How do Innovations affect Economic Growth? - Some Different Approaches in Economics

Transcription

1 Innovation Systems and European Integration (ISE) A research project funded by the Targeted Socio-Economic Research (TSER) program of the European Commission (DG XII) under the Fourth Framework Program, European Commission (Contract no. SOE1-CT , DG XII SOLS), coordinated by Professor Charles Edquist of the Systems of Innovation Research Program (SIRP) at Linköping University (Sweden). Sub-Project 3.1.2: Employment and Growth How do Innovations affect Economic Growth? - Some Different Approaches in Economics Submitted to the Commission: January 26, 1997 Birgitte Gregersen Björn Johnson IKE-Group Department of Business Studies Aalborg University Fibigerstraede 4 DK-9220 Aalborg Denmark bj@business.auc.dk and bg@business.auc.dk Abstract: This paper gives an overview of various approaches to how innovations affect economic growth: from standard macroeconomic equilibrium theory, growth accounting, new growth theories, catching-up theories, evolutionary theories based on theories of techno-economic paradigms and co-evolution to a systems of innovation approach. The paper focuses on approaches which sees growth as emanating from processes of interactive learning and on the relations between institutional and technological change. It is concluded that the main contribution from the systems of innovation approach to growth theory lies in its emphasis of the importance of institutions and institutional change and especially in the focus on interactions between institutional, organisational and technical change as the basic source of growth. Institutions and institutional change are at the very heart of the economic process and shape the whole process of growth. Without a thorough treatment of the institutional foundations of growth any growth theory is seriously flawed. For policy makers, who try to stimulate growth by supporting innovations, the focus should be on designing and implementing institutional changes which continuously supports technical and organisational learning and innovation, i.e. they should try to implement a more or less permanent process of institutional learning. Key-words: Growth Theory, Innovation, Institutional Change, National Systems of Innovation

2 2 Contents: 1. Introduction 3 2. Theories with a simplistic innovation concept A world in which innovations do not really matter The residual and transitional dynamics Why growth rates differ between periods and countries Growth accounting New Growth Theory 9 3. The phenomena of forging ahead, catching up and falling behind Theories with innovations at the centre Technological change as an evolutionary process Co-evolution of technology, organisations and institutions: 16 Techno-economic paradigms 4.3. Co-evolution between sub-systems: Matching and mis-matching Contributions from the systems of innovation approach 19 References 25

3 3 How do innovations affect economic growth? - Some different approaches in economics. 1. Introduction It is of course impossible to cover what economic theory has to say about how innovations affect economic growth in a short paper like the present one. It is a broad and difficult question which could very well be the topic of several state of the art articles of the Journal of Economic Literature kind. Therefore we do not have the ambition to cover the whole field and this papers should not be read as a survey article. There is a bias in both the selection and in the length of the discussion of different theories. Especially the approaches which we feel to be only weakly related to the systems of innovation approach to innovation analysis will be treated in a short and rather superficial way. Our purpose is to illustrate a development towards an approach which sees growth as emanating from processes of interactive learning and focuses on the relations between institutional and technological change. We want to raise the question of which role, if any, the innovation systems approach might have in economic theorising about growth and in relation to policies for growth. We will discuss a number of different approaches. We will start with theories which operate with a very simple and in our view questionable innovation concept, even if they may be quite sophisticated in other respects: (i) Standard macroeconomic equilibrium theory in which technical change is exogenous and all markets, including the labour market, clear. (Section 2.1.) (ii) Neo-classical growth theories, including new growth theories and growth accounting. (Section 2.2.) We will then go on to theories, which put emphasis on non-price factors and international relations and which have somewhat more refined concepts of technology and innovation: (iii) Catching -up growth theories. (Section 3) After this we will discuss approaches, which use the concept of institutions and institutional change and put the characteristics of technology and the process of innovation at the centre : (iv) Evolutionary theories in which the evolution is fuelled by technical change. (Section 4.1.) (v) Theories using the concept of techno-economic paradigms. (Section 4.2.) (vi) Theories about co-evolution between different subsystems in society. (Section 4.3.) Finally, we will discuss the usefulness of the concept of innovation systems and hint at some possible future contributions to the understanding of growth which come from the National Systems of Innovation and the Learning Economy approaches. (Section 5) Throughout the text we will define innovations on a general level as both the introduction into the economy and the diffusion in the economy of new knowledge or new combinations of existing knowledge. We do not try to identify sharp borders between introduction and diffusion of new knowledge. We will rather regard them as interconnected parts of an innovation process.

