The Effects of Trade Preferences for Developing Countries

Abstract

At the 1964 and 1968 United Nations Conferences on Trade and Development the developing countries pressed for, and the developed countries finally agreed to, the granting of preferential tariff treatment on the formers' manufactured exports. Such a discriminatory measure runs counter to the non-discrimination or most-favored-nation (MFN) principle set out in the first article of the General Agreement on Tariffs and Trade and a GATT waiver was necessary before the developed countries could introduce their individual schemes under the Generalized System of Preferences (GSP).' Since their introduction between 1971 and 1976, these schemes have aroused opposition not only from protectionist forces in the donor countries and from supporters of the non-discrimination principle, but also from economists arguing that the GSP yields little benefit to the developing countries who would do better to seek MFN tariff cuts on goods in whose production they possess a comparative advantage. The developing countries have, however, been if anything more skeptical of MFN tariff cuts since the introduction of GSP, because such cuts would erode the value of their preference margins. The relative merits of GSP versus MFN tariff cuts from the developing countries' perspective is partly an empirical matter, which could be settled by estimates of the trade creating and trade diverting impact of GSP. Such estimates require values for GSP beneficiaries' parameters, about which our knowledge is limited. In this paper I contend that the most influential estimates of GSP effects sidestep this requirement in a manner which guarantees the predominance of trade creation, whereas consideration of plausible parameter values and of the design of actual GSP schemes suggests trade diversion to be more significant. The standard partial equilibrium model with infinitely elastic supply (Figure 1) has proven popular for estimating the effect of non-preferential tariff changes, largely because of its meager data requirements. It is, however, inappropriate for preferential tariff changes because all imports come from the same source and all-or-nothing diversion of nonbeneficiaries' trade is predicted. This difficulty can be circumvented, while remaining within the