Output Measurement in Science and Technology. Essays in Honor of Yvan Fabian

178 Pages·1987·4.084 MB·English

by  C. Freeman (Eds.)

OUTPUT MEASUREMENT
IN SCIENCE AND TECHNOLOGY
Essays in Honor of Yvan Fabian
Edited by
CHRISTOPHER FREEMAN
Science Policy Research Unit, University of Sussex
1987
NORTH-HOLLAND
AMSTERDAM · NEW YORK · OXFORD · TOKYO
© Elsevier Science Publishers B.V., 1987
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any
means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner.
ISBN 0 444 70330 6
PUBLISHERS:
ELSEVIER SCIENCE PUBLISHERS B.V.
P.O. BOX 1991
1000 BZ AMSTERDAM
THE NETHERLANDS
SOLE DISTRIBUTORS FOR THE U.S.A AND CANADA!
ELSEVIER SCIENCE PUBLISHING COMPANY, INC.
52 VANDERBILT AVENUE
NEW YORK, NY 10017
U.S.A.
These papers were first published as a special issue of Research Policy, Volume 16, Numbers 2-4, August 1987, pp. 55-228 (Elsevier
Science Publishers B.V. (North-Holland, Amsterdam).
PRINTED IN THE NETHERLANDS
ν
Preface
For over a century now there has been an that the US government (through the National
increase in the scale of those industrial, scientific Science Foundation) began its first regular surveys
and technical activities which are generally known of R&D using definitions which were largely
as "Research and Development" or, more accu- adopted by most other OECD countries as they
rately, as "Research and Experimental Develop- followed the US lead in the 1950s and 1960s. The
ment". Starting in universities and scientific socie- OECD standardised these definitions in the 1960s
ties in the Middle Ages, scientific research is now in the so-called "Frascati Manual", named after a
a driving force of economic growth and interna- Conference at Frascati in 1963 which was con-
tional competition all over the world. It was the vened by Yvan Fabian, who had just been ap-
German chemical industry and the American elec- pointed to the OECD staff to head up the work on
trical industry which first realised the possibility, science and technology statistics.
that by organising their own "in-house" scientific It proved to be an excellent appointment. From
research and development they could gain a big the start until his untimely death in 1985, Yvan
competitive advantage by winning new markets Fabian devoted himself wholeheartedly to impro-
with new products and new processes. ving the accuracy, scope, timeliness and compara-
Following the early successes of these in- bility of these rather specialised but very im-
dustries, most large firms in all the industrialised portant statistics. He was extraordinarily helpful
countries now organise their own "R&D", and and considerate to all those people throughout the
this mode of technological competition is now OECD area (and outside it) who were starting up
spreading from the manufacturing industries to surveys for the first time, or trying to introduce
the service industries as these branches of the new types of statistics or improve the system. This
economy now develop their own software for was perhaps his most admirable quality and made
service innovations. his group a centre of research and new ideas and
As the professional R&D department became not just a routine statistics collection operation.
a regular feature of the industrial landscape in the From the very beginning in 1963 at the first
inter-war period, there was increasing interest in Frascati Conference, Yvan Fabian recognised that
the possibility of measuring these activities. It was the official R&D statistics were only the first step.
obvious that there were wide variations in the He was more aware than anyone else of their
scale of commitment of different firms, industries limitations and understood the importance of other
and countries to R&D, and managers, scientists, scientific and managerial activities in bringing
engineers and economists were all interested in the about technical innovations. Most important of
comparative performance of more or less "re- all, he knew perfectly well that R&D expenditures
search-intensive" firms or industries. and personnel statistics were only measures of
The first surveys were made by industrial feder- inputs and that it was the measurement of outputs
ations and by independent university-based scien- which was the real challenge.
