Classifying technology policy from an evolutionary perspective

Research Policy 30 (2001) 759–775

Classifying technology policy from an evolutionary perspective Uwe Cantner a,∗ , Andreas Pyka b,1 a

Friedrich-Schifler-Universitat, Jena, Wirtschaftswissenschaftliche Fakultät, Carl-zeiss-str. 3, D-07743 Jena, Germany b INRA-SERD, Université Pierre Mendès BP47, F-38040 Grenoble Cedex 9, France Accepted 17 April 2000

Abstract For the classification of technology policy in an evolutionary framework the taxonomy of mission- and diffusion-oriented policy design introduced by Ergas [Ergas, H. (1987), The importance of technology policy, in: Dasgupta, P., Stoneman, P. (eds.), Economic Policy and Technological Performance, Cambridge Univ. Press, Cambridge.] is useful. However, Ergas’ indirect method of identifying the respective policy style is only able to give a first and rough insight in the technology policy of a specific country. To improve on that, we developed a so-called direct method aiming at a sound characterization of single policy measures and giving a more detailed picture about policy orientation. To demonstrate the basic procedure of the suggested approach, it is applied empirically on the German technology policy of the last 2 decades. © 2001 Elsevier Science B.V. All rights reserved. JEL classification: 039 Keywords: Mission-oriented policy; Diffusion-oriented policy; Technological specificity; Market vicinity

1. Introduction In 1987, Ergas introduced the concepts of missionand diffusion-oriented policy designs to classify and analyze national systems of innovation. According to him, mission-oriented systems are characterized by centralization and the concentration of policy support on a small number of technologies and larger firms, unlike diffusion-oriented systems, which concentrate their policy efforts on increasing an economy’s capacity of innovating by concentrating on the scientific infrastructure, technology transfer and co-operation, ∗ Corresponding author. Tel.: +49-3641-943200; fax: +49-3641-943202. E-mail addresses: [email protected] (U. Cantner), [email protected] (A. Pyka). 1 Tel.: +33-476-825-831; fax: +33-476-825-455.

i.e. formal and informal relationships between different actors, etc. Although, on the one hand, this taxonomy is considered well suited for analyzing technology policy in an evolutionary context, on the other hand, it appears somewhat crude, especially regarding the characteristics Ergas identifies in order to assign a specific innovation system to one or the other policy design. To surmount these critics, we suggest to take a closer look at the specific policy measures applied. For this purpose, a new classification scheme is introduced, suggesting four categories spread out between the technology side (high and low technological specificity) and the economic side (high and low market distance). This scheme allows for more evident comparisons of different innovation systems. As a theoretical point of view an evolutionary perspective is taken looking at capabilities of the actors involved, the processes of technological competition

0048-7333/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 7 3 3 3 ( 0 0 ) 0 0 1 0 4 - 9

760

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

and co-operation as well as the connecting feedback during the various phases of innovation processes (in a broad sense). Demonstrating the basic procedure of the suggested approach the classification scheme is finally applied empirically on the German technology policy of the last 2 decades.

2. Theoretical foundations Asked for the most important driving forces of economic development, most economists do not hesitate to state that technical progress is the main source of quantitative economic growth and qualitative economic change. However, a necessary sequent question is, what kind of innovative activities can be expected in capitalistic societies, and how can policy support or even spur such innovation processes? In order to analyze these issues, some economists left the traditional equilibrium-oriented path of neoclassical economics and turned to an evolutionary theorizing. In particular, they argue that the evolutionary paradigm is more adequate for analyzing development processes initiated within the system, characterized by strong uncertainty and disequilibrating forces and which is composed of heterogeneous actors. 2 In this perspective, neoclassical economics oversimplifies the issues of innovation processes by drawing on a representative agent who attempts to reach an optimal equilibrium because he acts in a deterministic or quasi-deterministic environment and is endowed with perfect rationality. Contrariwise, in an evolutionary context, the complexity of interactions between heterogeneous agents guided by trial-and-error is considerably increased, and a benchmark for optimal solutions is missing — thus, uncertainty prevails and the optimality of solutions can only be assessed ex post. Therefore, one has to search for other justifications for policy interventions than sophisticated but nevertheless simple market failure analysis. 3 Such a new view on technology policy, although not well elaborated as yet, will certainly have to be framed by an understanding of innovation processes as search and experimentation (instead of immediate optimization). Policy measures will then aim at the propelling and sustaining forces 2 3

See e.g. Witt (1987,1993), Dosi and Nelson (1994). See Cantner and Hanusch (1997).

of these processes (instead of not yet achieved optimal solutions). 4 Among these forces, the capability, education, and skills of actors are of foremost importance. 5 Absorptive capacities 6 and receiver competencies 7 are required for both the generation and dissemination of technological know-how. In many cases, bottlenecks emerge just here, and any market incentives promoting or restricting innovative activities are subordinate. Whatever the policy measure, crucial points should be considered, not in a facultative but in a general and long-lasting manner. This emphasis on non-market factors, of course, is not to be understood as an argument for the total neglect of markets. Markets are here to be considered mainly as selective devices for awarding better and penalizing worse technological solutions or performances. Of course, in this context policy measures aim at the functioning of markets and still attempt to prohibit monopolization — not in a static allocative sense, but with a dynamic orientation, keeping in mind the role of heterogeneity as the fuel for technological evolution. 8 Moreover, the evolutionary perspective dispenses with the simple linear-sequential model of innovation 9 , where an exogenous inventive stage is followed by an innovation stage in which firms can draw on well-defined new technological opportunities, and finally also a diffusion stage, where the successful innovations spread over the whole economy is no longer adequate. Instead, some authors draw from the notion of collective invention 10 where actors in different stages of the innovation process influence each other, to and fro. Thus, in these environments so-called cross-fertilization effects 11 additionally propel technological change. Applying this concept of looped processes 12 to the major innovative actors in an economy, the concepts of national innovation systems 13 were introduced. There, different actors 4

