On Incentives in Technology Policymaking: What the EU can learn from the U.S. developments. ³

Wolfgang Gick ^{1 2}

Center For European Studies

Program for the Study of Germany and Europe
Working Paper Series #7.7

January 2001

Abstract

This paper focuses on the incentives political and bureaucratic actors face in the institutional setting of EU technology policy. In examining the implications and assumptions of neoclassical and evolutionary theories of technological change, it tries to answer why some approaches are difficult to translate into policy designs. By focusing on positive policymaking the paper examines why policy learning does not occur in certain institutional settings. A special focus is on the informational constraints that limit policy design. Since evaluation and oversight mechanisms have not been sufficiently developed and accepted within EU settings, there is much room for inefficiency, as a basic agency model with hidden information shows. If political planners have incomplete information about the state of the world, programs cannot be designed efficiently. Hence, a better link between evaluation and program design could reduce inefficiencies. Regarding this point, current discussions in U.S. policymaking seem to focus increasingly on program design effects and policy implementation.

1. Introduction

This paper focuses on a simple observable fact. It tries to explain why policies can persist over time despite substantial shortcomings having been identified. In many political settings there are several areas waiting for change despite the presence of considerable criticism. Moreover, it is apparent that such political institutions admit in general to critiques of their operation and emphasize their awareness of having such shortcomings. It is even common that technology policies in the EU, while publicly maintaining that they should have overcome the weaknesses of earlier programs, do not, in fact, change their instruments at all. This is even more astonishing because, during the past decades, there has been a growing literature focusing on innovation, growth, and technology policy.

Notwithstanding the research to be done in the field of policymaking, my argument is that the persistence of old instruments and programs is due to the incentives that political and bureacratic actors face in certain settings. Although I focus on EU issues, the approaches discussed in this paper can be applied to other settings. By using a simple agency framework I try to back some arguments that can be found in the policy literature and in Wilson's (1989) theory of bureaucracy.

Regarding the viewpoint of the evolutionary theories of technological and institutional change, the contribution of this paper can be seen as an attempt to explain why in a certain setting no coevolution of political and bureaucratic institutions occurs. Implicitly, it tries to contribute to the discourse between modern policy analysis and evolutionary theories started in Nelson and Winter (1982). It finds that little is yet understood about the policymaking process and the role of incentives. In offering an extended viewpoint of transaction cost politics it follows Dixit's (1996) rationale on the imperfectness of the real world: "[T]he state of the world may not observable ex post, or even if observed by the parties of the contract, may not be verifiable to an outsider whose job it is to enforce it." This largely follows Nelson's 1994 statement that learning processes in a technological evolution cannot simply be presumed to be effective enough and that human and organizational learning depends on incentives and pressures.

The paper focuses on a common group of policy approaches in technology policy programs: the so-called "cost-shared actions" in which the EU subsidizes a certain percentage of a firm's R&D expenditures. This type of technology policy has gained ground in the last decades in many OECD countries but has reached certain patterns of program design within the EU. Since firms apply for subsidies and governments usually are not the final consumer for R&D undertaken by the firms, there is a general need to control for the government's decision about which firms to subsidize.

The aim of the paper is to discuss solutions that could improve these policies. Any solution has to cope with the incentives given in an existing institutional framework. Notwithstanding the ongoing changes in EU policymaking since the Davingon Report (European Commission 1997) there is still the need to focus more on contractual settings and program design effects as current discussions on the reform of U.S. programs show. ⁴

The paper is organized into the following sections. The second section examines the different theories of technological and institutional change. In the third, I refer to the policymaking process, the suitability of these theories, and the definition of political learning. The fourth section analyzes EU technology policy, the decision-making process, and the impact of evaluations; it also offers a simple agency where the lack of information restricts the principal to design a suboptimal contract. model with full and with asymmetric information. In the subsequent conclusion I discuss the policy implications.

2. Neoclassical and evolutionary theories of innovation and technological change and their impact on policies ⁵

2.1. New Industrial Economics

One of the central neoclassical contributions has been D'Aspremont's and Jacquemin's (1988) model on R&D spillovers using a two-stage oligopoly game with quadratic costs of R&D to display decreasing returns. In the first stage, two identical firms conduct research leading to a reduction in unit cost; the firms are Cournot competitors in the second stage. Thus, D'Aspremont and Jacquemin compare the magnitude of cost-reducing technical advances achieved when firms conduct R&D competitively versus cooperatively. The key point is that the firms cooperating in the first stage may encounter a lack of incentive to undertake R&D due to a market failure stemming from uncertainty regarding the other firm's R&D activity. The solution that D'Aspremont and Jacquemin offer is the subsidizing of cooperative R&D. Moreover, social welfare is higher in any case in which subsidies apply. ⁶

The basic model has been modified and extended to include information sharing (Beath et al. 1990 and Kamien et al. 1992) and the more general view expressed in Suzumura (1992). Other approaches like Katz (1986) examine the relationship between R&D cooperation and product market competition while Spence (1984) focuses on the possibility of internalizing externalities in Research Joint Ventures (RJVs). Commonly, these approaches treat R&D as a cost and correspond with the normative and market-failure approach discussed below.

2.2. Evolutionary theories and the knowledge-based approach

Evolutionary theories of technological change refer to broad issues of firm-specific questions. A group of evolutionary theories refer to the possibility of creating technological opportunities and the capabilities to exploit them. Thus, evolutionary theories go far beyond linear models of technological change. Their focus, in general, is on the dynamic performance of an economy. This in particular raises the question of whether equilibrium theory is the right framework for describing knowledge production. Schumpeterian theories focus on the fact that economic and technical progress occur in the carrying out of new combinations, which happens in out-of-equilibrium situations where innovative entrepreneurs are temporarily able to gain monopoly rents. Nelson and Winter (1982) focus on firm behavior and performance, on the nature of competition, and on the primary processes driving short- and long-run economic change. A central component of their view of innovation is the portrayal of the discovery of alternatives and the choice among alternatives. In this realm the following topics play a role:

Technological trajectories and paradigms

Following Nelson and Winter, technological trajectories describe the cumulative and evolutionary features that mark the developments and changes experienced by technologies. Technological progress thus follows certain trajectories characterized by technological improvements. McKelvey (1996) attributes to this concept of a trajectory some predictive power, in that certain types of firms are more likely to develop certain types of technology than others.

Similarly, technological paradigms (Dosi 1982) embody a definition of the relevant problem and a pattern of inquiry. Innovative activities are strongly selective, finalized by precise directions, and cumulative in the acquisition of problem-solving capabilities. The solution is simply found inside the paradigm.

Variety and selection

Another key concept in evolutionary theory is the generation of variety. Following Saviotti (1996) firms try to differentiate themselves from rivals through a series of innovations. Selection processes limit this ability of the firm. As the firm consciously tries to improve its adaptation to the environment, the selection mechanisms involved might be Lamarckian.

The role of knowledge

Knowledge is often a public good to which access by other users cannot be restricted. This does not, however, mean that access is cost-free. Networks, tacit knowledge and innovation are interrelated topics. ⁷ Knowledge important to product or process innovation is not usually embodied in blueprints or operating manuals and, hence, is "tacit". The flows of tacit knowledge are important to firms, not only for their access to technical information problem-solving, but also for access to new knowledge generated through research. Saviotti distinguishes a range of variation of types of knowledge with industrial applications between two boundaries. ⁸

Networks

The centrality of research spillovers in the process of innovation is at the root of formal or informal "networks". In general, cooperation may cause serious appropriability problems if an information transfer is not compensated for. Thus, the creation of networks and the management of science and technology policy can be important issues if certain restrictions are met.

Path dependence

A powerful argument is the development of path dependence. Arthur (1994) and David (1985) argue that under certain conditions a single technology may achieve a decisive advantage over competing ones—independently if it proves to be the most efficient one in the long run. Arthur's explanation involves a series of assumptions, including learning effects and adaptive expectations. Explanations of why the QWERTY keyboard or the VHS video cassette made the grade are within this argument.

Absorptive capacity and the role of research spillovers

In the evolutionary literature research spillovers have been defined to "[i]nclude any original, valuable knowledge generated in the research process which becomes publicly accessible, whether it be knowledge that fully characterizes an innovation, or knowledge of a more intermediate sort." ⁹

The concept of "absorptive capacity," developed by Cohen and Levinthal (1989), relies on the firm's ability to assimilate and integrate new external knowledge into its knowledge base in order to use a large amount of the knowledge available. In doing so, one firm might then be more successful over a certain period than another firm that produces only one technological success over time.

National Systems of Innovation (NSI)

An extensive approach to cover and describe a country's future chances is the concept of a National System of Innovation which is based on two main conceptual foundations. The first covers the importance of macro-level institutions and "[t]he characteristics of the principal education, innovation and investment-related institutions, the degree of coherence between these and the way they link together". In many countries, governments are the largest purveyors of economy-wide externalities ¹⁰ . Another line of inquiry focuses on linkages between inventive activity and the economy, especially the extent and manner in which inventive activity affects economic development and competitiveness. ¹¹

In a geopolitical setting inventive activity often transcends national boundaries. There have been extensive studies on the impact of strategic alliancies and extent to which firms perform research abroad. Schott (1994:27) defines a transnational system as an institutionally regulated activity performed by actors who are significantly related within and across national boundaries. Section four examines the extent to which EU member states form such a system and, in consequence, whether or not EU technology policy will be able to create stable regional networks. ¹²

The second foundation concerns the conditions defining how private agents can capture particular returns stemming from successful interaction among institutions. This, of course, leads again to the concept of the distribution of knowledge in a society and the particular capability of a firm to exploit these processes over time.

