|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 373
Technology Diffusion, Public Policy,
and Indusmal Competitiveness
PAUL A. DAVID
The intertwining of the processes of technological innovation and
diffusion makes it important to consider how policies intended to
promote innovation may affect the rate awl the ultimate extent of
adoption of new technologies. Yet this is seldom done. By failing to
address systematically the issues concerning diffusion in our national
policy discussions, we hate surrendered the opportunity to see whether
it is possible to formulate any consistent set of goals, or to coordinate
the actions of the many public agencies that are engaged in de facto
setting of policies affecting the development of our technological
capabilities. We do not avoid making mistakes by proceeding in this
way, however. Rather, we avoid having to acknowledge the mistakes
and learning from them.
Technological progress today is widely perceived to be the force propelling
the American economy forward, the veritable prime mover of the country's
lon~,-run economic growth. Myriad advances in the systematic knowledge
of '`the useful arts" have been directly and indirectly responsible for the
enormous gains achieved in the measured productivity of the naiion's labor
and capital resources over the course of the past century. They have brought
radical improvements in the qualitative attributes of goods and services,
transformations that (however difficult they are to measure wide precision)
contributed palpably to enhancing He compeuaveness of our industries abroad
and the economic welfare of consumers at home.
Yet the United States does not have a well-articulated set of policy goals
with regard to the development and utilization of its technological capabil-
~ties, much less a coherent, integrated program directed to He attainment of
such goals. This much has been openly acknowledged by the President's
Commission on Industrial Competitiveness (see Young, in this volume). The
recently published report of the commission, in calling for creation of a
cabinet-level Department of Science and Technology, speaks of He need to
`'~ansform He current fragmented formulation of policies for science and
373
OCR for page 374
374
PAUL A . DAVID
technology into one that would be far more effective in meeting long-te~
national goals" (President's Commission on Industrial Competitiveness, 1985,
vol. 1:22)
Among the more obvious symptoms of this persistent condition is the
comparative lack of attention devoted to economic analysis of public policies
impinging on the diffusion of new technologies into actual use. The economic
determinants of technology diffusion is a field of research with which I am
most concerned, and the one that I will review in this chapter.
My discussion proceeds along the following lines. First, I address the
validity of the assessment jUSt offered and suggest some of the reasons why
policy issues regarding technology diffusion have come to be neglected in
this country, even as the subject now is receiving greater explicit consider-
ation elsewhere in the industrialized world. Next, I contrast this state of
affairs with the progress that has been made by theoretical and empirical
research on the microeconomics of technology diffusion during the past two
decades. The findings have led to many new insights into the demand-side
and supply-side aspects of this important class of dynamic processes. Bllt
these findings also have given economists a healthy respect for some of the
complexities of a subject they have yet to master fully.
This is not the place for a detailed survey of the significant contributions
to a voluminous technical literature, so I will only highlight some among
many economic factors determining (1) whether and when adoption of spe-
cific processes and process innovations is likely to be advantageous for the
users and (2) the costs Nat potential adopters face in order to secure the
specific information, equipment, and materials.essential to the effective use
of the new technologies.
Cursory as such a review must be, it provides a basis for me to comment
next on what I believe are important implications of "the new microeco-
nomics of technology diffusion" for the way that the formulation of public
policy In the area of technology should be approached. The review should
indicate also how we might begin systematically to assess Me impact of
present and proposed economic policies on private sector decisions affecting
the installation of new production methods and the acceptance of new goods
and services by consumers. It will be seen that the relevant range of gov-
ernmental actions is very broad, including the tax treatment of investment,
the funding of R&D, the education of scientists and engineers, regulation
and standards setting, as well as the monetary and fiscal measures shaping
Me macroeconomic environment I shall contrast the tangled mass of eco-
nomic policy interventions that are being pursued, seemingly without regard
for their impact on the diffusion of technological innovations, with the much
narrower "domestic technology transfer" programs that have been assigned
a formal mission to promote the domestic dissemination of technological
.
information.
OCR for page 375
TECHNOLOGY DIFFUSION. PUBLIC POLICY, kD INDU=RI~ COMPLETENESS 375
The upshot of my review is not some simple, all-purpose policy prescnp-
tion, but a call for more general recognition of the reality that choices among
products and production processes in He private economy are not being left
to the unimpeded workings of He market; and that the introduction of ad-
ditional policies intended to quicken, or perhaps to retard, the adoption of
specific new technologies cannot sensibly be justified except by explicit
assessments of the extremely varied and changing conditions that obtain in
different industries. Where technology diffusion is at stake, therefore, an
absolutely indispensable ingredient in the formulation of rational economic
policies is detailed assessments on an industry-by-industry basis. A model
for such studies exists in the series recently completed by the National Acad-
emy of Engineering (1982-1985)—which deals with electronics; steel; ma-
chine tools; automobiles; fibers, textiles, and apparel; pharTnaceutica1 products;
and civil aviation manufacturing.
One of the central empirical insights guiding the direction of recent eco-
nomic research is Hat new technology-diffusion and new technology-devel-
opment processes often are very closely intertwined. (See David and Olsen,
1984; Ireland and Stoneman, 1984a.) Separation of the two for purposes of
analysis may be convenient, but eventually their study must be reintegrated.
Moreover, policy actions undertaken in direct reference to the one will more
likely than not have significant effects on the other. The importance for
intelligent economic and technology policymaking of taking the more inte-
grated approach to He design of innovation and diffusion policy, which the
microeconomics of the problem demands, is a point I wish to stress here
To grasp its importance one need only consider the following four, illustrative
potential paradoxes of technology policy:
1. Efforts to speed up the rate of innovation in industries supplying capital
goods can create expectations of larger capital losses through obsolescence
for firms that consider adopting the new technology when it first appears.
Hence, the promise of faster innovation rates can delay actual adoption
decisions. (See Rosenberg, 1976; Ireland and Stoneman, 1984b.)
2. Tax and over subsidies for R&D can reduce He costs of "imitation"
and lead to expected wider diffusion of a new technology throughout an
industry. But if it is expected that every one will quickly adopt the technology,
He inducements to bear the costs of adopting it early are reduced. So R&D
subsidies may slow the initial speed of diffusion even if they do help dis-
seminate information about the new technology more widely and increase
the eventual extent of use. (See Stoneman, 1983; David and Stoneman, 1985.)
