Introduction

To some extent, ‘technological progress’ can occur in the absence of gross investment. In the models treated in chapters 11 and 12, technological progress was strictly of this ‘disembodied’ type. Other innovations require only small modifications of existing capital equipment and, therefore, negligible capital expenditures. However, certain innovations can be introduced only by scrapping old capital equipment and replacing it by new capital equipment that ‘embodies’ the new technology. Such changes in technology are brought about only by significant gross investment.

One may add to this ‘fact’ another ‘fact’ of some importance: once a piece of capital equipment has been installed, there is often very little scope for factor substitution, certainly not the scope posited in the ‘neoclassical fairy tale’. Two questions immediately arise. First, if capital equipment existing at a given moment is in fact heterogeneous, is it possible to define a capital aggregate such as that used in simple neoclassical theory? Second, if machine operation requires fixed or almost-fixed proportions, can one deduce the marginal equivalences upon which neoclassical theory rests?

The first question, while long recognized, was explicitly posed in 1953 by Mrs Robinson and Champernowne. Solow provided the answer. If a production function involves more than one kind of capital good, it can be written in terms of a capital aggregate if the production function satisfies the condition of Leontief's theorem on separable functions. In this context, the Leontief theorem requires that the marginal rate of technical substitution between different types of capital equipment be independent of the amount of labor used. This, of course, is a very restrictive condition.