4 4 Within this general definition we will distinguish between product- and process innovations and we will also consider organisational innovations and, sometimes, institutional innovations. Our purpose is to find out how different theoretical approaches explain these kinds of innovations and how they treat their contribution to economic growth. Because innovations affect productivity it is then often necessary to distinguish between output growth and employment growth. We use the term economic growth to denote long term changes in production capacity of the economy as a whole as distinguished from short term changes in the degree of capacity utilisation. We will not go into a discussion of the many conceptual and measurement problem connected to economic growth. We simply assume that economic growth can be at least adequately indicated by long term trends in aggregate production figures from national account statistics. 2. Theories with a simplistic innovation concept 2.1. A world in which innovations do not really matter Important parts of the economic theory which deals with the questions of production and employment are formulated in ways which make technical innovations rather uninteresting. This is the case with standard, textbook, macroeconomic theory. We discuss this topic here as a kind of extreme reference point; innovations are completely discrete and exogenous and time does not enter the analysis so that there is no innovation process, only unexplained technical jumps. In this kind of economy technology, together with resources, population, tastes and institutions, are taken as given in order to create a frame of analysis, which permits stringent short run analysis. Changes in technology are looked upon simply as changes in the background data of the analysis. They are totally exogenous. This means that technical change can throw the system temporarily out of equilibrium, but it will not affect its basic equilibrating forces. The mechanisms of technical change and its connection to the state, structure and change of the economy are not on the theoretical agenda. Consider, for example, the following standard text-book 1 diagram: 1 See Begg, Fischer and Dornbusch (1994), chapter 26.

5 5 Figure 1: Technical change as a shifts in the long run aggregate supply curve. price level las e las' sas e' sas' mds Y Y' output The macroeconomic demand schedule (mds) shows the combinations of the price level and real income/real production at which planned investment and consumption spending equals actual production at the same time as the money market is in equilibrium. The long run aggregate supply schedule (las) shows the output that firms wish to supply at each price level. It is derived from an assumption of labour market equilibrium in the sense that there is only voluntary unemployment (everybody who wants a job at the going real wage has got one). The short run aggregate supply schedule (sas) shows the prices charged by firms given the wages they pay. It is flatter than the long run aggregate supply schedule because of sticky wages. When wages adjust to labour market pressures the short run supply schedule shifts. In point e the long run aggregate supply schedule (las) crosses the macroeconomic demand schedule (mds) and the economy is in long run (and short run) macroeconomic equilibrium. Within this framework changes in monetary and fiscal policy, population, tastes, resources, and technologies will show up as shifts of the demand and supply schedules. A technological innovation, for example, increases the production potential at full employment and may be illustrated by a shift of the long run aggregate supply schedule from las to las. A process of adjustment, which eventually will bring the economy to new long run equilibrium in e, starts. Competition induces firms to lower prices in order to utilise the increased production potential. This increases real wages and leads to an excess supply of labour which presses money wages down. The short run supply curve shifts downwards. Lower wages and prices with constant nominal money supply means increased real money quantity and lower interest rates. Real consumer and investment demand increases and the economy moves down the mds schedule. This process continues until point e is reached. We have now got more production and lower prices, while the labour market, the money market and the commodity market is again in equilibrium. If the government considers the adjustment process to be too slow they may shorten it by, for example, a fiscal policy that shifts the mds curve to the right. Within this framework innovations can not be a part of the picture. It is an example of comparative static economic analysis into which a process as technical change does not easily fit In some presentations however, innovations are assumed to be a possible factor behind the trend growth of economic output. In long run equilibrium output is supposed to

