Comments and Questions

On “Islands in Trade: Disentangling Distance from Border Effects” by Jose Luis Groizard, Helena Marques, and Maria Santana. Short summary:The authors investigate if island regions have different trade costs from mainland regions using Spanish data on inter-regional trade between different regions in Spain. They apply a standard gravity equation, in which they control for island regions in the trade cost function. Then they decompose their results using Blinder-Oaxaca decomposition to disentangle the effects between island regions and mainland regions. The authors find that islands are disadvantaged due to higher (island-specific) fixed trade costs. Reducing these trade costs, would lead to a significant increase in inter-regional trade. The research question is very interesting and relevant to correctly specify trade cost functions and hence gravity equations. Especially the Blinder-Oaxaca decomposition seems to be a highly suitable approach as it allows a more detailed analysis of the importance of island border effects in the trade cost function. Comments:1. The identification of the island border effect depends on the correct specification of the transport cost function: a) It is not clear why the Island border effect leads to higher fixed trade costs while the variable costs are the same for air, land or sea transport. The authors might be able to give more specific (anecdotal) evidence for the higher fixed transportation costs. It would be interesting to interact the island dummy with the distance measure in the transportation costs equation (8) to allow for different variable costs. Actually the Blinder-Oaxaca decomposition would suggest this as the “Coefficients term C” for distance is very high. This should considerably improve the results of the gravity estimation.b) Omitted variables, such as the (time varying) industry structure of the regions might bias (or even drive) the results. For robustness variables such as size of the service and agricultural sector could be included. As well as historical ties of regions, e.g. Balearic Islands and Cataluña.2. The motivation to use distance and quadratic distance should be described in more detail, especially as the effects in the gravity estimation are not clear. The authors find that distance has a U-shape effect on trade for islands and an inverse U-shape effect for the mainland. At which distance is the turning point for these effects? If I use log(km) with the coefficients of the gravity equation in Table B1, I find that exports and imports for Island regions are strictly decreasing for reasonable values of distance (<4,000km), while imports and exports for mainland regions increase for reasonable values of distance (<4,000km).3. The Alchian-Allen effect might be an explanation for the lower trade of islands due to higher fixed transportation costs. The corresponding literature should be mentioned.4. It would be interesting to link the inter-regional findings for Spain to international trade. Do we find a similar effect for islands, such as Cyprus, Ireland, or Sri Lanka.