3. Delaying the imposition of technical standards in order to encourage
continuing R&D investment and further innovation can slow effective ap-
plication of technologies in which compatibility and network integration are
vital. It does not, however, prevent the emergence of de facto standards,
OCR for page 376
376
PAUL A . DAVID
which may eventually be discovered to have been suboptimal. (See Arthur,
1983; David, 1985.)
4. Strengthening patent and trade secret protections that convey significant
(temporary) monopoly power to suppliers of goods embodying new tech-
nologies, while simultaneously providing public funds for the dissemination
of information about the benefits to users of those technologies, can have
the effect of inducing the supplying firms to set initially higher prices for
their wares. Thus, the combination of a policy measure meant to promote
innovation and another measure meant to encourage diffusion may perversely
work to slow the pace of application. (See David and Stoneman, 1985.)
Rather than elaborating on these particular paradoxical propositions, I shall
try in the following to indicate the conceptual framework within which they
were derived. From that general perspective it may be seen that a number
of policies being advocated to promote greater competitiveness through in-
novation in the United States today could have unsuspected and costly side
effects for technology diffusion. Were they to be as effective in the long run
as their adherents claim, they might perversely slow diffusion and the growth
of productivity during, both the near and medium teen.
THE ADOPTION OF NEW TECHNOLOGIES AND THE DOG
THAT DID NOT BARK
It would be hard to exaggerate He economic significance of technology
diffusion. What determines improvements in productivity and product qual-
ity thereby enhancing the competitiveness of fimns and industries is not
the rate at which significant technological innovations are developed, but the
speed and extent of their application in commercial operations. However
feasible the designs for new products and production processes may be from
an engineering standpoint, it is He prospects for their diffusion into use that
ultimately impart economic value to this form of new knowledge.
Be that as it may, the plain fact is that adoption of technological innovations
has yet to acquire an aura of glamour in contemporary American society.
Certainly, both as a field of business endeavor and as a matter for scholarly
investigation, it lacks the radiance that currently surrounds the process of
designing and commercially introducing new technologies. (See Freeman,
1982; Mansfield et al., 1982; Nelson, 1982, for recent scholarly contribu-
tions.)
Many aspects of our history and national character have contributed to
forming a climate of opinion more favorably disposed toward governmental
support for the generation and commercialization of new products and pro-
duction techniques than it is toward programs aimed at influencing the timing,
and the eventual extent, of the use made of those innovations. As a people
OCR for page 377
TECHNOLOGY DIFFUSION,PUBUCPO~CY,~DINDU=~COMP=~IVENESS 377
we have for a long time displayed an unusual psychological receptivity to
novelty and change, on which astute foreign observers—from Alexis
de Tocqueville onward recurringly have remarked: There is an understand-
able cultural predisposition on the part of Americans to seek a basis for
consensus and unity in our shared hopes for the future, rather than in the
diversity of our respective experiences of the past.
Success arnon=, us as a rule is equated with "leadership"- in this case
with pioneering on the technological frontiers. To be an assiduous "fol-
lower," by comparison, seems somehow to have acquiesced in defeat, aban-
doning adventure for the haven of routine. Of course, anyone with actual
experience of the challenges and opportunities inherent in the transfer and
adaptation of modern industrial technologies will be quick to object that
nothin;, could be farmer from the truth of the matter. Popular images fre-
quently are distorted, but potent nonetheless. "Mere imitators ! " is the epithet
that a harried leader is likely to hurl over his shoulder at the ever more closely
pursuing pack.
It is perhaps not surprising that in the policy discussions instigated by the
worrisome "productivity slowdown" occurring in the United States since
the mid-1970s, as well as in the writings of academic economists, interest
increasingly has concentrated on understanding We causes of shifts in the
rate and direction of technological innovation. But notice that the almost
universal fixation on proposals intended to accelerate the pace of advance
by influencing the allocation of private sector funds for research and devel-
opment has had a curious side effect: it contributes to distracting the attention
of policymakers from the ultimate goals of application, on which commercial
R&D expenditures are predicated.
Innovation has thus become our cherished child, doted upon by all con-
cerned with maintaining competitiveness and renewing failing industries;
whereas diffusion has fallen into the woeful role of Cinderella, a drudge-
like creature who tends to be overlooked when the summons arrives to attend
the Technology Policy Ball. As a case in point, consider that in the report
of Me President's Commission on Indusmal Competitiveness (1985, vol. 1:51-
52), the words "diffusion" and "adoption" do not appear anywhere in the
summary of recommendations dealing with the nation's technological re-
sources. Here are the "three basic things" the commission has said we need
to do in order `'to make technology a continuing competitive advantage for
the United States": "~1) create a solid foundation of science and technology
that is relevant to commercial users; (2) apply advances in knowledge to
commercial products and processes; and (3) protect intellectual property by
strengthening patent, copyright, trademark and trade secret protections"
(vol. 1:181.
At the moment I am not concerned so much with the content of the
recommendations as with what is missing from the commission's list. It
OCR for page 378
378
PAUL A. DAVID
strikes me as noteworthy that this informed and influential group has rec-
ognized "needs" corresponding to the first two but not the third of the
elements usually identified in the compound we call technological change.
A less innovative document might well have been expected to offer rec-
ommendations encouraging activities under every category in the classic
tripartite scheme: (1) organized or in foal research leading to invention;
(2) development of commercial applications leading to the introduction of
an innovation, in the form of a new product or process; (3) imitation, or
selective adoption of the innovation, resulting in its diffusion into actual use.
Instead, where one looks for specific policies to accelerate the wider
diffusion of newly created technologies, there stands something quite dif-
ferent, and possibly antithetical. Concrete measures are recommended for
further strengthening economic incentives stimulating innovation by better
delineating and enforcing rights to exclude others from access to the new
knowledge thereby created. The closest the Young Commission comes to
addressing the issue of the rate of application of new production techniques
in U.S. industry is to recommend that private sector, educational, and gov-
ernment organizations "should initiate actions [otherwise unspecified] to
improve the development and use of manufacturing technologies to transform
R&D results into competitive products and services . . ." (President's Com-
mission on Industrial Competitiveness, 1985, vol. 1:521. In the riddle of the
report, entitled Global Competition, then, technology diffils~on policy has
been allotted the part of "the dog which did not bark in He night."