6 6 be on its trend level and the horizontal axis measures deviations from the trend output. This does not change the totally exogenous character of innovation in the analysis, however. Furthermore, since we are in a totally aggregated, one-commodity world, we can not take product innovation into account: There are by definition only process innovations. Of course macro economists are well aware of the existence of product innovations, but these are only theoretically relevant to the extent that they affect the production capability of the economy or the price level. The distinction between product and process innovations has no theoretical role to play in a totally aggregated analysis. Process innovations do not include organisational innovations in this framework. Or rather, the concept has no meaning. Full employment together with the capital stock uniquely determine the potential output and in such a macroeconomic production function approach there is no room for organisational factors. A firm is conceptualised as a production function, which is a technical input-output relation. However, it would be unfair not to mention that institutional innovations (understood as changes in the rules of the game ) are important in the macroeconomic policy discussion. Different institutional specifications of the labour market between Classical and Keynesian theories affect wage and price formation and, accordingly, how the economy adjusts to exogenous disturbances and changes of economic policy. The supply side policy discussion within macroeconomic theory is to a large extent about institutional innovations. Still, the institutional specification is rather crude and institutional changes are exogenous to the analysis. On the whole then, the conclusions are simple and reassuring: Technological change increases the potential output and there are reasons to believe that this potential will be utilised in the long run. Economic growth is stimulated and there is only voluntary (or natural) unemployment. If there is productivity growth due to technical change, output will eventually also grow so as to keep employment on its full employment level. If new technology is introduced to save labour in one part of the economy, wages and prices are reduced all over the economy so as to generate compensating demand for labour. In relation to innovation and growth there are no policy implications from this kind of theory; these phenomena are not on the agenda. Of course, macro economic equilibrium theory can be more sophisticated than this, but on the whole there is no room for innovations in it. It seems to be doing quite well without it and we have to turn to other kinds of theory in order to see how they may be introduced into economic analysis The residual and transitional dynamics. In the last section it was concluded that in macroeconomic theory technological innovations are exogenous but have the potential to increase output. It is only natural, then, to ask what the theories that explicitly deal with output growth have to say about innovations. Neoclassical growth theory is a suitable starting point for such a discussion. This theory (Solow, 1956) was based on standard neo-classical assumptions 2 and exogenous population growth and predicted that aggregate output in long run equilibrium would grow at the same rate as the capital stock and the population. This means that income per capita will not increase in the long run. This was obviously in conflict with empirical facts. Another empirical problem with the theory was that if the capital stock grows faster than the labour force you would expect, due to the character of the production function, the return on capital 2 Such as perfect competition, no externalities or public goods, decreasing marginal productivities and a homogenous macroeconomic production function of degree one.

7 7 investments to decrease in the long run. You would also expect capital investments to be more profitable in poor countries than in rich ones. None of these expectations are born out by the facts. Technological progress may be the solution to these puzzles, however, and Solow added this in the form of an exogenous term. The production function, then, got the following form: Q = L where Q is total output, K is the capital stock, L is the labour force and A is the exogenously determined technological factor. Usually A has been interpreted as a collective good which is freely floating between firms and countries and gives the same contribution to output growth everywhere. Differentiation gives: Q = (K, L) + LL Dividing with Q = A gives Q/Q = A/A + ( LL)/A which can be rewritten as: But Q/Q = A/A + ( K/K LL)/ L/L ( where is the capital share of income. If there are constant returns to scale the labour share is: LL)/ = 1- As a result the economic growth rate can be written as: Q/Q = A/A + K/K + (1-)L/L This means that production growth is determined by technical improvements and a weighted sum of the factor growth rates, where the income shares of capital and labour are used as weights. Technical change is now regarded as the main source of economic growth. Faster technical change leads to faster output growth and we are still in a neo-classical full employment world. There is no reason to expect long run employment problems. Since technical knowledge is supposed to move freely and instantly across national borders it should not lead to increasing income differences between countries. Furthermore, we are still in a one-commodity world in which there are no product innovations. Organisational innovations are obscured by the production function approach and the institutional specification of the growth process is almost non-existent. There are no clear policy conclusions of this theory since growth in income per capita in the long run depends only on exogenous technical change. It seems to be a good idea to stimulate technical improvements,

8 8 but there are no indications of how to do this. Obviously, this can only be the starting point for a serious analysis of the role of innovations in economic growth Why growth rates differ between periods and countries: Growth accounting and new growth theory The term A has served as such a starting point. It is called total factor productivity growth and may be written as: A/A = (Q/Q - K/K) + (1-)(Q/Q - L/L) which means that total factor productivity growth is a weighted average of the growth rates of the average factor productivities (Q/K and Q/L) This is the famous residual which is often supposed to reflect technical change and has been called a measure of our ignorance (Abramowits, 1956), because several studies have shown that growth of capital and labour do not explain very much of output growth (Denison, 1962, 1967). To find out what is behind output growth the residual has somehow to be reduced and its content has to be identified Growth accounting Two main roads have been followed to decrease the residual and get a better explanation of economic growth and the differences in economic growth between countries and periods (Fagerberg, 1995). One road is to augment the factor inputs themselves by adjusting them for shifts in structure and quality. The other is to add supplementary factors, like education, resource allocation and government regulation, as explanatory variables. Both roads have led to the explanation of a larger part of economic growth. 3 Important insights have been achieved about which factors that affect economic growth. Still, however, the role of technical change in economic growth is analysed in a rather crude way and we have not left the one-commodity world. Besides there are some methodological problems with these still basically neo-classical growth models. It is assumed that countries are on the frontiers of their production functions and output is as great as is possible under given resources and technologies. It is a property of the model that all explaining factors are independent variables. It is implicitly assumed that capital, labour and technical change have separate and additive effects on economic growth. If, for example, the measured impact on output growth from capital accumulation is 2% p. a., a reduction of capital accumulation to half of its previous level is supposed to reduce output growth with 1 % p.a. However, the growth sources feed from each other (Abramovitz, 1989, p. 23). There are, for example, obvious interdependencies between technological change and capital investments as well as between technological change and investment in human capital. This makes it impossible to isolate the contributions to growth from specific factors and it is not easy to discuss causality within this model. To understand growth it is necessary to understand the interdependencies between the different sources of growth. 3 Maddison (1987) uses the following variables: quality of capital, quality of labour, structural change, the catch-up bonus, foreign trade effect, economies of scale, the energy effect, labourhoarding/dishoarding, use of capacity effect, regulation and crime. This makes it possible for him to explain up to 75% of economic growth for some countries in the period

