International R&D spillovers revisited: Human capital as an absorptive capacity for foreign technology

Abstract

This paper examines aspects of R&D spillovers across countries, in particular, the role of international trade and human capital as the catalysts for international diffusion of technology. We present a new way of measuring foreign R&D stocks embodied in foreign intermediate goods and capital equipment, which we argue is free from the criticism of previous measures. With the pooled panel data spanning 1970 through 1995 for 103 countries, we find that the effects of foreign R&D on total factor productivity growth of both industrial countries and developing countries are substantial and that human capital is the most influential channel for absorbing foreign R&D spillovers.

Keywords:

• R&D spillovers
• human capital
• international technology diffusion

Acknowledgments

The authors thank Bill Neilson and seminar participants at the Bank of Korea, Ewha Women's University, Korea University, Sogang University, the Korean International Economic Association Conference, the Korean Econometrics Society Meetings, the International Economics Seminar, and the two anonymous referees for helpful comments and suggestions. In addition, Mikyoung Kim is thanked for research assistance.

Notes

¹See Borensztein et al. (1998) for a framework incorporating the role of FDI by multinational firms as a determinant of economic growth, and see Easterly et al. (1994) for a model of technology adoption through international trade and human capital accumulation.

²New growth models based on a technology-driven set-up, in particular, by Grossman & Helpman (1991a), Rivera-Batiz & Romer (1991), and Aghion & Howitt (1992), have brought about new interest in the beneficial interactions of international trade and technology diffusion for economic growth in the open economies. However, some other studies such as Devereux & Lapham (1994) and Feenstra (1996) point out that trade might result in a harmful effect on growth due to differences in initial conditions.

³See Mankiw et al. (1992) and Hamilton & Monteagudo (1998) for empirical analysis of the determinants of economic growth.

⁴It would be controversial that the secondary school enrollment rate might reflect flow (or change) of human capital rather than human capital stock. We consider two measures of human capital, the secondary school enrollment rate and the average years of schooling, finding a significant role of human capital as a determinant of TFP.

⁵The ratio of total import to GDP may not reflect the real import level of R&D-related intermediate goods from industrial countries. For example, when Japan's import share increased during the oil shock periods, it is hard to believe that the intensity of R&D spillovers for Japan increased as well.

⁶Coe et al. (1997) find that some variables are non-stationary with no cointegrating relation among the I(1) variables but the differenced data are stationary with the panel unit root tests of Levin & Lin (1993).

⁷See Coe & Helpman (1995) for a detailed explanation.

⁸The data for the capital income shares are from OECD Economic Outlook. For Austria, Denmark, Netherlands, and New Zealand, however, the capital income share data are not available for the full sample period. Thus, we use the average of the capital shares over the available period for these countries.

⁹Lichtenberg & van Pottelsberghe de la Potterie (1998) show that, for instance, with a potential merger of the 11 EC countries, under Coe & Helpman's computation scheme, the US and Japan's foreign R&D stocks increase by 108% and 25%, respectively. See Table 2 later.

¹⁰The potential shortcoming of this measure would be that the measure depends on the size of the recipient country. When a small country imports a small fraction of the exports of an industrial country, the amount of foreign R&D stocks of the small country from the industrial country is small. Another point is that the imports of manufactures are a better measure than the total imports for the channels of international R&D spillovers, a fact that is not examined here.

¹¹Heteroskedasticity often takes place in the analysis of cross-section data. In our panel data analysis, heteroskedasticity might arise due to the differences in TFP levels across countries. We can easily expect to observe greater variation in the TFP of large countries than in the TFP of small countries. The differences in variations of TFP might also be attributed to the degree of industrialization, levels of R&D stocks, and social-political situations.

¹²However, the openness measured by the fraction of imports from industrial countries to GDP did not show a significant relationship with the TFP growth, which is not reported.

¹³An anonymous referee kindly pointed out this fact.

¹⁴The figures in Table 5 may be misleading since the direct effect of the R&D investment on the domestic TFP growth is excluded.