Two problems with capital
The direct and the indirect substitution mechanisms are initially developed rather separately, and in the nineteenth century some authors appear conscious only or predominantly of the first one, others of the second one; but, with the singular exception (discussed later) of Walras and of his few followers (such as Pareto and Barone), capital is always conceived as a single factor “embodied” in the heterogeneous capital goods in proportion to their value.
The reason is that the rate of interest is tendentially uniform on all capital, and therefore the logic of the theory obliges to see it as the rental of a single factor. One might think that each capital good could be seen as a separate factor, with its own contribution to production; but all founders of the marginalist approach as well as the second generation share with the earlier classical theorists the “method of normal positions”, that is, the belief that it is fundamentally impossible (and also uninteresting) to explain the day-by-day market prices and quantities, influenced as they are by a multitude of accidental and transitory causes; what economic theory can and should try to explain is their averages and trends as determined by the gravitation toward persistent normal or long-period values. Then the treatment of each capital good as a different factor with its given endowment is made impossible by the rapid adaptability of the quantity in existence of each capital good to the demand for it, that deprives this quantity of the persistency required for its inclusion among the given factor endowments of an equilibrium determining normal prices. The quantities of the several capital goods must then be endogenously determined by the equilibrium, as the quantities associated with a uniform rate of interest on supply price. Therefore the only possibility is to view the several capital goods as “embodying” a single factor “capital” of endogenously determined “form”, that is, composition. Now, the reward of capital, interest, is proportional to the value of capital, so if, like the rentals of other factors, it is to be proportional to the quantity of the factor “embodied” in the heterogeneous capital goods, this quantity must be proportional to the value of those capital goods; hence the value of capital measures the quantity of capital; nor is there any other physical way to specify this quantity: weight or volume of capital goods have no univocal connection with productivity or remuneration.This poses two problems. A supply-side problem is how to specify the endowment of capital without logical circularity: the value of capital goods in terms of any numeraire depends on relative prices, but normal relative prices depend on the rate of interest, so a change in the rate of interest will change the normal value of the existing stock of capital goods even in the absence of any change in its physical composition, violating the need of the theory for data independent of what must be determined. A demand-side problem is that, even assuming the supply of capital to be somehow determinable without logical circularity, the theory needs to prove that the demand for capital is a decreasing function of the rate of interest, that is, that the two factor substitution mechanisms work for substitution between value capital and labour in the same way as for substitution between physically measurable factors; but this would require relative prices to be given, instead of being among the variables to be determined.