ABSTRACT
During the past 25 years there have emerged precise and increasingly general versions of the ancient but vague proposition that free trade is better than no trade. The ®rst of the modern achievements dates from 1972, when it was demonstrated that free trade is potentially bene®cial under assumptions of Arrow±Debreu±McKenzie type; see Kemp and Wan (1972) and Grandmont and McFadden (1972). And since then it has been shown that the same is true when markets are incomplete, when generations overlap, so that goods and agents are in®nite in number, and, conditional on existence, when increasing returns to scale and oligopolistic competition are admitted. Moreover, at each step except the last it has been possible to match the gains-from-trade proposition with the even more general Kemp±Wan proposition (Kemp and Wan, 1976; 1986) that, however tariff-ridden is an initial trading world, any two or more of the trading countries can form a customs union which worsens the lot of no agent whether inside or outside the union.1 It can now be shown that, even in a context of increasing returns and oligopolistic market structures, a nested pair of propositions is available.