9 9 To sum up: Economic growth accounting tries to pinpoint and measure the factors behind economic growth in order to make it possible to explain differences in growth rates between countries and periods. The role of the black box of technological innovations in economic growth is in comparison to the basic neo-classical model somewhat reduced but still very important; it has not been possible to cut the whole residual up into specific growth contributing factors and the many important interdependencies between them are left in the darkness. So, innovations are still, in a way, miracles. They are not explained in the model and there are no interconnections between technical and economic change. There is no distinction between product and process innovations and neither organisational nor institutional factors are taken aboard. It is still essentially a neo-classical, full employment, long run equilibrium growth approach. The model has nothing specific to say about employment, which can only be treated indirectly by relating it to output growth. From a policy perspective it is important to observe that growth is no longer produced by an anonymous and mysterious residual, but from empirically and politically meaningful factors like quality of labour, economies of scale, foreign trade, etc New growth theory The poor empirical performance of standard, neo-classical growth theory has led to efforts, within neo-classical theory itself, to make growth less exogenous. Growth should be explained within the theory and not follow mainly from exogenous technical change. The theory should be able to include the central idea that capital accumulation leads to technical change and technical change provides incentives for capital investments. In the first version of the so called new growth theory Romer (1986), following both Kaldor and Arrow, assumed that capital accumulation gives rise to technical change which, thus, follows as a byproduct of other economic activities. 4 These learning effects of the investment activities of the individual firm can not be appropriated by the investing firm itself; learning is external to the firm. The economy as a whole, however, benefits through external effects and the productivity of all firms increases. The tendency towards decreasing returns to capital accumulation is counteracted and long-run increasing per capita income may occur. In this way neo-classical growth theory becomes consistent with empirical facts. In a later version (Romer, 1990) technical change has become more explicitly endogenised. There is now a knowledge producing sector within the economy, which supplies other sectors with new technology. Innovations are the results of deliberate efforts by firms within an optimisation framework. Technical change has been included in the neo-classical sphere of rational-choice. Imperfect competition allows firms to appropriate results of their investments in new knowledge. There is no instaneous diffusion of new knowledge preventing firms to cover the costs of their investments in new technology. At the same time, however, diminishing returns to investment in new technology at the firm level are counteracted by spillover effects so that the productivity of subsequent innovation projects are raised all over the economy. New technology is, thus, supposed to have both a private and a public side in contrast to the earlier model in which it is a pure externality. In this model the rate of growth of an economy depends on how much it invests in new technology, typically in the form of R&D and on how appropriable new technologies are. Furthermore, if technological spillovers are concentrated within the home country and international spillovers are limited and if investments in R&D are different from country to country then there need not be any tendency towards long-run equalisation of growth rates. The policy 4 For short overviews see for example Fagerberg (1994, 1995, 1996) or Verspargen (1992)

10 10 conclusion is straight forward: inducing firms to invest in research, development and innovation may raise economic growth enduringly. With this theoretical development the neo-classical analysis of the role of technology in economic growth has been much improved: Technology is no longer a totally exogenous public good (manna from heaven for all firms) but is produced within the economy and has both private and public aspects. But we are still basically in a full employment world and it is only fair to observe that the treatment of the role of new knowledge in the economy is crude. Which different kinds of knowledge are important? How does investment in knowledge differ from capital investment? Where, how and why is new knowledge produced, diffused and stored? Would it, for example, be more relevant to think of a learning economy as a whole than of a learning sector of the economy? Which factors and relationships may account for differences between countries and periods in how new knowledge is developed and how it affects economic processes? Questions like these are not raised within the neoclassical tradition of growth theory, maybe because they require a more disaggregated, institutional, historical and systemic approach than neo-classical models can cope with. 3. The phenomena of forging ahead, catching up and falling behind. It may be convenient to start the review of theories with more complex conceptualisations of technological change and innovation with catching-up theories of economic growth. In its simplest form the catch-up hypothesis says that the growth rate of productivity for a country over a period of time is negatively proportional to the level of productivity at the beginning of the period. 5 The reason for this would be international movements of capital and technology from high to low average labour productivity countries. This would lead to a long term process of convergence during which the laggard countries would grow faster than the leading countries. Formulated in this way the catch-up hypothesis is not very different from the case of transitional dynamics of standard neo-classical growth theory which, for a period of time, permits countries to grow at different rates because of different initial conditions. Accordingly, this crude version of the catching-up hypothesis is in conflict with empirical facts. The hypothesis becomes more convincing when applied only to a limited number of countries in a specific historic period. This may be illustrated in the following way: Figure 2: The catching up hypothesis 5 The catching up arguments have been develped by several scholars, for example Gerschenkron (1962), Gomulka (1971) and Abramowitz (1979, 1989, and 1994)