tists such as Huxley and Bernal. But these early For this reason he insisted on including a sec-
estimates suffered from a lack of standard defini- tion on output measurement even in the first
tions and from low response rates, so that they Frascati Manual, which proclaimed the objective
generally underestimated R&D activities, particu- of output measurement, even though this was
larly the "D" part of "R&D". thought by many to be an impossible goal. Over
It was not until after the Second World War the next 20 years Yvan seized every opportunity to
promote the attainment of his goal. Recognising
that it was not possible for an international
Output Measurement in Science and Technology: Essays in organisation to take the initiative directly, he nev-
Honor of Yuan Fabian, Edited by C. Freeman ertheless did everything possible to encourage re-
© 1987, Elsevier Science Publishers B.V. searchers in universities to experiment with in-
VI Preface
novative ideas and also stimulated governments ous problems in science policy and the manage-
and industrial organisations to take new ini- ment of R&D. The papers were first pubhshed as
tiatives. As important steps in this prolonged a special triple issue of the journal Research Policy
campaign he organised a series of international in 1987 in honor of Yvan Fabian. I am certain
seminars in the 1980s, where the results of work in that he would have been delighted with the origi-
progress were presented. nal publication and happy to see the papers made
As a result of two decades of research and available to a wider audience in book form. I
experiment, much progress has been made in the know too that, like me, he would have hoped that
use of a variety of "output" measures, especially this publication would stimulate further efforts to
patent statistics, bibliometric and citation analy- improve these statistics and to establish them on a
sis, and numbers of innovations and their diffu- regular national and international basis.
sion. The papers pubhshed in this book are a
representative set of the results of empirical re- Christopher Freeman
search in all these fields. It is for the reader to Science Policy Research Unit,
judge how far the authors have succeeded in using University of Sussex
"output" statistics to illuminate their comparisons July 1987
between firms and countries and to analyse vari-
3
Introductory note
J. MARCUM
Director, Directorate for Science, Technology and Industry, Organisation for Economic Co-operation and Development, Paris, France
For many years Yvan Fabian was the mainstay which remained with him to the end. First he was
of efforts to develop better measures of scientific always thinking ahead to identifying new data to
and technological resources and performance in meet future policy needs. This involved keeping
an international context. On behalf of Alison up with academic work in the field and identifying
Young and his other colleagues, I welcome this new approaches which could be adapted to the
opportunity to provide a few reflections on his OECD surveys. Particularly, he maintained an
career at the OECD. interest in S&T output indicators throughout their
Early in 1985, just before Yvan died, the Com- period in the wilderness between the pioneer anal-
mittee for Scientific and Technological Policy rep- ysis in the OECD pilot study of 1961 data and the
resenting ministries and agencies from 24 OECD reawakened interest inspired by the National Sci-
countries was discussing possible areas of work by ence Foundation Science Indicators reports in the
the Secretariat in 1986. They gave highest priority late 1970s. Second, he always worked in close
to science and technology indicators ranking it cooperation with the experts in Member countries
ahead of all other projects. This was just one who are both the main providers of the data to the
example of how much the quality and policy rele- OECD and also amongst the most regular users.
vance of the work of the Unit which he had As a result the response rates to the OECD S&T
initiated and managed in the Directorate for Sci- surveys have always been high (20 out of 24
ence, Technology and Industry for over 20 years countries for the latest exercise). He had an im-
was appreciated in national capitals. peccable sense of timing for introducing new indi-
This work began when he joined the Organisa- cators on the international scene, knowing exactly
tion as a consultant in 1962 to organise the meet- when a workshop for academics would be useful,
ing held at Frascati in Italy to discuss the draft of when there would be sufficient interest amongst
the first OECD Manual for the collection of R&D policy makers to introduce a formal project and at
statistics with a group of national experts. Some of what point the topic was established enough to be
these, for example those from my own country, included in regular surveys. He exhibited all these
the United States, already had considerable expe- skills during the period of the introduction of
rience of R&D surveying whereas others were just output indicators into the OECD programme of
planning their first exercises. During the same work.
