Innovation policy in an open economy: A normative framework for strategic and tactical issues

121

Innovation policy in an open economy: A normative framework for strategic and tactical issues * Moshe JUSTMAN Department

o/ Economics.

Ben-Gurion

Unirersrr,v. Bwrshrca.

Israrl

Morris TEUBAL

Final version received December

1985

In this paper we present a normative policy grounded

in traditional

sponds IO the Schumpeterian

analysis of innovation

economic

principles

which re-

critique of previous efforts in the

neoclassical vein. Focusing directly on specific policy issues we pursue our analysis in IWO main directions, role of government tion.

in developing

and the optimal

investigating

an infrastructure

design of support

the

for innova-

schemes for

R&D

projects. The analysis suuests mental

that while the structural and dcvelop-

issues stressed by the Schumpeterian

fundamental

importance,

identification

social and private gains. the ‘*market vides an invaluable innovation

policy.

of scale.

supranormal

Focusing

our attention

the incidence

profits

offered

and

how

industrial innovation contribution

pro-

and formulating on such lactors as

imperfectly

and

the

competitive

and realistic basis for anticipat-

government

intervention

on behalf

of

is likely to be most effective. We view our

as a s~cp towards

neoclassical mainstream of innovation

between

approach,

of externalities.

by

markets, it provides a rational ing where

failure”

tool for understanding

economies

school are of

of disparities

bridging

the

gap

between

the

of economic analysis and the practice

policy.

1. Introduction

The neoclassical economic approach to justifying public intervention in the market-place uses the concept of “market failure” as its point of departure. Arguing that a disparity between the marginal value of an economic activity for society

as a whole and its marginal value for the private sector may lead the free market to a suboptimal allocation of resources, it suggests that in some cases such as this government intervention can bring about an improvement in market performance. Application of this approach to innovation policy was first developed in the early work of Nelson [14] and Arrow [2] and has since been continued in a series of theoretical studies. ’ These adopted Arrow’s working definition of invention as the production of information by the firm, and drew on his taxonomy of the sources of market failure inherent in the innovative process, in formulating their analysis. These were grouped by Arrow under the three headings: inappropriability, indivisibility and uncertainty. There has, however, been little application of this line of analysis to concrete policy issues, the notable exception of Nelson et al. [13] notwithstanding. ’ Kamine

and Schwartz

[lo] provide an extensive hihliogra-

phy of earlier efforts. More recent work includes Dasgupta and Stiglitz IS]. Lee and Wilde

[Ill.

Rcinganum

[19] and

Spence [221. inrrr afia. Nelson et al. stress the role of market failure in explaining the pattern of government

support for R&D

in the United

States. in the early 1960s. while recognizing the importance l

We

apprectate

Nelson.

comments

on a previous

K. Pavitt. two anonymous

draft

from

R.

referees and from par-

of other factors as well. However

ticipants in seminars given at the Science Policy Research

their analysis

today.

Unit, University

their analysis,

are now paramount

dustries. And

Nelson

of Sussex. and at the lnstituto Torcuato

Tella, and Fundacion

Mediterranea.

Argentina.

di

the special characteristics

of the US economy at the time limit the direct relevance of Open

economy

et al. do not provide

broad schemes of direct government Research Policy 15 (1986) North-Holland

0048-7333/86/%3.50

121-138

R&D today.

Q 1986. Elsevier Science Publishers B.V. (North-Holland)

issues, absent

of the type gradually

from

in high technology

in-

an analysis of

assistance to industrial

being implemented

in Europe

Indeed. this approach has been criticized sharply both by evolutionary economists in the tradition of Schumpeter. and by practitioners of innovation policy. They point to the structural paucity of the neoclassical models in describing the innovation process. their neglect of developmental elements of technological progress that transcend the scope of the individual firm. and their exaggeration of the possibilities for rational decision. While in general agreement with this critique, we feel nonetheless that the identification of disparities between the private and social gains that stem from the various economic activities that comprise the innovation process provides an essential analytical basis for formulating innovation policy. albeit one that must be reconciled with its Schumpeterian critique. The purpose of this paper is to develop a normative framework for analysis of this type, within a concrete policy context. This is pursued along two lines: analysis of the rote of government in the development of an infrastructure for industrial innovation - the strategic issues referred to in the title of this paper; and analysis of R&D project support systems the tactical issues of the title. Both are dcvclopcd with specific reference to their structural context and to the policy issues they must answer. The plan of the paper is as follows. In the next section we expand on the theoretical background of the analysis. We then go on to consider the infrastructure in section 3 and R&D support systems in section 4. The paper concludes with a brief summary of its policy implications.

2. Theoretical

background

The theoretical basis for our analysis is a synthesis of the basic neoclassical approach to the economics of technological innovation and its Schumpeterian critique. In the present section we develop this synthesis in general terms, first summarizing the basic neoclassical position, then outlining its Schumpeterian critique, and finally describing the nature of the synthesis which we propose to develop in the remainder of the paper. ,7.I. The neoclassical approach It is convenient for exposition to distinguish between two parts of the neoclassical position: the

basic principles set forth in Arrow’s [2] exposition: and the complementary analysis of more recent contributions. Arrow’s analysis of the allocation of resources for invention derives from his characterization of invention as the production of information; his choice of a closed economy context: and the nondevelopmental nature of his approach. These have led him to focus on three attributes of the innovation process: its inappropriability. its uncertainty. and its indivisibility. The inappropriability of the gains from invention derives from the low social cost of transferring information. and Arrow’s view of invention as the production of information. This leads to market failure of a type which is particularly difficult to resolve as it is manifested in two conflicting effects. On the one hand, the low social cost of transfurring information implies that once information is produced it should be distributed freely. This applies both to the final product of the innovative process and to intcrmcdiate results. indicating both insufficient dissemination of inventions and wasteful duplication of effort. On the other hand this same phenomenon also implies that the gains from innovation cannot be appropriated in their entirety by the innovator, leading to underinvestment in research and development. Clearly, dealing with these conflicting effects simultaneously raises serious difficulties. If incentives are strengthened by increasing appropriability. e.g., through the patent system, then the diffusion of innovations seems bound to suffer and more wasteful duplication is also likely. Conversely, increasing the dissemination of the results of innovative activity would seem to imply less payback for the innovator and consequently fewer incentives to innovate. Spence [22], analyzing this tradeoff in a rigorous formal model, suggests that a combination of cooperative research to reduce wasteful duplication, and government subsidies to bolster incentives, offers some remedy for this type of market failure. Recent experience in Japan with government intervention in the electronics industry along these lines. would seem to bear this out, at least to some extent. Arrow’s analysis of the role of uncertainty in economic activity has had a pervasive and abiding influence on many aspects of economic theory and