3. The realm of policymaking

There is a gap between policy design and evolutionary theories since the latter do not focus on incentives and instruments. In my opinion the neglect of a clear instrumental choice in evolutionary explanations has been often misunderstood as a tacit agreement for policymakers to keeping up current instruments and for interpreting any instrumental environment as sufficiently good, since some degree of learning might occur anyway. ¹³

One reason why evolutionary theories—despite their ability to contribute substantially to a new understanding of the processes of innovation, economic change and technological performance of an economy—still have a weak impact on redesigning policy programs is that they do not necessary to focus on the fact that these theories do not involve political actors and institutions and hence do not refer to shortcomings inside the world of politics and bureaucracy. This has simply not been their goal. Political institutions may coevolve, but they play only a low-key role in a dynamic process among firms. Instead, evolutionary theories contribute much to the question of technological change in institutional settings where the traditional policymaker plays a subordinated role, if any.

3.1. Policymaking processes

The real world involves policymaking. Thus, any theory, if it wants to enter the realm of policy analysis, needs to design a realistic framework which is able to answer policy questions and to analyze the instrumental choices involved. The rationality criterion required by the practitioner is simply the fact that any critique of the existing political or instrumental setting needs to assume, to some extent, that there are rationally better policies applicable than the ones currently used to overcome certain shortcomings. Theories that are unable to contribute to this question at all are unlikely to achieve any results inside the realm of political institutions. They are simply not applicable.

Rationality criteria are not superfluous to political settings even if they are insufficient to explain overall technological change.

One of the first and best publications referring extensively to the evolution of public policies in the framework focused in this paper has been Nelson's and Winter's (1982) evolutionary theory of economic change. Their discouraging thoughts on the evolution of policies reflect the trade-off between evolutionary thinking and the incentives in political institutions opposed to the picture drawn above. ¹⁴ Nevertheless, there is a need to bring both strands of literature together in order to redesign policy approaches. Due to the importance of control in organizations, efficiency criteria play an important role. This has been often disregarded in the way economists and evolutionary theorists deal with their policy implications. Vague formulations of policy learning do not help to design future policies as long as rationality criteria have not been addressed in the setting of political and bureaucratic institutions. Following Nelson's (1994) statement on should rather conclude "[t]hat one can not presume that individuals and organizations actually see or think their way through to an optimal behavior and then adopt it. Furthermore, neither learning processes nor selection pressures can simply be presumed effective enough so that behaviours actually observed at any time can safely be presumed optimal. On the other hand, humans and organizations do learn, if the environment permits them to do so and the incentives and pressures are there."

The reason for this neglect is that evolutionary theories need to argue outside of rationality assumptions if they want to explain when, in certain settings, the pursuit of rationality may not lead to a shift in institutional settings. ¹⁵ As Lemarié and Joly (1996) point out, rational models treating organizations as bodies of decision-making rules cannot analyze the transition between trajectories. If radical uncertainty applies, there is no policy conclusion to be drawn except the one that there is none.

In turn, the impossibility to explain technological trajectories by efficiency and rationality criteria should not lead to the conclusion that policymakers should abandon rationality for their instrumental design. Instead, as comparative approaches do, theories may explain when ill-suited policy designs lead to unwanted outcomes. Policy analysts cannot suggest political actors to adopt irrational behavior in order to achieve a better outcome if the organizational structure cannot cope with the instrumental choice suggested. In order not to limit themselves, many economists have thus been content to ignore the realm of politics or simply to assume a sufficient degree of policy learning. The point is that by observing coevolutions and organizational learning in general they cannot conclude that any coevolution determining the emergence of new policies is the right one. Instead, they have to develop a rationale that explains where the shortcomings preventing the right development come from. In any policy setting scholars should be aware that there is more scrutiny required than simply a general description of processes. Scholars set themselves too easy a task if they only tell practitioners that in the long run there will be some sort of organizational learning—if at the same time they are unable to explain why inefficient outcomes cannot be avoided by changing some variables in the institutional setting.

This, in turn, does not mean that policy learning does not matter. By focusing on incentives in political and bureaucratic agencies, this paper wants to contribute to an understanding of the question: under what conditions does learning occur? In doing so, the paper goes beyond simple behavioral assumptions on the variability of information sets and assumes, following evolutionary theories, that the incentives and the objectives of actors can vary considerably. ¹⁶

Instrumental ignorance

The same thought applies to the instrumental debate. As we must start with the assumption that, in a market economy, the market approach leads to efficient outcomes, there is certainly a need to justify market interventions such as subsidies. The economics of innovation does not exist in an instrumental vacuum. Scholars have to define the instruments to be used in any policy framework as soon they enter the realm of politics. If, for example, deregulation is more effective at furthering collaboration among firms and related research institutions, one cannot neglect this and simply suggest that subsidies be used to counteract the existing disincentives without analyzing the losses. An economist has to explain why other instruments offer no viable approach and why the proposed new approach leads to a second or even third best outcome.

One cannot separate instrumental issues from efficiency. This is a bridge that some scholars should never have crossed before they knew the impact of their words in the world of policymaking. In particular, scholars involved in consultation and evaluation processes and dependent on political actors should avoid giving rise to misunderstandings. ¹⁷ General arguments—e.g. that any subsidy is better than none because there will be at least some positive effects anyway towards the desired technological end—may create major confusion and are not helpful to the policy debate. To beliefe in an unquestioned continuation of policyes at the other end has proven to be even more dangerous. ¹⁸ Instead, economists should heed Smith's statement that high budget deficits in OECD countries feed into the drive for a more "effective" policy. ¹⁹

The theory of market failure politics

Regarding the design of EU technology policy programs a series of organizational issues have been neglected. Recall the policy conclusions derived from the D'Aspremont/Jacquemin model discussed above. The conclusion that subsidies are the solution is powerful precisely because it so well fits the market-failure approach adopted in politics.

The main reason why these groups of models still gain ground in policymaking approaches is their rather easy translation into an instrumental design. Political and bureaucratic institutions prefer "market failure approaches" because they fit into their understanding of rationality. Their long-lasting experience and the relative learning procedures show a close affinity to this approach.

This prompts the question of whether normative or positive approaches can explain the incentives of political actors and institutions. Curing market failures raises the power and budget of the involved governmental institutions. And, currently, Europe is looking back on several decades of "well-functioning" settings between government and markets where no evaluation techniques have signaled the political actors the needs for change and adaptation. Political and bureaucratic actors have been able to play their games under a veil of imperfect monitoring. The failure of political principals to engage in active oversight and control plays a major role in this issue as shown below.

In this way, D'Aspremont and Jacquemin are not at all to blame for having lead to a simplified policy. The model is perfectly "right" in the sense that it is appropriate for the market-failure approach. It is easy applicable to the existing bureaucratic and political settings.

The right question to ask here is not which coincidence among existing EU programs and the predictions of evolutionary economics might exist. Of course, most of the existing programs involve "cost-shared actions". But to argue it would be the goal of any program just to subsidize R&D cooperation in the long run would only set a seal upon existing shortcomings instead of moving technology policy towards the consequences of evolutionary theories.

One might argue of course that new EU programs and frameworks still work in the same way but use other—namely evolutionary—language. Since many evolutionary theorists do not develop their theories in order to lead to different instruments, it is likely that policymakers will not change their instruments if no further results stemming from the evaluation of earlier programs do not find their way back into the process of policymaking.

Far more scrutiny is necessary in this field. As Metcalfe states, technology "[p]olicy makers must address complex questions of institutions and their connectivity", and "[t]he contrast between the theory which underpins policy and the implementation of policy will seem acute." ²⁰ Thus, focus should not be on market failures but on institutional failures that refer to the myopia of the political and bureaucratic system. In summing up our thoughts on evolution and learning, we must take note of institutional settings which do not learn.

In some way, this refers to Patel's and Pavitt's (1988) notion of myopic systems. The distinction here is that there is no intention of criticizing systems which underinvest in R&D, but rather of referring to political actors who avoid control, that is, to the myopia in settings where learning could occur but the policymakers are constrained by political rules and interests. ²¹

3.2. The normative approach: the welfare agencies create

This old approach derives from social contract theory in which the government acts in favor of the people to maximize social welfare. ²² Applying this approach to the outcome of the D'Aspremont and Jacquemin model in policy perspective, it is apparent that by subsidizing R&D a government will be able to achieve a higher degree of social welfare. As long as a welfare function is applied where welfare increases with the utilities of the individuals, the result is straightforward. ²³ As discussed above, the market-failure approach is based on this normative view. Despite its unrealistic reduction of policymaking to a technical problem, and its shortcomings in the results of many programs, the market-failure approach has been very successful in the past. ²⁴

3.3. From the positive approach to organizational settings

To examine the reason for this compatibility, it is necessary to leave the normative world. Theories of optimal outcome do not govern the acts that determine the policymaking processes.