See Metcalfe (1995). See Carlsson and Eliasson (1995). 6 Cohen and Levinthal (1989), Cantner and Pyka (1998). 7 Eliasson (1990). 8 See e.g. Metcalfe (1995). 9 See e.g. Schumpeter (1912). 10 Allen (1983). 11 Mokyr (1991), Pyka (1999). 12 Kline (1985). 13 See e.g. Nelson (1994), Lundvall (1992). 5

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

and institutions (firms, independent inventors, universities, etc.) jointly and individually contribute to the exploration of new and the exploitation of given technologies. In this perspective, not only risk (i.e. weak uncertainty), but also strong uncertainty 14 necessitates policy intervention in order to support innovation processes. So, in addition to the question how to provide for venture capital, the problem of suitable or appropriate structures underlying and supporting these evolutionary development processes moves into the center of interest. Here, new trade-offs have to be solved, e.g. between heterogeneity as a necessary and rich source of innovation and standardization 15 , which is, on the one hand, useful for the diffusion of new technologies, but on the other hand, obviously reduces heterogeneity. From this evolutionary perspective, policy designs now have a straightforward process-orientation instead of solely “repairing” market failures, which is a target-orientation (benchmark-based). Following this, the well-known mission- and diffusion-oriented policy designs may be seen in a different light. Quite obviously, diffusion-oriented policy with its missing clear target perspective serves more to the aim of sustaining an appropriate degree of heterogeneity. At the first sight a mission-oriented policy appears quite the contrary, by favoring one specific development path, this being a severe disadvantage for alternative technological trajectories. 16 However, one might also argue that mission-oriented policy is able to bring ex-ante seemingly unrelated technologies together by exploiting cross-fertilization effects. In this sense, mission-oriented policy also has to be seen as a policy instrument for sustaining evolutionary development. 17 Within this analytical and interpretative framework (i) it is discussed whether the Ergas taxonomy of technology policy is appropriate for identifying evolutionary policy, (ii) based on the criticism raised 14

Knight (1921). Cowan and Foray (1997). 16 Ergas (1992) adds a further critical aspect to mission-oriented programs: the very likely high degree of technology complexity of these programs restricts participation in program execution to a few, technology sophisticated agents, thereby reducing heterogeneity from an actor’s perspective. 17 Conesa (1997). 15

761

the classification is extended using a so-called direct method, and (iii) this is applied to the German federal technology policy from 1975–1996.

3. Mission- vs. diffusion-oriented policy design — some taxonomical considerations In order to judge the efficiency of specific technology policy measures, Ergas (1987) introduces his often-cited taxonomy, where he differs between mission- and diffusion-oriented policy designs. Following his arguments, a national innovation system is connected to state interventions for two main reasons. On the one hand, in nearly all industrialized countries, technology and innovation are promoted due to considerations of national sovereignty and international competitiveness. On the other hand, technology policy measures are invoked by the awareness of significant ‘imperfections’ of the market, e.g. among others in neoclassical terms the public-good nature of technological know-how, and in more evolutionary terms, the intrinsic uncertainty of innovation processes. The attribute mission-oriented seems to be justified, in a rough sketch, in those cases of technology policy, which aim at national sovereignty, e.g. the development and introduction of new generic technologies. Here, a clear target is given and the respective funds are granted to fulfill specific technological aims. Whereas measures aiming at a broad spreading out of new know-how, including measures to promote the utilization of this know-how or even the respective transfer, are classified as diffusion-oriented. Quite typical is the furthering of so-called key technologies, which are assumed to exert positive influence on a broad group of industries and firms. Ergas (1987) had already emphasized the rough nature of every taxonomy, because a classification of countries or even specific technologies to one or the other policy design obviously disguises single measures, which follow the respective other policy design. Anyway, Ergas (1987, p. 52) claims: “But the focus of policy differs in the two groups, and this allows a clearer examination of the relation between technology policy and innovation performance.” Although one is readily able to agree to this statement, one might be tempted to look for an alternative to Ergas’ indirect method which should in a way reduce shortcomings

762

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

resulting from roughly classifying whole countries to a specific policy design. Before this could be accomplished, a first look will be on the classical Ergas procedure.

Table 1 Public funding for R&D in the high, medium and low R&D intensity industriesa as respective percentage of total public funding (approximate estimates) Country

R&D intensity

3.1. The indirect method According to Ergas (1987), the technology policy of a country can be classified as mission- or diffusionoriented by examining several criteria concerning (i) the state of the technology life-cycle of a specific technology, (ii) the share of public funded research institutions as well as private recipients, (iii) the specific design of the educational system, (iv) the opportunities for co-operative R&D, (v) standardization efforts, and finally (vi) the share of military research. These criteria are applied to a group of six countries (United States, France, United Kingdom, Germany, Sweden, and Japan). With respect to the first criteria, Ergas claims that mission-oriented policy concentrates on a small number of technologies in an early phase of the technological life cycle. Thus, a specific feature of mission-oriented measures is concentration: first, only a small number of technologies is selected for public funding and second, also most often only large corporations have the adequate infrastructure necessary for these programmes. “A few bets are placed on a small number of races; but together these bets are large enough to account for a high share of each country’s total technology development programme.” 18 On the contrary, diffusion-oriented policy aims at a broader spectrum of technologies in a more advanced phase of the life cycle. In order to distinguish phases of the technology life cycle Ergas draws on the assumption of declining R&D intensity going hand-in-hand with advances in that technology. His results are summarized in Table 1. By this Ergas implicitly assumes a direct relationship between the number of furthered industries or technologies and the specific policy design 19 applied. However, in order to classify technology policy, the decisive criteria is the design of the specific measure 18

Ergas (1987, p. 55). Of course a large number of firms furthered by technology policy contradict the aim of decreasing parallel research of a mission-oriented design. 19