has important implications for innovation policy as well. Thus Arrow emphasizes the importance of capital markets for shifting the risks inherent in invention and innovation. And he acknowledges the imperfections to which they are generally susceptible. Information plays a vital role in capital markets. yet its production is characterized by the same market failures noted above in reference to the inappropriability of innovative information, and its application is characterized by extensive economies of scale. This implies that some capital markets which would be useful from a social point of view do not exist because they are not sufficiently profitable for the market-maker. But Arrow also made an important contribution to our understanding of the inherent qualities that limit the capacity of capital markets to shift risks. Arrow makes the important point that the same instrument that shifts risk from inn~~v~~t~~rto investor may also dull the innovator’s profit motivc. Imperfect monitoring of the investor by the innovator, and the presumed divergence of purpose between the two must then lead to underinvestment. Arrow referred to this as the “moral factor”, and more recent usage has it as “moral hazard”. Clearly this factor must limit the scope and efficiency of government efforts to supplement the imperfect capital markets of the private sector. The third issue which Arrow raised is the indivisibility associated with the use of information in production. There are inherent economies of scale in the application of an innovation by the individual firm: a process innovation which cuts production costs by IO percent on a current production rate of 100 units per day may well achieve similar savings per unit on higher rates of production without increasing innovation and implementation costs. S&h economies of scale are likely to lead to a well recognized divergence from marginal cost pricing as well as to a suboptimal selection of innovations by the market. This last point is also reflected in more recent work by Spence [21] on product selection under monopolistic competition. Spence demonstrates formally that with economies of scale characteristic of monopolistic competition, product selection by the private sector on the basis of profitability will not necessarily be optimal from a social point of view; and he provides indication of the direc-

tion of the biases that are IikeIy to arise. 3 Yet another recent line of research in the neoclassical tradition of economics that lends support to public allocation of resources for R&D refers to infant industries, in which accumulation of experience plays a central role. 4 When such experience is difficult to appropriate, either because it is embodied in individual human skills, rather than in firms or equipment, or because there are no effective restrictions on imitation, market forces will develop the industry more slowly than is socially optimat. Other sources of market failure associated with infant industries are possible external effects on complementary industries 5 and disparities between social and private rates of discount. High technology industries often disptay infant industry characteristics in the early stages of their development. Finally. recent research on international competitive strategy, notably a suries of papers, by Spencer and Brander (e.g. 1983), has drawn attention to profitable opportunities for intervention on the part of national government in intcrnational markets in which conlp~tition is imperfect and supranormal profits can be earned. Division of thcsc profits bctweon market competitors in such instances cannot be determined a priori on economic grounds but depends on the strategic interaction between them, Spencer and Brander make the point that by subsidizing R&D, governments can ultimately reduce the marginal cost of production by domestic manufacturers and consequently increase their market share and profits. Other forms of output subsidy, direct or indirect, might serve the same purpose. However. R& D subsidies have the advantage of effecting subsidization at the earliest stage of the production process. They are thus less likely to eticit retaliatory responses than are more direct forms of output subsidy. Further analysis of intervention of this type is taken up in section 4.2. We shall make

further

USC of this line of analysis.

and

describe it in greater detail. in section 3 below. where it is employed

to provide a theoretical

basis for public support

of infrastructure. Again, portant

Arrow

provides

contributions,

a seminal among

many.

treatment.

Other

arc Corden

im-

[4j and

Spence 1231. Heller

and Starrett

this point.

[7j provide

a systematic

treatment

of

2.2. Critique and s_vnthesis Despite its wide currency in the mainstream of economic analysis. the neoclassical. “market failure” approach to studying issues of innovation policy has come under sharp criticism both from practitioners of innovation policy and from academic economists following in the evolutionary tradition of Schumpeter. 6 These emphasize the limitations of the market failure approach, or at least of the simple form of static market failure analysis that has been common in the past. They emphasize the limitations of an approach that characterizes firms as fully rational profit maximizing entities functioning in an environment with no fundamental uncertainty and substantial equilibrium. and engaged in the production of information. Incomplete information and fundamental uncertainty, they argue, are inherent in the innovation process and severely limit the scope of rational choice. Waves of technological innovation are often associated with long periods of discquilihrium in which evolution and development are the rule. and structural parameters that transcend both firm and industry are the primary detcrmimints of the rate and direction of technological and economic progress. Moreover, there is more to innovation than the production of information; much of it is firm-specific learning that can be appropri:lted by the firm. Clearly, these are relevant points of criticism which cannot be dismissed lightly, though they mily not apply with equal force to all neoclassical efforts at understanding the economics of innovation. In many respects the assumptions and perspectives of the market failure approach are still not good enough, and indeed, some major policy issues - such as how to assure a plentiful supply of Schumpeterian entrepreneurs - cannot yet even be addressed within the conventional framework of analysis. Moreover, the actual pattern of government support is shaped by other forces beyond “market failure” not least of which are its incomplete access to critical information and susceptibility to “capture” by interested parties. While recognizing these and other limitations of conventional analysis. we contend nonetheless,

6 See. e.g. Nelson and Winter [IS] and other rrcenl work by Nelson

1161.

that a systematic attempt at identifying activities associated with gaps between private and social profitability provides an analytical basis for innovation policy. and in this sense, further development and adaptation of the market failure approach is warranted. This development, however, need not be based on equilibrium analysis; on the its perspective may be explicitly dycontrary, namic and based on attempts at a detailed observation of real world phenomena. Thus our paper is not meant to supplant the large body of economic analysis in the tradition of Schumpeter. that has come to have a profound, if not dominating, influence on both the theory and practice of innovation policy. but to complement it. Entrepreneurs are without doubt an essential factor in achieving technological progress. and rationality in a dynamic. changing environment is without doubt bounded. However. the right economic inccntives c;m help entrepreneurs produce desired results: and rationality though bounded is most certainly not absent from their decisions. The fact that market failures are ubiquitous in relation to innovation [ISJ only reinforces our :irgumcnt. Since it is no longer enough to identify it market failure in order to conclude that a particular action is needed, there is substance to the stntement that the existing mode of reasoning cannot carry policy analysis very far. At least an explicit ranking of “market failures” in terms of the benefits and ‘costs of remedial action is required. Theory should then attempt to help us identify the various types of market failure which are likely to appear at the various stages in the development of a high-technology industrial sector; determine their likely relative importance: and recommend the kinds of policies to be applied. The latter recommendations should in turn explicitly take into account the constraints on government action. In short. while we agree that a normative analysis of innovation policy may be hindered by simple, static market failure analysis, there is room for other types of such analysis that recognizes the existence of phenomena such as “bounded rationality and slow-moving selection”. Throughout the analysis we will make the critical distinction between support of infrastructure and support of specific innovations (or R&D projects). This distinction is based on indivisibilities and may have far-reaching implications concerning the optimal combination of public and

M. Justman and .+I. Teubul / Innowrron

private decision-making in resource allocation. It is reflected in what we believe to be a most useful distinction between “tactical” and “strategic” instruments. Whereas the former are limited in scope and hence lend themselves to the partial equilibrium analysis of bureaucratic management. the latter will often have far-reaching effects that must be considered in the more general context of national priorities which must ultimately be decided at the political level. We would like to emphasize however that the present analysis is concerned primarily with necessary conditions for government intervention. They are also sufficient conditions only if government can do better than the free market in the pertinent dimensions. Clearly this need not always be the case. The absence of a theory of government failure as developed as the economic theory of market failure does not imply the absence of such failure. The ability of government to execute a desired policy as conceived is without doubt an important consideration in the adoption of that policy. Nonetheless, difficulties of implementation of the various innovation policies considered in this paper are generally beyond its scope. Finally, it is important to bear in mind that innovation policy is often implemented against a background of previous economic distortions that must be taken into account, and indeed often provide much of the political justification for such policy. This applies most commonly to foreign support for competing industrial sectors but may also arise within the domestic context of a tax and subsidy structure that prejudices investment in R&D, e.g. compared with industrial investment in plant and machinery. In the next two sections we turn to a detailed application of the principles outlined above to support for the development of an “strategic” infrastructure for innovation, and “tactical” support systems for individual R&D projects.