THE NEW MICROECONOMICS OF TECHNOLOGY DIFFUSION
AN OVERVIEW
Despite the drift of the spotlight of public policy interest away from the
subject in this country, during He past 20 years there has occurred a quiet
revolution In the economists' search for a deeper understanding of the pro-
cesses involved in the adoption of novel technologies. The "new micro-
economics of innovation diffusion" Hat is emerging from these research
efforts is based on bow theoretical insights and an accumulating body of
empirical evidence.
In this county Edwin Mansfield and his students at the University of
Pennsylvania have been responsible for carrying through the most extensive
and systematic program of collection and econometric analysis of modern
time-genes samples, Facing the extent of adoption of many specified pro-
duction technologies within firms and industries. (See e.g., Mansfield, 1968;
Mansfield et al., 1971, 1977; Romeo, 1975, 1977.) This work was directed
toward identifying common features and determinants of diffusion processes.
Like the earlier, classic studies by Gnliches (1957, 1960) of the adoption of
hybrid cam, Mansfield's work focused attention on the roles of expected
OCR for page 379
TECHNOLOGY DIFFUSION, PUBLIC POLICY, kD INDU=RI~ COMP=~7VENESS 379
profitability for potential adopters and dissemination of information within
the using industry. These were emphasized as of critical importance in over-
coming the obstacles placed in the path of rapid diffusion by uncertainties,
and the consequent risks for firms contemplating a large investment com-
mitment to the new technology.
In the United Kingdom, parallel econometric studies have been carried
out by Nabseth and Ray (1974) at industry, firm, and plant levels. These
have confinned the general role of profitability considerations in adoption
decisions, while showing that differences in technical characteristics of pro-
duction programs, product mixes, and institutional structures of firms are
key factors governing the diffusion process. Davies (1979) also has studied
the adoption of production innovations in British industry, finding that the
speed of diffusion is slower where the scale of production typical of firms
is smaller and where there is a longer expected payback period for fixed
investments embodying new processes.
New conceptual approaches to the microeconomic analysis of the subject
have evolved out of these path-breaking empirical investigations. The point
of departure in Mansfield's studies was a model of diffusion involving a new
technology having prespecif~ed engineering and economic characteristics, an
unchanging population of potential users who had to be persuaded of the
profitability of the innovation, and an objective economic environment in
which the only consequential change occumng was the gradual dissemination
of information. Looked at from this angle, the gradual increase in the extent
of an innovation's application across the fins and sectors of the economy
talces on the appearance of an adjustment process, which eventually ap-
proaches the restoration of equilibrium.
An alternative conception has been developed by considering the historical
and contemporary evidence that many new technologies are initially in~o-
duced in forms and under market conditions that make them appear profitable
in immediate applications only for some firms within We relevant industry;
indeed, perhaps only in the operations of some plants and departments within
those fins. Subsequently, however, as the new technology and its micro-
economic environment convolve, the extent of profitable application will
broaden.
Abundant confirmation of the modern relevance of the major empirical
premises on which this approach is grounded has been provided by Gold
(1979, 1981, 1983) and his associates at Case Westem Reserve University
Trough their detailed case studies of managerial decision processes pertaining
to the adoption of innovation. Furthermore, its importance has been con-
firrned repeatedly through the many individual historical diffusion studies
that have absorbed some of the best efforts of a generation of quantitative
. . .
economic h~stonans.
Analytical work in which I have had a hand (David, 1969, 1975; David
OCR for page 380
380
PAUL A. DAVID
and Olsen, 1984) and important contributions by Davies (1979), Stoneman
(1983), and others have led to the elaboration of the class of so-called
equilibrium diffusion models. These models have emphasized two funda-
mental points, which can be added to those brought out by Mansfield's
contnbutions. The first is that even if information relevant for rational de-
cision making about the innovation were instantaneously disseminated with-
out cost, there would remain many reasons to expect that states of equilibrium
would exist involving less-than-complete diffusion of the new technology
within the industry. The second point, following immediately from the first,
is that our attention should be directed to the various dynamic forces whose
influence gives rise to a "moving equilibnum" in the potential level of fully
informed adoption of innovations by rational, profit-seeking agents.
I shall want to carry on from this last point and discuss some of the supply-
side forces that have become recognized as crucial in driving technology
diffusion forward. (See Rosenberg,, 1972; Stoneman, 1976; Metcalf, 1981;
Sahal, 1981; Stoneman and Ireland, 1983; David and Olsen, 1984.) But
before I can do that, I must briefly address some fundamental aspects of the
demand side of the adoption of innovation.
Key Demand Factors in Technology Diffusion
The demand to take up new technologies whether embodied in inter-
mediate products, such as epoxy resins and ethylene/propylene rubber, or in
such equipment as computerized numerical-controlled machine tools, or in
complete industrial facilities, such as large-scale ammonia plants will not
be ubiquitous and instantaneous, if only because potential users do not find
themselves in identical technical and economic circumstances. They may
face different raw material costs, energy prices, and transport charges; they
may differ in regard to the makeup of technically related product arrays
produced using joint facilities; they may operate in different labor markets
and have different implicit or explicit contractual commitments with their
employees; they may encounter different teas for borrowing or different
opportunity costs of internally financing capital projects. All these may have
a bearing on whether a proposed change in production methods to incorporate
an innovation will appear worth undertaking when it first becomes available,
or at some subsequent point in time.
The preceding catalog does not yet exhaust the list of significant aspects
of heterogeneity within the population of "potential adopters"; not even the
list of important objective economic differences capable of generating a wide
distribution of responses to an innovation about which few technological
uncertainties remain. Two further basic aspects of the demand side of mi-
croeconomic adoption decisions must be recognized here.
