Determinants of the Commodity Structure of U. S. Trade: Reply

Lawrence Weiser and Keith Jay have made an important contribution to our efforts to sort out the relative significance of innovational activities, different types of labor skills, and economies of scale as determinants of the commodity structure of international trade. In particular, they provide new evidence substantiating the importance of technological progress as a basis for the U.S. comparative advantage position as well as support for the view that scale economies should be given greater weight than my study indicated. However, there are still a number of questions and problems to be answered before very firm conclusions can be drawn on these matters. In an effort to disentangle the relative importance of the innovational and skill factors, the authors calculate a measure of technical progress for various U.S. manufacturing industries between 1961 and 1967. However, their estimate of technical progress does not take explicit account of changes in the skill levels of industries during that period. Consequently, one still cannot be sure of the extent to which the measure represents innovational activity versus changes in the quality of labor. Although my study indicates that the size of plants in export industries is considerably larger than in import-competing industries, the scale variable did not show up as significant in the various multiple regressions that included such variables as capital/ labor ratios by industry, proportions of various skill groups, and R&D efforts. However, Weiser and Jay do obtain significant multiple regression results for the scale factor when they use Gary Htifbauer's measure of the scale variable and a different dependent variable than the one I uised. Their dependent variable is the ratio of U.S. exports of each industry to the industry exports of eleven leading exporting countries whereas my dependent variable was adjusted U.S. exports minus adjusted U.S. imports. It should be noted that when they used Hufbauer's scale variable and exports minus imports as the dependent variable, they failed to obtain a significant result for the scale variable. To a considerable extent, therefore, the issue seems to come down to what is the best dependent variable to use. Clearly, if one is interested in the best indicator of export performance (as they are), a variable including imports is not appropriate. However, trade theory should, I think, generally focus on net trade flows rather than just exports or imports since the policy variables in which we usually are concerned are framed in net terms, for example, balance-of-trade or net employment effects of trade policy changes. As the authors are aware, it might well be that their scale variable would not show up as significant if they in some way also included the U.S. world import share by industry as part of their dependent variable. It would be interesting, to extend their analysis of trade patterns along these lines as well as by taking account of skill changes in calculating a measure of technical progress.