Alternatives to the H-O-S Theory of Trade
Given the formidable list of assumptions itemized above, and the unlikelihood that any one of them (let alone all of them) would be realized in practice, it is not surprising that deficiencies in the predictive power of the H-O-S framework soon became apparent when empirical tests of the model began, starting with the Leontief paradox and the inability of the H-O-S model to depict intra-industry trade.
A particularly egregious assumption is that factors of production are homogeneous in quality and identical across countries. In Leontief’s original test and subsequent ones, one of the two factors is “capital”, and countries are differentiated according to whether they are “labour- abundant” or “capital-abundant”. Since there is no homogeneous and measurable puttylike capital that can give substance to this empirical test, Leontief and others following him resorted to the market valuation of a heterogeneous bundle of capital goods. This meant that market prices as well as quantities, instead of pure quantities, had to be used from the outset in estimating countries’ relative factor endowments. A similar critique applies to the assumed homogeneity of the factor “labour”.These and other critiques of the H-O-S model elicited a variety of responses among trade theorists. One was the birth of a “neo-Ricardian” theory of trade at the hands of Sergio Parrinello (1979), Ian Steedman (1979, 1987) and Lynn Mainwaring (1979, 1984), where the factors of production are homogeneous labour and a bundle of heterogeneous capital goods. Commodity prices in each trading country, both before and after trade, are determined by income distribution as well as technology. Since these models were inspired by the work of Sraffa (1960), they are sometimes referred to as “Sraffian”. The profit component in total cost reflects the lag occurring between the application of inputs and the resulting output (a lag mostly ignored in neoclassical trade theory).
One of the distributive variables, the wage or the profit rate, is assumed to be exogenously given. The advantages and limitations of neo-Ricardian models are well described in the “Introductory essay” (chapter 1) of Steedman (1979). The advantages claimed for them are the depiction of growing as opposed to stationary economies, and the allowance for the role of time and heterogeneous capital goods in the production of commodities. Their limitations are that they often focus on steady growth equilibria and do not allow for imperfections in competition. To ensure the attainment of steady growth, they omit inputs of land and other non-reproducible resources, a feature shared in common with neoclassical models whose authors are also mainly interested in elucidating the nature of steady-growth equilibria.In considering the appropriateness of neo-Ricardian models as possible reflections of Ricardo’s implicit trade model, recall that the primary aim of neo-Ricardian authors is rather to challenge the dominant H-O-S trade model. One of the basic issues they examine is whether the 2 ? 2 ? 2 Heckscher-Ohlin model remains applicable when the factor of production “land”, paired with “labour”, is replaced by “capital” consisting of a heterogeneous bundle of goods. While this is not the place to review the heated controversies on the theory of capital of the 1960s and 1970s (Harcourt 1972), the inclusion of heterogeneous capital goods among a country’s inputs is an important extension of the H-O-S theory in which capital is either ignored or assumed to be a putty-like substance on the same footing as land or labour. Neo-Ricardian authors do not claim to depict Ricardo’s comparative cost example in chapter 7 of the Principles based on the labour theory of value, and in fact point out the errors implicit in Ricardo’s procedure. Since positive profits usually cause prices to diverge from labour values, they show that, in contrast to Ricardo’s assumption, countries may specialize in the “wrong” commodity (that in which they have a comparative disadvantage) and experience a lower welfare than under autarky.
Consistency with Ricardo’s conclusions about the causes and consequences of trade thus depends crucially on his assuming “all the great variations which take place in the relative value of commodities to be produced by the greater or less quantity of labour which may be required... to produce them” (1817 [1951b]: 36-7).Another response to the inadequacies of the H-O trade model was to reject the assumption of perfect mobility of all factors among economic sectors, and assume instead that some are specific to the sectors in which they are located. This gave rise to the “Ricardo-Viner” or “specific-factors” model elaborated by Ronald Jones (1971), whose general features were suggested by Gottfried Haberler (1936: ch. 12). In Jones’s simplest 2 ? 3 ? 2 version of this model based on two sectors and three factors, each sector uses a type of capital (or land) specific to it as well as a “mobile” factor (taken to be homogeneous labour) common to both sectors.