Testing the Demand Side of Comparative Advantage Models
探讨了比较优势理论中需求侧的作用，指出现有实证检验多关注供给侧，而需求侧假设（如偏好相同）的稳健性未得到充分验证，对实证研究有重要启示。
Both the Ricardian theory of comparative advantage and the more modern version often associated with the names of Eli Heckscher and Bertil Ohlin (hereafter H-O) have been subjected to many empirical tests. But the results of the tests have been mixed, and on the whole, inconclusive.' While the Ricardian theory attributed the source of pretrade commodity price differences among nations to broadly specified production conditions as represented by factor(labor) productivity ratios, the H-O theory in its simpler version attributed the same to relative differences in factor endowments prevailing among different nations. The two theories are broadly similar, however, in the sense that they both identify the ultimate source of comparative advantage on the supply side. No crucial role is assigned by either theory to the demand side in the determination of comparative advantage.2 Indeed, the demand side is effectively neutralized, at least in the standard H-O model, by the strategic and commonly made assumption that consumer preferences are internationally identical and homothetic.3 But how crucial is such an assumption, particularly when there are many commodities with which almost all empirical tests of comparative advantage are concerned? Can the robustness of this assumption be established, and hence its empirical relevance, short of directly testing the assumption itself ? The existing literature has not dealt explicitly with these questions in the context of the comparative advantage issue. There have also been other difficulties associated with testing the respective theories *Associate professor of economics, Tulane University. I wish to acknowledge useful comments and suggestions I received from Martin Bronfenbrenner, Robert Stern, and the managing editor of this Review on earlier versions of the paper. I also gained from the comments of seminar participants at Tohoku University, the Economics Institute, and the Stockholm School of Economics. I alone remain responsible for any errors. 'As is now widely known, the pioneering work of testing the Ricardian theory of trade was undertaken by G.D.A. MacDougall (1951). This was followed, with substantively the same methodology, by Stern (1962), Bela Balassa, MacDougall et al., Mordechai Kreinin, and more recently with a somewhat different methodology, by Herbert Glejser. Kreinin's study is particularly suggestive, however, in that the remarkably stable empirical relationships which were found in the earlier studies with American and British data do not extend to several other countries which he considered. The first systematic attempt to test the H-O theory was due to Wassily Leontief, whose well-known finding on the U.S. capital position caused a major controversy. (See Robert Baldwin, for example, for a perspective review of the Leontief Paradox.) Leontief's work was followed, with widely mixed results, by Masahiro Tatemoto and Shinichi Ichimura, Donald Wahl, Ranganath Bharadwaj, Karl Roskamp, Michael Hodd, and Roskamp and Gordon McMeekin, among others, all of whom applied the input-output technique for observations from additional countries. More recently, however, a multivariate approach to testing the H-O theory, with diverse empirical specifications, has been undertaken. See, for example, Gary C. Hufbauer, William Branson and Helen Junz, Baldwin, Edward Leamer, Jon Harkness and John Kyle, Stern (1976), Branson and Nikolaos Monoyios, and Harkness. 2Ohlin (especially ch. 1) emphasized the interplay of demand and supply conditions in the determination of commodity and factor prices. The demand side appeared in many of Ohlin's passages, but it was never given as critical an importance as the supply side in his analysis of the cause of trade. 3Consumer preferences are said to be homothetic if indifference hypersurfaces have the same gradient for any consumption vector unique up to a scalar. The direct implication of homothetic preferences is that income elasticities of demand are identically unitary for all goods. It is of some interest to note, however, that John Stuart Mill also made a similar assumption in demonstrating his law of international values in a two-commodity context (ch. 18). As pointed out by John Chipman (pp. 484-85), Mill's assumptions (p. 155) were tantamount to positing a utility function (internationally identical) of the form, log U = 6, log Cl + 62 log C2, where the ci's denote the units of consumption, and the ai's are a set of parameters denoting consumption expenditure shares. The assumption of internationally identical constant consumption expenditure shares, often extended over many goods, was also employed in modern works on the classical theory of trade. See, for example, Thomas Whitin and Lionel McKenzie.