A different approach to the normative one which rules out its argument, is the contractarian one. Buchanan (1975) has contrasted theories of social choice with the rules governing political processes. The distinction between rules and individual actions in policymaking lead to an outcome in which policymaking is constrained by the working of the process within the rules that were established ²⁵ .

Our key question—similar to Noll (1988:1271)—must be put as follows: Are technology policymakers driven by a desire to create an optimum policy or are they driven by their own objectives? The crude positive approach refers to the second leg of argumentation and simply concludes that there are no degrees of freedom in the realm of policymaking and that normative analysis is irrelevant inside these processes.

The refined version of this approach, which is developed here, is that in the process of technology policymaking the political actors have their own incentives and pursue a state of the world in which their power increases. In any setting involving bureaucratic and political agents, one must be aware that agents may behave opportunistically, "[p]ursuing their own interests subject only to the constraints imposed by their relationship with the principal." ²⁶ In this paper I will underline these arguments with a simple model that shows how in a policy program a political planner lacking of the relevant informations over the state of nature is unable to design efficient policies.

As long as the overall institutional setting allows political agents to pursue their goals and no pressure is applied to these agents to change their behavior as empire builders, this state of affairs will continue. Concretely, the persistence of inefficient programs is merely due to a lack of monitoring that attributes more power to these agents than is necessary for them to have in order to fulfil their tasks. Hence, the solution consists of a policy learning process in which the useful information about the link between policy instrument and outcome can be redirected into the policy design process in order to lower the costs of control.

3.4. Understanding economic policymaking as a political process

There are several other general remarks pertinent to an assessment of incentive problems. Incentive problems do not arise in an Arrow-Debreu world in which individuals as well as organizations perform in the contracted way. Here a theory of incentives in policymaking needs to be applied in order to show how the political process performs a political action. Rules and acts and their adaptation cannot be analyzed in depth within simple positive settings.

Laffont and Tirole (1993:475) contribute to the process of policymaking and focus on the constraints policymakers face in their actions. They extend their general thoughts by including information asymmetries and referring to agency relationships between politicians and their delegates in the bureaucracy. Their contribution to the process is the focus on the constraints policymakers face in their actions.

Informational constraints and transaction costs

The classic argument that refers to an endogenous action is known as moral hazard. It applies when the acts of an agent cannot be sufficiently controlled by the principal. The agent faces an informational advantage over the principal, his effort is not observable. ²⁷ As argued in this paper, the lack of information leads to a less efficient policy.

It is worth adding that from a contract theory perspective the notion of transaction costs is strongly related to this point. If the principal cannot design a better policy due to the lack of information about the state of the world this may lead to the persistence of ill-structured incentives. Implementing different incentive schemes by changing the existing contracts may lead to better outcomes. However, in political settings contracts are difficult to change in the way required because the transaction and information costs are very high. North (1990) explicitly connects the state and the persistence of political settings to the lack of instrumental rationality of the participants. ²⁸

Credibility as a political constraint

Related to transaction costs is the question of credibility which has been explored by Persson and Tabellini (1990). In going beyond Williamson's concept they define economic policy as a game in which credibility constraints apply. They distinguish this case from transaction costs and define political constraints as applying when there is a conflict of interest between the policymaker and the citizens because of disagreement over the final goals of politics. In a manner similar to Kydland and Prescott (1977) they analyze political constraints that coincide with the interest of the society only to some extent. The reason for many developments in the political world lies in the policymaker's possibly being subject to a binding incentive constraint. When time inconsistences apply, this may lead to a lack of credibility.

Lack of comparison of political actions

To these views Dixit (1996) adds the realistic notion of "real time" to the policy process in order to overcome some of its overly rigid settings and to combine the viewpoint of contract theory to explain why rules and acts differ. In this way Dixit offers a dynamic approach: Policymaking is seen as a dynamic game with changing and uncertain conditions in which the rules are made by the participants.

One addendum is of further importance and, despite it has been discussed in different political theories, it needs to be analyzed in greater depth: many government agencies sustain a monopoly position. Their performance is hard to assess. ²⁹ One could even argue that bureaucratic institutions in general try to bias oversight mechanisms. In many settings, other actors from outside, other governments or their counterparts, may overcome the problem of applying control to the institutional setting.

3.5. Bounded rationality, political learning and the coevolution of political institutions

This concept also meets partially the evolutionary perspective proposed by Metcalfe (1994). His notion of the adoption of a behavioral theory of the firm and its focus upon learning processes cast the policymaker "[a]s a behaviourist engaging in adaptive policy making, seeking improvements in the technological performance of the economy in the context of immense micro complexity." ³⁰

Herbert Simon's concept of bounded rationality refers to the notion that individuals have insufficient cognitive capacities. This basically entails a critique of the neoclassical assumption of rationality. Though Simon's work changed over time, together with his critique of neoclassic theory, it is useful to examine which concept of "bounded rationality" should be applied to the behavior of political agents that we accept for our framework. ³¹ Agency relationships are more complex in the political context than in most interfirm relations. It is necessary to face the fact that any policy selection occurs more slowly than market selections do and it is important to ask about the circumstances. ³² This fact can be explained by the very complexity of regulations and actions. In particular, the lack of oversight mechanisms and of observability ex post may entail adaptations to be very incomplete over time. ³³

Policy learning

Smith's (1996) notion that policy changes rarely follow a coherent rationale and, instead, often take narrow approaches to the aims they seek to achieve, leads to a next argument. Political learning consists of changes in policy strategy on the institutional plane. May (1992) distinguishes two different types of political learning. Instrumental learning encompasses the technical plane. Here optimization can take place, through improvement of policy instruments, better understanding of implementation, the adaptation of instruments used. In contrast, social learning refers to the social construction of the problem on an institutional plane. ³⁴

Technology policy issues refer to both types of learning. As May states, both types are interconnected, and of course the adoption of new policies involves learning. But the question is still, to what extent subsequent policy changes reflect learning, and whether the mimicking of policies is undertaken, since some political actors might achieve reputation in copying programs from other governments. This latter issue has been known especially in the area of technology policy. MITI is also an impressive example of how a bureaucratic and political actor can gain power by imitating technology policy programs. The EU has, like other governments, intentionally tried to follow its achievements. Political actors have a clear interest to imitate programs from abroad as long as information costs are high and as long as still no evaluation results show the real impact of these policies.

In this light one must reinterpret the Commission's tendency to gain reputation by imitating programs from its foreign counterpart (MITI). In general, playing the role of a powerful agent inside Europe creates more distance and reduces the risk of being evaluated in a field where evaluation procedures are difficult to apply.

Second, imitation fits into the incentives of the specific political and bureacratic world ³⁵ in which the agent is allowed to pursue a certain technology policy that has been perceived as successful. The search for similar agents and for institutions which perform similar issues results in protection of the status quo.

What "political learning" means in the context of technology policy

The often cited Japanese VLSI project highlights the difference between mimicking and policy learning. VLSI is well-known as the principal vehicle enabling Japan to join the leading players in the semiconductor industry at the end of the 1970s. In reality, the VLSI program adopted many elements of the Research Association system in the UK.

Suzumura and Goto (1997) explain some of the substantial modifications that were added to the UK version. First, Japanese Research Associations (JRAs) were organized to solve specific technological problems and second, they were dissolved after having performed their task.

The third and crucial point is that the JRAs usually do not perform their research jointly. Such an approach may run risk of wasting resources in multiple discoveries, but it ensures clearly the appropriability of the firm's effort and reduces disincentives in joining R&D consortia. ³⁶

In other words the very "political learning" in the VLSI case has not been the mimicking of the former UK program. It has been MITI's implementation of a different contractual setting protecting the research outcomes of any research unit. This lead to a coalignment of interest of the participants in the Japanese R&D setting. As shown below, one can doubt if EU programs until the Forth Framework Programme have followed this idea. ³⁷

4. The Making of Technology Policy in the EU

At the EU level there is a mix of technology policies ranging from large-scale interventionist programs to those furthering generic research. As Sharp and Pavitt (1993:136) posit, the political success of ESPRIT in the mid-1980s led to a mushrooming of similar programs based upon the principle of precompetitive, collaborative research. Since 1987 the Single European Act, as a revision of the EEC treaty, introduced Article 130f, which stipulates the need for competence in the field of technology policy.

I henceforth refer to the indirect actions in the EU Framework programs that have been drafted as "cost-shared actions" at the beginning of the 1980s. It is worth noting that these initially strategic EU concepts remain unchanged and are readily observable. As mentioned at the beginning, the EU in its Framework programs supposedly has been driven to concentrate on a few but more effective single programs. Regarding the Third Framework program the Commission tried to reduce the specific programs. However, this is not true. By analyzing the Council's decision it is easy to show that all earlier existing programs have been continued in the new Framework program. ³⁸

From the practitioner's point of view cost-shared actions came to dominant importance because they allow the Commission to:

• perform different programs without increasing bureaucratic effort;

• give financial incentives to certain industries; and

• avoid being controlled by the Council in certain circumstances. ³⁹

By focusing on informational and transaction constraints in policymaking, I want to explain why inefficient policies persist. Again, I specificly refer to the EU concept of "cost-shared actions" that subsidizes R&D cooperation among the different member states.

