Macroeconomic and Schumpetarian Features of Japanese Innovations in the 1980s
MASARU YOSHITOMI
The purpose of this paper is twofold. One is to identify various channels whereby Japanese R&D activities and innovations contributed to manufacturing output growth in the 1980s. These channels include the relationships between R&D activities on the one hand and total factor productivity, business investment, quality and prices of products, and diversification of business activities by enterprises on the other. The importance of increased interindustry effects of R&D activities through input-output relations is stressed. The other purpose is to explain what accounts for the dynamic technological developments in Japan during the 1980s. Japan's Schumpetarian system appears to be responsible, but a more fundamental question is what makes it possible for Japan's economic system to be both dynamic in Schumpetarian innovation and consistent with Ricardian static comparative advantage.
TOTAL FACTOR PRODUCTIVITY, R&D AND INNOVATIONS
In the 1980s, an increase in total factor productivity contributed to the growth of real output by about 4 percent per year in Japanese manufacturing (see Table 1). This contribution was higher than in the 1970s and was comparable to that of the second half of the 1960s during the high-growth period.
Real output growth can exceed the weighted growth of capital and labor inputs. Weights are given by factor shares of output. The growth of output beyond that contributed by capital and labor inputs should be attributable to improvements in productivity of both capital and labor, that is, total factor
TABLE 1 Contributions of Factor Inputs to Output Growth in Japanese Manufacturing, 1970–1988
|
Annual Rates (%) |
|
|
1970–1980 |
1980–1988 |
Real gross domestic product (GDP) |
5.9 |
6.5 |
Capital inputs, increase rates |
7.1 |
6.0 |
Contribution to GDP |
2.4 |
1.9 |
Labor inputs, increase rates |
-0.1 |
0.8 |
Contribution to GDP |
0.0 |
0.5 |
Total factor productivity contribution to GDP |
3.5 |
4.1 |
SOURCE: Economic Planning Agency. |
productivity. Technically, the production function can be specified as a Cobb-Douglas function, with the coefficients of capital and labor constrained to be equal to their shares in output or income. 1 On this basis, total factor productivity can be estimated as residuals that cannot be explained by contributions of capital and labor input to output.
Industry comparisons in Japanese manufacturing suggest that the higher the R&D expenditure, the higher is the growth rate of total factor productivity. For instance, the electrical machinery industry, which registered the highest R&D expenditure per total wages and business investment among industries, enjoyed the highest growth rate of total factor productivity.
In manufacturing, business investment was strongly and favorably influenced by R&D expenditure. The elasticity of real business investment (divided by real total sales) with respect to real R&D expenditure (divided by real total sales) is found to be about 0.6 with the distributed lag of R&D over the present and preceding three quarters (see Table 2).
The stock of R&D capital rather than the flow of R&D expenditure should contribute to the growth of output as in the case of physical capital inputs. Furthermore, the net rather than gross stock of R&D capital is more meaningful, since amortization or depreciation of technology and knowledge is extremely rapid though it is difficult to measure. Since total factor productivity may be interpreted as a measure of the state of technological
TABLE 2 Estimates of Investment Function, Selected Manufacturing Industries Combined (log linear)
|
Independent Explanatory Variables |
|
|
|
|
Dependent Variable |
R&D per Sales |
Operating Rates |
Profit Rates |
Investment Rates |
R2 |
Investment per sale |
0.58 |
0.43 |
0.04 |
-0.03 |
0.91 |
(t values) |
(6.60) |
(2.08) |
(4.67) |
(-1.78) |
|
SOURCE: The Bank of Japan, Monthly Research, ''On the Efforts of R&D Activities in Recent Years,'' October 1990. |
progress that enhances the productivity of both capital and labor, its key determinant is the net stock of R&D capital. An International Monetary Fund study has found that in Japan the net stock of R&D capital rose at an annual rate of 9.25 percent in the 1980s, contributing nearly one percentage point to the overall real growth in gross national product (GNP) of about 4 percent per year in the decade.
It can be also observed from industrial comparisons that the higher the rate of increase in R&D expenditure, the higher is the rate of increase in labor productivity (output/labor), and hence the lower is the rate of increase in output prices.
Furthermore, the increased total factor productivity appeared to be profoundly associated with the quality improvement of products in the 1980s, since real GNP (which should reflect higher value added of output) in manufacturing grew faster than the production index (which simply reflects the volume of output) of manufacturing in the decade, while both figures grew more or less in parallel in the preceding decades.
In the 1960s and 1970s, Japan's manufacturing industry witnessed an increase in labor productivity (manufacturing gross domestic product (GDP/man-hour) that was accompanied by an increase in capital/output ratio (manufacturing GDP/real capital stock). In sharp contrast, however, the 1980s witnessed not only continued increases in labor productivity but also declines in capital/output ratio (i.e., increases in capital productivity; see Table 3).
