• No results found

Discussion

Chapter 2 Theoretical Approaches to the Economics of Innovation Economics of Innovation

2.7 Discussion

2.7.1 Criticism of Neo-Classical Approaches

The central concern of mainstream neo-classical economics is the interaction of self-interested, intelligent, individuals. Most neo-classical theory further assumes that “agents are not only intelligent, they maximize… And when they interact they achieve an equilibrium in which each individual is doing the best he can given what all others are doing” (Krugman, 1996, p. 2).

Individuals who were not willing to proceed within the constraints of these core assumptions have developed alternative approaches, criticism being focussed particularly on the ideas of maximization and equilibrium. Evolutionary theorists aim to drop the assumption that individuals maximize. They try instead to represent decision making as a process of groping through alternatives in which it may take a long time to get to a maximum and where the maximum reached may well be local rather than global. They also try to get away from the assumption of equilibrium, to an approach where the economy is always evolving. Some of these criticisms have been fleshed out by Metcalfe: “the [neo-classical] framework is lifeless, which makes it hard to incorporate innovation as the driving dynamic force in the competitive process” (1995a, p. 446); and the assumption of symmetric behaviour by identical firms is perceived to be unacceptable because:

“…innovation and information asymmetries are inseparable and thus innovation and Pareto optimality are fundamentally incompatible” (1995b, p. 413).

However acceptance of the absolute truth of these neo-classical assumptions is certainly not a pre-requisite for those who choose to work within that tradition; for example Kenneth Arrow pointed out problems with the assumption that all individuals have the same utility function; “People just do not maximize on a selfish basis every minute. In fact, the system would not work if they did” (1987, p. 233). While Paul Krugman explained that his research has taken him to the edge of the neo-classical paradigm:

When you are concerned, as I have been, with situations in which increasing returns are crucial; you must drop the assumption of perfect competition; you are also forced to abandon the belief that market outcomes are necessarily optimal, or indeed that the market can be said to maximize anything. You can still believe in maximizing individuals and

some kinds of equilibrium, but the complexity of situations in which your imaginary agents find themselves often obliges you - and presumably them – to represent their behavior by some kind of ad hoc rule rather than as the outcome of a carefully specified maximum problem. And you are often driven by sheer force of modeling necessity to think of the economy as having at least vaguely "evolutionary" dynamics, in which initial conditions and accidents along the way may determine where you end up (Krugman, 1996, p. 1).

Seen in this light, economists in the neo-classical tradition are those who prefer to make sense of the world using models based on the simplifying assumptions of maximization and equilibrium. Those who believe that knowledge can be advanced further outside these constraints use alternative traditions such as the evolutionary approach.

Neoclassical economists see individual and organizational action as largely the consequence of optimizing choices. Institutional and evolutionary economists see action as largely the consequence of following habits or customs…, or rules of thumb …, or routines … appropriate to the particular decision context (Nelson, 1994, p. 249).

A central criticism of the neo-classical growth models of the 1950s and 1960s (section 2.3.1) is the assumption that technological progress occurs in an unexplained (exogenous) manner. Solow’s model cannot be used to determine the causes of technical change because it is treated as being exogenous. It is therefore of little use in studying policy issues such as whether market allocation of resources to R&D is optimal, or the more fundamental question of what explains long-run growth:

the long-run per capita growth rate is determined entirely by an element – the rate of technological progress - that comes from outside the model…

Thus we end up with a model that explains everything but long-run growth, obviously an unsatisfactory situation (Barro, 1997, p. 4).

Dissatisfaction with this exogenous treatment of technological change led to the development of alternative approaches; endogenous and new growth theories (that sought to endogenise technical change within the neo-classical paradigm); and evolutionary and institutional approaches.

Another problem with early neo-classical approaches was that R&D is not a profitable activity in a world where knowledge is a pure public good and where perfect competition prevents firms that invest in R&D from extracting any extra profit. In order to make R&D profitable new growth theory assumes that at least a

PhD Draft 2004 Rev3 Aug 27 Final.doc 30-Aug-04 1:04 PM

portion of benefits are appropriable and markets are imperfectly competitive to allow for profit from innovation.

However Nelson (1997) questioned the newness of new growth theory suggesting that all of these ‘new’ assumptions had been well documented in the empirical literature for some years and were included in a review article by Abramovitz (1952). Nelson went on to ask why some findings from the empirical literature were included in the new models while others that are equally important were ignored. He was particularly concerned with the unsatisfactory treatment of uncertainty and institutions:

…virtually all detailed empirical inquiry into major technological advance has highlighted the inability of the actors involved to foresee the path of development, even in broad outline, or the major surprises that occurred along the path. In contrast the new models assume perfect foresight or if they admit less than that, that uncertainty about the future can be treated in terms of a well and correctly specified probability distribution of possible future events (1997, p. 31).