3. Infrastructure Innovation diffusion are whose scope the individual that we refer They are the

and its subsequent development and generally affected by preconditions transcends the needs and abilities of firm. It is to these preconditions as the infrastructure of innovation. structural elements of the economy

policy IR

an

openeconomy

125

that can make innovation both possible and profitable and as such are an essential focal point for national innovation policy. Moreover, economic analysis suggests that the independent development of an infrastructure through the workings of the free market may well leave room for improvement through public intervention. This is the topic of the present section which deals with the role of government in the strategic, long-term development of an innovation infrastructure. We begin with an exposition of the economic principles: then develop a taxonomy of the main elements of the innovation infrastructure; and finally apply the economic principles to each of these elements so as to better define the appropriate role for government policy in their development. 3. I. Economic

theory

Spcncc’s [21] analysis product selection under monopolistic competition provides a theoretical basis for normative analysis of government support of infrastructure development within a ncoclassical framework. This focuses on the discrete, i.e. “ lumpy”. nature of investment in infrastructure and the economies of scale underlying this lumpiness. These preclude its being supplied competitively, and raise the possibility of a substantial disparity between the private and public criteria for providing the goods and services that make up the infrastructure. The principal conditions for such a disparity to exist with regard to a particular investment is that the scale of the investment transcends the scope of the individual firm, so that internalization of all the benefits of investment through vertical integration is not possible. In addition, the demand curve for infrastructure services (or goods) must be downward sloping, reflecting difference in the value placed on these services by different buyers (or by the same buyer for different uses). Furthermore it clearly must not be possible to practise perfect price discrimination in providing infrastructure services; or to import perfect substitutes at no additional cost. Given these conditions, Spence’s formal demonstration that the market selects against new products characterized by large fixed costs and low elasticities of demand can be applied to those products or services that comprise the infrastructure of innovation.

126 P

Phi

PO

V

I Xo

C Figure

1. hinrkrt

\ MR

f~~ilure under increasing returns: The product

These general argumen&s are presented in a static context in the above diagram (fig. 1) depicting the conditions of supply and demand of a new product in ;1 closed economy under increasing returns to scale. For simplicity, the latter takes the form of a fixed introduction cost F rend 3 constant variable cost I’. Given the demand curve for the product - 5D - the optimal supply for the good by the private investor is X, (we assume he is a monopolist who cannot discriminate among users) and the price charged PO. Private and social profits - np and n, respectively - are given by: np=

-F+ahcd,

n,=(-FSahcd)+bec, where uhc~f = x0( PO - V) are private operating profits. while bet is the additional surplus generated for users of the product (an external economy from introduction of the new product). When the condition ubcJ - bee > F > ahcd holds it implies

Thus, supply of the product is socially but not privately profitable suggesting a possible role for government intervention. Clearly, it price discrimination can be practised profits will more closely approximate social gain; if perfect price discrimination is possible there is

x margin.

no disparity between the two. And if the product or service can be imported, say at a price of P, on the diagram. then social gain will more closely the disparity disappearing approximate profits, when the import price is equal to or lower than the monopoly price. In a dyn~~rni~ context the demand curve will be moving outward in response to technological or economic developments. Supply of infrastructure services by the private sector than becomes profitable when the demand curve meets the average cost curve (AC). Thus many investments that warrant government support at the early stages of industry development would later be undertaken through private initiative. But for many investments of this type the private initiative will come later than is socially optimal. These basic principles serve a dual function. They provide general support for an active innovation policy of a type quite distinct from that which can be derived from firm-oriented R&D models in the spirit of Arrow’s [2] analysis. And they can provide practical guidelines for implementing such a policy, as we shall try and show in what foliows. 3.2. A taxonomy

o,f in/rasmrclure

components

The infrastructure of innovation, in the broad sense that we use the term in this section, comprises a wide range of elements embodied in a

1. Finance 1.1 R&D

stage

H ‘-Financial

assessment and management

0 “-Venture 1.2. Implement

capita1 market: government

H

-Financial

0

-OTC

management

market;

2. Research and Development H -Scientific 2.1. Access to 0

state of art

of implementation

international

projects

stage (high risk. growth)

correspondence

network;

government

funding mechanisms

and technical knowledge

-University;

P ’ -Libraries

research institutes: Links with foreign research institutes and data bases; basic and generic research facilities (e.g. la~rato~es)

2.2. Experi-

H

-fnventive

mentation

0

-Applied

P

-Experimentation

H

-“Design

0

-Innovating

P

-Design

2.3. Design

of R&D

funding mechanisms

abilities. S&T knowledge research groups, testing facilities facilities

or innovation

capabilities”

- all of the above + market/economic

firms; market research organizations;

subcontracting

orienlation

of specialized

enperience

design work

facilities

3. ~lanufactu~ng 3.1. In the

t-i -Prototype

R&D

process

0 P

-Machinery

3.2. Exploiting

H

-Manufacturing

skills in relevant industry: capacity

0

-Firms

to pr(~u~ing/using

-Firms

manufacture

skills

geared to specialized high skills products of type needed for producing

prototypes for quick study

the fruits of

R&D

gearcz

university and skiflcti manpower

4. Marketing

P

-Manufacturing

H

-thigh tLwhnology marketing

0

-Trading

P

-Distribution

a ff. embo&d

in human capital;

h 0, embodied

in organiration

E P. embodied

in physical capital.

new

products

training

and

custom-made

~on~~~nents

(flexible.

open};

facilities

facilities that can produce/use

new products

skills, reputation

companies. export boards, market research organirations; and warrhousing

facilities (n;ttionnl

marketing

training

institutions

ytrxlwiil)

structure;

variety of forms and supporting different stages of the extended process of innovation and diffusion, These include both traditional elements of physical infrastructure such as power, transportation, and communications, as well as less tangible elements embodied in human capital, such as technicaf and scientific skills. They have in common economies of scale in their production and supply that transcend the needs and abilities of the individual firm so that they cannot be vertically integrated with the innovating firm and must be supplied externally. Table 1 presents the essential components of the infrastructure, organized in two dimensions: according to the medium in which each is embodied (human capital, organizational structure. physical capital); and the particular function in the innovation process which it serves (finance,

research and development, manufacturing, ’ marketing). The selection is not comprehensive nor is its,classification definitive but it provides a good idea of the specific phenomena with which we are concerned in this section. It also provides a framework for an integrated analysis of the infrastructure in its entirety. Such an analysis can help rectify imbalances in the infrastructure by diverting development efforts to its weakest links. in the remainder of this section we consider each of these elements in turn, concluding with

7

Strictly speaking manufacturing tion process but the exploitation domestic industry

is not part of the innovaof local innovations

is clearly a principal

tion policies insofar as they are incorporated industriat

policy. Cf. R. Williams

[ZS].

by

goal of all innovain an overall

some comments on the problem of achieving anced development of the infrastructure.

