The Competitiveness of Nations
in a Global Knowledge-Based Economy
Harry Hillman Chartrand
April 2002
Organization for Economic Co-Operation and Development
THE
KNOWLEDGE-BASED ECONOMY
|
INDEX SUMMARY I. T HE KNOWLEDGE-BASED ECONOMY:TRENDS AND IMPLICATIONS A. Introduction B. Knowledge and economics C. Knowledge codification D. Knowledge and learning E. Knowledge networks F. Knowledge and employment G. Government policies II. T HE ROLE OF THE SCIENCE SYSTEM INTHE A. Introduction B. Knowledge production C. Knowledge transmission D. Knowledge transfer E. Government policies Web 3 III. I NDICATORS FOR THE KNOWLEDGE-BASED ECONOMYD. Measuring knowledge stocks and flows E. Measuring knowledge outputs F. Measuring knowledge networks G. Measuring knowledge and learning H. Conclusion References |
29
III. INDICATORS FOR THE KNOWLEDGE-BASED
ECONOMY
Economic indicators are
measures that summarise at a glance how an economic system is performing. Since
their development in the 1930s, and particularly after World War II, the
national accounts and measures such as Gross Domestic Product (GDP) have been
the standard economic indicators of the OECD countries. Based on detailed
censuses that survey economic activity at the establishment level, they measure
broad aggregates such as total production, investment, consumption and
employment and their rates of change. These traditional indicators guide the
policy decisions of governments and those of a broad range of economic actors,
including firms, consumers and workers. But to the extent that the
knowledge-based economy works differently from traditional economic theory,
current indicators may fail to capture fundamental aspects of economic
performance and lead to misinformed economic
policies.
The traditional economic
indicators have never been completely satisfactory, mostly because they fail
to recognise economic performance beyond the aggregate value of goods and
services. Feminists challenge the concept of GDP because it fails to take into
account household work. Environmentalists maintain that traditional indicators
ignore the costs of growing pollution, the destruction of the ozone layer and
the depletion of natural resource endowments. Social critics point out
divergence between traditionally measured economic performance and other facets
of human welfare. In response to these criticisms, work is proceeding on
extending censuses to include a set of household activities, such as cleaning,
food preparation and child care. Attempts are being made to “green” the
national accounts through indicators which track depletion of forests and
minerals, and air and water pollution. Novel indicators have also been proposed
to measure social welfare more directly, taking into account crime rates,
low-income housing, infant mortality, disease and
nutrition.
Measuring the
performance of the knowledge-based economy may pose a greater challenge. There are systematic
obstacles to the creation of intellectual capital accounts to parallel the
accounts of conventional fixed capital. At the heart of the knowledge-based
economy, knowledge itself is particularly hard to quantify and also to price. We
have today only very indirect and partial indicators of growth in the knowledge
base itself. An unknown proportion of knowledge is implicit, uncodified and
stored only in the minds of individuals. Terrain such as knowledge stocks and
flows, knowledge distribution and the relation between knowledge creation and
economic performance is still virtually unmapped.
The methodology for
measuring GDP and most other macroeconomic indicators is specified by the United
Nations System of National Accounts, which are structured around input-output
tables that map intersectoral transactions. In the national accounts framework,
the gross output of each establishment is measured by its market value and
summed across sectors and/or regions. Net output by sector or region is obtained
by subtracting out intermediate purchases. National GDP is the sum
of
30
net outputs across sectors
and regions. To the extent that input-output proportions are stable, this
double-entry framework translates input statistics into output indicators. Thus
employment, strictly speaking an input, can also be interpreted as an indirect
indicator of the level of national output.
In the knowledge-based
economy, problems emerge with the conceptual framework of the national accounts.
Not least is the issue of subsuming knowledge creation into a measurement system
designed for traditional goods and services. The pace of change complicates the
task of measuring aggregate output and raises questions about the use of input
measures as output indicators. Factors which are not sufficiently incorporated
into the national accounts framework include qualitative changes in products,
the costs of change and rapid product obsolescence.
Knowledge is not a
traditional economic input like steel or labour. When traditional inputs are
added to the stock of economic resources, the economy grows according to
traditional production function “recipes”. For example,
more labour can increase GDP by an amount that depends on current labour
productivity, or more steel can increase production of autos, housing or tools
by predictable amounts according to the current state of the arts. New
knowledge, in contrast with steel or labour, affects economic performance by
changing the “recipes” themselves – it provides product and process
options that were previously unavailable.
While new knowledge will
generally increase the economy's potential output, the quantity and quality of
its impact are not known in advance. There is no production function, no
input-output “recipe” that tells, even approximately, the effect of a
“unit” of knowledge on economic performance.
Knowledge, unlike
conventional capital goods, has no fixed capacity. Depending on
entrepreneurship, competition and other economic circumstances, a given new idea
can spark enormous change, modest change or no change at all. Increased
resources devoted to knowledge creation are likely to augment economic
potential, but little is known as to how or how much. Thus the relationship
between inputs, knowledge and subsequent outputs are hard to summarise in a
standard production function for knowledge.
It is also difficult to
stabilise the price of knowledge by the trial and error discipline of repeated
transactions in the market. There are no company knowledge records nor census of
knowledge creation or exchange. In the absence of knowledge markets, there is a
lack of the systematic price information that is required to combine individual
knowledge transactions into broader aggregates comparable to traditional
economic statistics. In knowledge exchanges, a purchaser has to gauge the value
of new information without knowing exactly what it is he is to buy. New
knowledge creation is not necessarily a net addition to the economically
relevant knowledge stock, since it may render old knowledge
obsolete.
There are thus four
principal reasons why knowledge indicators, however carefully constructed,
cannot approximate the systematic comprehensiveness of traditional economic
indicators:
à there are no stable
formulae or “recipes” for translating inputs into knowledge creation
into outputs of
knowledge;
à inputs into knowledge
creation are hard to map because there are no knowledge
accounts analogous to the
traditional national accounts;
à knowledge lacks a
systematic price system that would serve as a basis for aggregating pieces of
knowledge that are essentially unique;
31
à new knowledge creation is
not necessarily a net addition to the stock of knowledge, and obsolescence of
units of the knowledge stock is not documented.
The problem of developing
new indicators is itself an indication of the unique character of the
knowledge-based economy. Were we faced with trivial modifications to the
traditional accounting system, a few add-on measures might suffice. To fully
understand the workings of the knowledge-based economy, new economic concepts
and measures are required which track phenomena beyond conventional market
transactions. In general, improved indicators for the knowledge-based economy
are needed for the following tasks:
à measuring knowledge
inputs;
à measuring knowledge
stocks and flows;
à measuring knowledge
outputs;
à measuring knowledge
networks; and
à measuring knowledge and
learning.
Students of the
knowledge-based economy have to date focused on new knowledge formation or
knowledge inputs. The principal knowledge indicators, as collected and
standardised by the OECD, are: i) expenditures on research and
development (R&D); ii) employment of engineers and technical
personnel; iii) patents; and iv) international balances of
payments for technology (Figure 4). Some of these activities are classified by
sponsorship or source of funding (government and industry) and by sector of
performance (government, industry, academia). Major emphasis has been placed on
the input measures of R&D expenditures and human resources. Despite
significant advances in recent years, these traditional indicators still have a
number of shortcomings with respect to mapping the knowledge-based
economy.
32
Indicators of R&D
expenditures show direct efforts to enlarge the knowledge base and inputs
into the search for knowledge. Indicators relating to research personnel
approximate the amount of problem solving involved in knowledge production. But
only a small fraction of all inputs into knowledge creation are attributable to
formal R&D expenditures and official research
personnel.
Successful R&D draws on
ideas from many different sources, including informal professional exchanges,
users' experiences and suggestions from the shop floor. In addition, current
indicators count formal R&D conducted by the public sector, academia and
large manufacturing firms, and tend to understate research expenditures by small
firms and service-sector enterprises. As data collection improves, the
importance of the services sector to R&D and innovation is only now being
fully recognised.
Patents, since they represent ideas themselves, are the closest
to direct indicators of knowledge formation; of all the traditional knowledge
indicators, patents most directly measure knowledge outputs (rather than
inputs). Patent data have certain advantages in that most countries have
national patent systems organised on centralised databases, the data cover
almost all technological fields, and patent documents contain a large amount of
information concerning the invention, technology, inventor, etc. There are
several ways to analyse patent data, including categorising patents by
geographic area and industrial product group. However, differences in national
patenting systems introduce bias which make comparisons difficult. In general,
not all new applications of knowledge are patented and not all patents are
equally significant. Patents also represent practical applications of specific
ideas rather than more general concepts or advances in
knowledge.
The technology balance
of payments measures international movements of technical knowledge through
payments of licensing fees and other direct “purchases” of knowledge, and
thus is more appropriately a flow measure than an input measure. But there is no
claim that the technology balance of payments measures the full flow of
technical knowledge between any two countries. International transfers of
knowledge through employment of foreign personnel, consulting services, foreign
direct investment or intra-firm transfers are important avenues of diffusion
that are not factored into these indicators. International joint ventures and
co-operative research agreements are also instrumental in the global diffusion
of knowledge.
D. Measuring knowledge
stocks and flows
In order to improve the
measurement of the evolution and performance of the knowledge-based economy,
indicators are needed of the stocks and flows of knowledge. It is much easier to
measure inputs into the production of knowledge than the stock itself and
related movements. In the case of traditional economic indicators, the
transmission of goods and services from one individual or organisation to
another generally involves payment of money, which provides a “tracer”.
Knowledge flows often don't involve money at all, so that alternative
“markers” must be developed to trace the development and diffusion of
knowledge.
Measuring the stock of
physical capital available to an economy is an awesome task, so that measuring
the stock of knowledge capital would seem almost impossible. Yet
measuring knowledge stocks could be based on current science and technology
indicators if techniques were developed for dealing with obsolescence. For
example, annual R&D inputs could be accumulated for various countries and
industries and then amortised using assumptions concerning depreciation rates.
In this way, measures of R&D stock relative to production have been used to
estimate rates of return to R&D investment. Similarly, stocks of R&D
personnel could be estimated based on annual increases in researchers in
particular fields, depreciated by data on personnel movements and
occupational
33
mobility. The patent stock
might be approximated using data on use and expiration of periods of exclusive
rights.
A more difficult challenge
is measuring the flows of knowledge, or the proportion of knowledge stock
which enters into the economy during some time period. Two proxy indicators are
most frequently used to measure knowledge flows: i) embodied diffusion,
or the introduction into production processes of machinery, equipment and
components that incorporate new technology; and ii) disembodied
diffusion, or the transmission of knowledge, technical expertise or technology
in the form of patents, licences or know-how.
Overall flows of
embodied knowledge, particularly embodied technology or R&D, can be
measured using input-output techniques. Technology flow matrices have been
constructed as indicators of inter-industry flows of R&D embodied in
intermediate and capital goods. This methodology allows separation of the
equipment-embodied technology used by a particular industry into the technology
generated by the industry itself and the technology acquired through purchases.
In this way, estimates can be made of the proportions of R&D stock which
flow to other industries and the extent to which industries are sources of
embodied knowledge inputs (Table 6). Analysis of embodied technology diffusion
shows that inter-sectoral flows vary by country. Countries also differ in the
amount of embodied technology acquired from abroad vs. that purchased
domestically (Sakurai et al., 1996).
Micro-level analyses of
embodied knowledge flows focus on the diffusion and use of specific technologies
in different sectors of the economy – an area of analysis which needs more
standardisation across countries in order to allow international comparisons.
Studies attempting to compare the diffusion of microelectronics in OECD
countries have encountered severe statistical problems in defining the
technologies, gathering data on use and calculating the share of total
investment (Vickery, 1987). Existing comparative data are sketchy; they show
generally that Japan and Sweden have the most widespread use of advanced
manufacturing technologies (AMT), followed by Germany and Italy who have
profited from AMT in their motor vehicle and
mechanical
34
engineering sectors.
Industry in the
More is known about
technology diffusion patterns in individual countries. Canadian surveys, for
example, have asked manufacturing firms about their use of 22 advanced
manufacturing technologies, including computer-aided design and engineering
(CAD/CAE), computer integrated manufacturing (CIM), flexible manufacturing
systems, robotics, automated inspection equipment and artificial intelligence
systems. Approximately 48 per cent of Canadian firms use these technologies,
mostly in the area of inspection and communications. The attempt to relate
technology use to performance showed that technology-using firms tended to have
higher labour productivity and to pay higher wages than non-users (Baldwin et
al., 1995).
Information technology
indicators are being developed which focus on the diffusion and use of
information technologies – computers, software, networks – by businesses and
households. These measures of technology flows, and factors facilitating and
impeding such flows, such as pricing, give an indication of the rapid growth of
the information society. For example, the OECD is compiling indicators of the
number of personal computers, CD-ROMs, fax machines and modems per household in
the OECD countries. Data show that the use of personal computers has more than
doubled in the last decade, with about 37 per cent of
The knowledge-based economy
is an interactive economy at both the national and international levels as
illustrated by emerging indicators of computer and communications network
infrastructure. Such measures show the ratio of households and businesses with
outside computer linkages, cable connections and satellite services. More work
is needed on indicators by country and region of the development of the
Internet, the world-wide web of computer networks; these include host
penetration, network connections, leased line business access, dial-up services
and price baskets.
Growth in the number of
computers hooked to the Internet has been phenomenal – from 1 000 in 1984 to 100
000 in 1989 to over 4.8 million in 1995. It is estimated that the number of
Internet users (as to official host connections) exceeded 30 million in 1995
(OECD, 1995b).
Flows of disembodied
knowledge are most often measured
through citation analysis. In scholarly journals and patent applications, it is
the practice that users of knowledge and ideas cite
their
35
sources. This makes it
possible to map the interconnections among ideas in specialised areas. For
example, the Science Citation Index provides a database for exploring
inter- and intra-disciplinary flows of knowledge in the realm of basic research.
Attempts have been made to map the interdependence of scientific ideas using a
citation index (Small and Garfield, 1985; Leontief,
1993).
In the future, computer
capabilities may make it possible to scan and analyse enormous volumes of text,
flagging complex similarities and differences and enabling us to identify
knowledge flows beyond the areas where formal citation is practised. Others have
traced the linkages among areas of applied technical knowledge through patent
citations, which are considered carriers of the R&D performed in the
originating industry (Table 8).
Based on a concordance of
Similarly, improved data on
international patent citations can help track technology flows on a global basis
as could further refinements of technology balance of payments measures. But
while the amount of knowledge subject to formal citation requirements includes
the entire content of scientific literature and all patented ideas, these areas
are only a limited part of the modern economy's knowledge
base.
E. Measuring knowledge
outputs
The standard
R&D-related measures do not necessarily show successful implementation or
the amount and quality of outputs. Nevertheless, these input and flow indicators
form the starting point for measuring knowledge outputs and for gauging social
and private rates of return to knowledge investments. Rough indicators have been
developed which translate certain knowledge inputs into knowledge outputs in
order to describe and compare the economic performance of countries. These
measures tend to categorise industrial sectors or parts of the workforce as more
or less intensive in R&D, knowledge or information. The measures are based
on the assumption that certain knowledge-intensive sectors play a key role in
the long-run performance of countries by producing spill-over benefits,
providing high-skill and high-wage employment and generating higher returns to
capital and labour.
36
For example, the OECD
maintains a classification of high-technology, medium-technology and
low-technology manufacturing sectors based on their relative R&D
expenditures or R&D intensity (ratio of R&D expenditures to gross
output). Computers, communications, semiconductors, pharmaceuticals and
aerospace are among the high-technology and high-growth OECD sectors and are
estimated to account for about 20 per cent of manufacturing production. Output,
employment and trade profiles can be drawn for countries, based on the relative
role of their high-, medium- and low-technology sectors. However, current
indicators of R&D intensity are now confined to manufacturing sectors and
have not been developed for the fast-growing service portion of OECD economies.
Nor do these indicators take into account R&D which may be purchased from
other industrial sectors, either embodied in new equipment and inputs or
disembodied in the form of patents and licences. More complete indicators of
total R&D intensity, including both direct R&D efforts and acquired
R&D, need to be developed (Table 9).
37
In a similar vein, early
studies in the
A related methodological
approach is to use employment and occupational data to categorise jobs according
to their R&D, knowledge or information content. One early study used
occupational classifications to assign jobs an informational component;
information workers included those in the primary information sector, a large
portion of the public bureaucracy and a few in remaining
sectors.
According to this study,
information activities accounted for 47 per cent of GNP in the
Occupational data has been
used to estimate the proportion of economic effort devoted to creating,
implementing and administering change. One study finds a variation among sectors
in the proportion of non-production workers in total employment, ranging from as
high as 85 per cent in sectors normally seen as high-technology to 20 per cent
or less in slower-growth, more traditional industries (Carter, 1994). There
appears to be a close connection between the proportion of nonproduction workers
and the rate of change in a sector; the major function of non-production workers
may be to create or react to change. In these sectors, more workers are engaged
in the direct search for new products and processes, in implementing new
technology on the shop floor and in opening new markets and reshaping
organisations to accommodate changes in production. As a result, a growing
proportion of costs are most likely the costs of change rather than the costs of
production.
Indicators are needed which
go beyond measuring R&D and knowledge intensity to
assessing
social and private rates
of return (Table 10). Rates of return
are generally estimated by computing the benefits (including discounted future
benefits) vs. the costs of innovation. For example, early studies of the
agricultural sector showed that public research was undervalued and that private
investment did not naturally respond to the prospect of large returns to
scientific research. One analysis estimated that social returns of 700 per cent
had been realised from US$2 million in public and private investments in the
development of hybrid corn from 1910-55 (Griliches, 1958). In another, the
median private return to the innovations studied was 25 per cent, while the
median social rate of return was 56 per cent (Mansfield et al., 1977). A
recent review of macro-level econometric studies of the
38
The importance of both
innovation and technology for productivity growth and long-term economic growth
is poorly understood; indicators are needed which capture the impacts of
technological progress on the economy and employment. Measuring rates of return
to R&D may be particularly challenging in the services sector where
productivity is especially difficult to measure.
Regression analysis can be
used to estimate the returns to R&D in terms of total factor productivity
growth. This is being attempted for both the manufacturing and services sectors
and for performed and acquired (or embodied) R&D. On average, across ten
OECD countries, the estimated rate of return of embodied R&D in terms of
manufacturing productivity growth has been estimated at 15 per cent and in the
services sector at over 100 per cent in the 1980s, illustrating the importance
of technology diffusion (Sakurai et al.,
1996).
Indicators are also being
developed of rates of return to R&D expenditures and acquisitions at the
firm- or micro-level. In one study, the top R&D executives of major American
firms were polled about the proportion of the firm's new products and processes
that could not have been developed (without substantial delay) in the absence of
academic research (Table 11). Extrapolating the results from this survey to the
academic research investment and returns from new products and processes, a
social rate of return of 28 per cent was calculated
(
39
40