period he commissioned a pilot study to analyse Yvan also believed firmly that although consid-
the whole range of S&T Indicators then available erable attention must be given to checking the
for selected OECD countries and designed the technical aspects of the data Member countries
questionnaire for the first OECD survey of re- submitted to the OECD surveys, " the proof of the
sources devoted to R&D. In 1964 the first ad- pudding was in the eating", i.e. in their utility in
ditional staff member joined him together with a analytical reports. Such studies written by the
group of consultants from Member countries and staff of the Unit appeared intermittently in the
in the following year the Science Resources Unit 1960s and 1970s, and were followed by the bien-
was officially set up. nial OECD Science and Technology Indicators
I detail this first round of activity on R&D reports of which the second appeared shortly after
statistics because it presents a number of char- his death. It fulfilled his long felt wish to combine
acteristics of Yvan's approach to indicators work analysis of R&D data with other technological
indicators such as patent and technological bal-
ance of payments data and with production trade
Research Policy 16 (1987) 57-58
North-Holland and investment statistics.
0048-7333/87/$3.50 © 1987, Elsevier Science PubUshers B.V. (North-Holland)
4 /. Mar cum / Introductory note
When I joined the OECD as Director of DSTI, Directorate's work on science and technology in-
late in 1983, Yvan was a long-established member dicators with our industrial statistics and with
of the directorate's senior staff. As a new Director developments in the area of information, com-
I particularly welcomed the opportunity to work puter and communications data, in a new thrust to
with a Unit head with such a solid reputation in measure the contribution of science and technol-
national capitals and with such a high level of ogy to economic growth. His ideas on output
personal dedication. I was also happy to discover indicators will live on in the context of efforts to
that despite his many years in an international develop new measures of technological and in-
bureaucracy he was always open to new ideas. At dustrial performance. We all very much regret that
the time of his death we were working on another he did not live to see the raising of his Unit to
major extension of his responsibility, this time to Division status in 1986 and to oversee the chal-
apply his knowledge and skills to integrating the lenging new projects which it is now undertaking.
5
Is Western Europe losing
the technological race? *
Pari PATEL and Keith PAVITT
Science Policy Research Unit, University of Sussex, Brighton BNl 9RF, UK.
Using a variety of indicators, this paper compares techno- scientific publications and citations, and invest-
logical levels in Japan, the USA and W. Europe. There is no ment in plant and equipment. As such, it inevi-
justification for assuming general European technological
tably owes much to Yvan Fabian's contribution
backwardness. Whilst it lags in electronics, and since 1975 has
both on improving the range and quality of R&D
had a relatively slow rate of growth of innovative activities, it
also has sectors of strength (chemicals, machinery, production activities, and in pioneering the development and
engineering), and countries whose relative commitments of use of many other indicators of scientific and
resources to innovative activities are at least equal - and
technological activities.
sometimes superior - to those of Japan and the USA (FRG,
Partly as a result of his efforts, our understand-
Sweden, Switzerland). Its main policy challenges are to in-
crease the rate of growth of innovative activities in lagging ing of the economic importance of technology has
countries, and to ensure pervasive applications of information moved well beyond its association with an unde-
technology. fined, yet large, statistical residual amongst the
Similarly, there is no convincing evidence that W. Europe is factors contributing to economic growth. Its role
relatively backward in converting technology into economically
in the international competitive performance of
efficient innovations. Some policy problems are significant in
the high wage countries, in productivity growth,
all three regions (skill shortages), whilst others are in the USA
and W. Europe (short-term planning horizons of top manage- and in the generation of investment and employ-
ment). Particular European problems are the effects of social ment opportunities is much more clearly under-
legislation on worker flexibility and mobility, of national regu- stood [18,49,66]. Its importance is also more widely
lations on intra-European competition, and of slow rates of
recognised amongst private and public policy-
growth of output on the growth of innovative activities. Japan
makers and the general public. Leaders of firms
and the USA have their problems, with the latter continuously
under challenge from the former, and increasingly specialised and of national governments proclaim the impor-
in defence - and raw materials - related technologies. tance of continuous improvements and changes in
their products and processes for their competitive
survival and growth.