bat-

3.3. Finurm Innovation requires financial support on three possible levels. In the initial stages of innovation venture capital must be raised to finance the first efforts of learning and experimentation that characterize this phase. Then growth capital must be found to fund the implementation and initial diffusion of the innovation. Finally, if the potential market for the innovation transcends national borders successful diffusion will require the support of international banking facilities. In~pticati~~ns of these needs for infr~~struct~ire dcvclopmcnt are first and foremost organizational. A venture capital market and/or some governmcnt funding mechanism are required if capital is to he alh~catcd for the early stages of inllov~lti~~n. An over-the-counter (OTC) market or paraflcl government mechanism is nscdcd to finance the suhscqucnt high growth stage of initial diffusion. And foreign credit ~Irr~ingcli~c!its arc rcquirsd if international diffusion is to succocd. such organization element involves conSitllXlhlC economics of scale in several respects: iitf~~r~t~~lti~~Iinods. risk spreading, and administration. Moreover foreign credit markets and credit alternative for facili tics arc rarely a feasible domestic innovators. Furthermore the high elemcnt of uncertainty inhcrcnt in the innovative process coupled with the moral hazard that arises from the asymmetry of information between innovator and investor would seem to preclude any comprehensive attempt at price discrimination. These conditions would therefore seem to support the inclusion of financial markets among the elements of infrastructure for innovation for which government assistance is possibly justified. And indeed, government intervention in the financial aspects of innovation has been observed in many countries in recent years. Such a function is served in thr Federal Republic of Germany by the Ministry for Kesearch and Technology (BMFT) which funds industrial projects aimed at enhancing the technologic~t competence of German industry, generally on a mutual basis with the private sector [16]. Spain’s Center for the Developmcnt of Intfustrial Technology. ;I part of the Ministry of Industry and Energy, extends loans t'ilch

for R& D efforts within industrial enterprises [lS]. And Israel’s industrial R&D fund. administered by the Office of the Chief Scientist of the Ministry of Industry and Commerce, extends grants on a matching basis to industrial firms for the execution of R&D projects [25]. These government mechanisms augment private sector venture capital in financing the early stages of innovation through direct intervention. They are generally reinforced by preferential tax treatment for R&D expenditures aimed at stimulating private investment. Government sponsored institutions aimed at improving foreign credit arrangements are less common (though subsidized government credit for large foreign purchases is a common form of indirect export subsidy). They are more likely to be needed in the newly industrializing countries where the private banking sector has not kept pace with industry. Korea’s Export Import Bank which provides long term credit for foreign purchase of Korean plants and turnkey projects is a good example of such an institution which has dcvelopcd to mcrt the special needs of local industry. Beyond the institutional and organizational arrangements. needed to meet the financial needs of technological innovation. indeed stemming from them. there arises a need for special skills or human capital. These are needed for evaluating the needs and prospects of new R&D projects; for the financial management of the critical high growth stage; and for making international credit arrangements. Because of the moral hazard inherent in any external investment in human capital (Urban f27]) the supply of skilled manpower to meet these needs is again an element of infrastructure that possibly merits government support. These training requirements can be met in several ways, applied in combination; through formal education apprenticeship in the private sector; and apprenticeship in the public sector. The more technical aspects of financing R&D are better served by formal education with some limited stewardship; the more creative aspects require a larger share of learning by doing. The public sector has an important role to play in providing the relevant business education. And it can also provide opportunities for gaining practical experience in financing technological innovation. if it is involved in these activities. Indeed, accumulation of such experience may be an im-

portant consideration favoring direct government financing of R&D in the earlier stages of technological development. ’ 3.4. Research and development Research and development is. of course. at the heart of the innovation process. We have chosen here to divide it into two stages, each of which requires its special type of infrastructure support. Generally, the first prerequisite for successful R&D is familiarity with the state of the art. While some innovations are so revolutionary as to obviate all previous experience this is, of course. very much the exception. The great preponderance of innovations builds on available scientific and technological knowledge. This is most clearly apparent in the electronics industry where many innovations are achieved through a new assembly of existing components. But it is equally true in other fields characterized by high levels of innovation, such as pharmaceuticals, where innovative rcscarch without a sound scientific basis in chcmislry is virtually unthinkable. Familiarity with the state of the art is cmbodisd first and foremost in human capital, and as such is clearly an elcmcnt of infrastructure by dint of the general imperfection of markets in human capital; investment in such markets is limited by an inhcren1 asymmetry of ,information which gives rise to an element of moral hazard. Moreover, such investments are likely to generate positive external effects. On an organizational level ongoing support for this element of infrastructure is provided by research-oriented university departments, and other research institutions engaged in basic and generic research as well as by the research departments of the larger innovating firms. Some measure of such research activity is essential for keeping abreast of developments in the slate of the art, even if it yields few such developments itself. It is embodied also in the physical facilities (libraries, laboratory R To some extent. frastructure

Government

supporting

viewed PS support essenkd

an infant

service (financial

a learning

of the financial R&D

industry

supplying

capital) to that process. That

clcmcnt such as this, indivisihilities

an is.

of an

over and ahove the

reflected in our application

model of product seklion.

in-

process may he

process is involved in the eslahlishmcnt

inlrastructure production Space’s

of

support

the industrial

of

equipment. computers) that support these activities. Public support for basic science is universally accepted as necessary and beneficial. though this is not so much the case for generic research. In this respect the Japanese experience is exceptional in its breadth as well as its measure of success. Japan’s Ministry of International Trade and Industry (MITI) has played an active role in funding and orchestrating large scale cooperative research efforts in various fields including colour television. very large scale integration (VLSI), computer design, and biotechnology. Japanese efforts in this respect were geared not for the development of specific products but to help Japanese manufacturers develop the capabilities to move in a variety of possible directions along a broad front [16,17]. Generic scientific and technological knowledge includes broad design concepts, general working characteristics of processes. knowledge of matcriaIs characteristics, aspects of quality control processes and testing methodology, and standards. Such knowledge is often not patcntablc; much of it is shared openly by scientists and engineers working in the field, whcthcr they arc located in universities. government or corporate laboratories. Research systems involved in this kind of work fit in a niche between fundamental research and applied R&D for industrial firms [I4]. The agricultural and medical sciences appear to have defined their niches appropriately, but public sponsoring of industrial generic research in the U.S. and in Europe has traditionally been lacking. ’ Recently things have been changing, however, both in Europe and increasingly in the U.S. For example, the newly established Microelectronics and Information Sciences Center at the University of Minnesota is conducting a joint research effort in the generic areas of integrated circuit design, microelectronics and systems architecture. This effort involves both industry and government, with researchers having access lo the facilities of the larger participatory firms, as is done in Japan [24]. While knowledge of the state of the art is a prerequisite for R&D, the essence of innovation is experimentation and design. These require specific scientific and technical skills of course. but also some measure of creativity and a propensity for P

NACA‘s

pre-World

War

II experiments

nology. which led to the DC-X

with aircraft

tcch-

is one notable exception.

risk: and they require a commercial 0rienta~ion that takes into account not only technological feasibility but also production costs and market demand. In short, the crucial stage of R&D requires the attributes generally associated with what has come to be known as the “Schumpeterian entrepreneur”. Much has been written about this essential element of the innovation process, with national success and failure attributed to its presence or absence, but little is known about the sources of its supply. Some importance must be attached in this regard to the basic psychological parameters of the economy. i.e. the “national mentality”, which is affected, in turn, by the country’s educational system. its class structure, and its social traditions. But it is far from clear how these influences act and we have nothing to add that might help clarify their effect. They arc beyond tht: scope of ecotromic analysis. However. we would argue that the supply of Schumpctcrian entrepreneurs can be affected by economic policy. Entrepreneurs arc, by definition, scnsitivc to profit ~~pp~~rtunities so that creating such new opportunities in technological innovo(ion attracts potcnlial entrcprcnrurs to try their ha~xl. The number of people in the population with strong personal motivation, a capacity for finding creative solutions to difficult problems, and ;t propensity to take calculated risks, may very well not be susceptible to manipulation by economic means. Dut where these talents are applied - in the armed forces, in civil administration, in trade, in science or the arts, or in technological innov~ition - is influenced by economic incentives. An attempt to exert such an influence was one of the main objectives of Israel’s innovation policy in its early stages. Matching grants for research projects administered by the Office of the Chief Scientist of the Ministry of Industry and Trade had the beneficial effect of creating a pool of Schumpetcrian entrepreneurs who continue lo pIay a key part in rurthering the country’s technological development. This effect clearly transcended the success or failure of the initial projects for which the funds were allotted [26]. Related to the generation of Schumpeterian entrepreneurs is the development through “learning by doing” of what may be termed experiencebased industrial innovation capabilities. “Doing” industrial innovation enhances these capabilities:

both in regard to R&D itself (e.g. the desired combination of analogue and digital components in an optical instrument) and to its relationship with subsequent steps of the innovation process, i.e. production and marketing. Moreover, it enhances the capabilities of new entrepreneurs to search for appropriate R&ZD projects. Thus. R&D project support. especially in the earIy stages of development of the high-technology industrial section, indirecf!v supports an important infrusmrcrum element for successful innovation - the devetopment of innovation capabilities. Private investors are unlikely to capitalize on all the social benefits derived from enhancement of these capabilities: the movement of individuals to other firms will generally carry not only specific informa!ion (Arrow’s template externalities) but person-embodied general capabilities as well. Government subsidy should be regarded as an attempt to take into account the existence of both types of extrrnalitics. At the early stages of growth in a hightechnology industry the stock of innovation capahifity in the economy is not yet sufficient. Therefore. work on almost any inn~~v~lti~~n will lead to a significant incrcaso in these capabilities. Morcover. the gencrul lack of experience will cause a large number of R&D projects to fail at this stage. Thus (relative to subsequent stages) a large share of the social bcnrfits from innovation will take the form of externalities and a large share ol these involve enhanced general capabilities. At the “growth” or “maturity” stages, an increasingly important share of the external brnefits from inn5vati0n will be of the “run of the mill” kind, stemming from innovation specific information flows and from additions to user surplus, with little effect on the pool of general innovation capabilities.

Manufacturing capabilities form part of the ii~~rastructure of successful innovation in two different ways. In the R&D stage specific manufacturing capabilities are often necessary for experimentation and design. Models and prototypes must be custom built in small quantities and meet stringent tolerance levels. often requiring unconventional materials, methods. or processes. These manufacturing services require close interaction with the innovator which largely precludes their

provision by foreign suppliers. Larger firms often solve the problem by developing in-house manufacturing capabilities but this is clearly not a feasible option for smaller innovators. Moreover, the need for secrecy limits the ability of smaller firms to hire these services from their large competitors. The second state at which manufacturing capabilities are crucial for success is in the implementation and diffusion of the innovation. Innovations contribute most fully to industrial development when they are embodied in products rather than sold or leased for foreign manufacture. This implies a need for a domestic manufacturing sector that can produce such products at competitive cost. Of course it also implies that R&D projects selected for government support must be chosen to accord with existing capabilities, as we elaborate in section 5, below. Thus, for example, in electronics and other industries, a crucial role is played in this regard by an infrastructure of specialist producers of parts or components. The manufacturing of new products may be significantly hindered at the early stages of high-technology industrial growth by the absence of such a network of producers. In such situations, local firms will need to produce a much larger share of the product than best practice, causing higher costs and delays in complementary innovations, and placing them at a competitive disadvantage vis-h-vis foreign firms with better access to component sources. This was the case in Israel’s electronics industry in the late 1960s and early 197Os, when local suppliers of custom-made printed circuits and other custom-made electronic components were absent. The indivisibilities in creating a capability to supply these specialized products or inputs is, prima facie, a reason for government support.

Marketing is a crucial element in diffusion. Economies of scale in marketing that transcend the needs of the individual firm and allow us to speak of a marketing infrastructure are most prevalent when diffusion of the innovation must transcend national boundaries and penetrate foreign markets. In such cases a concerted effort by groups of

firms. or a national effort. can capitalize on economies of scale in information on economic conditions and the legal and regulatory environment in foreign markets. Provision of general public information of this type is an accepted function of government in many countries. Beyond that. market research organizations as well as legal. commercial and even language specialists are important private sector elements of the infrastructure of marketing information. Agencies in foreign countries. for both sale: and service, are another aspect of marketing that offers an opportunity for exploiting economies of scale. Japanese trading companies are of course the prime example of a private sector response to this opportunity, and one that has been replicated in many other countries. They offer the advantages of goodwill, access to potential buyers, distribution facilities and technological capabilities that are generally beyond the scope of the single innovating firm, at least in the early stages of its growth. Furthermore. pcnctration of an export market often depends on some element of national reputation. The label “Made in Japan” is an e.xcellent example, Once a synonym for simple low cost products it has since become a symbol of efficiency in production, quality control, and technological progress and as such is a great asset for Japanese manufacturers. A nation’s commercial image is of course most strongly influenced by the quality and price of the goods and services it has to offer. It can be enhanced or formed more quickly through advertising campaigns, trade fairs and other forms of sales promotion, and penetration pricing. The first two of these measures are commonly the province of export boards (such as Israel’s Export Institute). The last of these has apparently been subsidized by Japan’s Ministry of International Trade and Industry (MITI) but is obviously more exceptional. Finally, the public sector has a role to play in developing the human capital for international marketing. This it can do by promoting language and regional studies pertaining to export target areas as well as the relevant legal and commercial specialties. Beyond that, national economic representations in foreign countries can also add to the aggregate local experience of commercial conditions in these markets.

3.7. Bulunce While each of the elements of the infrastructure described above must be examined separately the success of technological development is determined by the overall performance of the system. Therefore. considerations of economic efficiency require that the balance between the different elements also be considered. Ideally the educational system should produce just enough trained manpower, and no more than can be absorbed. generic research laboratories should spawn just the right number of new ideas and state of the art techniques; manufacturing capabilities should be available to capitalize on new innovations: and the marketing infrastructure should be sufficiently strong to handle the international diffusion of all viable innovations. This would insure that no resources were wasted. Needless to say, this is hardly feasible. in an environment of technological change discquilibrium is ths rule rather than the exception. Moreover, some elements of infrastructure arc more amcnahlc to government devclopmcnt than others suggesting that the creation of temporary imbalance may itself be a means for strengthening the infrastructure. I” By developing those gcncral clcmcnts of infrastructure for which it is better suited the public sector can create profit opportunitics for private entrepreneurs to complete the missing elements. Nonetheless, any strategic plan for development of the infrastructure must strive to keep these imbalances, both spontaneous and contrived, within appropriate limits. These limits derive, in the first instance, from the need to use a limited development budget in the most efficient possible way. They must also take into account however, the very serious considerations that arise from the realities of international factor mobility. In a free society that allows international movement of the factors of production, and especially of highly skilled labor, large imbalances in the infrastructure can create large disparities between

II1

In this connection

it is worthwhile

man’s unbsl~nced

growth

compelling

correcttve

IO recall Alberl

thesis. Some imbnlances

(market)

others are permissive (see [Y]).

forces while

Clirschinduce

the effects of

the marginal product of these factors at home and abroad. generating strong incentive for movement of these factors of pioduction across national borders. Clearly the direction of movement is crucial in these cases: a surplus of trained manpower can attract foreign investment or it can cause a “brain drain”. Thus in creating such imbalance careful consideration must be given to the projected direction of the factor flows.

