The pattern of international patenting and technology diffusion

Abstract

The article focuses on the impact of R&D expenditure on labour productivity using international patent applications as a technology diffusion indicator. Considering the relationship between research and productivity, the pattern of international patenting reflects the channel between the source and the destination of transferred technology. Accounting for nonstationarity and cointegration, I find that patent-related foreign R&D spillovers are present for a panel of 18 OECD countries. Moreover, nonG7 OECD countries benefit more from foreign rather than domestic R&D activities. Estimates also show that there is no significant spillover effect from bilateral trade, but confirm the impact of FDI on domestic labour productivity.

Acknowledgements

This article was written in part at New York University/New York and CIDE Institution/Mexico City with special thanks to Jonathan Eaton and to David Mayer-Foulkes and Cermeño Bazán Rodolfo for helpful comments. I also would like to thank two anonymous referees for their remarks. Financial support from the German Science Foundation (DFG) as well as the German Academic Exchange Service (DAAD) is gratefully acknowledged.

Notes

¹ Table B4 in Appendix B presents data on foreign patent application filled by nonresidents from 18 OECD countries for 2001.

² Coe and Helpman (1995) assume a Cobb–Douglas functional form with constant returns and define TFP as output divided by input factors according to their elasticity. To keep the analysis comparable, I assume aggregated output produced by a single input factor.

³ The reader is referred to the article for further details and discussions.

⁴ The 18 OECD countries are respectively: Australia, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Netherlands, Norway, Spain, Sweden, United Kingdom and USA.

⁵ See Appendix A for an analytical derivation as well as Table A.1 for further information.

⁶ Given depreciation rates for capital stocks between 5 and 15% used in comparable studies, I also run regressions assuming a depreciation rate of 5%. As expected, the results do not change and the main conclusions remain valid. Table B.3 in Appendix B lists the results.

⁷ The OECD provides FDI flow- and stock data in their International Direct Investment Statistics. However, data on inward and outward positions are not complete for some countries or periods. I therefore prefer to use FDI flow data and to calculate stock variables.

⁸ See Table A.2 in Appendix A for further information. The R&D deflator is derived from BERD figures published in million current US$ (PPP) as well as in million constant US$ (PPP) in the Main Science and Technology Database by the OECD. PPP data is from the OECD Economic Outlook Database.

⁹ Baltagi (2001) provides an excellent overview for nonstationary issues as well as cointegration.

¹⁰ A GAUSS code for the estimation techniques is freely available on the homepage of Chihwa Kao at Syracuse University, NY: http://web.syr.edu/∼cdkao/

¹¹ To allow for a limited degree of dependence across units, cross sectional averages are subtracted from the observed data without affecting the limit distribution of the panel unit root test, see Levin, et al . (2002).

¹² For further information as well as for analytical derivation see Kao and Chiang (2000).

¹³ Both papers estimate the impact of domestic R&D amongst others variables on TFP. However, the impact of domestic R&D on either total or labour productivity should not vary largely. The coefficient is 0.097 for Coe and Helpman (1995) using pooled OLS. For Kao, Chiang and Chen (1999) the coefficient is 0.084 by the use of OLS with bias correction or FM-OLS and 0.107 by DOLS.

¹⁴ The 13 OECD countries are respectively: Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Netherlands, Spain, Sweden, United Kingdom and USA.