Table 2 lists both specifications and results from both of these calculations, and compares Nash equilibrium outcomes to those under no intervention (free trade in the goods trade model). Because of the network externality accruing to own country residents, origin countries have an incentive to subsidize own consumption of internationally traded services, but destination countries have an incentive to apply a significant tax due to revenues transferred from abroad under the tax.

As a % of national income, results in Table 2 suggest that the welfare gains from moving to no intervention (the analogue of free trade) are larger in the two service type model than in the trade model. On the basis of these calculations we suggest that country size seems to play a more pronounced role in the services case than in the goods trade tariff war. The gains from liberalization are the considerably larger in the network related services case than in the goods case with a similar parametric specification.

We have also used the two service type model set out above to analyze the implications of global liberalization in telecoms using an empirically based specification of the modelling approach above. For these purposes, we consider a global four region telephone service network consisting of Canada, Europe, USA and the Rest of the World (ROW). We use this to compute the global welfare gains from liberalization adopting the network related service model presented above.

We also analyze the role played by the relative sizes of domestic and global networks in determining the division of gains, using data on the volume of cross-border telephone calls, and the relative size of caller and recipient regions. The size of the domestic network in Canada, for instance, is smaller than that of the USA, even though both countries have inter connected domestic and international networks. European countries have a well developed domestic network, while networks elsewhere in the world are, on average, not as developed as in the three other regional blocs in the model.