First, in projects characterized by larger fixed costs for state-of-the-art
OCR for page 381
TECHNOLOGY DIFFUSION, PUBLIC POLICY, ID INDU=~4 COMPLETENESS 381
plant and equipment, where offsetting savings in variable costs become sig-
nificant only at high throughput rates, a critical issue is the scale of output
that the enterprise can anticipate maintaining with the production facility in
question. Considerations of this kind were as much a factor bearing on the
adoption of the early grain harvesting machines, and later of gasoline farm
tractors, as it has been in the post-World War II diffusion of new plant
designs in the petrochemicals industry, or as it now is in regard to the
installation of second-generation industrial robots (see David, 1984, and
references).
The second point to note is that the decision to introduce a new industrial
process is often bound up with the determination to discontinue operation of
existing capital facilities. While an old plant may be technologically obso-
lescent, prevailing product prices in the industry may permit variable costs
to be covered and so make it rational for profit-maximizing firms to defer
the date of capital replacement. New technologies are placed at a distinct
disadvantage in competition with their predecessors whenever they come
embodied in or are technically interrelated with indivisible capital goods that
will burden the user with heavy fixed-cost charges (see Frankel, 1955; Salter,
19661.
This is so especially when the old techniques are embedded in extremely
durable physical plant with low maintenance requirements. The slow headway
made by the Solvay process, in displacing the use of the antiquated Leblanc
method for the manufacture of sodium carbonate in Britain during the last
quarter of the nineteenth century, has been shown by Lindert and Trace
(1972) to be explicable largely on just such grounds. Thus, the legacies of
past capital formation decisions, as well as the costs of new investment, may
combine to create differences which will determine We timing and pattern
of technology diffusion within firms and industries. Durable facilities sur-
viving from earlier epochs may pose barriers to the introduction of best-
practice methods ~at, in pathological cases, cannot be surmounted by the
workings of normal competitive market processes (see David, 1975:Ch. 5;
and David, 19851.
In light of We foregoing, it should be plain that a decision-process "failure"
Is not necessarily involved just because we notice that particular innovations
are being "neglected," or only haltingly and partially adopted by firms in
one place even Dough Hey have been applied effectively elsewhere. Such
lags are to be expected. For example, in a widely cited study of four important
U.S. industries (bituminous coal, iron and steel, brewing, and railroads,
spanning We period 189~1958 when we were not particularly worried
about the competitiveness of our staple industries), Mansfield (1968) found
that the complete diffusion of new process technologies among just the major
firms stretched out over durations of 10 years and longer in 9 out of a sample
of 12 process innovations; and for 5 of those same innovations, more than
OCR for page 382
382
PAUL A. DAVID
20 years was required for the complete transition. The causes, and quite
possibly also the remedies, may lie at some remove from Me immediate
locus of decision making within the "lag Bard" films. We need not surmise
that the managers are poorly informed about the potentialities of emergent
technologies, or psychologically wedded to familiar routines, or hesitant to
accept risks in search of greater profits; nor must we hasten to implement
organizational revolutions that will alter their ways if not their personalities.
Human beings even graduates of schools of business and of engineenng-
certainly do fall prey to all these weaknesses, which in some instances may
prove critical. Yet, there are ample grounds for doubting that sluggish man-
agenal reactions, made more sluggish still by uncertainty, are really what
lies behind the long diffusion lags so commonly experienced when process
innovations are involved.
Key Supply Factors in Technoloov Diffusion
con, _ ~~,
Part of the explanation for the length of time it takes for new technologies
to supplant old ones is to be found by considering the factors on We supply
side that drive forward the diffusion process. I can deal only briefly here
with these inherently complicated questions, so it may be excusable for me
to report what has been learned in the following, overly simplified way. The
terms on which users can acquire effective access to new technologies will
reflect one or more of three classes of cost: (1) the costs of securing and
evaluating information about Hem from others who possess it; (2) the costs
of obtaining the specialized materials or equipment in which new technologies
having particular performance characteristics are physically "embodied,' by
the supplying firms; and (3) the costs of specialized facilities, ancillary prod-
ucts, or services Hat are closely tied to the innovation for technical reasons
and thus will affect its performance.
The significance of the first two items on this little list seems obvious
enough, although it has come as something of a surprise to many economists
that the real resource costs of transmitting and absorbing all of the relevant
"unembodied" technical knowledge can be very substantial, even under
conditions of coordinated technology transfer such as Hose effected by
multinational fins. A study made by Teece (1976) of 26 technology-transfer
projects involving manufacturing ventures In chemicals, petroleum refining,
and machinery found ~at, on average, 19 percent of total project costs was
accounted for by these intangibles: pre-engineering information exchanges;
engineering costs associated win ~ansfemng He necessary designs; R&D
personnel unsized during He transfer phase; and pre-start-up Gaining, learn-
ing, and debugging.
We should recognize at this point Hat differences among fins in the costs
of absorbing technological information, even information disseminated on a
uniform basis by a supplier or an independent agency, may give rise to
OCR for page 383
TECHNOLOGY DIFFUSION, PUBLIC POLICY, AD INDU~^ COMPLETENESS 383
differences in ~nnovation-adoption behavior. Mowery (1983) suggests Mat
firms that maintain R&D activities are Hereby provided with an enhanced
capability for monitoring and assessing technologies originating elsewhere,
and Teece (1976) also contends that R&D contributes to the ability of trans-
ferees to absorb manufacturing technologies at minimum cost. To the extent
Mat this may be confirmed by other studies, especially studies of `' domestic
technology transfer costs," this effect could turn out to be a substantial if
generally unacknowledged source of the pnvately appropnable benefits de-
nved from expenditures on research and development. It certainly has been
overlooked in studies Mat have sought to assess the private and social returns
on R&D solely in terms of the value of Me innovations that are generated
(see Mansfield et al., 1977~.
The existence of R&D capabilities Mat enable some firms widen an in-
dusoy7 or outside its to ``reverse engineers new products and ``invent around, 7
patented processes contributes to reducing the costs of sanitation. The "~m-
itahon threat capacity" Hereby created even if it is not actually used
may act as a factor limiting the prices that patent holders will charge licensees
for access to He technology. Even firms that do not possess such R&D
capabilities may benefit from Heir existence in this way: This is a possible
externality (or spiBover effect) of company-financed R&D; it may contribute
to raising productivity levels In the industry by promoting tile diffusion of
available technologies, rather than the generation of further innovations.