1980

United States France United Kingdom Germany Sweden Japan

high

medium

low

88 91 95 67 71 21

8 7 3 23 20 12

4 2 2 10 9 67

a Defined as industries where the ratios of R&D to sales are respectively more than twice (high intensity), between twice and a half (medium) and less than a half (low) than those of the manufacturing average. Source: OECD, Ergas (1987, p. 54)

and not the number of technologies chosen. The indirect procedure neglects this. According to Ergas’ second criteria, a high share of public-funded research performed ‘in-house’ i.e. by public research institutions is another distinguishing feature of a mission-oriented policy. For evidence, the share of public-financed R&D performed within public institutions is invoked (see Table 2). Again, this is to be regarded as only a considerably rough procedure, because following Ergas’ definition of mission-oriented policy, it is the specificity of a measure and not the recipient that is constituent for the respective policy design. So, there are technology policy programmes aiming at technology transfer including public institutions e.g. universities which can be clearly classified as diffusion-oriented, as well as programmes aiming at the development of a specific technology by private firms (e.g. Transrapid in Germany), which are clearly designed as mission-oriented. Table 2 Share of government-financed R&D carried out within the government-financed sector. Source: OECD, Ergas (1987, p. 58) 1983 (%) France United Kingdom (1981) Germany United States (1981) Switzerland (1981)

46.8 38.9 31.6 25.7 24.7

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

Specific features of a technology policy design following a diffusion orientation can be found in the structure of the education system, standardization efforts and the furthering of co-operative research summarized under the heading decentralization. As to the education system, Ergas focuses on the interdisciplinary character of e.g. engineering schools on the one hand, and on the other, he emphasizes the combination of theoretical knowledge as well as practical skills, as found e.g. in the German apprenticeship system. “The most significant feature [of a diffusion-oriented policy] is probably the depth and breadth of investment in human capital centering in the dual system of education.” 20 In order to guarantee an effective flow of information and know-how, standardization can help to reduce the respective transaction costs. 21 In the same way co-operative R&D including close university–industry links, as well as formal 22 and informal 23 co-operation within firms are necessary devices for increasing the pace of the diffusion of new know-how and technologies. With this, Ergas concentrates on the technological infrastructure 24 responsible for information transfer as well as the building up of receiver competencies or absorptive capacities. Although in most cases the abovementioned elements are obviously closely connected to a diffusion-orientation, the reverse conclusion cannot necessarily be claimed to hold also. All difficulties in comparing national education systems aside, many mission-oriented programmes involve co-operation (e.g. in thermo nuclear fusion research private firms as well as public research institutes and universities are engaged), and sometimes they even take the initiative to create a new prevailing standard (e.g. HDTV). Again as a consequence, a clear-cut classification is impossible without taking the specific design of a technology policy programme into account. Finally, Ergas (1987) claims a high share of military R&D as a significant feature of mission-oriented policy dominated countries. For evidence, he draws on the shares of defense-related R&D in total government spending for R&D shown on Table 3. 20

Ergas (1987, p. 67). 21 See e.g. Foray (1995). 22 See Haagedoorn and Schakenrad (1990). 23 See Pyka (1997,1999). 24 See Tassey (1991), Justman and Teubal (1996).

763

Table 3 Share of defense-related R&D in government expenditure on R&D, 1981. Source: OECD, Ergas (1987, p. 54) 1981 (%) United States United Kingdom France Sweden Switzerland Germany Japan

54 49 39 15 12 9 2

Obviously, military research is mission-oriented par excellence. But it cannot be compared with most of the civil projects, as very different co-ordination mechanisms are invoked e.g. no true cost-control, etc. Therefore, in the direct method to be introduced below, military research expenditures will be neglected altogether. With the criteria introduced above, Ergas concludes to characterize countries according to their respective policy design. So, typical mission-oriented countries are the United States (high funding of R&D intensive industries, high share of military research), the United Kingdom (high funding of R&D intensive industries), and France (high funding of R&D intensive industries, high share of military research). Countries with a dominating diffusion-oriented policy are Germany (education system, standardization), Sweden (low share of defense-related R&D), and Switzerland (low share of defense related R&D, low share of government in-house R&D). However, the indicators of this indirect method applied by Ergas cannot give more than a first and rough idea of the technology policy of a specific country. In order to obtain deeper insights in the development and structure of technology policy, and thereby still drawing on the useful taxonomy of mission- and diffusion-oriented policy design, a so-called direct method is suggested and introduced in the following section. 3.2. The direct method The preceding section showed that the indirect procedure by drawing on specific characteristics of the innovation system can only provide a rough and

764

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

partially distorted sketch of technology policy classification. Without going into further detail of the design of the specific measures applied any classification tends to disguise the multifaceted character of technology policy. In order to improve on that a so-called direct method based on an inspection of the different policy measures specifically is proposed. The direct method consists of two steps. (1) A classification scheme is introduced allowing to group specific policy initiatives or programmes with respect to certain policy designs. (2) An in-depth inspection of the policy programmes and classifying them accordingly. Fig. 1. Classification matrix.

3.2.1. The classification scheme As to step 1, dealing with policy designs, it turns out that it is not possible to always classify certain policy programmes either into diffusion or into mission orientation. Some policies show a clear orientation towards furthering basic science and they do not intend to promote any concrete economic application. Nevertheless, during time, a number of those basic sciences-oriented policy programmes change their nature: Advances in the basic sciences often contain a — sometimes rather vague — promise on economic benefits in the future and the respective policy programmes then become redesigned. As they then get more specific with respect to their prospective final economic application, a classification into diffusion or mission orientation becomes possible. In order to cope with these issues and developments, it proves appropriate to apply a classification along the criteria of “vicinity to the market” and of “specificity of a policy measure.” By the former criterion, one is readily able to distinguish basic and applied research, in order to take account of the large share the furthering of basic research demands in the public budget. The second criterion serves as indicator as to whether the policy programme under consideration quite precisely describes the research target or whether this is rather open. This combined with a high degree of market vicinity obviously leads to the distinction between mission- and diffusion-oriented policies. Furthermore, combined with low market vicinity a distinction between basic I and basic II turned out as appropriate.

