Technology spillover effects and economic integration: evidence from integrating EU countries

Abstract

The article uses time series for the period 1981–2008 to estimate the impact of foreign technology spillover effects on Greece, Ireland, Portugal and Spain, representing the integrating European Union (EU) countries. I restrict technology diffusion to EU-12 countries and compare the results to unrestricted technology diffusion from a sample of 32 OECD countries. Accounting for nonstationarity and co-integration, the dynamic OLS estimator is used to estimate the impact of foreign R&D stock on labour productivity, taking into account patent-, trade- and FDI-related technology diffusion channels. I find empirical evidence for trade-related foreign technology spillover effects for Greece and Ireland if technology diffusion is unrestricted. Restricting technology diffusion to EU-12 countries, there are significant foreign technology spillover effects from European integration for Portugal (patent related) and Spain (trade and FDI related). Moreover, the domestic R&D stock and education are significant drivers for labour productivity in integrating EU countries. The empirical results are robust for different regression specifications and sources of technology diffusion.

Keywords:

• technology diffusion
• economic integration
• nonstationary time series
• co-integration
• dynamic OLS

JEL Classification:

• C22
• O30
• R12

Acknowledgements

This article has benefited greatly from the comments of JörnKleinert. I am grateful to the seminar participants from the Faculty of Economics at the University of Valencia and the University of Castellón for helpful suggestions. I am, of course, responsible for any errors.

Supplemental Data

Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/00036846.2014.920479

Notes

¹ Using real GDP data from the Groningen Growth and Development Centre (GGDC, 2010), for example, annual growth rates in 1981–2008 vary from 2.46% (Portugal) to 5.14% (Ireland).

² Coe and Helpman (1995) assume a Cobb–Douglas functional form with constant returns and define TFP as output divided by the elasticity weighted input factors labour and capital. However, TFP figures are susceptible to calculation and measurement errors and estimated coefficients might be less reliable due to inherent biases.

³ There are insufficient data for Chile and Estonia, which reduce the number of current OECD countries from 34 to 32.

⁴ Detailed information is given for the R&D capital stock in Table A.1 and capital stock in Table A.2 in the Appendix section.

⁵ The data set supporting the results of this article is available as Supplemental data.

⁶ I refrain from analysing model specifications in which the null hypothesis of co-integration was rejected by Table A.5 in the Appendix section. Moreover, I include dummies to patent-related foreign R&D capital stocks to account for structural breaks in the patent data in order to estimate their spillover effects. The regression equation includes unreported country-specific constants.

⁷ Alternatively, I also run regressions using patent-, trade- and FDI-related foreign R&D capital stocks from EU-15 countries (i.e. EU-12 countries, Austria, Finland and Sweden) for the same period 1981–2008. The empirical results are almost the same.