The third item on my list assumes particular significance for technologies
characterized by the presence of network externalizes, as Hey are called in
He economists' jargon (see Hanson, 1984; David, 1985; Katz and Shapiro,
1985~. Illustrations spring to mind readily in the field of telecommunications,
where decisions regarding terminal equipment are affected by He costs of
access to other pardes over existing transmission networks. In the case of
computers, the hardware costs of particular machines are only a part of He
story, for, die available range and price of compatible software also matter
In detennining the use that will be made of the technology. On a more
mundane level, conveniently located repair centers staffed win Gained tech-
nicians and adequately stocked spare-parts depots also constitute a form of
"service network" supporting users of specialized equipment from trucks
to vacuum cleaners. As In He preceding cases, service networks may be
provided by agents other than the suppliers of the equipment itself.
All Free classes of technology-access costs share a common feature that
is crucially important for the dynamics of diffusion processes. Win the
passage of dine, and in response to the widening extent of application of He
technology in question, each Is likely to undergo a decline. The specific
details vary among the classes, however:
1. Coordinated technology transfer appears to be a decreasing cost activity,
In the sense that its costs decline with each application of a given innovation
OCR for page 384
384
PAUL A. DAVID
or each startup undertaken by Me transferor (see Teece, 1976; Stoneman,
1983~. In addition, uncoordinated transfers of technical information can and
do occur through the accepted mobility of scientific and engineering personnel
within and sometimes between industries. The phenomenon was as relevant
for the rapid diffusion of production methods within the Japanese cotton
textile industry dunug the late Meiji Era (189~1912), as it can be seen to
be in Silicon Valley today (see Saxonhouse, 1985; Okimoto, 1984; Cohen,
1985~. Notice, then, Mat a given rate of personnel turnover in the industry
will have a greater effect in increasing the accessibility of information about
a new technology, the larger Is the cadre of scientific and engineering workers
who are being exposed to it at any one point in time. Consequently, this
~fonnation-d~ssem~nation mechanism will, at least for a time, be positively
reinforced by the widening adoption of the innovation as an increasing pro-
pori~on of the relevant skilled manpower pool comes into contact win those
who have already had an opportunity to acquire it.
2. Technologies are not static. They undergo instead a gradual, evolu-
tionary development that is ornately bound up with Me course of their
diffusion. This was a point made by Rosenberg (1972), but Me rest of us
have token about a decade to catch up with him. The initial versions of a
new product or process, even those Mat reach the market and find purchasers,
often suffer from numerous flaws in production or design. Identification and
remedy of such defects are, in many instances, dependent on Me accumulation
of feedback infol~ahon from users a process to which Rosenberg (1982)
recently has affixed Me label "learning by using." By considering that
performance improvements in commodities whose costs remain unchanged
may be rendered equivalent win reduced costs of products of a constant
kind, we can draw a direct analogy win the more familiar and widely
documented phenomenon of "learnung by doing," or irreversible dynamic
scale economies. (See Arrow, 1962; David, 1975:Ch. 2 md citations therein.)
Along both kinds of leaking curves Me accumulation of experience in
production and In utilization that diffusion itself makes possible is seen to
govern and sustain a continuing flow of incremental innovation. And such
innovations, by lowenug the effective '~cost" of successive vintages of the
embodied technology, reciprocally extend its penetration into new markets
and areas of application.
3. Network technologies exhibit some of the same dynamic features in
their development as do "freestanding" product and process innovations
As a network's coverage is extended by linking up additional '~subscribers,"
Me cost of providing basic services to each user will decline, and Me potential
qualitative advantages of being "hooked up" with a widening circle of users
tend to increase. The essential economic problem posed by such systems,
however, is that integration requires some measure of technical compatibility
or standard ton and thus imparts to them Me characteristics of a public
OCR for page 385
TECHNOLOGY DIFFUSION, PUBLIC POLICY, kD INDUCED COMPLETENESS 385
good (see Kindleberger, 19831: you cannot be linked with me by videotext
without my being thus linked with you. This means that the benefits derived
(or Me effective costs incurred) by any one prospective system user may be
dependent on the willingness of others to incur the costs of achieving com-
patibility with the same network.
Where Were are alternative emerging network technologies to choose among,
as is currently the case with teletext and videotext (see Tydeman et al.,
1982), the public goods problem tends to work to retard the adoption of any
one of them; potential subscribers will be inclined to wait in the hope that
others will bear the costs of achieving compatibility with them. "Free-riding
behavior" of this kind works to prevent the true demand for public goods
from being revealed by competitive market processes. Therefore, if de facto
standardization does not occur through the market dominance of one among
the available alternative systems, there may be a case for public standard
setting to foster diffusion of the technology. A difficulty, of course, lies in
choosing among technological alternatives at an early stage of their devel-
opment, while great uncertainty still surrounds both the potential technical
capabilities of rival systems and the eventual performance charactenstics Mat
users will value most highly. Furthermore, there is Me distinct danger that
premature imposition-of standards will close off opportunities for profitable
investment in research, design, and development of still over technological
alternatives.
Economics lives up to its reputation as "the dismal science" by telling
us again and again that the world is arranged so Mat one cannot have ev-
er~ing: here we need to recognize the existence of a trade-off between Me
more certain gains of greater benefits of diffusion today and the chance of
having more beneficial innovations tomorrow.
Once we have moved beyond a static conception of technologies, Mere is
another point of great importance to be made about the dynamics of tech-
nology choices under conditions of decreasing costar system scale econ-
om~es. It is a point that has emerged clearly in the recent fundamental research
on the subject by my Stanford colleague, Brian Arthur (19831. The existence
of irreversible, dynamic scale economies—such as those generated in "learn-
ing by doing" and "learning by using" means that small, initial advantages
(or disadvantages) can readily cumulate into larger ones. This opens Me
possibility Mat a particular product design, process technology, or system
can become "locked in," and rival technologies can become '`locked out"
through the working of competitive market processes. In other words, even-
n~al de facto standardization is Me most likely outcome, indeed, under some
conditions it is a virtual certainty as Arthur (1985), based on Arthur et al.