11 11 Growth of real GNP per capita catching up countries front countries long-term laggards Relative productivity gap If we consider the period after 1945 up to the beginning of the 1990 s, USA would represent the front and most of the industrialised countries (and for the later part of the period also the newly industrialising countries) would be catching-up countries, while underdeveloped countries would be long-term laggard countries failing to catch up. Interpreted in this way the catching-up hypothesis more or less stands up to empirical facts. (Abramovitz, 1989) Catching-up growth, then, is not an automatic process. This may be because new technology is not freely floating between countries. Abramovits (1989, 1994) has used the term social capability to refer to a country s ability to utilise a technological gap and catch up in relation to the technology leaders. Obviously it becomes important to find out what the social capability depends on. Abramovitz (1994) mentions the level of general education and technical competence, the institutions which support modern, large scale business and, broadly, the incentive system. The concept of social capability is important because it introduces organisational and institutional factors alongside technological factors into the analysis of economic growth. In spite of this, the catch-up literature has as yet not made much use of institutional theory in its growth analysis and it may be argued that the importance of the institutional factor has not been fully acknowledged. International transfer of technologies requires institutional adaptations and innovations and an element of international transfer of institutions tends to accompany economic growth (North, 1990, Dalum et al, 1992). The concept of social capability should not only refer to the imitation of techniques and products used abroad, but also taken to include the capability to adapt and make minor innovations and quality adjustments to the absorbed technologies and to make them compatible with local standards and tastes. It should also include a capability for organisational and institutional adaptations in order to get a better fit between the imported technologies and the existing institutional framework. Work organisation may have to be adjusted, employees might need education, communication patterns within and between firms may have to change, etc. Furthermore, it should include a capability to import or borrow suitable organisational forms and institutions from abroad. Changes in the education system, infrastructural reforms, financial innovations and so on, might be needed in order to absorb the new technologies productively. Finally, the broader concept of social capability should include the capability to adapt both the imported technologies and institutions to the prevailing over-all institutional and cultural framework of the country.

12 12 Looked upon in this way it is clear that processes of organisational and institutional change play important roles in economic growth. It is also clear that economic growth is an international process, which is difficult to understand in a model for a closed economy. The growth process is different between front countries and laggards and even if it may be much more difficult to forge ahead than to catch up, the latter is far from automatic. A technology gap only constitutes a rather abstract possibility for catch up. Both a social capability and a social effort in the form of investments, education, R&D, etc., are required for a lagging country in order to close a productivity gap by catch-up growth. Furthermore, catch-up by simple technical imitation is not possible. Innovations (technical, organisational and institutional) are also necessary. Fagerberg ( 1994) concludes that, empirically, there is general support for models which includes both the scope for catch-up, proxied for example by GDP per capita, and efforts to close the gap, proxied for example by investments and education. The catch-up literature generally emphasise the specific and context dependent character of the development and diffusion of technologies. There are both sectoral and institutional stamps on innovation processes and since countries differ both with respect to structure of production and institutional set-up, long lived differences in growth rates are to be expected. The analysis is explicitly historical in the sense that the development of new technology depends on existing technology, which can only be understood as formed by the historical development of the economy s specialisation pattern and institutional structure. An element of technological and institutional path dependency is introduced - both in the growth towards the front by lagging countries and in the inability to start a real catch-up process by many underdeveloped countries. In this approach the concept of innovation has become more important in the analysis of technical change. Since the production structure of a country matters for its growth process we are no longer in a one commodity world and it is opened up for a distinction between product and process innovations. Because processes of catching-up, falling behind and forging ahead may lead to different growth rates and growth patterns in different periods of time, the catch-up hypothesis may shed some light on both long-term changes in employment and on structural employment problems. The structure and growth of employment may be a question of social capability as well as social effort for example in terms of different kinds of investments. There is a quite general policy implication of this. Policies for growth should probably be multi-stringed and broadly based. There are no quick technological fixes. As we have seen social capability is about much more than a large number of engineers and social effort is about much more than investments in high-tech sectors. This is of course not very specific, but it is a step towards a new perspective on growth policies which may be helpful in the longer run. However, even if it is fair to say that the catch-up growth hypothesis has introduced a more sophisticated technology concept it is also true that it has raised far more questions than it has answered - especially about the character of the concept of social capability. Furthermore it focuses much more on catch-up growth than on the related processes of forging ahead and falling behind. This means that it concentrates on what is, relatively, easy to understand and circumvents the most difficult questions. (See section 8 below). Its main contribution may be that the organisational and institutional prerequisites for technology development and technology diffusion have been placed firmly on the scene. An important policy implications is that technological fixes will probably not do the trick. Technical change must be seen in relation to organisational and institutional factors.