1. Purpose and background In such circumstances, international evalua-
tions of technological levels, particularly amongst
This paper makes systematic comparisons of Japan, the USA and W. Europe, have properly
the levels, trends and distribution of technological become part of political and analytical debate.
activities in Japan, the USA and Western Europe, The focus and conclusions of such evaluations
based mainly on data on R&D and patenting tend to be specific to each region. In Japan, em-
activities, but also on innovations and their diffu- phasis has been placed on national weakness in
sion, surveys of judgements of industrial peers, the development of basic science and technology
[63]. In the USA, there is both general concern
about the declining trade balance in so-called high
* The research for this paper has been undertaken as part of technology sectors, and specific fears of the strong
the programme, funded by the UK Economic and Social
Japanese challenge in manufacturing and electron-
Research Council, on Science, Technology and Energy policy.
ics technologies [9]. It is in W. Europe that the
We are grateful to C. Freeman and R. Nelson for helpful
comments on an earlier draft. most serious concerns have been expressed. A
specific weakness in electronics technology has
often been seen as the reflection of a more general
Research Policy 16 (1987) 59-85
North-Holland weakness in the development and exploitation of
0048-7333/87/$3.50 © 1987, Elsevier Science Pubüshers B.V. (North-Holland)
6 P. Patel and K. Pavitt / Is W. Europe losing the technological race?
new and unfamiliar technologies, and linked to technological leads and lags across countries and
the apparent inability of the W. European econo- regions is bound to be a complicated one, with
mies to create jobs and growth as successfully as considerable variations across sectors, types of
those of Japan and the USA. This state of affairs activity, and time.
has come to be called "Eurosclerosis". λ To sum up, thorough international comparisons
However, such a diagnosis should be treated of technological activity are desirable, because of
with some scepticism. Similar assumptions have the economic and political importance of technol-
made in the past about Western Europe and they ogy, because of the complexity of what is actually
turned out to be mostly and badly wrong: in happening, and because perceptions and policies
particular, that "The American Challenge" of the should be based on systematic information, rather
1960s, as epitomised in Servan-Schreiber's book of than on special pleading.
the same title, [64] would result in Europe's
technological dependence and industrial decline.
In the event, W. Europe as a whole improved its 2. Assumptions and scope
industrial performance compared to the USA, over
the subsequent ten years, whether in terms of Different conclusions from the same measured
output, production, exports, or of technology as indicators of scientific and technological activities
reflected in research and development (R&D) ex- can also result from different concepts and models
penditures [46]. What was to happen could have of the nature, sources and dynamics of technol-
been detected at the time, but diagnoses were ogy. Our definitions, concepts and measures are
often insufficiently detailed, and did not take suf- based on a wide range of empirical studies under-
ficient account of trends over time. Given the taken over the past twenty years [for recent
considerable improvements in statistical informa- summaries, see 17, 36, 44, 54, 55, 59]. We shall
tion since then, there is no excuse for similar make them explicit, and discuss their implications
faults in analysis today. for measurement and interpretation.
Variations in the conclusions emerging from
evaluation are not simply a consequence of inade- 2.1. The nature of technology
quate data. Thus, using virtually the same sources,
officials of the US National Science Foundation We define technology as knowledge that con-
and of the OECD have recently come to very tributes to the creation, production and improve-
different conclusions about the relative state of ment of economically and socially useful products
science and technology in the USA and W. Europe and services. Such knowledge thus relates not only
[4a,27]. Two factors might help explain such ap- to physical artefacts, but also to forms of organi-
parent inconsistencies of view. First, there are the sation for their production, distribution and use.
special interests that either consciously or uncon- Our analysis will concentrate on knowledge re-
sciously any analyses serve to support. The ap- lated to physical artefacts rather than to organisa-
parent behaviour and threats of foreigners can be tion, since it is easier to measure, and since it
used as a spur to national action, as a lever for develops in ways that are parallel and complemen-
more subsidy, protection or international co-oper- tary to organisational knowledge.