(1983, 1985), has shown.
Thus, the lesson borne home to us is Mat even in the absence of govern-
OCR for page 386
386
PAUL A . DAVID
mental intervention to impose technical standards, even without formal rec-
ommendations of industry associations, and in even in markets in which dlere
is no dominant technology supplier (like IBM) or sanctioned monopolist (like
the old AT&T), we are quite likely to get technology "standards." They
will come to us through the long-run operation of the forces of decreasing
cost that ~ have reviewed in looking at the supply side of We diffusion process.
The issue is what kind of technology standards will we get, and what seem-
ingly small "accidents" of industrial history, or temporary public policy
twists during the early phases of the diffusion of rival technologies, will
leave an unintended but nonetheless indelible mark on Be future? If you
need a concrete, everyday illustration of the workings of the "lock in"
process and the kind of technological outcomes to which we can be led by
seemingly small accidents of history, take a look at the awkwardly arranged
QWERTY keyboard of your typewriter and your personal computer (see
David, 1985).
CONCLUDING PERSPECTIVES ON PUBLIC POLICES
In light of Be ongoing conceptual reorientation whose main aspects I have
just reviewed, the terminology of "diffusion" itself now seems less and less
helpful as a metaphor for Be phenomena it is being used to label—although
by now it is inextricably lodged in Be economists' jargon. The modern
allusions Hat the word "diffusion'' carries to the behavior of gas molecules
suggests that new technologies are somehow (randomly) finding their way
into application in new locations. But our new explanatory approaches derive
from the contrary vision of economic agents purposively acting to acquire
and apply innovations.
We are skill well short of achieving a full and deep understanding of
technology diffusion at a purely theoretical level, and a great deal of pains-
taking empirical research remains to be undertaken in order to determine the
ranges over which venous proposed models may be said to apply. What has
been achieved is, in a sense, only the first stage of Be integration of the
subject into the mainstream of modern microeconomic analysis.
But even this much is not without significant implications for our thinking
about technology policy. To begin with, it greatly facilitates identification
of a wide array of technical, infonnaiional, and market factors as impinging
on either Be supply side or the demand side of technology transfers, and it
facilitates investigation of those factors' likely influences on economically
rational decisions about when to undertake Be application of specific new
technologies. In this way, the recognized assortment of effective "policy
levers" can be augmented.
Further, the framework being fashioned leads one naturally to Wink Eat
Were is an "optimal" (expected) waiting time before a new technology is
OCR for page 387
TECHNOLOGY DIFFUSION, PUBLIC POLICY, AD INDU=RI~ COMP=7TWENESS 387
adopted for use by a given producer or consumer. It will no longer do to
say Hat something must be amiss with the way private decision processes
are working simply because everyone does not decide to adopt technologies
as soon as Hey are commercially introduced. More rigorous economic cn-
teria, showing that there is a divergence between privately and socially
optimal "waiting times," must be met in order to justify conclusions that
the collective outcome of such individual decisions deferring adoption is
unsatisfactory, in the sense of proceeding "too slowly" and therefore leaving
the extent of use "too limited."
Both implications constitute bases for initiating a thorou~h-going re-
examination and reformulation of public policies and programs that affect
technology diffusion. Here, however, I shall venture no farmer than to offer
some broad perspectives on the current policy scene.
Formal programs directed toward encouraging wide application of new
technologies in the civilian economy are often associated in the public mind
and in the mind of legislators with "diffusion policy." For better or for
worse, the approach to technology diffusion that those programs represent
has not received extensive public funding in this county (see Rhode, 19851.
The Cooperative Extension Service program of the U.S. Department of
Agriculture long has remained the major claimant for such federal funding
as has been devoted to domestic technology transfer. Although there has
never been a thorough economic evaluation of the costs and benefits at~b-
utable to this program, it has established for itself a reputation for "field
success" Hat has carried over into academic circles. (See Mosgavero and
Shane, 1982; Ruttan, 1982; Rogers, 1983.) This led other federal agencies
such as He National Aeronautics and Space Administration and He Depart-
ment of Energy—to attempt to imitate its methods on a much smaller scale
and in quite different circumstances. Yet, in comparison with federal ex-
penditures for R&D, even with the small outlay for civilian R&D, the Co-
operaiive Extension Service program has remained trivial in its budgetary
. .
c ,lmenslon.
My purpose here is not to lament the small budgetary scale of federally
financed technology-~ansfer efforts, or to bemoan He Reagan administra-
tion's evident reluctance to imitate Western European governments, which
several years ago began developing industrial policies with greater emphasis
on measures aimed at '~promoting He diffusion of new technologies" in
specified areas (see Stout, 1981; David and Stoneman, 19851.
What ~ wish to stress instead, is that there is far more to public policies
and actions affecting technology diffusion Han the information-dissemination
programs modeled on the Agricultural Extension Service. Innovation-
adoption decisions made by the individual firm are decisions about invest-
ment, in essence and usually in substance. As such, they can be critically
influenced by monetary and fiscal policies Hat affect the costs of borrowing
OCR for page 388
388
PAUL A . DAVID
and the after-tax rates of return on new technology-embodying capital goods;
by the influence of macroeconomic policies on the prospective rate of uti-
lization of fixed plant and equipment; by the impacts of energy policies on
the prices of inputs complementary with capital facilities, by educational
policies that may quicken or slow the growth in the supply of workers with
specialized skills required by new technologies; by antitrust policies and
regulations that shape the market structure in the industries that may take up
the innovations; by patent laws and rulings that influence the costs of access
to new technologies; by the entire panoply of policies directed ostensibly
toward generating future industrial innovations at a faster pace, and the
expectations of obsolescence risks that these engender in regard to the in-
novations already at hand.
We in the United States, therefore, are already in the business of making
de facto public policy choices regarding technology diffusion. We have been
doing it on a big scale, using a wide array of instruments, without fully
facing up to the fact and taking heed of what we are about. In this we
resemble the character in Moliere's play The Misanthrope who discovered
he was a doctor in spite of himself. We have not spared ourselves the effects
of He actual policy measures. Rawer, for the sake of avoiding the discomfort
of addressing systematically the issues concerning diffusion in our national
policy discussions,-we have surrendered the opportunity to see whether it is
possible to formulate any consistent set of goals, or to examine what would
be entailed in trying to coordinate He actions of the many different public
agencies that are now participating in a de facto technology policy-setting
process. We are not avoiding making mistakes by proceeding in this way.