4.1. EU Technology Programs and observable shortcomings

The pork barrel argument

As stated earlier in this paper, many incentives in the realm of technology policy can be described from a pork barrel perspective. ⁴⁰ Cohen and Noll argue that politicians face incentives to use technology programs to reward constituents instead of focusing on the correction of other shortcomings. The argument might still have an influence here, but the political power of interest groups to further specific programs is higher in strategic areas and does not explain the persistence of inefficient programs when firms have an interest in R&D success. The reason why I do not adhere to the pork barrel argument as an explanation for the emergence of EU technology policy programs is simple: cost-shared actions furthering R&D cooperations have a broad impact throughout the EU and offer a rather small payoff to interest groups compared to special programs furthering leading-edge technologies.

Moreover, the application procedures are particularly costly to the firms. Following different studies in the EU member states ⁴¹ this has been a significant reason why many firms abstain from applying. Instead, what exists is an organizational constellation in which the Commission's interest in enhancing its reputation in the field of technology meets the firms' interest in subsidies. The initiative, however, could hardly be attributed only to the industry or its pressure groups. Rather, Schmidt's notion that EU policy formulations balance business interests might illustrate to some extent the technology policy case. ⁴²

Can multiple goals and divided structures further efficiency?

Next we must note that the EU tries to meet regional and structural objectives. However, cohesion and cooperation might only be wishful thinking that seldom achieves any measurable result in a general regional framework. Agents with power to deviate from group interests and superpose their own objectives over existing national settings cannot do better when they want to create an empire.

Another point refers to the internal organization of technology policymaking. The EU Commission's structure and the tasks of its different DGs (Directorates General) are highly heterogenous. ⁴³ The above cited and highly praised ESPRIT program has been allocated to DG III, while most of the other programs are administrated by the recently reorganized DGs XII and XIII. ⁴⁴

Disincentives created by shortcomings in RJV contracts

Many R&D programs in the EU are far from efficient; because of bad contractual settings in the function-performance link they risk disincentives. Kastrinos (1996) refers to shortcomings in the standard contract used in EU R&D programs regarding intellectual property rights. The key point is that each partner cannot prevent other partners from exploiting knowledge generated within the project. Here a trade-off between function and performance applies if the contract does not take into account the costs of adoption when performance criteria are determined. ⁴⁵ In refering to practical RJVs, Kastrinos subsequently shows that appropriability considerations have not been sufficiently encompassed in the Commission's contracts.

In one example contractual shortcomings led to severe problems in a generally successful RJV when a small firm, depending on two large software houses which had been able to keep control of the specifications of the innovation, was not able to appropriate the result of the RJV, as the software houses were enabled by the incomplete contract to exploit the knowledge generated within the RJV against the firm. After the end of the RJV, the user firm had to buy an expensive new system from its former partners who, in the meantime, had commercialized a new version and left the user firm with a slower and less effective product.

The stability of networks

As referred to at the beginning of this paper, I focus on the interrelation among national and transnational networks. Let us first assume that the EU might be able to establish a new network, with links accross the entire Community. Then the question arises: what conflicts between the National Systems of Innovation and the EU network will occur, and to what extent will the efforts of the EU be balanced by other forces allocated at the NSI level. As Stubbs and Saviotti (1994) put it, we have to ask whether the EU networks will ever be stable across the entire community. This is of course a question of opportunity costs and to what extent the resources allocated at the EU level might impair the performance of NSIs. ⁴⁶

Does it make sense to encourage research collaborations?

Lichtenberg (1997) concludes that, in general, the idea of strenghtening R&D networks may be unnecessary. Policies attempting to encourage international research parterships may be ineffective or misguided. Studies on knowledge transfer show that industries tend to cluster if knowledge spillovers are prevalent. This, in turn, suggests that, in other industries where knowledge spillovers play a less important role, a technology policy supporting networks might not even be necessary.

The low importance of EU regions in a geopolitical link

To the above mentioned question about how the EU might establish a stable network, Schott (1994) offers an interesting answer. He focuses on the extent to which collaboration among nations is embedded in a geopolitical link and measures this link through its particular intensity of trade. Schott then tests the hypothesis that collaboration in invention among nations is embedded in and promoted by the geopolitical links among them. Subsequently, he identifies six geopolitical regions where such an integration could be measured and found out that collaboration among EU member states was at the lowest range of all the six cases of geopolitical integration he was able to identify. ⁴⁷ This, of course, is a strong argument against the hypothesis of embeddedness of the EU regions in the contest of technology policy.

Does any widespread R&D subsidizing have any effect at all?

Verspagen (1997) offers a convincing analysis: simply subsidizing research in all EU regions does not lead to measurable success. Verspagen shows that, during the 1980s, different European "regional clubs" existed, each of them differentiated by productivity, GDP per capita, and unemployment. He concludes first that in the EU advanced technological clusters are much smaller in terms of number of regions than advanced economic clusters. Second, his results show that advanced technological regions are not likely to be geographically connected to other regions. The results of his analysis is striking: It is nearly impossible to stimulate innovation in regions that are relatively backward economically. In his words "[s]imply subsidizing (private or public) research in those regions would not solve the problem because one cannot develop (high-)technology from scratch." ⁴⁸ He concludes that policymakers would be better off concentrating on one or a few central regions, where facilities such as public research institutes or universities are abundant.

Avoiding wasteful research versus ensuring appropriability

As just mentioned in the VLSI case, a crucial element of RJV contracts is to ensure appropriability. There is still no evidence that the desire to avoid wasteful research can meet this requirement. The EU Commission's attitude on coordinating and bundling Europe's technological resources can be considered an outdated one and refers to the resource-oriented thoughts of earlier innovation theories mentioned above.

Globalization and the limits of policies

There is another stark argument against the EU philosophy of broadly furthering R&D cooperation across EU borders. R&D and production are becoming increasingly geographically fluid. This renders long-term employment objectives encompassed by EU TP more and more useless and undoubtedly raises questions over the efficacy of EU-funded pre-competitive R&D that has been in operation since the beginning of the 1980s. ⁴⁹

Furthermore, by limiting the observation to globalized R&D networks among firms, it should become clear that common national and EU-wide technology policy goals furthering innovations in certain regions could become obsolete in a future where firms embedded in global R&D networks may easier get their technical innovations from abroad instead of reinventing the wheel in a costly process involving public funding.

Subsidiarity and the role of the EU Commission

This raises the question if the concept of subsidiarity in an EU-wide setting of technology policy can make any sense at all. Sturm (1996) argues that there is a rising risk of deliberately misunderstanding the subsidiarity principle in allocating programs. This generally leads to the choosing of ill-suited instruments and objectives. As bureaucracies are unable to foresee technological developments, they consequently incur to inefficiencies when funds are allocated to the programs.

The outcome of further debates on subsidiarity at the technological plane is uncertain: the amendmends contained in the Maastricht treaty (Art. 130h, 2) now read that "[t]he Commission may take any useful initiative to promote the coordination...". Starbatty and Vetterlein (1994:60) thus argue that an existing power of the EU Commission will create a shift in the debate. National ministries and the EU Council might have to accept the Commission's tendency to attract more power and to rule out implicitly existing competencies of the member states. Sectors like the aeronautical and space industry, especially, which achieved successful cooperation without the EU Commission in the past might be an unprotected "delicacy" for the Commission which could develop a new concept of necessary actions in this field.

In the past the EU Commission has been clever enough to gain technological competency in return for granting them funds. This supports the above mentioned statement that EU technology policy issues do not encompass the crude "pork barrel" issue but instead entail a constellation in which missing oversight allows a stable, mutually beneficial relationship of reputation and monetary benefits between the Commission and the firms.

The Commission's role in identifying critical areas in EU technology policy has often been noticed. Citing the "extraordinary" importance—which is of course unmeasurable — the EU Commission can easily gain power over large areas of technology policy. ⁵⁰ Following Caswill (1996:48-49) we might conclude that the Commission continues simply to maximize its budget and its own scientific program with the result that most of the impact of European funding is little more than "important rhetoric."

4.2. Oversight and evaluation: When and why political actors do not want to accept the outcome

In short, "[t]he gap between the outcome of evaluations and political ideologies is remarkable." ⁵¹ Although evaluations are often initiated by an ideological debate, it will take far more to integrate the outcome of these evaluations into the political decision-making process. Similarly, to the design of policy programs, evaluation also generally takes place in a too isolated manner, motivated by concerns of ideology rather than efficiency.

Europe is well known for having a large gap between its expansion of research programs and the empirical evidence that underlies these actions. Following Roessner (1989:311) evaluations often focus on objectives which are not under the control of the decision-makers. This, in turn, leads to behavior that favors political action that remains stable rather than one in which evaluation plays a critical role in the political process. Despite Roessner's conclusion citing the need to integrate the evaluation of R&D programs, one must note that political organizations seldom accept their results if they conflict with the existing incentive scheme the organization offers.