ation, or as a justification for greater deregulation, We depart clearly from the widely held assump-
liberalisation or flexibility. In such circumstances tion that technology is a form of "information",
the gap between the apparent and the real be- that has the properties of being costly to produce,
haviour, resources and efficiency of foreigners can but virtually costless to transfer and to use (and
often become quite large. Second, there is the fact, by implication, widely applicable in the first place).
to become apparent in this paper, that there are On the contrary, technological knowledge is often
now three regions of the world that are competing tacit (i.e. cannot be made fully explicit in the form
along the world technological frontier. Given the of instructions or codes of operation), and mostly
differentiated nature of technology, the picture of specific to firms and to particular classes of prod-
uct and production process. This knowledge is
1 For persuasive and influential examples of these views, see generated in large part not through "research",
[2,13,16,27]. activities, but through full-time or part-time in-
P. Patel and K. Pavitt / Is W. Europe losing the technological race? 7
novative activities, undertaken in firms to develop balance between "science push" and "demand
and improve specific products and production pull". Second, they ignore the very considerable
processes [4,24]. Depending on the type of tech- variation amongst sectors, products and technolo-
nology, industry, firm or innovation, such activi- gies in the rate, direction and determinants of
ties might be defined mainly as "Design", "Devel- technological change. Thus, whilst scientific re-
opment" or "Production Engineering". And with search has enabled radical innovations in chem-
the increasing efficiency and diffusion of informa- icals and electronics, its major function in most
tion technology, software is becoming an increas- other industries is to help provide trained techno-
ingly important locus of technology, emerging logical personnel, and to be one of the sources of
mainly from "systems groups" and "systems background information for technological prob-
houses". lem-solvers [20,56]. Similarly, whilst investment
Both technology acquisition and related in- activity is the driving force behind the induce-
novative activities are essentially cumulative ment and commercialisation of innovations in
processes. Given the firm-specific and differenti- steel-making and many other process industries, it
ated nature of products, processes and related is almost entirely a derived function of success in
technological knowledge, firms do not engage in product innovation (with a strong element of "sci-
comprehensive and complete search activities, but ence push") in sectors like pharmaceuticals [44].
explore technological and market zones contigu-
ous to their existing activities: what firms try to do 2.3. Implications for measurement and interpreta-
technologically in the future is strongly condi- tion
tioned by what they have been able to do in the
past. These characteristics of technology and its de-
terminants have a number of implications for our
2.2. Sources of technological change analysis. The first is that they help explain the
considerable variation amongst firms, in the same
Similarly, we reject the universal applicability national and international market environment, in
of two extreme models of the process of innova- the level and composition of their innovative ac-
tion and technological change. The first can be tivities. In part, these reflect the well known and
described as the "science push" or "linear" model, considerable ex ante uncertainties surrounding
where "R" lead to "D" and then to "innovation" innovative activities. They also reflect what Atkin-
(i.e. first commercialisation) and then to "diffu- son and Stiglitz pointed out some time ago [3]:
sion" amongst the potential population of users. It when technology is localised in firms and cumula-
is sometimes assumed in such models that "inven- tive in development, decisions about investments
tions" - as measured through patenting activity - in technology reflect both past patterns of innova-
are an intermediate "output" of R&D activities, tive activity, and expectations about the future,
and that the nature of an innovation and of re- both of which are firm-specific.
lated technology remain the same throughout the Second, given that technological change is
process of diffusion. firm-specific and cumulative, the relative techno-
The second model can be described as "de- logical strengths and weaknesses of firms and
mand pull", and assumes that the rate and direc- countries do not change rapidly over time. This
tion of technological change are inevitable and characteristic is given statistical confirmation in
determined by-products of other forms of eco- section 4.2 below. It means that technological
nomic activity: in particular, investment in plant patterns and trends in the recent past are a relia-
and equipment is assumed to be both the means ble guide to events in the not too distant future. It
through which inventions are commercialised, as also means that our statistical comparisons can be
well as the mechanism for inducing innovative seen as an attempt to measure and compare levels
activities upstream in related capical goods. and composition of " technological accumulation"
Whilst both these models do reflect observable or "technological capital" across regions and
dimensions of technological change, they have two countries.
major limitations. First, they ignore the impor- Third, the process of technological accumula-
tance in innovative activities of interaction and tion is not synonymous with the processes of
8 P. Patel and K. Pavitt / Is W. Europe losing the technological race?
either physical capital accumulation, or scientific (2) output and trade in "high technology" prod-
progress, even if it overlaps with both of them. uct groups: the definition of what are "high
Thus, although we include both in our compari- technology" product groups is very sensitive
sons, we concentrate our analysis on innovative to definitions of "high technology" (e.g. R&D
activities and technology. intensity or patent intensity; see [68]), and
Fourth, the clear distinctions in the "linear" neglect very important intersectoral flows of
model amongst invention, innovation and diffu- technology [53,60]. As a consequence, non-
sion do not reflect the more complex and interac- electrical machinery and automobiles are often
tive reality: "invention" is often induced in order excluded from "high technology" product
to solve problems of innovation, whilst diffusion groups, even though they have been central to
almost always involves further technology and the development and diffusion of electronics
innovation, if only to adapt the innovation to the based production technology [1], and are, as
specific skill, factor and product markets of adopt- we shall see, important sectors of W. Europe's
ing firms. Measured aspects of innovative activi- technological strength;
ties - such as R&D or patenting - are undertaken (3) indicators of productive efficiency (e.g. total
in relation to the invention, commercialisation, factor productivity) and of competitive ef-
diffusion and adaptation of innovations. Their ficiency (e.g. normalised shares of world ex-
attribution to any one part of the process has no ports); these indicators are only indirect mea-
empirical foundation, nor has the assumption that sures of technological capacities, since they
patents are an intermediate "output" of R&D reflect a number of other determining factors;
activities. Our own assumption will be that (4) statistics of R&D personnel and on national
innovative activities and technological accumula- receipts and payments for technology: al-
tion cover the whole spectrum of invention, though these are direct measures of innovative
innovation and diffusion, and that the various activity and technology, they have a number
statistics we use are proxy measures of such activi- of faults. International differences in defini-
ties. tion and measurement of R&D personnel
Finally, although considerable progress has make them unreliable for purposes of com-
been made over the past fifteen years in the mea- parison. Monetary receipts and payments for
surement of scientific and technological activities technology reflect only a small part of total
(see, in particular, the Science Indicators publica- national exports and imports of technology,
tions of the US National Science Board [71], and mainly because they do not measure technol-
the conferences organised by Y. Fabian and his ogy exports and imports embodied in goods,
colleagues in the Science and Technology Indica- nor those resulting from imitation through
tors Unit at the OECD [40]), there is no unique "reverse engineering" [32].
and superior measure of innovative activities and
Our comparisons of technology and innovative
technological levels. Each has its strength and
activities will be based on expenditures on R&D
weaknesses, and any analysis or comparison should
(section 3), patenting in the USA (section 4),
use a range of indicators adapted to its purpose.
specific innovations and their diffusion (section
Precisely what range of indicators should be
5), judgements by technological peers (section 6),
used in our analysis remains a matter of judge-
and scientific publications and citations (section
ment. We have decided that we should not use the
7). The particular strengths and weaknesses of
following four sets of indicators, even though they
each measure will be discussed in each section. In
have been used in similar comparisons by other
conclusion, we summarise the main patterns and
analysts:
trends emerging from the various comparisons,
(1) output and trade in electronics products: al- and discuss their implications for policy and for
though such products incorporate and reflect future analysis (section 8).
much technological change and innovative ac-
tivities, they account for about 30 percent of
3. R&D activities
the total at most, and often do not reflect the
technologies necessary for the application of As in other spheres of economic activity in the
electronics in other products and processes; twentieth century, the production of innovations,