One is almost bound to make mistakes in these matters. Instead, we are
avoiding having to acknowledge the mistakes we make and limiting our
ability to learn from them.
How better should we proceed? Are there some broad implications for
public policy discussions that can be drawn from the hurried tour I have
conducted of "the new microeconomics of technology diffusion"? Let me
close by suggesting these three:
1. The cases for encouraging wider adoption of new technologies must
be considered on their respective meets; generalizations in this area are more
apt than not to be misleading, and indiscriminate promotion of the maximum
possible extent of adoption is not a desirable goal for public policies. The
socially optimum extent of diffusion sometimes will exceed Hat which private
markets generate, as when static public goods effects are strong. But this is
not always the outcome.
2. Policies designed to quicken the rate of adoption of available new
technologies must be framed win due attention to the private and social
costs, as well as the benefits associated win establishing a more rapid pace
of adoption. In this sphere, `'faster" will not invariably be `'better."
OCR for page 389
TECHNOLOGY DIFFUSION, PUBLIC POLICY, AND INDUSTRIAL COMPET777VENESS 389
3. Assessments of the sat~sfactonness of We rate of diffusion cannot be
undertaken without regard for the many complex and conflicting relationships
Rat obtain between decisions about the adoption of new technologies and
He commitment of private resources in pursuit of further technological ad-
vances. Innovation policies and diffusion policies have for too long a time
been formulated and evaluated separately, as if the two processes were quite
independent. It is time to treat them within an integrated approach to en-
hancing economic welfare by improving the technological and other bases
of our industrial competitiveness.
ACKNOWLEDGMENT
The author is indebted to Douglas Puffers, Paul Rhode, and Joshua Ro-
senbloom for able research assistance on projects supported under a grant to
the Technological Innovation Program of the Center for Economic Policy
Research at Stanford University, upon which this chapter has drawn.
REFERENCES
Arrow, Kenneth J. 1962. The economic implications of leaniing by doing. Review of Economic
Studies, 29 (June):155-173.
Arthur, W. Bnan. 1983. On Competing Technologies and Histoncal Small Events: The Dynamics
of Choice Under Increasing Returns. IIASA (Laxenburg, Austna). Paper presented at the Tech-
nological Innovation Program Workshop, Stanford University, Deponent of Economics, No-
vember, 1983. Revised version available as Technological Innovation Program Working Papers,
No. 4 (January 1985). Center for Economic Policy Research. Calif.: Stanford University.
Arthur, W. Bnan, Yun M. Ermoliev, and Yun M. Kaniovski. 1983. On generalized um schemes
of the polya kind. Cybernetics 19:61-71.
Arthur, W. Bnan, Yun M. Errnoliev, and Yun M. Kaniovski. 1985. Strong laws for a class of
path-dependent um processes. In Proceedings of the International Conference on Stochastic
Optimization, Kiev, 1984. Munich: Spnuger-Verlag.
Cohen, David. 1985. Turnover Statistics for Electronics Engineers and Electronics Industry Exempt
Personnel in Local U.S. Labor Markets. Silicon Valley Research Project Memorandum, Center
for Economic Policy Research, Stanford Universiny, March.
David, Paul A. 1969. A Contnbution to the Theory of Diffusion. Center for Research in Economic
Growth Research Memorandum, No. 71. Stanford University.
David, Paul A. 1975. Technical Choice, Innovation, and Economic Growth: Essays on American
and British Experience During the Nineteenth Century. London and New Yorlc: Cambridge Uni-
versity Press.
David, Paul A. 1984. The reaper and Be robot: The diffusion of microelectronics-based process
innovations in historical perspective. Technological Innovation Program Working Papers, No. 2.
Center for Economic Policy Research. Calif.: Stanford University. Forthcoming in J.-J. Salomon,
ea., Science, Technology, and Society, Paris, 1985.
David, Paul A. 1985. Clio and the economics of QWERTY. American Economic Review 75,
2 (May):332-337.
David, Paul A., and Trond E. Olsen. 1984. Anticipated automation: A rational expectations model
of technological diffusion. Technological Innovation Program Working Papers, No. 2. Center
for Economic Policy Research. Calif.: Stanford University.
David, Paul A., and Paul L. Stoneman. 1985. Adoption-subsidies vs information provision as
OCR for page 390
390
PAUL A DAVID
instruments of technology policy. Technological Innovation Program Working Papers, No. 6.
Center for Economic Policy Research. Calif.: Stanford University, April. (Forthcoming in Eco-
nomic Journal, 1986.)
Davies, Stephen. 1979. The Diffusion of Process Innovations. London: Cambridge University Press.
Frankel, M. 1955. Obsolescence and technological change. American Economic Review 45 (June):298-
319.
Freeman, Christopher. 1982. The Economics oflndustrialInnovatzon. 2ded. London: Frances Pinter.
Gold, Bela. 1979. Productivity, Technology and Capital. Boston: Lexington Books.
Gold, Bela. 1981. Technological diffusion ire industry: Research needs and shortcomings. Journal
of Industrial Economics (March):247-269.
Gold, Bela. 1983. On the adoption of technological innovations in industry: Superficial models and
complex decision processes. Ch. 10 in The Trouble with Technology: Explorations in the Process
of Technological Change, S. MacDonald, D. McL. Lamberton, T. D. Mandeville, eds. London:
Frances Pinter.
Gnliches, Zvi. 1957. Hybrid cam: An exploration in the economics of technological change. Econ-
ometrzca 25 (October):501-522.
Gnliches, Zvi. 1960. Hybrid corn and the economics of innovation. Science 132 (July):275-280.
Hanson, Ward A. 1984. Bandwagons and Orphans: Dynamic Pricing of Competing Technological
Systems Subject to Decreasing Costs. Paper presented at the Technological Innovation Program
Workshop, Stanford University, Department of Economics, January, 1980.