Since 1992 evaluations encompassing the EU Framework programs as well as single projects have been binding, they began with the evaluation of the Second Framework program (1987-1991). The EU Commission recently published its evaluation of the Third Framework program. However, these evaluations currently do not have any influence as a ceiling for current or future fundings, and bear the character of an ex-post evaluation without further importance.

Decision-making

Contrary to the general description by Pollack (1997:119) the oversight mechanisms for controlling EU technology policies are more complicated. His idea to "clip the Commission's wings" via revision of the Council does not apply because the default condition is not the expiration of the regulation. The principal cannot wait thus simply for any built-in pressure to apply.

Second, the rule of unanimity only applies to the Framework program. Since the Single European Act was enacted, the rule of unanimity has been replaced by a qualified majority voting for single programs. Since log-rolling is a procedure often used in the Council's decision-making processes, we must realistically conclude that any politically less important issue from the viewpoint of the member states will not be altered at all. Technology policy issues will unlikely be "clipped" if the responsible member states have to pay a high price. Moreover, the voting member states often do not want to join a losing minority or for ideological reasons be regarded as blocking an European idea. Thus, curious as it may appear at first glance, the Council often votes unanimously on these issues.

The role of evaluations in the political process

It is also necessary to refer to different mechanisms of member-state control, in particular the Commission of the European Communities and the various oversight mechanisms available to the political principals. We must see that the institutional framework by accepting the pursuit of old goals of technology policy and not interacting if certain requirements are not met, leads to persistence over time. The reason for this sort of "locking in" has much to do with the incentive the actors face when they set up a program. If the agent cannot be put into an incentive scheme because the final outcome is hardly measurable, much space arises for inefficiencies.

On a control engineering plane, uncertainties about what works and what does not work in innovation programs can be particularly favorable to support a "lock in". Developing an evaluation system that delivers useful results is difficult.

On an ideological plane, even when an evaluation exists, the given incentive schemes inside the setting of political institutions prevent the actors from choosing a better framework for regulation. ⁵²

4.3. A simple model to back the arguments

Within the broader conceptual framework of complete contracting, at least two larger groups of approaches may be applied to analyze the shortcomings of contractual settings with political and bureaucratic actors.

First, common-agency models explain the existence of low-powered incentives in bureaucracies and thus of the persistence of inefficency ⁵³ . The outcome of their thought leads to a series of argument discussed below. Dixit (1996) highlight two properties of common agencies that directly correspond with EU policy settings in which the EU Commission is the agent performing the policy for many political principals (member states, EU Council). In such a setting the agent cannot be properly controlled and gains information rents. Second, the multiprincipal setting leads to further inefficiencies and to the persistence of the status quo (Dixit (1996: 157-171).

A second group of models covers the realm of bureaucracy regulation. ⁵⁴ Similar to regulations of a monopolist with unknown costs, these models try to find an optimal contract in order to delegate an activity to a bureaucrat. A general model and simplified model presented here follows the theory of mechanism design in a setting with hidden information. It will be argued that to a large extent the question of policy design can be reduced to the question which settings allow the implementation of an optimal contract between politicians, bureaucratic actors and firms.

For a series of cases, a basic model one-period model following Mas-Colell et al. (1995) and Macho-Stadler / Perez-Castrillo (1997) delivers an explanation why first-best contracts cannot be implemented when the political planner lacks information about the state of the world.

One important prerequisite for mechanism designin public bureaucracies is ex-ante contract acceptance by the agent. He chooses the relative effort level when nature reveals him the state of the world. The agent accepts ex ante and learns the state of nature he has been given. The contract design has to take into account that the principal cannot distinguish both types of the world.

Such a basic model is not limited to R&D policies where the output is difficult to describe ex ante, but also to a series of other policy settings, e.g. a regional setting where a politician (principal) tries to further the development by different measures for small businesses covering consulting or technology transfer activities. As we will see below, the setting leads to the problem that on the one hand, the agent needs to design a self-selection contract since private information exists. This contract, in order to be optimal under asymmetric information, needs to distort the effort of the agent working in the bad state of nature. On the other hand, the principal has no incentive to distort the agent's wage. As long as the principle has no full information about the state of nature the best he can do is to offer a second-best contract.

The principal's task is the program design which will induce the agent to take the single appropriate action. He hence tries to make different actions attractive under different states of the world, a favourable one, , and a bad state of nature, . The agent's decision depends on the hidden state of . ⁵⁵

The timing of the contract between politician and bureaucrat in the one-period setting is as follows:

We assume that the principal does not observe the condition neither the effort e provided by the agent, she can only observe the outcome of the policy measure +e. The total effort observable by the principal denotes E=+e. p denotes the probability that occurs. E is the total effort provided by the agent and y(e) his disutility when providing e. We have E = —e, and always > holds.

Following the standard literature on optimal contract theory ⁵⁶ we now consider a setting with a risk-neutral political principal and a risk-averse bureaucratic agent. In the simpliest way, the principal will offer the agent the following contract menu:

with the following restrictions:

where l denotes the participation constraint of the agent, m the incentive compatibility constraint corresponding to the good state of nature, and g the incentive compatibility constraint for the bad one.

The four first-order conditions denote:

It is easy to show that under full information the principal does not have to respect the incentive compatibility constraints m and g. ⁵⁵ She, knowing the type of nature, offers the agent the efficient wage and the latter provides the effort requested.

Before finding the optimal contract under asymmetric information, we check the implications of a bureaucratic setting where—like under full information—. One can argue that in many cases a bureaucrat has the same earning, no matter what effort he supplies. ⁵⁸ Such a situation can often be observed within public administration. It can be shown that this way of designing a contract cannot lead to an optimal contract in the case of two types of nature. ⁵⁹

In short, the optimal contract under asymmetric information involves different payments and must lead to an optimal effort if the good state of nature applies. As the case in which m=0 and g>0 as well as m>0 and g>0 can be rejected due to inconsistencies, the solution is found when m > 0 and g = 0. ⁶⁰

Put briefly, the solution implies the following: By keeping in mind that > always holds, we get for the good state of nature , for the bad state of nature the result is . As and on the other hand , in the bad state we always get an insufficient level of effort due to the lack of information of the principal. In other words the mechanism design in order to deliver an efficient solution necessarily leads to a too low effort in the bad state if asymmetric information applies. ⁶¹

5. Conclusion: Which institutional change could be helpful to redesign technology policymaking?

5.1. How to change the policy setting?

This result underpins a series of arguments in the ongoing policy discussion. If existing institutional settings do not allow for a better contract to be implemented due to informational restrictions one could either correct the informational shortcomings or—if this does not coincide with the political interest—simply change the institutional setting. In other words, the conclusion to be drawn from this model is to find an institutional framework with ensures more proximity between evaluation results and the incentive scheme inside the institutions.

"Telling the truth" versus "wishful thinking": Establish platforms in the overall institutional setting where experts are allowed

In general one can note, following Wilson (1989:130), that there is still not enough openness in the consultation process observable inside the EU or within its member states. This is not new, but it is striking. Europe lacks reliable platforms where efficiency criteria are allowed to enter publicly the policy consultation process publicly. Political discussions still reveal the interest of the participants in wishful thinking, they do not refer to factual constraints. This meets Wilson's distinction between "procedural fairness" and favoring "certain interests over others." ⁶²

U.S. economists look back to decades of successful measurement developments starting with the Griliches (1958) paper on social returns in the hybrid corn case. As Tassey (1996:44) shows, there are now four different groups of particular metrics for economic impact studies commonly adopted in the U.S. .The difference with the EU is that these approaches have gained ground inside the administrative world. Thus, ironically, one can expect that EU institutions will be increasingly forced to care about evaluations as a number of leading U.S. researchers focus clearly on EU technology policy programs. ⁶³ During the last few years a growing interest on technology indicators and evaluations leading to „best practices" can be identified at the EU level and inside the ministries in the different member states.

Delegate R&D issues to other agents

A general implication suggests to change the institutional setting by either empowering a different principal, by changing the agent's tasks or by delegating the task of technology policy to other actors where control is less costly. Greater oversight and more evaluation, if accepted by the political actors, will enhance the possibility to design a better contract.

It has already been discussed ⁶⁴ to attribute power to EuroHORCS, the European Heads of Research Councils, which could influence the decision-making process in EU technology policy. EuroHORCS have close contact with the European Science Foundation. The latter has faced a redefinition of its mission and has been trying to follow a more strategic role for the past few years. The first question is however, to which extent a new player will pursue more efficiency and if its power will be large enough to voice existing shortcomings publicly. The existing incentive scheme shown above suggests a stable relation between the Commission and other players.

Second, political feasibilities may lead to changes in the long run. To which extent the EU Commission might desist from parts of its program depends on the power of the the political principal—the Council. If in the long run budget restraints might become a more important issue it might occur that EU technology policy could be altered. One might argue that EU technology policy is less important to the Commission than other pillars (e.g., Commom Agricultural Policy).