4. Direct support for R&D

projects

Development of a sound infrastructure is a necessary condition for industrial innovation. But there appears to be general agreement - among theorists and practitioners alike - that the economic performance of innovative industries will also benefit from a program of direct support for specific R&ZD projects. The support of practitioners is evident from the wide currency that such programs have gained. Theoretical support is manifest in a long line of analyses of the positive externalities generated by innovative activity, from Nelson’s [ 141 and Arrow’s [2] early work to the more recent contribution of Spence [22] along the lines indicated in section 2. Such analyses have focused primarily on the need for R&D subsidies and on their optimal level. We take this a step further, by bringing these well recognized theoretical considerations to bear on the practical issues of designing a functional efficient R&D project support system. These issues are captured in the following four questions: What are the individual criteria that a project must meet to receive support? Should these criteria be applied neutrally or should the system be targeted to specific industries? Should firms or nonprofit institutions be preferred as the locus of industrial innovation (or nei [her)? How far into the project life cycle should public support extend? A fifth ‘practical issue of substantial importance, the appropriate instruments that should be employed in supporting R&D projects (direct subsidy, loans, tax relief, government procurement, etc.), is left for future study.

4.1.

Criteria

for

project

approval

On a general level we have been arguing that project subsidies should be allocated according to economic rather than in direct measure to anticipated technological or commercial success. This implies viewing subsidies as a means for modifying the economic behavior of firms and individuals so that it is in greater consonance with the general welfare of society. Thus a subsidy might be used to make unprofitable ventures that are socially worthwhile, profitable, or to increase R& D outlays when less R&D is performed than is socially desirable. But there would appear to be little reason to subsidize R&D in projects that appear profitable, ex ante, without the aid of a subsidy, and where the scope of R&D is either optimal or unaffected by subsidies. This would seem to indicate some form of social cost-benefit analysis, that might be compared with parallel calculations of private profitability. However. practical application of such an approach raises substantial difficulties that are not easily rcsolvcd. These are inherent in its informutional rcquircmcnts regarding the anticipated costs and bcncfits from innovation, both private and social. which cannot generally be met within reasonable bounds of either time or money. Thus with regard to private profits the difficulty stems in the first instance from objective limits on information, which in the early stages of innovation may consist of little more than an educated guess. And this is aggravated by the strong incentives that firms have to misrepresent what they do know, in their dealings with the authorities, so as to gain as large a subsidy as possible, and by the dependence of the authorities on information supplied by firms applying for aid. As for the measurement of social benefits these are even more nebulous. An individual R&D project may be insignificant in itself. and yet be an important link in a developing chain of innovations. Indeed, Nelson and Winter have consistently made the point that one of the important functions of R&D subsidies is to encourage experimentation and preliminary work leading to a richer set of feasible R&D projects (151. Obviously. in placing a value on the marginal benefits from future research projects (yet unknown!) one cannot pretend to any degree of precision. Yet another difficulty that arises in assessing

prospective R&D projects singly derives from possible competitive interaction between innovations aimed at overlapping (or even identical) markets. In such a case innovation generates negative externalities and as the theoretical work of Barzel [3] and Hirschleifer [8] indicates this may give rise to excessive innovation. Finally. the stochastic dependence between prospective R&D projects raises yet another set of issues. Financial theory teaches us that when such dependence exists an optimal portfolio of risky assets will take it into account. seeking to balance the risk so as to eliminate as much of it is possible. ” This complex problems, i.e. the optimal composition of a national portfolio of R&D projects, has not yet been dealt with analytically, to the best of our knowledge. All of the above suggests that the allocation of public support for R&D projects cannot be detcrmined, in practice, with full adherence to theoretical rigour. Instead, qualitative indicators of conditions likely to justify support. viz. large disparitics between private and social gains, should hc sought. These would include the type of extcrnalitics suggested by Arrow’s analysis, e.g. the inappropriability that derives from the limitations of patent protection and secrecy in guarding against imitation. and the existence of a consumer surplus, uncaptured by producers. And they should also take into account such macroeconomic benefits as increased demand for domestic labor, enhanced value of local natural resources, and improvement of the balance of trade. Moreover, in the open economy context often characteristic of high-technology industries, strict attention must be paid to the national incidence of externalities generated by the innovative process. It matters very much who will produce the good that results from the innovation and who will use it. And it matters who owns such assets as it enhances and such as it renders iess valuable. Two polar examples should serve to illustrate this point. Consider, at one extreme, a small Israeli subsidiary of a multinational corporation developing integrated circuits which are then manufactured in the Far East and sold to an American automobile ”

The portion generally

of risk that can be eliminated

referred

mental literalure.

to as “unsystematic

in this way is

risk” in the funda-

manufacturer. Business prospects may be excellent but these will be accounted for in investor’s decisions in any event. Local externalities. however. seem slight. limited perhaps to those stemming from the labor requirements of the R&D effort. Immediate inter-firm spillovers are also likely to be minimal in the absence of a local automobile industry. ” At the other extreme, the development by Israeli innovators of irrigation equipment for arid zones has generated external effects at several levels. There is extensive local manufacture of the equipment by a well developed domestic plastics industry: there is widespread use of the innovation by local agriculture: and the produce is at least partly consumed by the domestic market. Moreover. the small scale at which production becomes efficient and the relative ease of imitation make for a competitive industry in which spillovers are hard to prevent, and the appropriation of benefits by innovators is consequently limited. Indeed. taking this line of reasoning a step further, there may well bc some rcscarch activities which should be discouraged. hccausc of the adverse cxtcrnal effects they arc likely to have on important national assets. This might apply, for example. to innovative research in the field of nuclear power generation, in a country with a large oil surplus. It is widely recognized that the sale of existing technologies though profitable for the individual firm may well be detrimental to the national economy, and the same reasoning applies to the generation of new technologies. Moreover, while the magnitude of the externalities that a research program is likely to generate cannot generally be predicted to any reasonable degree of precision the incidence of these externalities may well be quite clear even at the very earliest stages of the program. This would then provide a useful criterion for discriminating between research programs competing for public subsidies. Projects likely to produce benefits for domestic consumers through domestic manufacture would rate the highest priority while those geared primarily for licensing to foreign manufacturers would rate the lowest. Projects of the latter kind may well be of great benefit to the country ”

However. effort

support may he justified

to strrngthcn

section 3 ahove).

as part of a long term

the local integrated circuit industry

(cf.

but this is likely to be so only if they are also profitable to the private investor, in which case subsidy is unnecessary. The intermediate case. of goods produced domestically for export may also offer substantial externalities. Jobs are created, often with higher than average wages. And foreign currency is earned. improving the balance of payments. In a country with a large trade deficit and, concomitantly. a large foreign debt, this has the beneficial external effect of improving the terms on which it receives credit, and strengthening its currency. This is a general argument for supporting export-oriented industries. There is an additional argument that applies more specifically to innovative industries that compete in world markets. This is taken up in the next sub-section. 4.2. Ntwtrdity We use the term neutrality. here, to refer to the uniform application of universal criteria lo all projects applying for R&D subsidies, regardless of the industry, technological area or product class to which they belong. In early stages of economic dcvelopmcnt, and generally in the absence of reliable information on the commercial prospects and externalities likely to be associated with different projects a neutral subsidy policy will take the form of a flat rate of subsidy to all projects meeting some minimal criteria of competence and integrity. In later stages such policy will be modified in accordance with information gathered regarding the relevant economic variables. This might induce a shift from the formal neutrality of the early stages without implying abandonment of the principle of neutrality. Thus, given the costs of a thorough examination of all projects submitted, the use of signals or rules of thumb may well be justified on economic grounds, at least as initial screening devices. These rules of thumb might be aimed at capitalizing on the strengths of the country’s industrial infrastructure; enhancing the value of its natural resources: providing a productive channel for underemployed human capital; etc. Their form;lation might naturally rest on a classification of projects by industry, technology, or product class. inter dia. without there being a departure in principle from a policy of neutrality. A policy of neutrality in the broad sense de-