Relative price changes between capital and labor promote substitution of capital for labor. In the 1980s, such substitution advanced, thanks to absolute stability or even absolute declines in prices of capital goods in the face of wage increases. This decline in the deflator of capital goods was particularly evident for electrical machinery and, to a lesser extent, general machinery, in contrast to construction machinery whose deflator actually
TABLE 3 Capital/Output Ratios by Industry in the 1980s
|
1980 |
1988 |
Total industry |
1,530 |
1,941 |
Manufacturing |
1,917 |
1,878 |
Nonmanufacturing |
1,354 |
1,977 |
SOURCE: Economic Planning Agency, National Accounts. |
rose. Declines in the prices of semiconductors and computers greatly contributed to these absolute price falls.
Thus, numerically controlled machinery nearly tripled in the 1980s. The number of installed industrial robots in Japan at the end of the 1980s was far greater than that of the United States and Germany (see Table 4). FMS (flexible manufacturing system) and CIM (computer-integrated manufacturing) have also been introduced in recent years.
These technological developments have made it possible for enterprises to produce much greater varieties with smaller volumes of each product in a shorter product cycle, to satisfy diversified individual demands of users and consumers in markets. The product cycle of capital goods has also shortened, and the machinery lease industry has grown rapidly.
Together with these developments, the aforementioned strong business investment has resulted, through greater addition of new investment to existing capital stock, in younger vintage of installed capital. The vintage of
TABLE 4 Industrial Robots in Operation, End of 1989
capital began to get older after the end of the high-growth period. In the second half of the 1980s, however, the vintage of capital stock peaked at just over nine years and started to decline somewhat in the manufacturing sector.
Computerization and information technologies have promoted greater business investment not only in manufacturing but also in nonmanufacturing industries such as finance, insurance, transportation, telecommunications, and distribution (wholesale and retail). In fact, the capital/labor ratio increased more rapidly in the nonmanufacturing sector than in manufacturing in the 1980s, contributing to an improvement in the low level of labor productivity in nonmanufacturing relative to manufacturing in Japan (see Table 3).
R&D also makes it possible for an enterprise in a declining industry to diversify into other industries and to compete with existing enterprises in the industries new to the entrant. In the 1980s, such proliferation of Japanese business participation in more promising industries was evident particularly for declining industries such as textiles, nonferrous metals, and iron and steel.
Total factor productivity in a given industry can be affected not only by its own R&D activity but also by the improvement of the quality of inputs into the industry. The latter is mainly the result of R&D activities of input-supplying industries. The fruits of input suppliers' R&D efforts, as embodied in intermediate and investment goods, can thus be captured by their users. This interindustry technology flow can be calculated by using input-output tables, as shown in Table 5.
TABLE 5 Total R&D Intensity, Direct and Indirect (%)
|
1975 |
1980 |
1985 |
1986 |
1987 |
1988 |
1989 |
Agriculture |
0.600 |
0.612 |
0.708 |
0.789 |
0.846 |
0.861 |
0.788 |
Mining |
0.712 |
0.687 |
0.705 |
1.020 |
1.348 |
1.339 |
1.404 |
Construction |
0.806 |
0.808 |
1.044 |
1.187 |
1.234 |
1.272 |
1.313 |
Foods |
0.830 |
0.841 |
1.043 |
1.136 |
1.286 |
1.261 |
1.317 |
Textiles |
0.955 |
1.096 |
1.429 |
1.355 |
1.834 |
1.671 |
1.832 |
Papers and pulps |
0.885 |
0.840 |
1.312 |
1.307 |
1.413 |
1.521 |
1.595 |
Chemicals |
1.928 |
3.718 |
5.343 |
6.190 |
6.466 |
6.940 |
7.130 |
Petroleum and coal |
0.766 |
0.795 |
0.828 |
1.272 |
1.540 |
1.662 |
1.673 |
Ceramics |
1.200 |
1.292 |
1.996 |
2.583 |
2.621 |
2.488 |
2.691 |
Iron and steel |
1.493 |
2.147 |
3.692 |
4.141 |
4.170 |
4.272 |
4.362 |
Nonferrous metals |
|
1.584 |
2.789 |
3.139 |
2.815 |
3.464 |
2.662 |
Metals |
1.108 |
1.480 |
2.188 |
2.324 |
2.193 |
2.271 |
2.255 |
General machinery |
1.854 |
2.680 |
3.121 |
3.628 |
3.814 |
3.509 |
3.822 |
Electrical machinery |
2.157 |
4.565 |
6.431 |
7.134 |
7.297 |
6.995 |
7.572 |
Transport machinery |
1.616 |
3.444 |
5.225 |
5.593 |
5.674 |
5.976 |
6.130 |
Precision machinery |
1.512 |
3.176 |
5.337 |
5.311 |
5.870 |
6.318 |
5.913 |
SOURCE: Keizai Kikakucho (Economic Planning Agency), Heisei San Nendo Keizai Hakusho (Economic White Paper 1991), (Tokyo: Okurasho insatsu kyoku, 1991). |
The direct input of R&D into a given industry increased in all industries except agriculture from 1980 to 1988, particularly in the four machinery sectors and in chemicals. More interestingly in terms of indirect inputs of R&D into a given industry from all other industries, every industry benefited more from the rapid technology advance of other industries. In particular, the transport and the iron and steel industries benefited most from such indirect technology advances. In contrast, the electrical and precision industries enjoyed relatively fewer benefits from other industries, since they received proportionately less inflow of embodied R&D. These two industries have the highest total direct and indirect intensity of R&D, but mainly thanks to their own high input of R&D (Table 5).
