Economists of all descriptions have accepted that new products and new processes are the main source of dynamism in capitalist development. But relatively few have stopped to examine in depth the origins of such innovations or the consequences of their adoption. Most have preferred, in Rosenberg’s (1982) apt description, not to look ‘inside the black box’, but to leave that task to technologists and historians, preferring to concentrate their own efforts on ‘ceteris paribus’ models, which relegate technical and institutional change to the role of exogenous variables.

The classical economists were generally more ready to look inside the black box; Adam Smith and Marx in particular both showed a deep interest in the relationship between scientific research, technical innovation and the market. Smith (1776), pointed already in the 18th century to the growth of specialization in scientific research and to the links between innovation in the machine-building industries and scientists (‘philosophers’ or men of ‘speculation’ whose task is ‘to observe everything’). Marx and Engels (1848) probably more than any other economist assigned to technical innovation the driving force in economic development and competition – ‘the bourgeoisie cannot exist without constantly revolutionizing the means of production.’

But in the first half of the 20th century Schumpeter was almost alone among leading economists in following and developing this classical tradition. Consequently those economists such as Nelson (1977, 1982) and Rosenberg (1976, 1982) who have concentrated much of their attention on the economics of innovation are often referred to as ‘Schumpeterian’ or ‘neo-Schumpetarian’, even though their ideas may considerably diverge on many topics.

It is to Schumpeter that we owe the threefold distinction between invention, innovation and diffusion of innovations, which has now become the generally accepted convention in analysis of technical change. Invention is generally defined as a novel idea, sketch or model for a new or improved product, process or system. It need not necessarily imply any empirical test of feasibility or prototype experience, but as Jewkes (1958) suggests, it usually does convey the first belief that something should work and often the first rough test that it will in fact work.

Nevertheless, Schumpeter was right to stress the distinction between invention and innovation. There is an enormous difference between ‘working’ under laboratory conditions and working under commercial conditions. Schumpeter used the expression ‘innovation’ to connote the first introduction of a new product, process, method or system into the economy. (This is generally taken to include military or health care applications as well as the more purely commercial innovations.) As Schumpeter pointed out, there is many a slip between cup and lip in the development of an invention to the point of commercial introduction. Problems in scaling up from laboratory scale to works scale lead to the demise of many apparently sound ideas and unanticipated ‘bugs’ are the rule rather than the exception in the exploitation of inventions. Many (perhaps most) inventions are patented, but most patents are never actually used commercially except perhaps as bargaining counters.

Some ambiguity still surrounds the definition of ‘innovation’, since the word is used both to indicate the date of first introduction of a new product or process (e.g. the float glass process was innovated in 1958) and to describe the whole process of taking an invention or set of inventions to the point of commercial introduction, as in ‘management of innovation’ – a process which may take many years of development work, trial production and marketing.

In fact the date of launch of an innovation is seldom as precise as might appear at first glance, since false starts and modifications to the design of a radical new product or process are commonplace. Thus many different dates can be found for the innovation of well-known products, such as the radio or the electronic computer. National bias plays a part too, as well as definitional problems.

This point is an important one when we come to consider the third aspect of technical change in the Schumpeterian framework – the diffusion of innovations. Although almost all economists would agree that the diffusion of innovations through a population of potential adopters is crucial for the achievement of productivity gains and successful competitive performance more generally, they would also agree with Rosenberg (1976), that the product or process which is being diffused is itself usually subject to further change during the diffusion process. Indeed, this has been one of the main criticisms of some studies of diffusion in the 1960s and 1970s (Metcalfe 1981) which tended to make the static assumptions of an unchanged product diffusing through an unchanged environment. Nevertheless, this does not invalidate Schumpeter’s analytical distinction, which has proved extremely fruitful both in theoretical and empirical work, as shown notably in the major international conference on diffusion of innovations in Venice in 1986.

When Jewkes and his colleagues (1958) made their original study of the sources of invention, they rightly complained that economists had made very little contribution to the study of invention and innovation, and Rogers (1962) could legitimately make a very similar complaint, in relation to the study of diffusion of innovations. However, in the next quarter of a century the picture changed considerably. Following the impetus given especially by Mansfield (1968, 1977) numerous empirical studies in Europe, America and Japan covered much of the territory which Schumpeter sketched out in a preliminary way. Unknown and uncharted territory still remains, however, and its exploration is by no means straightforward (Dosi 1985).

Thus, for example, we now know a good deal about the conditions surrounding success and failure in the competitive struggle of private firms to innovate, but far less is known about the types of government policies which are most likely to encourage innovators and promote their success. The study of the latter is inhibited by the difficulty of isolating any specific single measure, such as a tax incentive, development subsidy or procurement initiative from other more general influences on the behaviour of the firm and numerous factors specific to individual firms (Rothwell and Zegveld 1981).