Based on these two categories one achieves at four broader fields of technology policy, which are given by the matrix in Fig. 1. 25 3.2.2. Characterization of the classes With respect to classifying policy programmes along this matrix, the respective classes or cells of the matrix are characterized as follows. Basic I. Basic I summarizes basic research in a narrow sense, meaning there is no commercial orientation. Basic II/Vision. However, still far away from a market introduction, some specific technologies, e.g. thermo nuclear fusion researches are publicly funded. Here, public funding is combined with a certain degree of technological specificity because a large potential for further development is expected in these fields. Therefore, it is Basic II/Vision as the respective technology policy goes hand-in-hand with a kind of technological vision or farsightedness or expectation. However, due to the enormous financial efforts necessary to promote innovation and the intrinsic uncertainty in these fields, private actors are not expected to engage themselves here by their own. Thus, policy intervention intends to further those technologies so that they may reach higher market vicinity in the future. 25 The respective classification matrix shows some similarities with the so-called Pasteur’s Quadrant introduced by Stokes (1997). However, Stokes differs between the motives of researchers, in particular understanding and application, which is why his scheme is not appropriate for our purposes of classifying technology policy.

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

With decreasing market distance the classes of mission- and diffusion-oriented policy designs apply which differ in the degree of technological specificity given the single policy interventions. Consequently, the following can be distinguished. Diffusion-orientation. A low degree of technological specificity allows for a broad field of heterogeneous technologies to be furthered in a diffusion-oriented style. There is no specific description with respect to technology to be pursued or the economic application to be achieved at. Mission-orientation. Policy programmes characterized by a relatively high degree of specificity are considered a mission-oriented policy design. Both the technological objective and the economic application of a certain technology or technology field are quite clearly described and the policy measures are well designed accordingly.

4. Empirical application The direct method introduced above has been applied to data describing the technology policy of Germany for the period from 1975 to 1996. Before presenting the empirical results, first a brief introduction to the data used is provided; second the respective criteria of assigning specific policy measures to the policy classification scheme are discussed. Additionally, upcoming problems of assignment — which are still associated with the direct method — are also worked out. 4.1. Database The data are taken from the annual issues of the Ministry of Research and Technology, Bundesministerium für Bildung und Forschung, (BMBF), for the years 1984, 1986, 1988, 1990, 1993, 1996 and 1997. From these Table 8 serves as a coherent and topic-oriented scheme out of part VII (Statistics): “Ausgaben des Bundes für Wissenschaft, Forschung und Entwicklung nach Förderbereichen und Förderschwerpunkten” (state expenditures on science, research and development classified by areas and main subjects). Table 4 gives an overview of the promotion areas.

765

The numbers delivered are the aggregated expenditures of the federal state (Bund) for the specific promoting measures (or ministries) where a further sub-division within the main subject is not provided. Since not all data are available for the whole period under investigation (1975–1996) and because the statistics of 1984 is based on a different classification, the missing values are approximated by appropriate measures. For 1997, there are no actual data available yet so that the planned numbers were taken. In order to make the amounts spent comparable between the years of investigation, they have been deflated/inflated to prices of 1991 using index numbers taken from the report of the Sachverständigenrat (1996), the German Council of Economic Advisors. 4.2. Criteria of classification The data we use are the state financial funds, which were spent on the respective promotion areas. The task to be performed is to assign these promotion areas into the classification matrix of the direct method, i.e. basic researches I and II, mission-oriented policy design, and diffusion-oriented policy design. For this purpose, in part 3 of the annual report, the specific promotion areas and main promotion subjects are evaluated with respect to their objective, the results achieved so far and the applied measures. In order to check and adjust the procedure, personal interviews with the persons in charge at the Federal Ministry of Research and Technology were of considerable help. As a result, a specific procedure of classification and certain criteria has been set up according to the following three main considerations. (1) The first consideration in this respect is to distinguish the promotion areas of Table 4 in three principally distinct kinds of focus: the funding of specific institutions; the promotion of certain research projects and technologies; the implementation of measures to improve the conditions for developing and innovating technologies, so-called indirect policy measures. For each of these objectives a different way of classification has to be taken. (2) Following that, the second consideration is to be aware of the fact that even technologies with a broad account of application and economic value continuously require basic research. Thus, by inspecting those technologies, one has to carefully look at

766

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

Table 4 Promotion areas as stated in the annual issues of the Ministry of Research and Technology Promotion areas A A1 A2 A3 A4 A5 A6 B C C1 C2 C3 D E E1 E2 E3 E4 E5 F F1 F2 F5 F7 F8

Research organizations; restructuring institutional research in the new Bundesländer; universities Max-Planck-Organisation (MPG) Deutsche Forschungsgemeinschaft (DFG) Fraunhofer Gesellschaft (FhG) Restructuring institutional research in the new Bundesländer Universities Mainly university-related special programmes Large-scale equipment for basic research Marine research and Polar research research; marine technology Marine research Marine technology Polar research Space research and space technology Energy research and energy technology Coal and other fossil energy sources (power plant engineering and combustion research) Renewable energies and energy conservation Nuclear energy research Decommissioning of nuclear facilities; risk sharing Thermo nuclear fusion research Environmental research; climate research Ecological research Environmental technologies Water Climate and atmosphere Safety

what stage or phase of the innovation process (i.e. invention, innovation, diffusion) the respective policy measure is applied. (3) As a third and final step, it has to be figured out whether and to what degree policy makers make precise decisions or prescriptions on the direction research has to take and on the results of that research to be expected. Thus, one has to carefully investigate whether technology policy actively directs progress or whether it serves as a moderator.

