Measuring insularity as a state of nature : is there a case of bad geography?

According to both empirical and theoretical literature the most immediate case of "bad" geography is not having a direct access to the sea. A large proportion of less developed countries are landlocked economies. Being an island is also associated to a role for geography in the economy. Islands are surrounded only by sea, which is the main geographical difference from coastal countries. The total land discontinuity rises costs by cutting out some alternatives in the connection system of the island. But at the same time insularity may lead to growth opportunities mainly linked to tourism specialization. The first goal of this work is to collect available physical geography data on islands and elaborate measures of insularity for all countries in the world. Then we want to evaluate if and in which cases insularity can turn into "bad" geography and when instead it appears to be associated with better economic performance at the country level. We find that two dichotomies are important: states which are islands (full insularity) perform worse than countries which have islands (partial insularity). Within the group of island-states, isolation is the crucial dimension associated with poor economic performance, less the dispersion of the land in many islands. Instead, being coastal and having islands is associated to better results than having only a direct access to the sea. It is the subsample of countries with islands the one which performs better in the heterogeneous group of coastal economies.