A Survey of the Theory of Process-Innovations

The topic considered here is innovations, not inventions: the entrepreneur is viewed as facing a list of known but as yet unexploited inventions from which he may select. How this list is itself drawn up and continuously augmented is an issue that deserves separate treatment.' Any analysis of the rate at which techniques improve in an economy cannot ignore the pace and scope of inventive activity. But my subject is not what Kaldor has aptly called the degree of technical dynamism in an economy-the flow of new ideas and readiness of the system to absorb them-but rather the pattern of technological progress when the economy is in fact technically dynamic. Innovations fall into two classes: process-innovations and product-innovations. The terms are self-explanatory. The distinction is to some extent an artificial one: the introduction of a cost-reducing process is sometimes accompanied by a change in the product mix, while new products frequently require the development of new equipment. In practice the two are usually so interwoven that any distinction between them is arbitrary. Nevertheless, in principle, novel ways of making old goods can be distinguished from old ways of making novelties. Since the index-number problem has so far doomed all theoretical analysis of innovations which alter the quality of final output, the refusal to discriminate between product- and processinnovations would close the subject of technical progress to further analysis. A process-innovation is defined as any adopted improvement in technique which reduces average costs per unit of output despite the fact that input prices remain unchanged. The new technique may involve drastic alterations in equipment, but this is not a necessary feature of the definition: the mere reorganisation of a plant may be as factor-saving as the introduction of new machines. We have to guard ourselves against a widespread misunderstanding at this point. An innovation represents an addition to existing technical knowledge. Since the production function already takes account of the entire spectrum of known technical possibilities-known, in the sense of being practised somewhere in the system-innovating activity ought to denote the adoption of hitherto untried methods. But few indeed are the successfully adopted innovations which do not have a long history of unsuccessful trials, and even the imitation of previously tried techniques almost always involves a creative response . This difficulty has caused some authors to define technical progress as any change