This paper chooses to use classic Ricardian gravity model to estimate the distance between cities, based on the static feature of the data available. It is a good choice based on the known estimated tax and non-tax barrier. However, given that we can know what kind of goods is traded between those cities, is there any chance we can adapt the monopoly competition between those cities, over one or two certain kind of goods? This would be dependent on the sparsity of traded goods of course, but it would allow more precise analysis from the perspective of trade.
This paper chooses to use classic Ricardian gravity model to estimate the distance between cities, based on the static feature of the data available. It is a good choice based on the known estimated tax and non-tax barrier. However, given that we can know what kind of goods is traded between those cities, is there any chance we can adapt the monopoly competition between those cities, over one or two certain kind of goods? This would be dependent on the sparsity of traded goods of course, but it would allow more precise analysis from the perspective of trade.