G

R&D in the service of health

H I I1 I2 I3 I4 I5 K L

R&D to improve working conditions Information technologies Computer sciences Communication Electronic devices Micro systems Production engineering Biotechnology Materials research; physical and chemical technologies New materials Physical and chemical technologies Hypersonics Traffic Geosciences and raw material supplies

L1 L2 M N O O1 O2 P P1 P2 Q R S T T1 T2 T3 T4 T8 T9 U V W

Geosciences (especially deep drillings) Raw materials Regional sciences Urban development Construction/architecture R&D in the food sector Agriculture, forestry and fishery Educational research Innovation infrastructure R&D employment subsidies Knowledge transfer Venture capital Other indirect means Rationalization else (indirect) Specialized information Humanities Technological assessment and consequence

These considerations in mind, the following principle rules of assigning policy measures to the classification scheme have been worked out. Institutional funding. As to the institutional funding, which is oriented towards universities and research institutions, it is the principal objective of the respective institution, which decides on the classification. By and large institutional funding is assigned to basic research — universities, MPI are appropriate examples. The focus here is clearly not on innovation but on

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

invention, although in the case of so-called technical universities this is doubtful — a data disaggregation to perform this distinction unfortunately is not available. An exception for assigning institutions to basic research is made when their focus is explicitly not on basic research such as A3, the Fraunhofer–Gesellschaft (FhG) with its specific objective to sustaining applied research and technology transfers from the basic research — in this case a classification into the cell diffusion-oriented policy design is appropriate. Promotion areas. For the different promotion areas, an in-depth analysis of the subjects (i.e. organizations, individuals, etc.) and the research objects has been performed. (1) As to the subjects, when they belong to universities and research institutions the criteria above apply. An example is the promotion of the BML (ministry of agriculture) of the Bundesforschungsanstalt für Landwirtschaft (public institute for research in agriculture) in Braunschweig within the promotion area R, agriculture, forestry and fishery, which is classified as basic research. (2) As to research objects of the different promotion areas, the inspection of the descriptions of their respective sub-projects, their main objectives, and their results achieved so far has been performed along the following lines. (a) With respect to basic research the main guideline is the definition of basic research as given by the OECD: “Basic research is the experimental or theoretical work undertaken primarily to acquire new knowledge of the underlying foundation of phenomena and observable facts without any particular application or use in view.” Thus, the research funded is far away from any phase of innovation; it rather belongs to the phase of invention without any application or even economic value to be expected — as yet. The sub-area F7, climate and atmosphere, out of research area F, environment and climate, is an appropriate example. (b) The difference between basic research I and basic research II is rather often only of a minor degree. Whenever the careful inspection of the description and objectives of the promotion areas to be classified as basic research delivered various hints that policy makers for the long-term hope that an applied technology with economic value might come out of this, an assignment to basic research II has been made. Points

767

in case are various sub-projects under the heading D, space research and space technology. (c) Any policy measure in specific promotion areas, which are not suitable for basic research, may be classified into diffusion-oriented design. This is appropriate whenever the decision on which specific direction the research should be oriented is not taken by the policy makers in a centralized manner but rather contrary is decentralized. Thus, those policy measures, although they aim at a certain technology field, do not give, however, a detailed specification of technological aims to be pursued. An example in this respect is the promotion area L1, new materials, for which — from a technical point of view — no clear technological objective of research is specified. The policy focus here is clearly on innovation. On equal terms, sub-projects of the main promotion areas attempting to accelerate technology transfer are defined as a diffusion-oriented policy design. They enjoy an explicite consideration under indirect policy measures below. (d) Any policy measures which are not basic research and which specify technological objectives are classified into mission-oriented policy design. Contrary to diffusion orientation, the decision on the specific direction and even (appreciated) result of research is made by the policy makers in a centralized manner. A most prominent example in this respect is the magnetic levitation train Transrapid promoted within the promotion area N, traffic. Other examples for this are certain national prestige projects, projects which are undertaken under the heading of sovereignty and projects will have a more or less military objective (e.g. E3, nuclear research; M, aeronautics and hypersonic). Obviously, here as for the diffusion orientation the focus is on the economic application of the technologies pursued and thus on innovation. Indirect policy measures. Indirect policy measures aim at improving the conditions for innovation in and application of technologies. The specific technological and promotion areas funded herewith cannot be identified on the basis of the data available. Although it is not farfetched to assume that both mission- and diffusion-oriented promotion areas gain from this indirect funding, these measures are classified entirely as diffusion-oriented because the decentralization of the decisions in what direction to do research or which technology to apply is a main characteristic. Indirect policy measure are designed as:

768

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

1. technology unspecified measures for R&D in firms (e.g. potential-oriented measures, such as T1, R&D employment subsidies) or 2. technology specific measures to further broaden application and diffusion of so-called key technologies (e.g. T3, information technology). Moreover, the main promotion areas U, specialized information which mainly contains measures to improve the informational infrastructure and W, technology consequences, are considered as indirect policy measures because in a considerably broad sense both areas serve to further R&D decisions of firms. 4.3. The problems of inclusion and ex-post/ex-ante classification The classification criteria suggested cannot be applied without particular qualifications. Especially for performing a comparison of different time periods — as done in the following — or an international comparison — to be pursued in future work certain problems of interpretation and assignment must be taken into account. The following are the main problems to be considered. (a) Policy mix and separating policy designs. For a number of promotion areas and sub-projects an unequivocal assignment of the policy style is not possible as we find a policy mix. Since the statistics used often show a relatively high degree of aggregation, in those cases, a detailed separation of policy types cannot be accomplished. A viable solution to the problem is to assign the aggregate to that type of policy design, which is assumed to command the largest fraction. As an example serves the already mentioned assignment of technical universities as basic science, although a considerable share of their activities are oriented at applied research. (b) Distinguishing between mission and diffusion design. Depending on the hierarchic position of the decision-maker on the policy side, even a straightforward diffusion-oriented measure can be classified as mission-oriented. On the level of those decisionmakers who decide on the distribution of funds, the policy measure would be classified rather as a missionoriented policy design; contrariwise from the point of view of the respective ministry or the policy program