Empower a different principal

Starbatty's and Vetterlein's (1994) conclusion suggests an empowering of the EU Parliament. They especially criticize the weak role of the Parliament and the attribuition of power to the Commission following Art. 189c of the Maastricht Treaty: A Commission which acts cleverly can turn the EP's interest groups against those of the national research administrations by involving them in earlier stages of the decision-making process. This, in turn, weakens the Council's position in controlling the Commission. Their proposal is, thus, to enlarge the EP's power to supervise the decision-making process.

5.2. Institutional changes and the Davignon Report

It should be mentioned that the 1997 published "Five-year assessment of the European Community RTD framework programmes" (European Commission 1997) coincides largely with the arguments and outcomes presented above. The Davignon Report clearly suggests

• changes in the legal setting of the framework program as current control measures hinder adjustments towards new needs, ⁶⁵

• to further the strategic character of EU measures by changing the voting rules inside EU institutions,

• to shorten the approval processes from five to two years,

• to cut the Commission's power to commit the whole program budget during the first three years and to allow the council the choice every year to fund new programs,

• to improve management ability by delegating the task of implementing the programs to the Commission and simultaneously controlling the Commission's detailed handling of implementation in a way which "[c]orresponds to best management practice in Member States and enterprises", ⁶⁶

• and—finally and most important—a better monitoring of the Commission by establishing a new and stronger link between Commission, Council and Parliament. This is suggested to be done by the creation of a new "Union Committee" which could replace the existing Programme Committee structure.

5.3. What recent developments in the U.S. indicate

As explained at the beginning, I do not follow a simplified analysis to transfer institutional findings to EU settings. Indeed, there has not been any striking evidence in the past that U.S. developments have been much more efficient in general. But U.S. development increasingly show attempts to overcome these rigidities by addressing the right questions. If the goal of the game is to improve efficiency, to establish impact assessments, and to create a relationship between evaluation outcomes and program redesign, there is indeed strong evidence that recent U.S. developments have been successful. In some cases also changes in the institutional settings have been achieved in a much more rigorous way than in the EU.

First, in the case of the ATP Program subsidies have been cut back significantly. Political forces inside Congress have been successful in arguing that market failure arguments do not play any role and that subsidies are inappropriate. This did not lead to a termination of a program; however, ATP funds have been cut back significantly. Currently, following officials, a "temporary equilibrium" seems to have been established between the different Pros and Cons to ensure that the ATP can proceed in a restricted, but rather efficient way. ⁶⁷

Second, evaluation is becoming an increasingly strong force in deciding on the funding of technology policy programs. The incentive structure inside the U.S. setting is more dependent on the outcome of evaluations. This may have to do with the historic commitment in the U.S. towards efficiency and budget control. As Roessner (1989) points out, this might be a legacy of the Great Society programs of the 1960s. During these programs, a percentage of program funds was spent for evaluation. In general, lock-ins ensuring the agent's power and avoiding application of evaluation results are less likely in the U.S., as there is enough openness in the efficiency debate.

Third, the impact of program design effects on the result becomes an increasingly important topic for further research in the U.S. . Following Wallsten (2000), program incentives should be designed to support only research that would not be undertaken without the subsidy and hence offer positice net social benefits but negative net profits for the innovating firm (marginal R&D projects). In his evaluation of the Small Business Innovation Research (SBIR) program Wallsten (2000:97) concludes that there is a need of dealing with the endougenous nature of R&D subsidies in a policy setting since the empirical result clearly shows that SBIR awards crowd out firm-financed R&D spending.

5.4. Extensions

There is still much evidence that cost-shared actions will continue to be a common instrument in future technology policymaking. In a global environment in which firms depend less on own R&D but increasingly buy innovations from outside, the revision of policy designs is at stake. As current empirical studies show, program design effects are a highly relevant topic. ⁶⁸

As to the EU, the Commission is not at all to blame for shortcomings and overregulations inside its member states. Indeed, some of the instrumental shortcomings of the programs can be explained to a great extent by the limits the EU faces since a series of instruments cannot be used at the EU level. Second, one could hardly believe that the incentive schemes would be automatically adjusted to a more efficient outcome at the EU level while there are even more institutional shortcomings persistent inside the most EU member states. Thus, it is simply the setting of the actors and the incentives within the constellation that lead to the outcomes observed. As Stubbs and Saviotti (1994) state, much of the critique regarding the practical shortcomings can "[h]ardly be laid at the door of the EC programmes, which account for only a very small part of total R&D effort across the companies and countries of the EC." ⁶⁹

Nevertheless, the examples given in the paper show that the lack of information and of embedding the results of evaluations into the decision-making process, inefficient policies can be persistent over time. This reminds one of the fable of the emperor's new clothes. The emperor may still proceed along the road, but it is increasingly the task of new institutional theories and theories of organization to build a policy design on the empirical findings and thus to play the child's role telling the public when the emperor is unclothed.

---

References

Arthur, W. B. (1994). Increasing Returns and Path Dependence in the Economy, Ann Arbor: University of Michigan Press.

Baker, G. P. (1992). "Incentive Contracts and Performance Measurement", in: Journal of Political Economy 100, 31: 598-614.

Baron, D. P. (1989). "Design of Regulatory Mechanisms and Institutions", in: Schmalensee R. and Willig R. D., eds., Handbook of Industrial Organization, North-Holland:Elsevier Science Publishers, pp. 1347-1447.

Beath, J., Poyagou-Theotoky, J., Ulph, D., (1990). "Information Sharing and Organization Design in a Research Joint Venture with Spillovers", Discussion Paper 90/265, University of Bristol, Department of Economics.

Bernheim, B.D. and Whinston, M. (1986). "Common Agency", in: Econometrica 54, 4: 911-930.

Braczyk, H.-J. ed. (1997). Regional Innovation Systems: The Role of Governances in a Globalized Economy, London: UCL Press.

Branstetter, L. and Sakakibara, M. (1997). "Japanese Research Consortia: A Microeconometric Analysis of Industrial Policy", Paper presented to the NBER Science and Technology Policy Meeting, mimeo.

Buchanan, J. (1975). "A Contractarian Paradigm for Applying Economic Theory", in: American Economic Review, 65,2: 225-230.

Buchanan, J. and Tullock, G. (1962). The Calculus of Consent. Ann Arbor: The University of Michigan Press.

Buisseret, Timothy J., Cameron, H. M. and Gheorghiou, L . (1995). "What difference does it make? Additionality in the public support of R&D in large firms", in: International Journal of Technology Management, vol. 10, 4/5/6, : 587-600.

Caswill, C. (1996). "Subsidiarity—a British View", in: Sturm, R. ed.,Technology Policy in Europe (in German), Baden-Baden: Nomos, pp. 47-52.

Cohen, W. M. and Levinthal, D. A. (1989). "Innovation and Learning: The Two Faces of R&D", in: Economic Journal 99: 569-596.

Cohen, L. R. and Noll, R. (1991). The Technology Pork Barrel. Washington D.C.: The Brookings Institution.

Coombs, R., Richards, A., Saviotti P.P., Walsh V. (1996). Technological Collaboration: The Dynamics of Cooperation in Industrial Innovation, Cheltenham: Edward Elgar.

D'Aspremont, C. and Jacquemin, A., (1988). "Cooperative and Noncooperative R&D in Duopoly with Spillovers", in: American Economic Review 87: 1133-1137.

David, P. (1985). "Clio and the Economics of QWERTY", American Economic Review 75: 332-337.

Davey, M.E. (1997). "Federal R&D Funding Trends in Five Agencies: NSF, NASA, NIST, DOE (Civilian) and NOAA", CRS Report for Congress, Congressional Research Service, The Library of Congress, Washington.

Dosi, G. and Marengo, L. (1994). "Some Elements of an Evolutionary Theory of Organizational Competences", in: England, R., ed., Evolutionary Concepts in Economics, Ann Arbor: The University of Michigan Press, pp. 157-178

Dixit, A. K. (1996). "The Making of Economic Policy. A Transaction-Cost Politics Perspective", in: Sinn, H. W. (ed.)., Munich Lectures in Economics, Cambridge, MA: MIT Press.

Eggertsson, T. (1996). "A Note on the Economics of Institutions", in: Allston, L.J., Eggertsson, T., and North, D.C. eds., Empirical Studies in Institutional Change, Cambridge: Cambridge University Press.

European Commission (1997). Five-year assessment of the European Community RTD framework programmes, Report of the independent expert panel chaired by Viscount E. Davignon and the Commissions's comments on the panel's recommendations, Brussels.

Fabisch, G. (1996). "The Impact of the European Research and Development Policy on the German Research System", in: Sturm, R., ed., Technology Policy in Europe (in German), Baden-Baden: Nomos, pp. 125-130.

Feldman, M. and Lichtenberg, F. R. (1997). "The Impact and Organization of Publicly-Funded Research and Development in the European Community", mimeo.

Gheorghiou, L. and Meyer-Krahmer, F. (1992). "Evaluation of Socio-Economic Effects of European Community R&D Programmes in the SPEAR Network", in: Research Evaluation 2,1:5-15.