The new neo-classical growth models treat firms in a highly simplified way and barely address institutions aside from the “competitive” (or monopolistically competitive market (p. 32).

Systems approaches respond directly to Nelson’s concerns. The ‘state of the art’

in this field of research is discussed in the next section, followed by some conclusions regarding the theoretical approach adopted in this thesis.

2.7.2 Systems of Innovation; the ‘State of the Art’

Studies of national and sectoral systems of innovation have yielded many valuable insights. It is undoubtedly true that innovation typically involves interactions with a range of actors and that it is affected by institutional and organisational settings.

Studies using the systems perspective have greatly increased our understanding of the similarities and differences across countries in their innovation systems and have contributed significantly to policy making. But their findings are often country or sector specific and subject to change over time. It may also be asked whether the systems approach was always necessary for achievement of those

PhD Draft 2004 Rev3 Aug 27 Final.doc 30-Aug-04 1:04 PM

insights except in so far as they employ historical, holistic and interdisciplinary perspectives24.

The difficulty with the national innovation system (NIS) concept as a theoretical framework seems to lie in its generality and inclusiveness. Few would argue with the proposition that innovation is affected by “the links and relationships among industry, government and academia in the development of science and technology” (OECD, 1997b p. 11). But there seems to have been little progress in developing a set of general principles relating innovation to a set of specific characteristics of institutions and their links. Nor has much progress been made in the development of general methods for measuring and analysing systems of innovation (SI), or in the synthesis of general conclusions about national systems of innovation.

The literature often seems to be stuck ‘at first base’; still grappling with basic issues of definition and measurement. Karaömerlioglu (1998) and Carlsson et al (1999) amongst others have made practical suggestions on how systems of innovation should be defined but this has not led to the development of a more standardised SI methodology. Patel and Pavitt’s (1994) list of indicators (section 2.6.5.1) has been widely used, but the literature to date has contained little explicit discussion of the function of each system, or of what constitutes system inputs and outputs. As a result little progress has been made in measurement of the performance of the total system, which should be the main focus (Carlsson et al., 1999).

Radosevic suggests that SI research has entered a temporary phase of diminishing returns where “empirical research has not resulted in added theoretical insights and theoretical research seems to have been incorporated into the broad concept of institutions” (1998, p. 76). He suggests that institutional diversity and lack of a common conceptual framework are part of the problem, with the result that presentations of systems of innovation are never exhaustive and directly

PhD Draft 2004 Rev3 Aug 27 Final.doc 30-Aug-04 1:04 PM

24 These are two of the nine characteristics of the SI approach identified by Edquist and Hommen (1999, p. 65).

comparable. Further comment on the current state of SI work is provided by Cooke:

given the demands of testing [the talisman concept of NIS] from an interactive rather than a linear perspective, it is perhaps not surprising that the main international study reported by Nelson (1993b) is weak on generalisable findings, whereas the main theoretical study edited by Lundvall (1992b) is almost entirely lacking in empirical content (Cooke et al., 1998).

Various authors have made suggestions on how to break the impasse. Bell and Albu (1999) and Lundvall and Christensen (1999, p. 2) agree that research should shift from studying production systems to knowledge systems. This would bring empirical research more into line with systems theory but will tend to make quantification more difficult since knowledge systems are much harder to measure than production systems.

An alternative way forward has been to break national systems down into more manageable proportions. This approach has had some success and Cooke et al.

(1998, p. 1565) comment that “where theoretical and empirical work are well integrated the focus tends to be sectoral”. Work by Carlsson (1997; 1991; 1995) is extensively cited and highly praised, but is not immune from many of the problems which have been associated with the national SI approach.

Zysman believes the answer comes from quite a different quarter: “… the answer does not come from within economics, but rather from historical and institutionally oriented political science” (1996, p. 662). Zysman explains economic and technological development through a painstaking study of the historical, political and institutional developments in each country. He criticises the narrow SI concentration on elements of science and technology influence, although his arguments may be seen as consistent with a broad SI definition.25 Unfortunately “historical institutional analysis has low predictive power and it is difficult to generate much theoretical insight from it”(Radosevic, 1998, p. 79) because of the specificity of particular historical conditions.

PhD Draft 2004 Rev3 Aug 27 Final.doc 30-Aug-04 1:04 PM

25 e.g. Lundvall’s (1992a) definition; see Table 2.2.