the same policy measure will be classified as diffusion-oriented. (c) The problem of ex-ante and ex-post classification 26 . When comparing technology policy measures of different years one has to consider that measures that in the present perspective were diffusion-oriented, were mission-oriented in earlier times. In order to deal with this problem, it is unavoidable to consult the respective years’ political reports and program descriptions; nevertheless, the problem cannot be completely avoided. (d) International projects. A number of projects in certain promotion areas are funded on the international level (in Europe on the EU level, in particular). An assignment of these promotion measures to the one or the other policy design turned out to be quite difficult; in some projects specific countries are only involved by spending money without gaining any influence on the design of the project; in another project, decisions are made by several parties each with equal rights; in a third one, only the country under consideration has the lead in a certain project.

5. Results 5.1. Policy portfolio In order to give a first idea of the topography of German technology policy, the classification matrix of Fig. 1 above has been filled in with data of the most recent year 1996 of the investigation. Fig. 2a shows the resulting policy pattern. The different promotion areas are reflected as bubbles where the size of the bubble represents the relative financial amount invested. This illustration already gives a first hint on the distribution of funds on the respective policy designs in that year. The largest share of expenditures of technology policy is located in the fields with high market distance, i.e. those fields representing basic research. Also, it becomes obvious that in Germany, policy is guided by the vision of large development potentials for space research and space technology in the promotion area D (here also the commitment in multinational 26 We are grateful to one of our referees to have hinted us on this problem.

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

769

Fig. 2. (a) Classification matrix for 1996. (b) Classification matrix for 1975.

programmes as ESA have to be mentioned), as well as thermo nuclear fusion research (E5) and biotechnology (K), where the latter more and more gains in market vicinity. Although, space technologies have reached a certain market vicinity in the fields of commercial satellites, etc., nevertheless, the main focus still has been in areas quite far away from marketable technologies. The bulk of expenditures is spent on pure basic research in the first quadrant, e.g. financing of universities (A5), Max-Planck-Organizations (MPG), etc. Some promotion areas such as agriculture (R) or R&D in the service of health (G), although they might be considered as focussing at marketable technologies, as we look, however, at the level of the federal state, the funds spent are mainly institutional

support. Consequently, the assignment follows the principle objective of the respective institution which more than often is basic research (e.g. in the case of agriculture about two-thirds of the overall expenditures were given to the Bundesforschungsanstalt für Landwirtschaft (research institute for agriculture)). A look at the upper two cells of the classification scheme reveals a clear diffusion-orientation of German technology policy in 1996. The bulk of measures is characterized by a low degree of technological specificity. Besides the indirect measures (T), which demand a large share of the diffusion-oriented policy measures, immediate diffusion-oriented promotion areas like renewable energies (E2) show up here, where a widely spread diffusion of solar technology and its application in several areas were funded. Moreover,

770

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

Fig. 2 (Continued).

the same assignment holds for the so-called new technologies e.g. new materials (L1), information and communication technologies (I), etc. The large promotion areas with clearly specified technological targets, assigned to the mission-orientation, are nuclear energy research (E3), hypersonic (M) and environmental technologies (F2) 27 , as well as traffic (N) with the Transrapid. From this first sketch of policy measures in Germany, it might be concluded that the diffusionorientation to be the major focus. The next steps of analysis looking at the development of policy orientation over time will further sustain this result. A first 27 German technology policy in 1996 in the promotion area F2 environmental technologies is clearly focussed on specific technologies as waste incineration where e.g. the construction of prototype systems are furthered.

step is a look at the policy pattern in the first year of the period of investigation, 1975. In this respect, Fig. 2b, containing the 1975 classification matrix 28 , reveals that the diffusion-orientation was not always dominating German technology policy. Here, with respect to the mission-oriented policy design, the promotion area E3 nuclear energy research dominated, demanding the largest share of subsidies in the applied fields with a nearly 15% of the overall research. The diffusion-oriented field contains only the support of the FhG, the indirect measures (T) as well 28 It has to be mentioned, that the 1975’s matrix has to read carefully as here, the problem of ex-ante and ex-post classification is of significant importance. In order to avoid this problem at least partially, the 5th Forschungsbericht of the Bundesregierung of the year 1975 was consulted in order to evaluate the single promotion areas.

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

as the just rising information and communication technologies (I), which by that time comprises solely the institutional support for the Gesellschaft fuer Mathematik und Datenverarbeitung (society of mathematics and data processing). A closer comparison of the two years additionally shows interesting results with respect to the development of specific research fields and their funding. Most of the technological fields in the basic II area in 1975 have undergone a significant progress and developed towards marketable technologies. Among others, the promotion areas alternative energies (E2), new materials (L1), and physical and chemical technologies (L2), in 1975 still on the stage of basic research and characterized by a considerable market distance, have moved into the diffusion-oriented cell. Other promotion areas like biotechnology (K) and environmental technologies (F2) since 1975 significantly gained market vicinity; technology policy reoriented from a focus on fundamental insights only to a mission-oriented funding of specific marketable applications. 29 By the way of conclusion one may state here that a technology policy accompanying and partly initiating such development from the area basic II/vision to applied areas are to be considered as successful — at least in the sense of identifying early on the important technologies of the future. Up to the reader are conclusions on promotion areas which during a period of 21 years contrariwise did not succeed to move up to applied technologies, as e.g. E5, thermo nuclear fusion research. The discussion of Germany’s principal technology policy design will now be extended to an analysis of the development over 20 years from 1975 to 1996. By this the focus of investigation is on the share different policy styles occupy. For this purpose two comparisons will be made, the relative development of funding on basic and an applied research and the balance between diffusion- and mission-oriented policy designs.