Gick, W. (1996). "Modelling R&D Spillovers Between Firms." Paper presented at the EAEPE 1996 Conference "Work, Unemployment and Need", Antwerp.

Gick, W. (1998). "Chances and Limits of EU R&D Subsidies: The Impact of 'cost-shared actions'" (in German) in: ifo studien 44,1:25-45.

Griliches, Z. (1958). "Research Costs and Social Returns: Hybrid Corn and Related Innovations", in: Journal of Political Economy 46: 419-431.

Groenewegen, J. and Vroomen, J. (1997). "Theory of the Firm Revisited: New and Neo-institutional Perspectives", in: Magnusson, L. and Ottoson, J., eds., Evolutionary Economics and Path Dependence, pp. 33-56.

Hayek, F.A. (1988). The Fatal Conceit: The Errors of Socialism, London, The University of Chicago Press.

Hodgson, G. M. (1993). Economics and Evolution. Ann Arbor: The University of Michigan Press.

Holmström, B. and Milgrom, P. (1991). "Multitask Principal-Agent Analyses: Incentive Contracts, Asset Ownership, and Job Design", in: Journal of Law, Economics and Organization 7 (special issue): 24-52.

Howells, J. and Wood, M. (1993). The Globalization of Production and Technology, London: Belhaven.

Kamien, M.I., Muller, E. and Zang, I. (1992). "Research Joint Ventures and R&D Cartels", in: American Economic Review 82: 1293-1306.

Kastrinos, N. (1996). "Conflicts of interest in R&D collaboration: Negotiating the Function-Performance Link," in: Belcher, A. et al., eds., R&D Decisions: Strategy, policy and disclosure, London: Routledge, pp. 189-199.

Katz, M.L. (1986). "An Analysis of Cooperative Research and Development", in: Rand Economic Journal 17,4:527-543.

Klodt, H. (1994). "Foundations of Research and Technology Policy" (in German), Working paper no. 664, Kiel Institute of World Economics.

Konrad, K. and Torsvik, G. (1997). "Dynamic incentives and term limits in bureaucracy regulation", in: European Journal of Political Economy 13: 261-279.

Kydland, F. and Prescott E. (1977). "Rules Rather Than Discretion: The Inconsistency of Optimal Plans", in: Journal of Political Economy 85:473-490.

Laffont, J.-J. and Maskin, E. (1982). "The Theory of Incentives: An Overview", in: Hildenbrand, W., ed., Advances in Economic Theory, Cambridge: Cambridge University Press, pp. 31-94.

Laffont, J.-J. and Tirole, J. (1991). "Privatization and Incentives", in: Journal of Law, Economics and Organization, 7 (special issue): 84-105.

Laffont, J.-J. and Tirole, J. (1993). The Theory of Incentives and Regulation, Cambridge MA: The MIT Press.

Lemarié, St. and Joly P.B. (1996). "Mimetic Behaviour and Differentiation in the Various Stages of Technology Creation", Paper presented at the EAEPE 1996 Conference "Work, Unemployment and Need", Antwerp.

Lerner, J. (1999). "The Government as Venture Capitalist: The Long-Run Impact of the SBIR Program", in: Journal of Business 72:285-318.

Lichtenberg, F. R. (1996). "The European Strategic Program for Research in Information Technologies (ESPRIT): An Ex-Post Analysis". Paper presented at the CEPR/WZB Conference, Does Europe Have an Industrial Policy?, Berlin, Germany, 19-20 April 1996, revised version: September 1996.

Link, A. N. (1995). "Evaluating program performance: the case of Federally-funded collaborative research", in: International Journal of Technology Management, 10,7/8: 847-852.

Lundvall, B. A. (1992). National Systems of Innovation, London: Pinter.

Macho-Stadler, I., and Perez-Castrillo, D. (1997). An Introduction to the Economics of Information: Incentives and Contracts, Oxford: Oxford University Press.

Mansfield, E. (1986). "The R&D Tax Credit and Other Technology Policy Issues", in: American Economic Review, 76,2:190-194.

Mas-Colell, A. et al. (1995). Microeconomic Theory, New York: Oxford University Press.

May, P. J. (1992). "Policy Learning and Failure", in: Journal of Public Policy 12, 4: 331-354.

McKelvey, M. (1996). Evolutionary Innovations: The Business of Biotechnology, New York: Oxford University Press.

Metcalfe, J. S. (1994). "Evolutionary Economics and Technology Policy", in: The Economic Journal 104: 931-944.

Meyer-Krahmer, F. (1995). "Technology Policy evaluation in Germany", in: International Journal of Technology Management 10,4/5/6: 601-621.

Moe, T. M. (1991). "Politics and the Theory of Organization", in: Journal of Law, Economics and Organization 7 (special issue):106-129.

Moore, J. (1988). "Contracting Between Two Parties with Private Information", Review of Economic Studies 55: 49-669.

Mowery, D.C. and Rosenberg, N. (1989). Technology and the Pursuit of Economic Growth, Cambridge: Cambridge University Press.

Narin, F., Hamilton, K.S. and Olivastro, D. (1997). "The Increasing Linkage between U.S. Technology Policy and Public Science", mimeo.

Nelson, R. R. ed., (1993). National Innovation Systems. A Comparative Analysis. New York: Oxford University Press.

Nelson, R. R. (1994). "The Coevolution of Technologies and Institutions" in: England, R. W., ed., Evolutionary Concepts in Economics, Ann Arbor: The University of Michigan Press.

Nelson, R. R. and Winter, S. G. (1982). An Evolutionary Theory of Economic Change. Cambridge: The Belknap Press of Harvard University Press.

Niosi, J./ Bellon, B. (1995). "Une interprétation évolutionniste des politiques industrielles", in : Revue D'Économie Industrielle 71: 213-226.

Niosi, J. and Bellon, B. (1996). "The Globalization of National Innovation Systems", in: De la Mothe, J. and Paquet, G., eds., Evolutionary Economics and the New International Political Economy, pp. 138-159.

Noll, R.G. (1989). "Economic Perspectives on the Politics of Regulation", in: Schmalensee R. and Willig R. D., eds., Handbook of Industrial Organization, North-Holland: Elsevier Science Publishers, pp. 1254-1287.

North, D.C. (1990). "A Transaction Cost Theory of Politics", in: Journal of Theoretical Politics 2,4: 355-367.

OECD (1992). Technology and the Economy: The Key Relationships, Paris.

Patel, P. and Pavitt, K. (1988). "The International Distribution and Determinants of Technological Activities", in: Oxford Review of Economic Policy 4.

Peltzman, S. (1976). "Toward a More General Theory of Regulation", in: Journal of Law and Economics 19:21.

Persson, T. and Tabellini, G. (1990). Macroeconomic Policy, Credibility and Politics, Chur: Harwood Academic Publishers.

Pilorget, L. (1995). "Evaluation of Two Support Programs Promoting International Technology Transfer Between SMEs", in: International Journal of Technology Management, vol. 10,7/8: 867-878.

Pollack, M. (1997). "Delegation, Agency, and Agenda Setting in the European Community", in: International Organization 51,1 (Winter 1997): 99-134.

Quirk, J. and Saposnik, R. (1968). Introduction to General Equilibrium Theory and Welfare Economics, New York: McGraw-Hill.

Rasmussen, E. (1989). Games and Information, Oxford: Blackwell.

Reger, G. and Kuhlmann, St. (1995). European Technology Policy in Germany (in German), Heidelberg: Physica.

Roessner, J. D. (1989). "Evaluating Government Innovation Programs: Lessons from the U.S. Experience", in: Research Policy 18:343-359.

Salanie, B. (1997). The Economics of Contracts: A Primer, Cambridge: MIT Press.

Saviotti, P. P. (1996). Technological Evolution, Variety and the Economy, Cheltenham: Edward Elgar.

Schacht, W. H. (1997). "Industrial Competitiveness and Technologicy Advancement: Debate over Government Policy", in: CRS Issue Brief, Congressional Research Service, The Library of Congress, Washington.

Schmidt, V.A. (1996). "European Integration and Institutional Change: The Transformation of National Patterns of Policy-making", in: Göhler, G., ed., "Institutionenwandel", Leviathan, special issue 16:143-180.

Schott, T. (1994). "Collaboration in the Invention of Technology: Globalization, Regions and Centers", in: Social Science Research 23:23-56.

Senker, J. and Faulkner, W. (1996). "Networks, tacit knowledge and innovation, in: Coombs, R., Richards, A., Saviotti P.P. and Walsh V., Technological Collaboration: The Dynamics of Cooperation in Industrial Innovation, Cheltenham: Edward Elgar, pp. 78-97.

Sharp, M. and Pavitt, K. (1993). "Technology Policy in the 1990s: Old Trends and New Realities", in: Journal of Common Market Studies 31,2:129-151.

Smith K. (1996). "New Views of Innovation and Challenges to R&D Policy", in: Belcher, A. et al., eds., R&D Decisions: Strategy, policy and disclosure, London: Routledge, pp. 101-124.