hf. Jusmun

ad

M. Tcubul / fnnwarton

fined above is justified. in our view, when govemment intervention in R&D is aimed at correcting for externalities. Israel’s successful experience with a neutral. policy of research subsidies would seem to bear this view out. Departure from such a policy might be dictated, however, by large lumpy investments that are likely to significantly strain the country’s limited resources. Such investment decisions are inherently non-neutral and must be judged in a wide context that can take into account their strategic implications. Obviously, the strategic dimensions of large R&D investments are most commonly felt in small economies. For example. the development of modern fighter aircraft in countries such as Israel and Sweden drains vital engineering skiffs from other sectors of the economy. They are also sorely felt in developing countries undertaking large. high-technology projects such as the nuclear programs in India, Pakistan, and Iwn. inter cl&t. But they can also arise in large, developed countries, as in the cast of the United States space program in the 1960.s. ” Finally, a deliberate departure from neutrality might be justified by strategic considerations of the type outlined in section 2, above, irrespective of project size. These would target a specific, imperfectly competitive international market still in its formative stages for national development and possible dominance through early subsidy of industrial R&D. Such subsidies would clearly be non-neutral, in any sense. This form of targeted policy bears some resemblance to the type of industrial research policy, commonly referred to as “picking winners”. We should like to stress, however, that it is much more stringent in its requirements: it aims to identify economic rather than technological winners. It stipulates a market in its formative stages, which when mature will have room for only a small number of national industries. And the location of production fltcifities must follow the location of R&D activities. Moreover, the possible supranormal profits that may be gained must be weighed II

II does not strike us as coincident;ll

that R&D

investments

with

are often

defense re-

national

lated. The

strategic

dimensions

obvious dangers inherent

which put a disproportionate

in such investments,

number of eggs in the same

polrcyin an openeconom_v

135

against the risks inherent in a bureaucratic decision involving an assessment of the future prospects of an industry in its early stages of devefopment. 4.3. The locus of industrial

R&D

In practice. support systems for industrial R&D projects have been directed to both firm-based research and non-profit institutions. Our detailed knowledge of Israeli experience with both types of system I4 and general acquaintance with the experience of other countries indicates that while nonprofit research is most suitable for many projects in agriculture. health, and defense, the bulk of industrial research is best served by a system geared to supporting research in firms. This stems primarily from the responsiveness of firm based research to the prospects of commercial success. These prospects generafly bear a closer affinity to economic welfare than do the technological criteria to which non-profit research orgnnizntions arc more strongly oricntcd. Furthcrmorc. directing R&D project subsidies to the individual firm creales indirect demand pressures for the other necessary ingredients of successful economic development of high trchnology industries, viz. advanced manufacturing facilities, marketing capabilities, and engineering and managerial skills. The innovative firm receiving project subsidies thus serves as a crucial feedback mechanism, in the complex process of adjusting and balancing the entire spectrum of direct and indirect support policies. Moreover, it must be noted that while the biases of the individual firm in allocating resources for R&D have received the preponderance of attention of economic theorists such biases must surely also exist in non-profit institutions. That we know fess about the biases inherent in these institutions is not a recommendation in their favor. However, the above mentioned advantages notwithstanding, there appear to be at least three fields of endeavor where a substantial proportion of R&D project subsidies must be directed to non-profit institutes. These are: agriculture. health, and to a lesser extent defense. All share a rich potential for scientific discovery, and. in addition, each appears to have its own specific reasons for

basket. are more easily accepted when they can be justified by national security interests.

I’ This

is summarized

in Teubal

[ZS].

the large role it offers for bon-profit research. In agriculture it would appear to be a combination of a highly fragmented market structure and the inherent difficulties in approp~ating the gains from many of the specific innovations arising from agriculture R&D. In health it is the unique nature of the product and the concomitant traditional dominance of non-profit institutions in conducting medical research. And in defense it would appear to be a combination of the final product being a public good, the need to be constantly on the forefront of scientific knowledge. and again the special intensity of demand for the final product. Indeed, in the latter two cases there is a cfear divergence between business considerations and economic welfare which induces a preference for a good deal of non-profit-oriented research. Finally, past experience indicates that the natural commercial orientation of the individual innovating firm is likely to be reinforced if the insti[L~tion ~ldministering R&D support is also economically or commercially oriented. Thus a policy of R&D support is more likely to follow commercial or economic principks

if it is under

the auspices of the Ministry of Industry. Conversely its administra~i~~rl by the Ministry of Soience and Technology is likely to give greater weight to scientific and technological merit. 4.4. The scope of project supporf R&D is only one stage in the innovation process. Manufacture and marketing are subsequent phases that are also essential for successful application and diffusion of innovations. This raises the question of whether public support for innovation should extend beyond the development stage. Such support is generally not extended in practice, however there are arguments pro and con and these should be weighed in the context of specific projects. The main argument for extending support beyond the R&D phase is that supporting R&D activities along creates an imbalance in the system. A surfeit of inventions is generated, many of which are consequently not implemented. This argument can be interpreted in two different ways. The first is that the manufacturing or marketing infrastructure cannot assimilate all the inventions that emanate from the innova(ing sector. This was considered in section 3, above, and we will not

repeat the discussion here except to say that such circumstances indicate development of the relevant elements of the infrastructure, or realignment of the R&D support system so that it generates innovations that match the strengths of the infrastructure. Support for the manufacturing and marketing stages of technologically successful R& D projects may or may not be the economically efficient way of achieving this. A second type of imbalance occurs at the firm level between R&D outlays and expenditures on production and marketing whose objective is implementation or commercialization of R&D results. Here we must distinguish the case where external benefits to society not capturabie by the private firm are generated at the invention stage (e.g. spillovers of technological knowledge) from the case where the externalities result from commercialization (e.g. externalities from the “innovation” derived from additions to user surplus). In the former case, preferential subsidization of R&D vis-l-vis implementation is justified in principle. Moreover, this conclusion is probably rcinforccd by the presence of capital market imperfections, which severely limit the market’s ability to spread the risks of R&D. In the latter case, the target of government support should be the innovation process ns a whole and not only R&D, or any other stage. Earmarking the subsidy for R&D alone will produce a distortion in that the innovating firm will tend to replace high cost marketing and production effort by iow cost (but socially dear) R&D activities. Thus when the externalities from innovation are a result of commcrcialization there seems to be no reason to alter the overall balance attained by market forces between R&D and implementation. This may require, however, close coordination between the R&D project support system and other pre-existing support schemes that bear on the innovation process. I5 Finally, the case has been made that the manufacture and marketing of innovations merit public support because innovators exhibit an inherent bias against such activities, being more technologically oriented. Such support would then arouse

”

However. for R&D

in an cvolutionctry

content,

prcfermtial

(nr other stages of the innovation

necessary to compensate distortions.

for the rumdative

support

process) may be effects of past

their awareness. This is an argument which must be applied carefully. Awareness can be aroused through the dissemination of information by extension services of the kind successfully employed in promoting the diffusion of agricultural expertise. Widespread public support for large scale marketing efforts that cannot find a commercial sponsor seems an expensive way of going about this.

