An Empirical Investigation of the Heckscher-Ohlin Theory: A Comment

Abstract

Constant-elasticity-of-substitution (CES) production functions are often assumed to be an adequate tool when investigating international trade. Earlier in this journal Michael Hoddl reported an empirical investigation of the Heckscher-Ohlin theory, with the conclusion that his results strongly suggested the importance of factor-intensity reversals over the range of relative factor prices spanned by the countries considered. Hodd calculated the capital-labour ratios (CIL ratios) for goods produced in Great Britain (UK) and exported to the United States (US), and the CIL ratios for goods produced in the US and exported to the UK. He further investigated the CGIL ratios for exported goods when also produced in the other country. With these data it can be shown readily that Hodd's results cannot be interpreted in the way he has done. This implies, in considering the two-country case of the UK and the US, that the CES production functions cannot be applied as a tool in the explanation of factor intensities in international trade. The reason for assuming CES production functions is to rescue the suspect premise of internationally identical production functions for the same products. Consider two countries I and II, the factors L and C, and two goods X and Y. It is assumed that there are CES production functions (internationally identical for the same goods), with the production function of Xhaving a higher elasticity of substitution than that of Y. This means that the isoquants of X are opened to a wider degree than those of Y. (See Figure 1.) Further, it is assumed that country I is rich in capital relative to country II. To apply CES production functions it must be assumed that one country has a factor endowment which leads to a CIL ratio higher than tan Ca;2 and conversely for the other country. When factor endowments are such that both countries have CIL ratios either higher or lower than tan a, the conclusions of the HeckscherOhlin theory follow.3 But then there would be no reason for assuming cEs production functions.