Ireland, Nolan, and Paul Stoneman. 1984a. An Integrated Approach to the Economics of Tech-
nological Change. Paper presented at Warwick Summer Workshop on the Economics of Tech-
nological Change, July, 1980.
Ireland, Women, and Paul Stoneman. 1984b. Technological Diffusion, Expectations, and Welfare.
Paper presented at the Technological Innovation Program Workshop, Stanford University, De-
partment of Economics, February 1984.
Katz, Michael L., and Carl Shapiro. 1985. Network externalities, competition, and compatibility.
American Economic Review 75(3):420440.
Kindleberger, Charles P. 1983. Standards as public, collective and private goods. Kyklos 36
(Fasc. 3):377-396.
Lindert, Peter D., and Keith Trace. 1972. Yardsticks for Victorian enterpreneurs. Ch. 7 in Essays
in a Mature Economy, D. N. McCloskey, ed. London: Methuen.
Mansfield, Edwin. 1968. Industrial Research and Technological Innovation. New York:
W. W. Norton.
Mansfield, Edwin, J. Rapoport, J. Schnee, S. Wagner, and M. Hamburger. 1971. Research and
Innovation in the Modern Corporation. New York: W. W. Norton.
Mansfield, Edwin, J. Rapoport, A. Romeo, E. Villani, S. Wagner, and R. Husic. 1977. The
Production and Application of New Industrial Techwlogy. New York: W. W. Norton.
Mansfield, Edwin, A. Romeo, M. Schwartz, D. Teece, S. Wagner, P. Brach. 1982. Technology
Transfer, Productivity, and Economic Policy. New York: W. W. Norton.
Metcalf, J. S. 1981. Impulse and diffusion in the study of technological change. Futures 13(5):347-
359.
Mosgavero, Louis N., and Robert S. Shane. 1982. What Every Engineer Should Know about
Technology Transfer and Innovation. New York: Marcel Delcker.
Mowery, David C. 1983. Economic Meow and government technology policy. Policy Sciences
16:27~3.
Nabse~, L., and G. F. Ray, eds. 1974. The Diffusion of New industrial Processes: An International
Study. London: Cambridge University Press.
National Academy of Engineenng. 1982-1985. Committee on Technology and International Ecm
nomic and Trade Issues. The Competitive Status of Ike U.S. Auto Industry (1982), The Competitive
Status of the U.S. Fibers, Textiles, and Apparel Complex (1983), The Competitive Status of the
OCR for page 391
TECHNOLOGY DIFFUSION, PUBLIC POLICY, ADD INDUSTRIAL COMPETITIVENESS 391
U.S. Machine Tool Industry (1983), The Competitive Status of the U.S. Pharmaceutical Industry
(1983), The Competitive Status of the U.S. Electronics Industry (1984), The Competitive Status
of the U.S. Civil Aviation Manufacturing Industry (1985), The Competitive Status of the U.S.
Steel Industry (1985), Washington, D.C.: National Academy Press.
Nelson, Richard R., ed. 1982. Government and Technical Progress: A Cross-lndustry Analysis.
New York: Pergarnon Press.
Okimoto, Daniel I. 1984. Conclusions. Ch. 6 in Competitive Edge: The Semiconductor Industry in
the U.S. and Japan, D. I. Okimoto. T. Sugano, and F. B. Weinstein, eds. Calif.: Stanford
University Press.
President's Commission on Industrial Competitiveness. 1985. Global Competition: The New Reality.
Report of the President's Commission on Industrial Competitiveness, John A. Young, chairman.
Washington, D.C.: U.S. Government Printing Office.
Rhode, Paul. 1985. Federal Government Policies on Technology Diffusion. Paper presented at
Technological Innovation Program Workshop, Stanford University, Department of Economics,
January.
Rogers, Everett M. 1983. Diffusion of Innovations, 3rd ed. New York: Free Press.
Romeo, Anthony A. 1975. Intenndustry and interfile differences in the rate of diffusion of an
innovation. Review of Economics and Statistics 57:311-319.
Romeo, Anthony A. 1977. The rate of imitation of a capital-embodied process innovation. Econ-
ometrica 45 (February):63-69.
Rosenberg, Nathan. 1972. Factors affecting the diffusion of technology. Explorations in Economic
History 9 (Fall), 2d senes.
Rosenberg, Nathan. 1976. On technological expectations. Economic Journal 86:523-535.
Rosenberg, Nathan. 1982. Learning by using. Ch. 6 in Inside the Black Box: Technology and
Economics. New York: Cambridge University Press.
Ruttan, Vernon. 1982. Agricultural Research Policy. Minneapolis: University of Minnesota Press.
Sahal, Devendra. 1981. Patterns of Technological Innovation. Reading, Mass.: Addison-Wesley.
Salter, W. E. G. 1966. Productivity and Technical Change. 2d ed. Cambridge: Cambridge University
Press.
Saxonhouse, Gary R. 1985. Mechanisms for Technological Transfer in Japanese Textile History.
Paper presented at Social Science History Workshop, Stanford University, Department of Eco-
nomics, March, 1985.
Stoneman, Paul L. 1976. Technological Diffusion and the Computer Revolution: Me U.K. Expe-
rience. Department of Applied Economics Monographs, No. 25. London: Cambridge University
Press.
Stoneman, Paul. 1983. The Economic Analysis of Technological Change. Oxford: Oxford University
Press.
Stoneman, Paul, and Norman Ireland. 1983. The role of supply factors in He diffusion of new
process technology. Economic Journal, Supplement (March):65-77.
Stout, D. K. 1981. The case for government support of R&D and innovation. In C. Carter, ea.,
industrial Policy and Innovation. London: Heinernann.
Teece, David. 1976. The Multinational Corporation and the Resource Cost of Inrernatioru21 Tech-
nology Transfer. Cambndge, Mass.: Ballinger.
Tydeman, J., H. Lipinski, R. Adler, M. Nyhan, and L. Zwimpfer. 1982. Teletext and Videotext in
the United States: Market Potential, Technology, Public Policy Issues. New York: McGraw-Hill.
OCR for page 392
Representative terms from entire chapter:
technological innovation