29 One promotion area, E1 coal and other fossil energy sources has moved from a dominating mission — to a dominating diffusion-orientation in the observed time period. Whereas in 1975 the promotion of specific technologies as coal gasification and liquefacation were in the center of interest, in 1996 the furthering of the diffusion of environmental carefully filter-technologies are summarized in this promotion.

771

5.2. Basic research funding versus technology policy in a narrow sense The first investigation is concerned with the comparative development of funding basic and applied research, the latter considered as technology policy in a narrow sense. In order to get an idea about the significance of measures aiming at basic and on applied research, respectively — distinguished by the degree of market vicinity as stated on the vertical axis of the classification matrix — Fig. 3 states for each year the sum of mission- and diffusion-oriented funds on the one hand and the sum of funds for promotion areas in the cells basic I and basic II/vision on the other hand. In general, both funding of basic research and technology policy in a narrow sense raise in real terms — although the share of those expenditures in total government spending declined from 11.9% in 1975 to below 10% in 1996. For the first 10 years, 1975 to 1985, policy measures focusing on basic research areas almost amount to 50% of total funds; from 1985 onward this share is even increasing to about 60%. The sharp increase since 1991 is caused by the German re-unification, demanding large funds for re-structuring institutional research in the Neue Bundesländer. This comparative development indicates that in terms of shares German research and technology policy continuously retreated from funding applied research. 5.3. Mission- versus diffusion-orientation The second step of investigation is concerned with comparative development of diffusion- and mission-oriented policy styles. For this purpose, Fig. 4 states in absolute terms the development of funds devoted to mission- and diffusion-orientation where with respect to the latter the indirect measures are first stated separately — for the reasons given in Section 4.2. Comparing mission- with diffusion-orientation shows a clear dominance of the former policy design until the early 1990s. One might be tempted to interpret this as a contradiction to Ergas’ classification of Germany as a diffusion-oriented country (in the early 1980s to which his data apply). However, of course, the Ergas result is a relative one, where Germany is compared to other countries. Our result is based

772

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

Fig. 3. Development of expenditures according to market vicinity.

on comparing the absolute values in Germany. But this accepted, it nevertheless is quite remarkable (a) that the absolute funding in mission-oriented policy has declined so considerably in real terms, and (b) that diffusion-oriented measures in terms of funding became dominant. Taken the indirect measures as furthering the appropriate conditions for innovative activities and adding them to diffusion orientation accentuates this result even more, with a takeover in 1987. Doing this we also come closer to Ergas’ original classification made in 1987. The decline of mission-oriented funding in absolute terms is mainly due to drastic reduction in funds spent on Nuclear energy research (E2). The increase

in diffusion-oriented funding can be traced back to (i) the continuous increase in the number of promotion areas classified as diffusion-oriented and to (ii) the considerable steep increase in the funding of indirect measures (mainly promotion area T). The relative development of policy-orientation and the final dominance of diffusion orientation clearly shows up in Fig. 5, where the respective relative shares of these two broad categories — diffusion-oriented plus indirect and mission-oriented — are depicted. Here, the dominance of mission-oriented policy ends in 1986. So, since the mid 1980s, an overall diffusion-orientation — measured in a relative way — of German technology policy can be claimed. This

Fig. 4. Development of expenditures in mission-, diffusion-oriented and indirect measures.

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

773

Fig. 5. Relative shares of mission- and diffusion-oriented policy.

dominance seems to become even stronger as a further diverging development seems to take off since the mid of the 1990s. This increasing dominance can be traced back to recent promotion areas like new materials (L1), information and communication technologies (I), etc., in which the design of the policy instruments more and more refrains from specifying technological directions and targets, leaving it to the creativity of market and non-market actors. Taking this development one may ask for the reasons of this policy change. Of course, political reasons (e.g. with respect to a changed opinion to nuclear energy research) are to be taken into account. However, furthering diffusion and thus heterogeneity is always an appropriate measure, when the policy-makers do not have a clear idea about the direction of further development. However, the insight that the innovativeness of a country depends on collective innovation, too, might also have induced this policy shift. In this sense, policy does not have to lead and to direct the development but only to manage or moderate it — or only to be one actor in the collective process. This issue is the last to be investigated.

of public funds and whether the research projects are organized in a so-called Forschungsverbund, i.e. research collaborations where several actors, market and non-market, are involved. Fig. 6 shows the respective shares of co-operative R&D projects in the total number of co-operative research projects, again classified according to the different policy styles. Whereas the share of co-operative R&D in the fields of basic research and missionoriented policy measures are comparatively low — together they claim only about one-quarter of all funded co-operative projects — the bulk of co-operative research projects can be found within the group of diffusion-oriented policy measures. However, if this number is weighted with the amount of money spent for single measures, the situation

5.4. Collective invention The database applied allows to test Ergas’ hypothesis of co-operative R&D as a significant feature of diffusion-oriented policy design. Within the single promotion areas, the so-called Förderkatalog of the BMBF also gives information about the recipients

Fig. 6. Shares of co-operative measures 1996.

774

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775

changes significantly especially favoring large-scale basic-research projects.

