انقلاب فناوری اطلاعات و رشد بهره وری در ژاپن

The objective of this paper is to examine the contribution of IT investment to productivity growth in Japan. To this end, we first construct a panel data set of value added, labor inputs, IT investment, IT capital stock, and non-IT capital stock in 22 industries (including 13 manufacturing industries) from 1980–2000. We then decompose labor productivity into four terms and find that the recent slowdown in productivity is caused by the absence of a reallocation effect of labor. Further, we estimate TFP growth functions and labor productivity functions, examining network externalities of IT capital. We find that TFP growth in the 1990s is affected by positive inter-industry spillover effects from the supply side. J. Japanese Int. Economies18 (3) (2004) 362–389.

IT investment is expected to play a key role in the revival of the Japanese economy. However, due to the lack of data on IT capital by industry, there are few academic papers which have examined the contribution of IT capital stock to Japanese economic growth and, in particular, the spillover effects of IT capital. Constructing such data in terms of value added, labor, non-IT capital, and IT capital in twenty-two Japanese industries, we examined both the contribution of IT capital to economic growth and the spillover effects. Our findings can be summarized as follows. (1) IT investment in Japan increased rapidly at a rate of 14.8% per year from 1975 onwards and reached 22 trillion yen in 2000. The ratio of IT investment to total investment reached 25% in 2000. (2) In the 1990s, the accumulation of IT capital in the non-manufacturing sector (banking and insurance, transportation and telecommunications, etc.) was higher than that in the manufacturing sector. (3) The ratio of IT capital to total capital stock was higher in the US than in Japan in many industries. (4) Following Sonobe and Otsuka (2001), we decomposed labor productivity growth into intra-sectoral capital deepening, efficiency effects of capital deepening, efficiency effects of labor shifts, and intra-sectoral TFP growth. The decomposition shows that the recent slowdown in labor productivity growth was caused by the decrease in efficiency effects of the labor shift. (5) Using the panel data set, we estimated the spillover effects of IT capital. Using input-output relations, we divided inter-industry spillover effects into two types by transmission route: the first type is transmitted from the input side, whereby productivity in a specific industry increases through the purchase of intermediate goods; and the second type is transmitted from the output side, whereby the productivity in a specific industry is affected through the demand for products of the industry. As a result of the estimations, we find positive inter-industry spillover effects. The effects are stronger and more significant in the 1990s than in the 1980s. The results of our estimations for the 1990s show that positive spillover effects from demand side are also significant. Our study explains well many features of the Japanese economy of recent years. The weak performance in growth implies that the negative reallocation effects in the labor market offset TFP growth in the late 1990s. Our estimation results imply that TFP growth is affected by the spillover effects from the supply side and the direct effects from the demand side. These findings suggest that in order to achieve a revival of the Japanese economy, policies promoting labor reallocation and stimulating IT investment in the upstream industries, in addition to conventional demand side policies need to be implemented. However, three major points remain. First, our data on IT stock by industry does not include software stock, because we constructed the data set based on 68SNA. If we used software stock data by industry based on 93SNA, we might find clearer inter-industry spillover effects. Second, our study does not consider the recent rapid decrease in communications costs and the effects of the investment tax credit scheme which started in fiscal 2003 year. To examine these effects, we need to estimate an IT investment function. These tasks are left for our future research.