The Location of Overseas Research and Development and Intellectual Property Protection

Abstract

Do multinationals distinguish between locations on the basis of the local intellectual property laws, in conducting their overseas research and development? We seek to address this question using data on US multinationals spanning the period 1977–2004. Particular care is taken to capture the host location enforcement environment of intellectual property protection. For our sample and time frame, we find that stronger intellectual property protection is not important in determining overseas research and development by these multinationals. The results are robust to disaggregation of the protection measure into its component indices, as well as to disaggregation of overseas research and development by industry. Instead, host country market size and human capital resources are found to be consistently important in explaining the location of overseas research and development.

Acknowledgements

My gratitude to Peter Berck, Julie Cullen, Gordon Dahl, Bronwyn Hall, Ted Miguel, Robert Staiger, Brian Wright, and seminar participants at the University of Goettingen, University of Hannover, University of Bonn, the IEA World Economic Congress at Istanbul, and the Conference of the European Policy for Intellectual Property at Maastricht.

Notes

 1. This is not to deny the importance of other factors such as reallocation of resources to more productive enterprises (Pavcnik, 2002; Hsieh and Klenow, 2008), but this may require policy-level innovation. While this may not accord with the traditional definition of innovation, it is nevertheless true that certain countries (for example, the United States) allow patenting of business methods.

 2. Things might get more complicated in a dynamic context, where a strengthening of IPR might have ‘second round effects’ on the proportions of direct investment, exports and technology licensing, which will have its own implications for spillovers.

 3. Even so, the implementation was staggered over several years, and concerning certain aspects of protecting innovations with respect to plants and animals, many countries opted for a sui generis form of protection that is considered weaker than patent protection.

 4. A related question is the distribution of rents from stronger protection. McCalman (2005) shows that although the distribution of benefits is likely to be skewed in favour of the developed countries, there is potential for all countries to benefit from the stronger protection.

 5. There is a substantial literature which studies overseas R&D of firms, but does not consider how that responds to variations in IPR (see Granstrand et al., 1993).

 6. The data available pertain to the overseas activity of the sum total of affiliates in a given country (see the data section).

 7. Membership of the North American Free Trade Agreement (NAFTA) is considered as well (Park, 2008); which is problematic, because it is unclear which trade agreements influence a country’s level of protection and which ones do not, and whether this effect is stronger for a country with seven trade agreements compared to one with only six trade agreements.

 8. The Area 2 sub-index considers the ‘legal structure and security of property rights’ across countries, on the basis of survey responses of ‘experts’ from those countries. Information on these responses has been collected from several sources such as the Global Competitiveness Report, World Bank data on ‘Doing Business’ and ‘Governance Indicators’, and the International Country Risk Guide (see Gwartney et al., 2008 for details).

 9. Given the current state of data across countries (in particular developing countries), it would be virtually impossible to augment the Ginarte-Park index to reflect additional enforcement and implementation dimensions, using objective data alone. For instance, in the case of preliminary injunctions, one would have to consider all relevant litigation in each sample country in each time period, to establish differences in the de facto implementation of this provision across countries! Further, given that litigation typically spans several years, that parties go in appeal against judgments, and that initial judgments (passed, say, within one’s sample period) may be overturned by those passed later (outside the sample period), all render the task of capturing the de facto enforcement dimension on the basis of objective criteria well-nigh impossible, even if the data were available.  The Area 2 indicator of Gwartney et al. (2008) presents a plausible alternative. It provides information on several aspects of the effectiveness of the legal system in general in each country. Though broader than a direct assessment of IPR enforcement, it does provide the best available quantitative indication of how the overall legal environment of the IPR system in each country operates. The correlation coefficient between the G-P index and the ‘Area 2’ sub-index is 0.62 for our sample, which is substantial, yet not so high as to make redundant either index – for a very high correlation would imply that the two indices do not provide independent information and, therefore, do not need to be used together (see Park and Lippoldt, 2008, for an alternative view).

10. A location may be used to serve more than one market, but detailed information on the jurisdiction of each ‘hub’ is unavailable.

11. We also tried host country GDP and GDP per capita to represent market size. Although the results were similar to those using the SALES variable, the latter were more consistent. Moreover, GDP and GDP per capita also reflect many things other than market size.

12. One would also like to account for other taxes such as withholding taxes on company profits, but these data are not available.

13. Firms may not enter a host-country market at all if IPR protection is weak, so that variation in the model may be biased by the implicit selection of those firms that do find it worthwhile to enter. However, the aggregative nature of our data prevents accounting for this effect.

14. Data on ICOV, IDUR, IMEM, IREV and IENF were made available by Walter Park, Department of Economics, American University.

15. If we use Ln IPGP and Ln A2MOD instead of the composite protection index Ln IPGPMOD, the results are the same as those reported in Table 2 – in particular, both Ln IPGP and Ln A2MOD turn out to be insignificant, while the market size, human capital and interaction terms remain significant.

16. Not only does this drastically reduce the degrees of freedom, it would also require the estimation of truncated regression models, for which fixed effects estimators are infeasible.

17. A finer disaggregation is not possible with our data, even though there is some evidence that in certain industries the multinationals limit the ‘core R&D’ to their home/advanced country markets, and conduct only the ‘non-core R&D’ in developing countries (Xue and Zheng, 2006; Quan and Chesbrough, 2010). It is possible that the effect of intellectual property regimes on overseas R&D may differ across these different types of overseas R&D.

18. Although R&D in chemicals and pharmaceuticals is considered the most responsive to intellectual property protection, it is conceivable that our results reflect the possibility that pharmaceutical firms restrict ‘core R&D’ to their home/advanced country markets.

19. Thus, if ICOV_t refers to 1975, ICOV_t-1 refers to 1970.

20. We do not lose observations on account of using IPGPMOD _t-1, because data on this variable were available before the starting year of our sample (1977).

21. To the extent that multinationals restrict ‘core R&D’ to home/advanced country markets in certain industries, for instance chemicals and pharmaceuticals, it is possible that stronger IPR in developing countries may not induce higher overseas R&D in these industries anyway.