In my last subject of studies I am analyzing trade costs in a gravity model, e.g. multilateral trade resistance and bilateral trade resistance terms. The gravity model assumes world trade. However, I want to analyse the effect of these trade resistance terms only on a subset of countries, , e.g. trade costs of war on LDC's. I therefore would drop all developed countries as exporters and would only analyse LDC's as exporters. Developed countries would be included as importers though. Would it be statistically misleading or would that regression work ?
This regression may work. Two issues, however.
- How do you identify the "trade costs of war" given that trade and war are potentially endogenous?
- Dropping the developed countries, which are the main exporters, may affect the estimates of the bilateral and multilateral trade resistance terms. But this is easy to check by comparing the estimates with and without the main exporters.