In this Review (1981), we attacked a proposition (P1) of Eitan Berglas (1979) and others, that unilateral tariff reduction (UTR) is necessarily superior to customs union (CU), provided scale economies and changes in terms of trade are ruled out. We put forward a much weaker proposition (P2) (p. 706) that UTR is sometimes superior, sometimes inferior. In his 1983 paper, (p. 1142), Berglas concedes that our 1981 Figure 2 (which he calls example El) illustrates CU superiority. He thereby concedes our main point: P2 is correct; P1 is not. A question remains: how interesting is the domain where UTR is superior?' We argued (1981) UTR is superior under narrow assumption Al (partner B's tariffs can be ignored) or A2 (no tariff by outsider C nor transport costs in trade with C). But a CU is superior in our main example (Figure 2), where neither Al nor A2 holds, and partners A and B trade with mutual benefit in the price wedge between C's import and export prices. In addition to wrongly claiming we were illogical and incorrect,2 Berglas argues (1983) that there are two other assumptions in his 1979 paper which, taken together, are also sufficient to establish UTR superiority. However, he misstates them. They are not, as he says, A3 (the CU does not affect the direction of trade) plus A4 (all three countries trade). Instead, they are A3 plus a much more restrictive A5: C trades every good with the CU. (This assumption, in Berglas, 1979, Figure 1 and Table 1, is distinguishable from A4 only when there are more than two goods.) Why must A5 be assumed, not just A4? Without A5, trade between A and B can occur in some goods within C's price wedge. In short, Berglas establishes UTR superiority by A5, assuming that A and B can't trade in the wedge where CU provides mutual benefits. (Our wedge becomes increasingly important in the n-good case. Consider cement, for example.) Moreover, Berglas's A3 rules out changes in trade patterns, and therefore Viner's concepts of trade diversion and trade creation which introduced the modern CU debate. Like a case based on Al or A2, the A3 + A5 case for UTR superiority is not interesting. Even if it were, one more special case does not establish the general principle that UTR is necessarily superior to a CU, any more than one more example where protection raises welfare would establish a general proposition that protection necessarily raises welfare.

During the past decade and a half, an important part of the literature on customs unions has dealt with the question of whether a country might obtain the gains it would achieve from a customs union (CU) in an alternative way, by a unilateral tariff reduction (UTR). (UTR may involve a partial reduction in tariffs, or a reduction all the way to zero.) A widely accepted conclusion (see Eitan Berglas, p. 329; C. A. Cooper and B. F. Massell, 1965b, pp. 745-47; Roma Dauphin, ch. 2; Harry Johnson, p. 280; Melvyn Krauss, pp. 417-19; and Peter Robson)' is that UTR does indeed hold out the prospect for all the gains from a CUwithout the disadvantagesif two important simplifying assumptions are made; namely, that we ignore economies of scale and the effects of a customs union on the terms of trade.2 In the words of Berglas: It is important to note that if a [preferential] trade agreement does not affect the terms of trade, then it does not allow for any mutually beneficial policy opportunities which are not open to each of the member countries separately [through UTR] (p. 329). If this conclusion is correct, it is very important, in that it undercuts the earlier literature on customs unions. The question asked by Jacob Viner in his pioneering work whether a CU represents a net gain or a net loss in economic efficiencybecomes unimportant, except insofar as a customs union is based on terms-of-trade effects3 or economies of scale,4 since a CU can be summarily rejected in favor of UTR. The UTR case would mean that, for economists, the puzzle is not to identify the efficiency gains (or losses) from a CU, but rather to explain why countries form customs unions in the first place (Berglas, p. 329; Cooper and Massell, 1965b, p. 247; and Johnson, p. 270). Indeed, in his survey of CU theory, Krauss identifies the problem raised by Cooper and Massellof why countries form customs unionsas . . . the theoretical issue of the past decade [the 1960's] just as in the prior one the major issue, as explicitly defined by Jacob Viner (1950), was whether a customs union represented a movement towards freer trade or greater protection (p. 413). The typical reply to the Cooper-Massell question is: Countries tend to form a CU for noneconomic reasons (Berglas, pp. 329-30).5