13 13 4. Theories with innovations at the centre 4.1. Technological change as an evolutionary process The catch-up literature has emphasised that technical change is a process that largely has to be studied in a historical context. A next step might be to regard it as an evolutionary process. It may be fruitful to think of knowledge as evolving. Boulding (1978) has even suggested that something very much like knowledge is the answer to the question of what evolves? The idea that science and technology are evolving is not a new one. Many writers have noticed the similarity between mutations in biological evolution and new ideas, hypothesises and theories in science and they have noticed the presence of different selection mechanisms in relation to these and that the development of scientific knowledge (thus) is an uncertain and open-ended process which is impossible to foresee in any detail. It has also been observed that technology often develops through trial and error processes which have many evolutionary characteristics; they are to some extent blind, uncertain and open ended at the same time as there may be a certain amount of path dependency involved. (See Nelson, 1995, for an overview) Diversity and diversity creation are crucial factors in evolutionary processes. In some lines of growth theory the importance of the generation and existence of diversity was acknowledged rather early. It has for a long time been realised that aggregate income and productivity growth is connected to structural change in production. This connection puts the innovation process into focus. In Pasinetti s (1981) economic model, for example, increasing productivity due to technical advance tends to generate unemployment as consumer demand for different goods reach saturation levels. This stagnation tendency can be counterbalanced, however, by a product innovation process and a consumer learning process, which introduces new or improved products into the system and, thus, increases consumer demand. The economy depends on a diversity generating mechanism, which in broad terms is a kind of learning mechanism, to maintain long term growth. But Pasinetti was not the pioneer in this way of thinking. Kuznets was one of the first modern economists to underline, theoretically as well as empirically, the many structural changes associated with economic growth. He analysed primarily the shifts from agriculture to manufacturing and, later, services including government activities, but also shift from stagnation industries to growth industries on lower levels of aggregation, Kuznets (1971) related these structural shifts to the process of technical change and, thus, in his analysis of the over-all growth process a diversity generating mechanism is of central importance. Without the generation of new diversities, growth would soon come to an end, since the growth rates of particular products almost always are subject to retardation (Abramovitz, 1989). If one regards technological development as an evolutionary process, innovations become a central concept. Innovations help keeping the evolutionary process going by introducing new variety. This means that it becomes important to get a better understanding of what is actually going on when firms innovate. You need to distinguish between different kinds of innovations (product- and process innovations, incremental and radical innovations and so on) and you need to clarify the incentives, the sources and the methods of innovation. Organisational innovations are also important in this context. While it is widely recognised that organisational change may be both a consequence of technological change and a precondition for it, there is not much agreement about how good organisational learners

14 14 different kinds of firms are. For example, in which situations can different kinds of firms cope with radical organisational change and in which situations do organisational change mainly come from new firms? It must be noted here that also institutions, defined as sets of common habits, routines, established practices and rule which regulate the interactions between individuals and groups (Edquist and Johnson, 1996), can be thought of as evolving. The institutions of today have developed out of those of yesterday and even where deliberate institutional design by for example governments are involved institutions usually only change in small groping steps; the process is both path dependent and open ended and coincidence and chance play important roles. Even if everyone seems to agree that theories about institutional change are rather underdeveloped, some scholars regard institutions and institutional change as extremely important in the process of economic growth. For Douglass North (1981), for example, institutional change (mainly in terms of new property rights) is the main factor behind modern economic growth as such. Differences in institutional set-ups are the main reasons for differences in growth rates between nations. Mancur Olson (1996) also considers institutions and policies to be the main factors explaining why some countries stay poor and some grow rich. In an evolutionary perspective the importance of institutions and the connections between institutions and economic change become obvious. Institutions affect the creation, storing, distribution, use and destruction of knowledge since they shape the visions, the interactions and the decisions of economic agents; they form the processes of variety creation and selection in relation to technical and organisational change. Almost all learning processes are social and interactive rather than individual and knowledge is affected and changed through an institutionally embedded process in which the institutions themselves change as the result of the interactions leading to new knowledge. The institutions of the economy form the communications and interactions between people within as well as between organisations, and this feed the processes of technical, organisational and institutional change. Institutionally shaped interactive learning, inevitably, brings new resources into the economy as well as it destroys parts of its existing resources (Johnson, 1992). On a more general level institutions have been regarded as some kind of basic element in social evolution; they form selection environments for innovations and they store and transmit knowledge from one period to an other. Especially routines are often regarded as basic elements in both institutional and evolutionary economics. Furthermore, it is a kind of general assumption in evolutionary economic theory that evolutionary processes are embedded in institutions. It is not enough, however, to treat technological, organisational and institutional innovations as separate sources of economic growth. These sources of growth feed upon each other and you need to consider the connections between them. The idea of co-evolution between technology and business organisation is central in the writings of for example Alfred Chandler(1990). It is now a well established part of innovation theory that new technologies may require new organisational forms for their utilisation and that new ways of organising a business activity may affect technical learning and innovation in firms. It is also becoming increasingly clear that institutions and institutional change profoundly affect the interactions between organisation and technology. This is a main aspect of the idea of new technoeconomic paradigms to which we return in the next section.

