Introduction

The classification of inventions is straightforward. Further, determining the capital-using or labor-using character of any particular invention would also be straightforward if one could accurately observe the proper variables, i.e. the change in the marginal rate of technical substitution at a given capital–labor ratio or the change in the capital–output ratio at a given rate of interest. But, unfortunately, these changes cannot be accurately observed. At best, one can only determine, within statistical probability limits, the biased or neutral character of a stream of inventions over time.

Yet one can observe particular inventions and draw non-statistical inferences concerning their factor-saving effects. This, I would suggest, has been the cause of great mischief. The inventions typically observed are those that cause smaller machines and human labor to be replaced by larger machines: locomotives replaced wagon trains, and digital computers have more recently replaced desk calculators and file clerks. Even Pigou was presumably misled, for he wrote: ‘Probably, however, the majority of inventions in the narrower sense would have to be reckoned as “labour-saving”, because as Cassel has observed, “almost all the efforts of inventors are directed toward finding durable instruments to do work which has hitherto been done by hand.”’

Hicks agreed: ‘…there is no reason to question his [Pigou's] view that inventions have a decided bias in the labour-saving direction. It is indeed difficult to find cases of important capital-saving inventions—wireless is, of course, the standard case, but beyond that, although there can be little doubt that capital-saving inventions occur, they are not easily identified…’

Despite the weight of authority supporting these inferences, they are totally without foundation.