Having argued that the country division of the gains from international liberalization in network related services will typically differ from that for trade in goods, we next proceed to a discussion of the size of the gains from such liberalization.

We analyze the size of the gains from liberalization in a cross country policy game by comparing Nash equilibrium and free trade outcomes. The network related service policy game differs from a conventional tariff game (Johnson (1954), Gorman (1957)) in which both countries apply tariffs to their own imports. In this game, both countries have the ability to tax service flows between countries; the origin country through regulation of the service flow as it leaves its own border, and the destination country through regulation of flows from its own border onward within the country of the eventual recipient of the network related service9. Thus, in the two country case four tax rates t. are involved in the game; the tax rates charged by country j on services originating in country i.

We use our two service type model set out in section III above, with the further difference here being that tax rates charged by the two countries on network related services are endogenously determined in a Nash equilibrium rather than set exogenously. In the process, we determine the two country two tax rate reaction functions

To provide a point of comparison to tariff games in goods trade, we also compute Nash equilibria in a two country tariff game in which S2, instead of being a network related service, is treated as a conventional commodity, much like G. In this calculation, all share and elasticity parameters are as in the two service type model calculations reported on in Table 1, and in a Nash equilibrium tariff rates on imports are determined by commodity and by the country of source. In this case, levels of only two instruments (tariffs on imports by each country) are endogenously determined, rather than four instruments as in the network service case.