15 15 It might be argued that the evolutionary argument rules out policy intervention into the process of economic growth: An evolutionary process is too open-ended, uncertain and subtle to be tinkered with by politicians. However, even if there may some insight behind this argument, there are still important roles for government policy-making in connection to a process of knowledge evolution and learning, for example in connection to education, R&D, access to knowledge, intellectual property rights, etc. This will be discussed at the end of the article. Taking aboard the idea of product and process technologies and organisational forms as something which evolve is of course in deep conflict with neo-classical growth theory, including new growth theory and growth accounting. Neo-classical growth theory views economic growth as a long term moving equilibrium and see technological change as a rather well behaved process, which agents are able to treat in a rational way. This is very different from the main thrust of evolutionary thinking. Growth models which depend on technological change viewed as an evolutionary process go back to Nelson and Winter (1974). In such models the actors are firms influenced by different kinds of routines. There are routines for standard production activities, for investments and for searching for better ways of producing. By such searching routines learning and innovation has been built into the firm. Firms which are better learners will grow relative to their competitors at the same time as profitable innovations tend to be imitated by other firms. Models of a collective of such firms is capable to generate the standard empirical macroeconomic facts which growth models generally are supposed to do: growing total output, labour productivity and real wages, growing capital intensity and constant capital rent. We are not aware of models which generate specific employment growth paths which differ from their output growth paths, however. Economic growth is still about production capacities and employment is a question of utilisation of this capacity. As it is in neo-classical growth models. At the same time these models are only generating evolutionary processes in a limited sense. Radical technical, organisational and institutional change do not occur and specialisation patterns remain relatively stable. Formal theory, it seems, is not yet capable of handling full fledged evolutionary process. We therefore now turn to some appreciative theorising ( Nelson, 1995) which take up the interrelationships between technological, organisational and institutional innovations The co-evolution of technology, organisations and institutions: Techno-economic paradigms. The concept of a new techno-economic paradigm has been developed by Carlotta Perez and Christopher Freeman (Perez, 1983, 1985; Freeman & Perez, 1988) to indicate a radical shift in the entire technological and institutional basis of the economy. Such shifts change the whole character of economic growth and lead to long waves in the economic development. Different eras (or regions of time as Kenneth Boulding (1991) called them) are dominated by different basic technology systems, which profoundly affects production and management throughout the economy. In each era it is possible to identify a key factor, which through elastic supply and falling prices deeply affects the economic process. It is also possible to identify common-sense guide-lines for both organisational and technical decisions, for government policies and for consumer behaviour. The key factor and the common sense guide-lines affect each other and lock each other into a specific development logic or growth