Spence, M. (1984). "Cost reduction, competition, and industry performance", in: Econometrica 51:101-121.

Spiller, P. T. (1990). "Politicians, Interest Groups, and Regulators: A Multiple-Principals Agency Theory of Regulation, or 'Let Them Be Bribed' ", in: Journal of Law and Economics XXXIII:65-101.

Starbatty, J. and Vetterlein, U. (1994). "Research- and Technology Policy of the European Union" (in German), Discussion Paper no. 39, Economics Faculty, University of Tübingen.

Stephan, P. (1996). "The Economics of Science", in: Journal of Economic Literature XXXIV,3 (September 1996): 1199-1235.

Stigler, G. (1971). "The Theory of Economic Regulation", in: Bell Journal of Economics and Management Science 3,2.

Stubbs, P. and Saviotti, P. P. (1994). "Science and Technology Policy", in: Artis, M. and Lee, N., eds., The Economics of the European Union, Oxford: Oxford University Press. pp. 65-101.

Sturm, R., ed. (1996). Technology Policy in Europe (in German), Baden-Baden: Nomos.

Suzumura, K. and Goto, A. (1997). "Collaborative R&D and Competition Policy, Economic Analysis in the Light of Japanese Experience", in: Waverman, L., Comanor, W. S. and Goto, A., eds., Competition Policy in the Global Economy: Modalities for Cooperation, London: Routledge, pp. 197-223.

Tassey, G. (1996). "Rates of Return from Investments in Technology Infrastructure", 96-3 Planning Report, National Institute of Standards and Technology, U.S. Department of Commerce, Washington.

Tirole, J. (1994). "The Internal Organization of Government", in: Oxford Economic Papers 46,1:1-29.

Tisdell, C. (1996). Bounded Rationality and Economic Evolution, Cheltenham: Edward Elgar .

Tsipouri, L. (1992). "Evaluating the Economic Effects of R&D in Less Favoured Countries: The Notion of Complementarity", in: Research Evaluation 2,1:27-35.

Twight, Ch. (1994). "Political Transaction-Cost Manipulation: An Integrating Theory", in: Journal of Theoretical Politics 6: 189-215.

Verspagen, B. (1997). "European 'Regional Clubs': Do They Exist, and Where Are They Heading? On Economic and Technological Differences Between European Regions", Paper presented to the conference 'Economic growth and change: A comparative perspective', Cagliari, 19—21 June, 1997, jointly organized by the University of Cagliari, CRENoS and Northeastern University, Boston, Department of Economics. mimeo.

Vetterlein, U. (1991). Designing a systematic evaluation measurement on the Technology Policy of the European Communities (in German), Baden-Baden, Nomos.

Wallsten, S.J. (2000). "The effects of government-industry R&D programs on private R&D: the case of the Small Business Innovation Research", in: Rand Journal of Economics 31,1: 82-100.

Williamson, Oliver E. (1990). "A Comparison of Alternative Approaches to Economic Organization", in: Journal of Institutional and Theoretical Economics 146:61-71.

Wilson, J. Q. (1989). Bureauacracy. What Government Agencies Do and Why They Do It. New York: Basic Books Publishers.

Zucker, L.G. and Darby, M. R. (1997). "Academic Scientist-Entrepreneurs and Commercial Success in Biotechnology", mimeo.

 

Endnotes

Note 1: Centerfor European Studies, Harvard University, 27 Kirkland St., Cambridge 02138 MA, gick@fas.harvard.edu.  Back

Note 2: Department of Economics, University of Jena, Carl Zeiss St. 3, Jena 07743 Germany.  Back

Note 3: This is a revised version of my 1997 working paper written during my first affiliation with the Center for European Studies. Since many of the organizational changes to EU Technology Policymaking suggested in the meanwhile published Davignon Report correspond largely with the arguments I have presented in the earlier version, I try to include its conclusions in my discussion. I would like to thank Avinash Dixit, Peter Hall, Oliver Hart, Frank Lichtenberg, Paul Pierson, Larry Rausch and Jean Tirole for conversations that have been particularly helpful in bridging the gap between theory and policy, as well as Andrew Tylecote for comments on the earlier draft. All remaining errors are mine.  Back

Note 4: See e.g. Wallsten (2000).  Back

Note 5: I limit myself to two different groups of theories. For useful discussions see Klodt (1995).  Back

Note 6: For a discussion on an extended model see Gick (1998).  Back

Note 7: See Senker and Faulkner (1996).  Back

Note 8: Saviotti (1996: 171).  Back

Note 9: Cohen and Levinthal (1989:571).  Back

Note 10: Coombs et al. (1992).  Back

Note 11: Lundvall (1992).  Back

Note 12: A concept of regional innovation systems is described in Braczyk (1997).  Back

Note 13: For direct policy implications derived from evolutionary theories see OECD (1992).  Back

Note 14: Nelson and Winter (1982: 380).  Back

Note 15: See also Hayek (1988).  Back

Note 16: See Tisdell (1996: 302).  Back

Note 17: Starbatty and Vetterlein (1994) make it clear: It is unlikely that scholars refuse cooperation if they can gain reputation, power, and money by being embedded in such a setting.  Back

Note 18: See Pilorget (1995).  Back

Note 19: Smith (1996:107).  Back

Note 20: Metcalfe (1994: 941 and 931).  Back

Note 21: This refers in general to the second case May (1992:349) describes.  Back

Note 22: Dixit (1996:6).  Back

Note 23: Quirk/Saposnik (1968:105). For a further discussion of the political consequences of the Greenwald-Stiglitz theorem see also Dixit (1996:8-11).  Back

Note 24: Laffont/Tirole (1993:5).  Back

Note 25: See also Dixit (1996:13).  Back

Note 26: See Pollack (1997:108). Another explanation I do not follow here is the normative view that agencies and political organizations exist because of the need to obtain information about the state of the world. See Laffont and Tirole (1993).  Back

Note 27: See Laffont/Tirole (1993:1).  Back

Note 28: Dixit (1996:32). See also Twight (1994).  Back

Note 29: Tirole (1994:5)  Back

Note 30: Metcalfe (1994:933).  Back

Note 31: See Moe (1991).  Back

Note 32: Niosi and Bellon (1995).  Back

Note 33: See also Dixit (1996:53).  Back

Note 34: See Oltra (1996:12).  Back

Note 35: See Spiller (1990).  Back

Note 36: See Suzumura/Goto (1997:208-210).  Back

Note 37: See European Commission (1997).  Back

Note 38: See Starbatty and Vetterlein (1994:47).  Back

Note 39: As for the competences of the Commission see Starbatty/Vetterlein (1994:50-52).  Back

Note 40: See Cohen and Noll (1995).  Back

Note 41: See, e.g., the German IMPACT study done by Reger/Kuhlmann (1995).  Back

Note 42: Schmidt (1996: 169).  Back

Note 43: See Starbatty/Vetterlein (1994: 22).  Back

Note 44: For a description which scrutinizes the interest mediation and decision processes inside the Commission see Starbatty/Vetterlein (1994).  Back

Note 45: Kastrinos (1996:194).  Back

Note 46: Stubbs/Saviotti (1994:168).  Back

Note 47: Schott (1994: 45)  Back

Note 48: Verspagen (1997:15)  Back

Note 49: Niosi and Bellon (1996:144) focus on the precompetitive character of EU intra-Community collaboration.  Back

Note 50: See especially Starbatty's and Vetterlein's (1994) description of the decision-making process  Back

Note 51: Meyer-Krahmer (1989,330).  Back

Note 52: Stubbs and Saviotti give an excellent example on how the evaluations in smaller and less effective programs are performed, including the general difficulties to assess the pay-off to Framework R&D in financial terms.  Back

Note 53: See Bernheim and Whinston (1986).  Back

Note 54: See Konrad and Torsvik (1996).  Back

Note 55: For a longer discussion see the strand of literature on the revelation principle, and on separating and pooling equilibria as described in Salanie (1997), Rasmussen (1989) and Moore (1988).  Back

Note 56: See Macho-Stadler and Perez-Castrillo (1997), and Salanie (1997:139)  Back

Note 57: The Kuhn-Tucker-conditions cannot be satisfied in this case when incentive compatibility constraints appear.  Back

Note 58: See Macho-Stadler and Perez-Castrillo (1997:142).  Back

Note 59: As soon as the incentive compatibility constraints are binding, and , which leads to and hence to a suboptimal outcome.  Back

Note 60: For a detailed discussion see Macho-Stadler and Perez-Castrillo (1997:142).  Back

Note 61: ibid.  Back

Note 62: Wilson (1989:131).  Back

Note 63: See e.g. Lichtenberg (1996), and Feldman and Lichtenberg (1997).  Back

Note 64: See Fabisch (1996: 130).  Back

Note 65: European Commission (1997:14)  Back

Note 66: European Commission (1997:15)  Back

Note 67: As Schacht (1997:13) points out, the 104th Congress tried consciously to eliminate or curtail many public R&D projects that had previously enjoyed bipartisan support.  Back

Note 68: See Lerner (1999), Wallsten (2000).  Back

Note 69: Stubbs/Saviotti (1994: 165).  Back