Knowledge that Transforms

In defending the twin assumptions that commodities can move freely among countries but primary factors are completely immobile internationally, trade theorists generally point out that without the factor-immobility assumption the distinction between international trade theory and domestic production and exchange theory disappears.' However, since the regional pattern of trade as well as the geographic distribution of productive factors becomes indeterminate when it is assumed that both goods and factors are perfectly mobile within a country, domestic production and exchange theorists usually assume all economic activity takes place at one point in space. Consequently, factor movements and their interrelationships with commodity flows have not been analyzed within the mainstreams of either international trade theory or domestic production and exchange theory. Instead, the subject has become a subsidiary topic of economic theorystudied mainly by location theorists, by economic historians, and, more recently, by economists interested in development theory. The leading trade economist who has tried to change the typical practice of separating the treatment of commodity and factor flows is, of course, Bertil Ohlin.2 As he states in the Preface, a major purpose of his treatise is: To analyze the domestic and international movements of factors of production, and particularly their relation to commodity movements.3 Although Ohlin's work is rich in insights on this subject, the general impact of his work has, ironically, been to reinforce the traditional approach of trade writers. For although Ohlin stressed that labor and capital are neither completely mobile or immobile internationally, he in effect assumed in his simplified trade model that knowledge was completely mobile and, therefore, that production functions were everywhere the same.4 It then remained for Samuelson to show that, by adding a few seemingly reasonable assumptions, factor prices become equalized through trade.5 Despite Samuelson's warning that the actual disparity in factor prices among countries meant that these assumptions were not so innocuous after all, the factor-price equalization model has tended to become the cornerstone of international trade theory. And, since the same world production possibilities are attainable in this model with commodity trade alone as with commodity plus factor trade, the tradition of ignoring factor movements has been further justified. Recent events, especially in connection with the operations of international firms, have, however, made it increasingly inappropriate to ignore the interrelations between output and input flows. Trade economists could in the past partly justify their position on the grounds that different decision-making units were usually involved in commodity and factor flows and that in the nineteenth century a large share of factor flows were directed at the production of noninternationally traded services, e.g., canal and railway services, or of commodities effectively unavailable in the developed countries, e.g., tropical products and certain minerals. But, today we frequently observe the phenomenon of an international firm weighing the alternatives of producing a particular commodity in one country and then shipping it to the market of another country or transferring technology and productive factors to this latter country and manufacturing the product there. The possibility of various patterns of trade in intermediate inputs makes the set of feasible alternatives facing the international firm even more complex. In order to understand better the nature of current international commodity and factor flows and to be able to deal more adequately with the policy issues they raise, we should return to Ohlin's broad vision of studying these flows simultaneously. It is also important that we consider the institutional form that these flows take. Fortu1 See, for example, G. Haberler, The Theory of Trade (London: William Hodge, 1936), pp. 4-5. 2 Bertil Ohlin, Interregional and Tr-ade (Harvard Univ. Press, 1952). 3 Op. cit., p. viii. 4 0p. cit., p. 557. ' Paul A. Samuelson, International Trade and Equalization of Factor Prices, Econ. J., June, 1948.

In a recent article in this Review, Maurice Allais proposed a model of demand for money based on the 'psychological rate of expansion' (p. 1129). His formulation of model has certain arithmetic implications that appear implausible. The psychological rate of expansion (Allais' z) is comparable to Phillip Cagan's rate of change in prices (Cagan's E, p. 35).1 Although Allais, like Cagan, relates his psychological rate of expansion to remembrance of past events, he is not specific about its precise economic meaning (pp. 1129-30, 1155).2 In particular, there is some confusion whether his variable z or Z= 250z is appropriate counterpart to expected rate of change in prices.3 However, it seems clear, despite some effort by Allais to differentiate his product, that z, an estimate of anticipations of growth in total outlays, is his basic construct while Z is introduced [t]o facilitate econometric analysis. . (p. 1132). Allais' equation (2.28) defines Z by

I would like to thank Paul Zarembka for pointing out what is apparently a very careless mathematical error in my paper. In actual fact, however, my carelessness was not so much a failure to make a correct algebraic substitution as a failure to explain briefly why I changed the form of my analytical equilibrium model equation (8) from that directly implied by the underlying behavioural model-i.e. why I did not use Zarembka's equation (8a). The model set forth in my paper represented an attempt to provide a concise and mathematically rigorous formulation of a phenomenon which was described verbally, in considerably more detail, in an earlier paper published in the Yale Economic Essays. Unfortunately, in my desire to be concise I carelessly forgot to point out in the sentence before equation (8) that for analytic as well as policy purposes I was separating the employment probability variable, 7r(t), from the percentage urbanrural wage differential variable, a(t), so that each could be treated independently-i.e., the sentence should have read, Next we specify an aggregate labor supply equation which is a simplified version of equation (1) in the sense that only a one-period time horizon is assumed and the probability variable ('r) for analytical and policy purposes is treated separately from the wage differential variable. Now, having expressed my mea culpa for this carelessness, let me turn to Zarembka's correction and show why he also has been very careless in greatly exaggerating the quantitative significance of my apparent mathematical error. I shall then show why I feel that my equation (8) is a much better way of formulating the labor supply function than is Zarembka's (8a). Briefly, Zarembka shows that if I had made the proper substitution in equation (8) on the basis of my earlier equations (2) and (3), then my equilibrium employment rate could be closely approximated by