While a set of general principles may, in fact never be developed Metcalfe suggests that the institutional approach will continue to provide essential insights which are different to the market failure approach to technology policy.

“Fundamentally, the effective operation of an evolutionary innovation system depends on the effective coupling between firms with other knowledge-based institutions to jointly enhance processes of learning and creativity” (1995a, p.

497).

Krugman and Romer respond that evolutionary theorists must try to formalise their understanding into simple models. If they do not, then they run the risk of being bypassed. Krugman draws a parallel with the ‘high development theorists’26 of the 1950s; rather than compromise their insights so that they could be put into the available simple models, the high development theorists explicitly rejected the tendency in the profession towards rigor and formal modelling (Romer, 1993, p.

552). “Along with some others, notably Myrdal, Hirschman didn't wait for intellectual exile: he proudly gathered up his followers and led them into the wilderness himself. Unfortunately, they perished there” (Krugman, 1994).

PhD Draft 2004 Rev3 Aug 27 Final.doc 30-Aug-04 1:04 PM

2.8

Conclusions

In the preceding sections of this chapter, the main theoretical approaches to the economics of innovation have been described and reviewed. Each of these approaches has its advantages and its faults. The approach taken in this thesis does not fall under any single framework since, as Stoneman has pointed out: “the choice of framework is [often] a matter of ‘horses for courses’ and the most appropriate framework depends crucially on the questions being asked” (1995, p.

9).

Overall, the approach draws strongly on the evolutionary school – as summarised by Giovanni Dosi in his 1988 review in the Journal of Economic Literature27:

26 “…high development theory can be described as the view that development is a virtuous circle driven by external economies -- that is, that modernization breeds modernization. Some countries

… remain underdeveloped because they have failed to get this virtuous circle going. (Krugman, 1994)

27 See appendix 2 for an explanation of Dosi’s terminology.

In the new view, appropriability, partial tacitness; uncertainty; variety of knowledge bases; search procedures, and opportunities; cumulativeness and irreversibility … have been recognised as general features of technological progress. Relatedly, the endogenous nature of market structures associated with the dynamics of innovation, the asymmetries among firms in technological capabilities, various phenomena of nonconvexity, history of dependence, dynamic increasing returns, and the evolutionary nature of innovation/diffusion processes are some of the main elements of technological change (Dosi, 1988, p. 1164).

Theorists operating outside the mainstream have produced a wealth of appreciative theory28 and qualitative description; work that has the distinct advantage that it has the ring of truth to it. However the disadvantages of departure from the neo-classical paradigm are well recognised. Evolutionary and SI theorists have not been able to distil their understanding into simple models and even though they have a framework that denies the assumptions of maximization and equilibrium often make use of these assumptions as modelling devices that allow them to cut through complexity29. Work in the evolutionary tradition differs from mainstream analysis because it places ideas or discoveries at the centre of the framework. The difficulty with this approach is that ideas have not proved susceptible to mathematical description; indeed as Metcalfe noted “the idea of a production function for knowledge is not going to be very useful because ideas are not commensurable” (2000).

Acceptance of the simplifying assumptions of the neo classical framework does not imply that they are wholly (or even partially) correct just that this is a useful way of simplifying the complex world and moving forward. The neo-classical framework is most developed and has the advantage of general principles e.g.

Pareto criteria which are not available under the evolutionary approach. However there are occasions where research questions can best be framed within an evolutionary framework; particularly those that relate to systems of innovation.

The SI approach provides a more holistic way of looking at the world than the traditional neo-classical approach. While progress has been slow in developing

PhD Draft 2004 Rev3 Aug 27 Final.doc 30-Aug-04 1:04 PM

28 “Nelson and Winter introduced the term appreciative theory to describe theory that is less abstract, more descriptive and closer to practice and the real world context”. (Romer, 1993, p. 554)

ways of measuring and comparing systems this does not cast doubt on the validity of the concept. What seems to be required is more integration between the systems approach and other research into the economics of innovation. The SI concept and evolutionary theory on which it is based are useful supplements to neo-classical theory. They provide an alternative framework for thinking about innovation and encompass many of the features that seem to be essential to a full explanation of the economics of innovation. For these reasons the empirical research described in this thesis has been conducted mostly within the neo-classical framework but draws freely on the knowledge gained through other traditions whenever this is appropriate.

PhD Draft 2004 Rev3 Aug 27 Final.doc 30-Aug-04 1:04 PM

29 Krugman (1996) comments that “those who try to do economics without these organizing devices have a propensity to produce sheer nonsense”.