6. Conclusions For the classification of technology policy in an evolutionary framework, the taxonomy to distinguish between mission- and diffusion-oriented policy design introduced by Ergas (1987) is useful. In particular, this scheme gives a clue whether the policy-makers, in a mission-oriented way, aim at the realization of specific goals, thereby neglecting the development potentials arising out of heterogeneous technological approaches, or whether they are considered to be guided by diffusion-orientation in an overall unspecified search and experimentation process driven by the creativity of a multitude of actors. Anyway, Ergas’ indirect method of identifying the respective policy style is only able to give a first and partly crude insight in the technology policy of a specific country. To improve on that, we develop a so-called direct method aiming at a sound characterization of single policy measures and giving a more detailed picture about policy orientation. By applying this direct method on data describing German technology policy over the last 20 years, we outline a significant reorientation in the design of policy measures. First, for the period since the mid 1980s, we identify a dominant focus on basic research funding compared to funding of applied research and innovation itself. Second, with respect to applied research and innovation, for the same period a steadily rising dominance of a diffusion-orientation in technology policy in Germany can be stated, which also gives a hint on a kind of policy learning. Whereas up to the mid 1980s large-scale mission-oriented programs were in the center of interest, significant problems and failures in these promotion areas lead to an increasing emphasis on diffusion-oriented and indirect measures. Third, with respect to funding collective invention and research co-operation, for 1996 and with respect to the number of projects, diffusion-oriented co-operations take a share of close to 75%. We interpret these results as a clear reorientation of German technology from specifying and defining research objectives to a more institutional and indirect funding of a diversity of actors and projects with competition — non-market (as in the university sec-

tor) as well as market based — working on selecting the comparatively best solutions. For the areas of applied research and innovation this reorientation shows up in a switch form mission dominance to a straightforward diffusion-oriented policy focus. The change of policy design detected here is more of interest in its own respect than with respect to the analysis by Ergas. In order to get an answer to Ergas’ policy assignment problem in an international comparison, it would be necessary to apply our direct method to data from other countries. This might validate or falsify the Ergas classification when we look at the data early in the 1980s and it would allow us to track the development of the respective policy orientations over time in a comparative way. This is one point on the future research agenda. Policy obviously is not an end in itself but the success or failure of specific policy measures determines among others a country’s technological potential and its degree and intensity of exploitation. In this respect, it would be interesting and useful to define measurers for the success of policy interventions and to accordingly compare different policy styles. This is another issue for future research. Acknowledgements For helpful comments we thank Maureen Mc Kelvey, J. Stan Metcalfe and two anonymous referees. References Allen, R.C., 1983. Collective invention. Journal of Economic Behaviour and Organization 4, 1–24. Cantner, U., Hanusch, H., 1997. Ansätze zu einer schumpeterianischen Wachstumspolitik. Ifo-Studien 43 (2), 287–308. Cantner, U., Pyka, A., 1998. Absorbing technological spillovers, simulations in an evolutionary framework. Industrial and Corporate Change 7 (2), 369–396. Carlsson, B., Eliasson, G., 1995. The nature and importance of economic competence. Industrial and Corporate Change 7, 369–394. Cohen, W.M., Levinthal, D.A., 1989. Innovation and learning, the two faces of R&D. Economic Journal 99, 569–596. Conesa, E., 1997. Mission-oriented research programs and the evolutionary dilemma between diversity and standardization. Paper Presented at the EAEPE Conference in Athens, Greece, November 6–9, 1997. Cowan, R., Foray, D., 1997. The economics of codification and the diffusion of knowledge. Industrial and Corporate Change 6, 595–622.

U. Cantner, A. Pyka / Research Policy 30 (2001) 759–775 Dosi, G., Nelson, R.R., 1994. An introduction to evolutionary theories in economics. Journal of Evolutionary Economics 4 (3), 15–72. Eliasson, G., 1990. The firm as a competent team. Journal of Economic Behaviour and Organization 19, 273–298. Ergas, H., 1987. The importance of technology policy. In: Dasgupta, P., Stoneman, P. (Eds.), Economic Policy and Technological Performance. Cambridge Univ. Press, Cambridge. Ergas, H., 1992. A Future for Mission-Oriented Industrial Policies? A Critical Review of Developments in Europe, OECD working paper, Paris. Foray, D., 1995. Coalitions and committees: how user get involved in information technology standardisation. In: Hawkins, R. (Ed.), Standards, Innovation, and Competitivness: The politics and Economics of Technical Environments. Edward Elgar, Cheltenham, UK. Haagedoorn, J., Schakenrad, J. et al., 1990. Strategic partnering and technological Co-operation. In: Dankbaar, B. (Ed.), Perspectives in Industrial Organization. Kluwer Academic Publishing, Dordrecht. Justman, M., Teubal, M. et al., 1996. Technological Infrastructure Policy (TIP): creating capabilities and building markets. In: Teubal, M. (Ed.), Technological Infrastructure Policy. Kluwer Academic Publishing, Dordrecht. Kline, S.J. et al., 1985. Innovation is not a linear process. Research Management, 36–45, July–August.

775

Knight, F.H., 1921. Risk, Uncertainty and Profit, Chicago. Lundvall, B.A., 1992. National Systems of Innovation, London, Pinter. Metcalfe, J.S., 1995. The economic foundations of technology policies: equilibrium and evolutionary perspectives. In: Stoneman, P. (Ed.), Handbook of the Economics of Innovation and Technological Change. Blackwell, Oxford, UK, pp. 409–511. Mokyr, J., 1990. The Lever of Riches. Oxford Univ. Press, New York. Nelson, R.R., 1994. National Innovative Systems. Oxford Univ. Press. Pyka, A., 1997. Informal networking. Technovation 17, 207–220. Pyka, A., 1999. Der kollektive Innovationsprozeß — Eine Theoretische Analyse Informeller Netzwerke und Absorptiver Fähigkeiten. Duncker & Humblot, Berlin. Schumpeter, J.A., 1912. Theorie der wirtschaftlichen Entwicklung, Duncker & Humblot, Berlin, 8. Auflage, 1993. Stokes, D.E., 1997. Pasteur’s Quadrant, Basic Science and Technological Innovation. Brookings Institutions, Washington, DC. Tassey, G., 1991. The function of technology infrastructure in a competitive economy. Research Policy 20, 345–361. Witt, U., 1987. Individualistische Grundlagen einer evolutorischen Ökonomik. Mohr Siebeck, Tübingen. Witt, U., 1993. Introduction, In: Witt, U. (Ed.), Evolutionary economics, Edward Elgar, Chelteuham, xii–xxvii.