SCHUMPETARIAN FEATURES OF JAPAN'S INNOVATIONS
All in all, these dynamic technological developments contributed greatly to overcoming the recessionary effects of the drastic appreciation of the yen and to upgrading industrial and trade structures in Japan in the 1980s. What accounts for such dynamic developments? Japan's "Schumpetarian system" was responsible.
There is a widespread notion that Japan's success in industrial development cannot be understood in the context of traditional economic notions of comparative advantage. It is often claimed that Japan has "created" comparative advantages for strategic industries. A distinction is made between Ricardian or allocative efficiency on the one hand and dynamic or Schumpetarian efficiency on the other. It is asserted that there can be a real conflict between short-term Ricardian efficiency (specializing in the production of, say, textiles and black and white televisions) and long-term dynamic efficiency (say, specializing in high-income elasticity products such as color televisions and word processors). Accordingly, it is claimed that Schumpetarian, not Ricardian, efficiency clearly determined economic policymaking in Japan and that this distinction has made Japan's industrial development successful.
The fundamental question imposed by this argument is whether there is any conflict between the static efficiency of resource allocation based on comparative advantage (which would require the elimination of monopolies) on the one hand and Schumpetarian innovation and the resultant dynamic evolution of comparative advantage on the other.
There are two important aspects to Schumpetarian innovation. One is that technical change is not an accidental by-product of "residuals" of economic activities, but the result of deliberate efforts on the part of enterprises through R&D competition and organizational reform. The other is that in the basic Schumpetarian framework, such innovation or new technical and organizational knowledge is at least temporarily appropriable by allowing
innovative firms to establish monopoly positions. Over time, however, new technologies become public goods through imitation by rivals. Thus, the incentive for innovation depends on the expectation of the innovator that he will be rewarded with such temporary profits.
Technological advance has interesting "dual" features, namely, (1) it possesses the characteristics of a public good (i.e., use by one firm does not preclude its use by another) and (2) it is largely provided by private firms that do R&D. In the neoclassical production function with technological change, output can increase more than proportionately with increases in capital and labor. This extra output growth is attributed to total factor productivity, as noted above. In this production function, however, total factor productivity or the technological advance is treated as if it were a pure public good. If that were the case, any technological advantages could not be compensated for in the market. Innovators who succeed in applying new technology to commercial production should be rewarded by being able to appropriate extra profits, since innovation is not an unintended side effect of other activities. If firms succeed in innovation in terms not only of new production processes and new products but also of organizational and managerial reforms, they will be rewarded by extra profits thanks to achieving higher productivity, lowering production costs, securing higher quality, and meeting customers' new demands for differentiated products. However, the Schumpetarian process of innovation does not stop here. Intense competition among rivals inevitably leads to spillovers from innovation. Eventually, the technical advance and organizational changes that have contributed to the innovation will become public goods, thereby enhancing the overall economic well-being.
This Schumpetarian world is, therefore, an extremely competitive one, preventing firms from reaping any permanent monopoly profits on the basis of innovation. This dynamic competition results in the equally dynamic development of national resource endowments themselves in the form of increasing abundance of R&D and skilled labor inputs per unit of output, relative to other national resources. For this reason, the Schumpetarian dynamic evolution of comparative advantage is not at all inconsistent with the Heckshire-Ohlin trade theory, once one admits the dynamic and endogenous creation of national resource endowments themselves through deliberate policies at both enterprise and government levels.
In general, growth and Schumpetarian technology efficiency cannot be obtained by totally ignoring Ricardian comparative advantage. Comparative advantages are bound to evolve naturally as an economy accumulates capital and skills. In other words, the economic determinants of national comparative advantage naturally evolve, as the relative abundance or scarcity among endowed production resources (land, raw materials, labor, capital, skills, and R&D) dynamically changes through economic development.
As demonstrated by various studies, Japan has registered the dynamic changes in comparative advantage from unskilled labor-intensive to capital-intensive and further to R&D-intensive manufactured products.
Japan's innovation experience suggests that intense competition among a relatively large number of enterprises in a given industry not only encourages Schumpetarian innovations by rewarding innovators with above-average profits but also expedites spillovers of the innovation through competition and imitation among rivals. The dynamic evolution of comparative advantage through Schumpetarian innovations supported by R&D activities of private enterprises must be consistent with the static comparative advantage at a given point in time. This is because the relative endowments of domestic resources dynamically evolve over time through increasing inputs of R&D and physical capital relative to labor and natural resources. At a given time the domestic resource endowments should determine comparative advantage through efficient resource allocation due to intense competition.