5. Summary In this paper

we develop a conceptual frameand assessment of technology policy. This framework is a synthesis of the neoclassical “market failure” approach to normative economics, and the Schumpeterian view of innovation as a process of development. It is presented in the context of the concrete policy issues which it must address, In drawing these issues we distinguish between two principal components: development of an infrastructure for innovation: and support for specific R&D projects. We view the former element as a strategic activity with wide ramifications requiring coordination at the highest national level. The latter provides tactical support and can therefore be managed, within broad guidelines, at the bureaucratic level. We define the ‘innovation infrastructure as a combination of elements that support the innovation process while exhibiting economies of scale that transcend the needs of the individual firm. These elements fall into four principal categories, research and development, finance, manufacture, ‘and marketing. We then go on to describe the role of each component in the innovation process and the reasons why public intervention may be warranted to ensure that this role is performed adequately. Such intervention must promote the availability of the necessary skills and capabilities, organizational structures, and physical facilities needed at each stage of the innovation process and which the market will not find it profitable to provide. Moreover, it must carefully control any imbalances in the infrastructure both to avoid wasting resources and to prebent the outflow of valuable factors of production. The other key element of innovation policy. a support system for specific R&D projects, is work for the formulation

moulded in large part by the answers to four key questions which we consider in this study. They are: (1) what are the individual criteria that a project must meet to receive support? (2) should they be applied neutrally across industries? (3) where is the preferred locus of industrial innovation? (4) which stages of the innovation process should be supported? Our detailed answers to these questions stress the advantage of economic criteria for project approval. over technological and commercial criteria. These focus on the domestic incidence of positive externalities as the main justification for supporting specific projects and suggest a policy of neutrality, in principle. in the administration of project support systems. Such systems, however. may also have strategic dimensions. In the early stages of market development project support systems have an important role to play in the building up of basic design capabilities and other relevant skills. In this respect project support aids in strengthening the infrastructure. Beyond this, strategic support for targeted industries may be warranted under specific conditions of impcrfcct competition in an international market still in its formative stages. Finally, our work identifies several important issues for future study. Chief among these a better analytical perspective of the dynamics of infrastructure development, a quantitative approach to determining the optimal size of the various budget elements supporting industrial innovation, and a comparative analysis of the various instruments of innovation policy; subsidies, loans, procurement, patent protection, etc. Acknowledgment

This article is based on a study conducted at the Jerusalem Institute ol Israeli Studies in cooperation with the Samuel Neaman Institute For Advanced Studies in Science and Technology and the Israel Industrial Center For Research and Development. References [I] K.

Arrow.

Economic

sources for Invention.

DlmWion

0/

Princeton. [2] K.

Arrow.

Doing.

Welfare

and the Allocation

in: R. Nelson

(cd.)

Acfioify

(Princeton

Economic

lmplicalions

fncenfice

of Re-

The Rare and

University

Press,

NJ. 1961). The

Reciew o/ Economrc Studies

of Learning

(June 1962).

by

[3] Y.

Barrel.

Opumal

Timink

Economrcs and Srwsrrcs [4] W. Corden.

Tru&

PO/~

don Press. Oxford. [5] P. Dasgupta

of

Social and Policy Studies. Yale L’mversity,

RZI.ICH. o/

1171 M.

34X-355.

and Ecunomrc bt’rlfure (ClarenIndustrial

[18] K.N.

SING-

Economies o/ ReJeurch Polrcr rn

Working

Steven

Lin.

Externolrrres

Theo?

(Academic

Paper,

Ben-Gurion

the

On the Nature and

[19] J. Reinganum,

novations:

[ZO] R. Rothwell

to

Inventive

Activity,

[22]

Press. New Haven CT.

A Survey,

,4mertcun

561-574.

7%e Strure,qv oj Econonuc

and N. Schwartz.

Market

lournul

[23] M.

1956). Structure

of Economic

Lee and L. Wilde,

R. Nelson.

Market

Qwrrer!v

and T.

Washington,

R.

DC. The

Research. Joww/ Nelson

Journtrl of Conomicv

Hout.

Jupurrese /d~~~rtu/

No. 585 (January.

94 (I 980)

Reduction,

Economewica

Spence. The

[ZS] M. Teuhal.

and

Public

Pobq~ (Brookings

Simple

XXI [26]

1980).

Techrro/o,~~~.EcoInxtituticm.

Learning

of In-

Patent Protec-

Economrrrrco

Economics

of

Basic Scientific

M.

50 (1982)

An

Euol~rrwn~rr)? T/won

University

oj

Press. Cambridge,

Pohcies in Support

o/ //i,qh

Techno/o,qv /,I-

Paper No. 1011 (revised) (Institution

for

Innowrrun

and

1981).

Fined Costs and MonopoSfudres 43 (1976)

Competition

Curve

and Industry

101-121.

and Competition,

Be//

49-70.

and the Role of Government, untl Socrul Chun~e

11 (I 982)

The

in Ixael.

Mtnervu

Performance Firms:

through

Time

of

Methodology

and

an

Rewcrrch Police 11 (19X2) 333-346. Just.ifications

for Industrial

Policy.

and L. Klein (eds.) Imfutrrrul Policies /or

Cr~n~per~r~r,ene.v.~.(Lexington

Books.

19X3)

3.

Wrbtphal.

World

R&D

Theoretical

[2X] R. Williams. Korea’s

Technology

1984) 172-197.

in: F.G. Adams Chapter

in Science Policy: The Promotion

Industrial

lligh-Technology

illustration.

1291 L. Working

Teuhal.

Young.

Grow,th cd

o/ Polirrcul Econort~~ 67 (195Y) 297-306. S. Winter.

fndurfrlul

52 (1984)

12 (1981)

Neutrality

(Summer

1271 P. Urban.

19X2). [ 161 R. Nelwn.

the

163-1X0.

Polrcy.

1967).

Economrc Chun~e (llarvard

&trtes.

Cost

1241 G. Tassey. Infratechnologies

Structure and Innovation:

M. Peck and E. Kalachcck.

mwwc Grnw~h urd

[IS]

in

University

Promonon

Review o/ Economy

Journul o/Economics

and In-

Llrrrclrctrc I3

Spence.

of Sophisticated

Nelson.

R&D

Bank. Washington.

Pinter. London.

Te~~hnologrcol Forecurrrng

Policy Studies Institute.

R.

M.

Performance.

Der*e/opnrenr

l-37.

I. Magaxiner

[ 141

and W. Zegveld.

217-235.

A Reformulation.

[ 131

198-t).

Yale

Game of R&D:

(211 M. Spence. Product Selection.

1976).

42Y-436.

[ I2]

(World

Behaviour.

Public Poliqv. (Frances

o/ Economy

Reward

University

M. Kamien

[I I] T.K.

A Dynamic

listic Competition.

Hirschman.

(1975)

Xlimeo.

Gowmmenr

Typescript

tion and Competitive

Univer-

of Externalities,

iCfrururemmr

Press. NY,

Economic RCCICH. 61 (1971) (Yale

Innocutron.

The Private and Social Value of Informa-

[8] J. Hirschleifer.

[lo]

of

671-687.

[7] W. Heller and D. Starrett.

[9] A.

Industry.

Rao and C. Weiss.

durtrd

Bell Journul o/ Econonttcs I I

sity. 1984.

and

Electronics

Coordination

DC). The Wel/are

an Open Econonty.

tion

Government

(1983). Uncertainty.

11-28.

[6] M. Justman.

in:

Peck.

Japanese

1974).

and J. Stiglitz.

lure and the Speed of R&D. (1980)

Innovattons.

50 (1968)

‘Technology L.

Kimand

Acquisition

Policy. typescript (1983). C.

Dahlman.

of Technological

Bank. Washington.

DC.

March,

Reflections Capability.

1984).

on (The