation.