II. RECENT RESEARCH:
INVESTMENTS IN THE CREATION AND DIFFUSION OF KNOWLEDGE

AN INTERNATIONAL COMPARISON

Bert Minne^*

_

Abstract This paper provides data on the expenditures on the creation and diffusion of knowledge in ten OECD countries. It draws on the main themes of the neoclassical theories of endogenous growth: creation and diffusion of know-how and education. The paper argues in favor of a broad concept of knowledge. According to a definition based on data availability, the ten OECD countries spend about ten percent of their national income on their knowledge base. This figure, however, is likely to considerably underestimate these expenditures. Samenvatting Dit artikel geeft de uitgaven aan de creatie en verspreiding van kennis in tien OESO landen. Het is gebaseerd op de hoofdthema's van de neo-klassieke endogene groeitheorie‰n: productie en verspreiding van kennis en onderwijs. Dit artikel pleit voor een ruim begrip "kennis." Op basis van een definitie die gebaseerd is op data-beschikbaarheid geven de landen ongeveer tien procent van hun nationaal inkomen uit aan hun kennisbasis. De werkelijke uitgaven zijn waarschijnlijk aanzienlijk hoger.

Introduction

Knowledge and human capital are increasingly important factors of production. Hence, stocks of knowledge and human capital should be added to the stocks of tangible assets and labor resources as determinants of a nation's productive capacity. The national accounts register spending on know-how and human capital as intermediate inputs or consumption rather than as investment; this contrasts with the treatment of investments in tangible assets. Expenditures on research and development (R&D) are sometimes treated differently from other intangible assets. For instance, many years ago Kendrick (1972) already pointed to the need to incorporate R&D in investment. Moreover, Griliches (1958) estimated the impact of the R&D stock on output. Recently attempts have been made to link R&D expenses as satellites to the national accounts of the US (US Department of Commerce) and the Netherlands (Statistics Netherlands).

This paper argues that R&D is only one of the many types of investment in knowledge. Which knowledge concept should be adopted? Which available expenditure indicators fit this concept? How much do OECD countries invest in their knowledge-base? What are the main open questions for research in this area? This paper addresses these questions. It starts with a brief discussion of the theory of endogenous growth, which provides a solid foundation for measurement. Indeed, this theory emphasizes the important roles of the stocks of knowledge and human capital and the diffusion of know-how in generating economic growth. However, this theory fails to provide an empirical definition of knowledge. It is here that the main contribution of this paper lies.

Three themes of endogenous growth theory

The theory of endogenous growth provides three themes for the knowledge-based economy: investments in new knowledge, investments in human capital, and intermediate expenditures aimed at the diffusion of know-how.

Knowledge industry
Romer (1990) and Grossman and Helpman (1991) developed models describing a knowledge-producing industry which generates economic growth by producing new ideas. Each new idea requires a fixed amount of intangible investment (such as R&D). This amount declines in the course of time because the knowledge-producing industry can exploit an ever-increasing stock of know-how. The knowledge industry is human capital intensive and employs the existing stock of knowledge as an input. The size of the knowledge industry, its skill intensity and its efficiency determine the price of new ideas. Cheaper new ideas raise macroeconomic productivity and consumer welfare due to greater choice and better quality.

Human capital
The second theme involves expenditures to increase the stock of human capital. Human capital formation can be regarded as a distinct education industry (Mino, 1996, and Rebelo, 1991). This industry stimulates technological progress because spending on education increases the stock of human capital, thereby improving the economy's capacity to produce new ideas and a country's comparative advantage in skill-intensive production. Moreover, the education industry uses human capital as an input. Indeed, its expenses consist largely of teachers' wages.

Knowledge diffusion
Complementary expenditures aimed at gaining access to and diffusing the stock of knowledge is the third theme. Spillovers from the existing stock of knowledge are crucial inputs in the production of new knowledge. In this respect, public and firm-specific know-how can be distinguished. Grossman and Helpman (1991) and Jones (1995) assume that knowledge is public and hence freely accessible. In this case, permanent growth is feasible if the elasticity of new ideas with respect to public know-how equals one. In contrast, if know-how is firm-specific (Smulders and van de Klundert, 1995), only employees of an enterprise can exploit its spin-offs: the barriers to gain entry to other firms' specific know-how are infinite. In that case, the production of new ideas will be concentrated in large corporations because only these corporations can exploit the economies of scale from the indivisible stock of their own know-how and experience. In fact, both knowledge as a pure public and as a pure private good are extremes. Exploiting knowledge is generally costly, but these costs are not infinite. This paper examines these costs empirically.

Empirical data for the Netherlands

Knowledge industry
Spending on new knowledge should encompass all expenditures needed to launch new products or processes. Accordingly, the knowledge industry covers more than business enterprise R&D, which is limited to exact and natural sciences conducted in the business laboratories of manufacturing firms. Registering only business R&D thus neglects the creativity, know-how and experience in the rest of the economy, including the service sector and the government. This paper therefore broadens knowledge production beyond R&D by firms.

Also the production of know-how at universities, research institutes, software houses and marketing departments belongs to the knowledge industry. Governments largely finance university research, and research institutes carry out research for governments as well as firms. Application of software facilitates the production of new ideas and the supply of products and services that better fits individual tastes of the customers (e.g. by means of CAD/CAM software). Thus, the purchase of standard- and tailor-made software delivered by software houses is also an investment expenditure in knowledge production.

Marketing belongs to the knowledge industry as well. This, however, may be controversial. To illustrate, Wolff (1987) regards marketing as an unproductive activity since its output "performs a circulation function" (p. 42). Nevertheless, marketing can be classified as part of the knowledge industry for three reasons. First, marketing is required to inform potential customers about new products. Second, marketing departments make R&D more effective by providing information about consumer tastes to the development and design centers. Third, marketing may improve the probability of a successful product launch, thereby lowering the costs of market entry.

The knowledge industry needs also tangible investments. Scientific instruments and laboratories are included in R&D expenditures. Moreover, computer hardware is required to generate new ideas quickly and efficiently.

Which data on knowledge production are available for the Netherlands? According to the empirical definition in table 1, the Netherlands invested almost 30 billion guilders (5,1% of GDP) on intangibles in 1992 . Business enterprise R&D (NLG 5.5 billion) was only a fraction of that amount. On top of intangible investments, the Netherlands spent NLG 5 billion (0,9% of GDP) on investments in computer hardware.

_
Table 1 Expenditures on the knowledge base in the Netherlands (billions of guilders), 1992
Knowledge industry R&D by Business Enterprises 5.5 R&D by Universities 2.7 R&D by Research Institutes 2.0 Software 5.3 Marketing 11.8 ________+ Total Knowledge Industry 28.6 Human Capital Public primay/secundary education 19.6 Public tertiary education 5.7 Private education 2.0 Business enterprise education 3.4 ________+ Total Human Capital 33.5 Knowledge Diffusion Foreign patents 2.5 ________+ Total Intangible Investment 63.4 Total Tangible Investment 114.3 ________+ TOTAL INVESTMENT 177.7

Manufacturing industries differ considerably in their R&D and marketing intensity. The electronic-, chemical- and specialized machine industries face the best technological opportunities in the exact and natural sciences. Consequently their R&D intensities are high (Table 2). The electronics industry is a prime developer of original software embodied in new products. In the Netherlands, almost a third of all product innovation is regarded as software (CBS, 1995b), concentrated in the electronics industry. Philips Research claims to spend55% of its R&D in developing new software.

_
Table 2 Top 10 manufacturing industries in R&D, the Netherlands, 1993a
R&D share in gross production in percent 1 Electronics/ computers/ sientific instruments 10.5 2 Basic chemicals 6.5 3 Textile machines 3.3 4 Soap, detergents 2.7 5 Paper machines 2.7 6 Agricultural machines 2.3 7 Coaches and trailers 2.1 8 Metal working machinery 1.8 PM Aircraft NA PM Pharmaceutical NA a These R&D data cover only the operating expenses of R&D and R&D which is contracted out. Therefore, the wages of R&D personnel and investment in laboratories and scientific equipment are excluded. These expenditures are not available at this low level of aggregation. Total R&D expenditures are approximated by multiplying the operating expenses bij 1/0.4 where 0.4 is the share of operating costs in total business enterprise R&D according to: CBS, Speur- en ontwikkelingswerk in Nederland [Research and Development Work in the Netherlands]. Source: CBS, Produktiestatistieken [Production Statistics].

Table 3 contains data on sectoral advertising intensity. The food industries, including the beverage-, tobacco- and snack industries, spend a lot of resources on advertising. These industries face stagnating sales in terms of liters or kilograms. Nevertheless, higher value (and greater welfare) is achieved through more extensive brand choice. Also the detergent industry is a big advertiser. Only the detergent industry and the CD-video industry invest a lot in both R&D and advertising.

_
Table 3 Top 15 manufacturing industries in advertising, the Netherlands, 1993
Advertising share in gross production percent 1 Beer 11.4 2 Tobacco 10.9 3 Soap, detergents 10.9 4 Spirits, wine 8.0 5 Publishers 5.3 6 Food (other) 5.1 7 Coffee, tea 5.1 8 Vegetable- and food processing 4.8 9 Chocolate 4.1 10 Snacks 3.6 11 CDs, video-film 3.4 12 Potato chips 3.3 13 Paint 2.9 14 Ornaments 2.8 15 Sports equipment 2.7 Source: CBS, Produktiestatistieken [Production Statistics]. Not listed due to confidentiality: margarine, pet food, soft drinks

The empirical definition of intangible investment employed in Table 2 still considerably underestimates efforts needed to bring new products and ideas to the market. In particular, this definition does not take into account expenditures on e.g. test production, engineering, almost all product development in the services industries and commercial application of social sciences and humanities, such as management styles and organization of logistic systems.

Human capital
Education can be measured relatively easily. Indeed, OECD countries have measured their expenditures on public education for quite some time. Moreover, enterprises spend on education (see table 1). The Netherlands invested NLG 33.5 billion (5,9% of GDP) in education in 1992. Of this, public education accounts for by far the largest part.

Education spending underestimates spending on human capital. In particular, companies pay for training on the job to adopt existing technologies. These costs are probably rather high (Jovanovic, 1995). Moreover, opportunity costs of learning are not taken into account: the students could have provided labor instead of going to college. Accordingly, the education data presented underestimate the true costs of education.

Knowledge diffusion
Economic theory typically assumes that diffusion of know-how does not absorb scarce resources because it assumes that knowledge is either a public good, which is freely accessible, or a firm-specific private good, which cannot be transferred to outsiders. In fact, knowledge diffusion is possible but requires scarce resources. Important examples are

payments on patents;

costs of the institutions stimulating the exchange of ideas between universities and enterprises or the diffusion of know-how to small firms;

organizational costs needed to make a R&D alliance a success;

the media, which diffuse know-how among their readers, listeners and TV-watchers; and

expenditures on telecommunication equipment.
For the Netherlands, data are available on the payments on foreign patents. Unfortunately, various sources provide inconsistent amounts. National statistics show an amount of NLG 2.5 billion, while the OECD produces NLG 10.8 billion (see table 1, note). The Netherlands invests NLG 2.9 billion in telecommunication equipment. The data included in Table 1 reveals that in 1992 at least 63 billion guilders (11.2% of GDP) was spent on intangible investment. In addition, the Netherlands spent 8 billion guilders on tangible investments directly related to its knowledge base. Hence, total expenditures on the knowledge base amount to at least half of total private and public tangible investment (which amounted to NLG 114 billion in 1992). This indicates that the expenditures on the knowledge base are important from a macroeconomic point of view.

International comparison

This section compares spending on knowledge among ten industrial countries in the period between 1985 and 1992.1 Relative spending provides information on the compara-tive advantages of these countries in knowledge-based production.

In order to facilitate international comparison, we left out expenses on private education and education by enterprises, because most countries lack the relevant data. For the same reason, we limited marketing expenditures to advertising spending. In order to obtain the best results, we employ data gathered by the OECD.

Table 4 shows that countries spend between 8 and 11% of their GDP on intangible investment defined as the expenditures on public education, R&D, licenses, software and advertising. Sweden invested most in intangible assets, while Austria and Germany invested the least. Some countries experienced a declining intangible investment to GDP ratio between 1985 and 1992, particularly Germany and Belgium. One can observe a geographical shift in Europe in favor of the Nordic countries.

_
Table 4 Total intangible expenditures on the knowledge base in several industrial countries, 1985 and 1992a
1985 1992 in percent of GDP Sweden 10.1 11.0 Norway 8.6 10.8 US 9.6 10.2 Netherlands 10.0 9.6 Belgium 10.1 9.4 Denmark 8.9 9.4 France 8.9 9.2 UK 9.3 8.8 Austria 8.2 8.5 Germany 8.9 8.0 Source: CBS and CPB (1995), Tables 4 and 13 a Total expenditures on public education, R&D, licenses, software and advertising.

Table 5 reveals that expenditures on public education dominate expenditures in knowledge diffusion. On public education, Norway is first and Sweden in second place. Germany performs better than the data suggest, because it pours large amounts of funds into its business apprentice system, which is not counted as public education.

_
Table 5 Expenditures on knowledge diffusion and public education, 1992
KNOWLEDGE DIFFUSION PUBLIC EDUCATION Payements on foreign patents Telecommunication (tangible) in percent of GDP Sweden 0.05 0.50 6.45 Norway 0.15 0.50 7.30 US 0.10 0.40 5.15 Netherlands 0.45b 0.50 5.45 Belgium 1.20 0.35 5.25 Denmark 0.10a 0.30 5.95 France 0.20 0.45 5.20 UK 0.25 0.35 4.60 Austria 0.20 0.70 5.45 Germany 0.50 0.90 3.40 Source: CBS and CPB (1995) a 1990 b See note Table 1. This figure is based on the national CBS-statistics.

Sweden and the US take the lead in the knowledge industry,(see Table 6). The US is the largest spender, ranking first in business enterprise R&D, advertising and spending on computer hardware. The Anglo-Saxon countries (US, UK) are forerunners in advertising, and the Netherlands in software.

_
Table 6 Investment in the knowledge industry, 1992
R&D Firms R&D Universities and Research Institutes Advertising Softwarea Computersa in percent of GDP Sweden 2.05 0.95 0.70 1.05 0.70 Norway 1.05 0.90 0.75 0.85 0.60 US 2.05 0.75 1.25 1.00 1.00 Netherlands 0.95 0.90 0.90 1.10 0.75 Belgium 1.10b 0.55b 0.60 0.75 0.65 Denmark 1.00b 0.70b 0.80 0.85 0.80 France 1.50 0.90 0.70 0.80 0.60 UK 1.45 0.75 1.10 0.75 0.75 Austria 0.90 0.65 0.75 0.60 0.55 Germany 1.70 0.80 0.90 0.65 0.60 Source: CBS and CPB (1995) a 1993 b 1991

With regard to investment in knowledge diffusion, Belgium invests heavily in foreign patents. The position of the Netherlands depends on the data source: it ranks third according to national statistics. The OECD database, however, suggests that the Netherlands outperforms even Belgium. Germany is first in telecommunication, probably due to heavy infrastructure investments linked to the re-unification.

Northwestern Europe (Germany, the Netherlands, Belgium and Great Britain) spends less on their knowledge base than the Nordic countries and the US do. This does not necessarily imply that the prospects for economic growth are worse in these regions, because of two reasons. First, the applied empirical definition does not include all spending on the knowledge-based econo-my. Second, nominal expenditures do not provide information on the efficiency of education and knowledge production and diffusion.

Future research

This paper argues in favor of a broad concept of knowledge. All know-how that can enhance productivity and welfare should be taken into account. Data gathering should thus focus on a broad field. However, education and the creation and diffusion of know-how are not well defined from an empirical point of view. An international agreement on the definition of the knowledge industry and the types of spending related to knowledge diffusion would be welcome in order to assess the prospects of economic growth and model and test relationships in the knowledge-based economy. Furthermore, deflators should be developed in order to separate nominal development into a volume and a price component.

References

• CBS, 1995a, Research and development work in the Netherlands 1993 (in Dutch: Speur- en Ontwikkelingswerk in Nederland 1993), Voorburg.
• CBS, 1995b, R&D and software research at companies in the Netherlands (in Dutch: R&D en software-onderzoek bij bedrijven in Nederland), Voorburg.
• CBS and CPB, 1995, Intangible investments in the Netherlands: determination of an international position (in Dutch: Immateri‰le investeringen in Nederland: een internationale positiebepaling), Voorburg.
• Griliches Z., 1958, "Research costs and social returns," Journal of Political Economy, pp. 419-431.
• Grossman G.M. and E. Helpman, 1991, Innovation and growth in the global economy, MIT Press.
• Jones C.I., 1995, "R&D-based models of economic growth," Journal of Political Economy, August, pp. 759-783.
• Jovanovic B, 1995, "Research, schooling, training and learning by doing in the theory of growth," Paper presented at Econometric Society meetings, Tokyo, August 1995.
• Kendrick J.W., 1972, "The treatment of intangible resources as capital," Review of Income and Wealth, pp. 109-125.
• Mino K., 1996, "Analysis of a two-sector model of endogenous growth with capital income taxation," International Economic Review, February, pp. 227-251.
• OECD, 1995, Basic science and technology statistics, 1995 edition, Paris.
• Rebelo S.T., 1991, "Long run policy analysis and long run growth," Journal of Political Economy, 99, pp. 500-521.
• Romer P.M., 1990, "Endogenous technological change," Journal of Political Economy, 98, pp. S71-S102.
• Smulders S. and Th. van de Klundert, 1995, "Imperfect competition, concentration and growth with firm-specific R&D," European Economic Review, 39, pp. 139-160.
• US Department of Commerce, 1994, "A satellite account for research and development," Survey of Current Business, November, pp. 37-71.
• Wolff E, 1987, "Growth, accumulation, and unproductive activity," Cambridge University Press.

Notes

*. tel. 31-70-3383368;e-mail:b.minne@cpb.nl.
1. See CBS and CPB (1995) for the yearly data in 1985-1992.

| Top |