16 16 model. Because it is not technologies alone, but rather a constellations of technological, organisational and institutional factors, which form the growth model of a period the term techno-economic paradigm is used for such a period. A techno-economic paradigm introduces a bias into both technical and organisational innovations. This bias becomes increasingly embodied in capital equipment as well as in habits of thought and it tends to lock-out technical and organisational alternatives. However, after some time a techno-economic paradigm will run out of steam, mainly as a result of increasing tensions between technical change and the institutional and organisational set-up of the economy and unless and until a new techno-institutional fit is established, economic growth will remain slow. As example of techno-economic paradigms may be taken the fourth Kondratief long wave from the 1930s to the 1980s, sometimes called Fordism. The key factor was cheap energy, especially oil. Work organisation was predominantly hierarchical building on scientific management. Mass production of standardised products was an important organisational ideal and mass consumption of consumer durables was an important part of consumer ideology. Organised labour markets with centralised long term wage agreements and active macroeconomic stabilisation policies were other aspects of the Fordist growth model. It was a resource and energy intensive growth model, which was quite successful in terms of employment and in terms of output and productivity growth rates. The most important factor which broke the Fordist logic was the development of information and telecommunication technologies (the present techno-economic paradigm - the fifth Kondratief wave - may be called the age of micro-electronics and computer networks) but other factors were also involved: The oil-based energy and materials-intensive mass production system, which had been so successful in the post-war period and indeed during the war itself, began to bump against limits of various kinds. Dissatisfaction became widespread with working conditions in large mass production factories (as so evident in France in 1968), with the quality of many consumer durables, and the philosophy of consumerism itself, with the destruction of urban and rural amenities and the quality of life brought about by mass use of automobiles, the threats to wildlife from mass tourism and unrestrained commercialism, and the poor quality of mass media and mass education. (Freeman, 1992 a, p. 217) Analysis which builds on the concept of techno-economic paradigms is interesting from an evolutionary point of view, because it emphasises the co-evolution of technologies, organisational forms and institutions. Inherent in this way of thinking is that each long wave has its own specific characteristics which influence the way different types of innovation affects economic growth and employment. It follows that the connections between innovation and growth of output and employment do not only depend on if the economy is in a long term downswing or upswing, but also on the specific, historical characteristics of the concrete wave. History matters and it is not meaningful to search for a time-independent theory of innovation and growth. In some periods technical change may have powerful impacts on economic growth while these effects may be more modest in other periods. In some periods the employment creating effects of innovations may dominate over the employment saving effects. In other periods it may be the other way around. The policy implication is that the focus should be on the set of factors which make it possible for the economy to introduce, diffuse and utilise the new key factor, i.e. information-and telecommunication technologies in the case of the fifth Kondratief wave. For example, rather than concentrating on the size of the public sector and the power of trade unions which was

17 17 the case in the first diagnoses by the OECD of the growth slow-down in the beginning of the 1970s the interest should also be directed towards rigid organisational structures in firms, inflexible co-operation patterns within and between firms and inadequate science and technology policies which became the case later in the Sundquist report (OECD, 1988). Again, a more developed understanding of the role of technological and institutional change in economic growth seems to call for a more broadly based policy perspective Co-evolution between sub-systems: matching and mis-matching. The concept of techno-economic paradigms makes use of the idea of co-evolution of specific subsectors of society. These subsectors may match and mis-match with each other which make the process of growth and development both time and space specific. This way of thinking about growth has lately been developed further by Chris Freeman. He identifies five, partly overlapping, relatively autonomous spheres or sub-divisions of society. Each of these sectors may develop in their own, relatively autonomous way. The five sub-systems are science, technology, economy, politics and general culture and each of them is of great importance for economic growth. At the same time they are interacting with each other and this may lead to problems of lack of synchronicity and harmony or alternatively of harmonious integration and virtuous circle effects on economic growth. (Freeman, 1996, p.13) Positive congruence and interaction between the subsectors support economic growth while mis-match retards it. Consequently these spheres and their relatively autonomous development should be studied in their own respects as well as in their interactions with each other when analysing economic growth as a historic and evolutionary process. Only in this way it is possible to understand the characteristics of the specific periods of growth and development, i.e. the periods of acceleration and retardation. The relationships between innovations and growth will be deeply affected by this over all process of social evolution. If one wants to understand economic growth as an evolutionary process in historic time it is not enough to focus on the typical, or normal, or average growth path. Deviations from the average may be a key to the understanding of the process as a whole. Using terms coined by Abramovits 6, Freeman (1996) discusses falling behind, catching up and forging ahead between nations as separate but connected categories in an international growth process. Each category has to be studied for specific countries and specific periods. At the same time it is necessary to identify the main recurrent phenomena, in order to make it possible to make generalisations as elements of theory building. Of the three growth categories forging ahead is the most difficult to explain since the explanation depends on unintended factors, creeping up behind the backs of decision-makers, much more than the explanation of catching-up, in which known trajectories are more important and intentionality is more present. Besides, there are many more historical cases of catching-up than of forging ahead and consequently a more solid empirical background for generalisations. Freeman discusses the British industrial revolution as the most well known example of forging ahead in modern economic growth. He looks at the main explanations which historians have put forward and he summarises the picture in terms of the following aspects: i) Acceleration of growth from 1780s. ii) Exceptional growth and falling prices of cotton textiles. iii) Invention and innovation. iv) Capital accumulation. v) Entrepreneurship. vi) Agriculture. vii) Mobility of labour and work organisation. viii) Market demand. ix) In- 6 See for example Abramovits (1989) where the phenomena of falling behind, and forging ahead are discussed in connection with the catch-up hypothesis.