In the analysis of competitive attempts by individual firms to innovate the problems of multiple causality has been partly overcome by the use of statistical techniques in paired comparisons of success and failure, as for example in project SAPPHO (Freeman 1982; Rothwell et al. 1974) and similar studies in several countries (e.g. Szakasits 1974). By and large these studies agree in highlighting the main factors leading to successful innovation performance: the depth of understanding of the needs of potential users of the innovations and the steps taken to obtain this knowledge (external communications network): the research and development capability to eliminate or minimize ‘bugs’ prior to launch of the innovation; internal communications adequate to ensure effective links between those responsible for R&D, marketing and production within the firm; entrepreneurs or ‘business innovators’ with the status and experience to ensure the necessary mobilisation and coordination of resources within the firm. Studies of failure have been particularly illuminating in demonstrating the tendency of some technical innovators to neglect user needs and the lack of communication between various departments in some large firms (Burns and Stalker 1961). However, they also show that even in cases, when firms appear to follow all the ‘rules’ and ‘best practices’ which lead to good innovation performance, technical and market uncertainties may frustrate their best efforts.

Indeed, the empirical studies of the management of innovation and firm behaviour have undermined the traditional neoclassical theory of the firm. Imperfect information, uncertainty, complex institutional linkages, cumulative in-house technology, and searching modes of behaviour are characteristic of innovation, rather than the tidy, rational, optimizing calculations and perfect foresight postulated by neoclassical theory (Dosi 1984, 1985). For this reason contributors to innovation studies have also made major new contributions to a revised theory of firm behaviour, which take into account the findings of the stream of empirical research (Nelson and Winter 1982; Dosi 1984).

Less clear-cut conclusions have emerged with respect to the influence of size and concentration on innovative performance. Schumpeter (1928, 1942) is often known for his emphasis on the advantages of large size and monopoly on innovative performance, whilst traditional theory has continued to stress the advantages of competitive market structures. Clearly large size can facilitate innovative efforts in areas where development costs are unavoidably high because of number and complexity of components, as for example in spacecraft, nuclear reactors, or electronic telephone exchanges. The R&D threshold entry barriers in such areas can sometimes be so high as to limit effective competition to only a few large organizations throughout the world; and often innovation costs are partly met by state subsidies.

Even those economists, such as Jewkes et al. (1958), who have stressed the role of individual inventors and small firms at the stage of invention, have accepted that often development costs are so high that large firms tend to predominate when it comes to innovation. Many of the case studies described by Jewkes et al. illustrate this point, since the small firms or individuals who initiated the inventive work were often obliged to seek the help of larger organizations or were taken over by them before they could launch the new product or process on the market.

However, revolutionary advances in technology, for example the micro-chip, can sometimes lower entry barriers dramatically. In those areas where smaller firms can afford the entry costs they appear to perform relatively well in competition with larger firms. Thus the SAPPHO project did not show size as a variable which discriminated systematically between success and failure.

Schumpeter (1912, 1928) had himself recognized the advantages of new small innovator–entrepreneurial firms, but believed that the general trend of capitalist development and the rising costs of in-house R&D would lead increasingly to the management of innovation by larger bureaucratic organizations. Galbraith (1972) developed this notion of the ‘technostructure’ in large firms in his ‘New Industrial State’. However, empirical evidence suggest that small firms have continued to maintain, or increase their share of innovations, even though large firms do indeed now account for more than two-thirds of R&D and of all innovations (Townsend et al. 1982). The share of small firms in innovations is apparently greater than their share of R&D expenditures, and this phenomenon has been explained partly in terms of motivation and good internal communications leading to greater efficiency in the conduct of R&D, and partly in terms of the ‘spin-off’ of technical innovators who have left large government, industrial or academic laboratories with the idea for an already partly developed product.

The debate continues but with increasingly general acceptance that both very large and new entrepreneurial firms enjoy advantages in distinct types of invention and at different stages of the evolution of new technologies. The previously observed tendency for R&D intensity to decline in the largest firms has been denied by Soete (1979), who maintains that more recent evidence supports the Schumpeterian hypothesis.

Schumpeter’s contention that technological competition was more important than price competition with invariant conditions of production has also found increasing confirmation from empirical and theoretical work in the sphere of international trade. Since Hufbauer’s (1966) original demonstration of the role of technical innovation in the explanation of patterns of international trade in synthetic materials, evidence has accumulated to confirm that ‘neo-technology’ theories have greater explanatory power in relation to international trade performance generally than the Heckscher–Ohlin factor proportions theory (Soete 1981).

The notion that cumulative patterns of advantage in know-how, skills and innovative capability may underlie some of the persistent differences in comparative international trade and productivity performance has also found confirmation in many national studies of innovation and economic development (e.g. Pavitt 1980). These suggest that institutional innovations in education and training systems, as well as in research institutes and organizations have historically played an important part in building up cumulative technological capability. Thus, for example, German strength since the late 19th century in the chemical and engineering industries has been related to the establishment of the ‘Technische Hochschulen’ and other new developments in German universities, as well as the establishment of in-house R&D in the leading German chemical and electrical firms. Similar arguments have been advanced with respect to Japanese industrial training and technological innovation systems and the more recent outstanding successes of the Japanese economy (Freeman 1983).

To sum up, empirical studies of innovations and their diffusion have provided mounting evidence that mainstream neoclassical theories of firm behaviour, competition, international trade and consumer behaviour are seriously deficient in their assumptions and conclusions. However, the ‘neo-Schumpeterian’ tradition in economics has only begun the task of substituting a more satisfactory theoretical foundation which would take both technical innovation and institutional factors fully into account (Dosi 1985).

See Also