Abstract
This article investigates the determinants of technology licensing, focusing on how country-specific characteristics affect technology holders’ incentives to sell their proprietary technologies through licensing alliances. An empirical examination of licensing is done using a unique panel data set of licensing transactions involving companies in the EU. The strength of Intellectual Property Rights (IPR) protection, the degree of economic freedom, the level of country risk, and the number of patent granted in the country are found to be important determinants of inter-firm technology licensing. In addition, firms with prior independent experience as a licensor and public companies tend to license technology more.
Acknowledgements
We thank Professor Nicholas Vonortas and the Center for International Science and Technology Policy (CISTP) at The George Washington University for their research support. All remaining errors are our responsibility.
Notes
1 One license per entrant. We assume that once licensors commit the number of licenses, all other firms can observe it.
2 .Therefore, the second order condition is satisfied.
3 < 0.
4 These include technology licensing, joint venture, joint marketing, manufacturing, R&D, Original Equipment Manufacturer (OEM) and supply agreements.
5 We assume that determinants of cross-border strategic alliances may be qualitatively different from those of domestic alliances.
6 This study ignores licensing deals among individual investors, nonprofit organizations, governments and universities.
7 We dropped companies from three EU countries including Estonia, Latvia and Slovenia because these countries do not provide all the necessary information for our econometric analyses.
8 Sum of exclusive and nonexclusive licenses. It includes cross licenses.
9 We also looked at random effects Poisson and negative binomial models. The results from the Poisson model were very similar to the results for the Logit model in terms of signs and statistical significance. In particular, the coefficients on the IPR variables were positive and statistically significant. The random effects negative binomial model, however, failed to converge. The negative binomial random effects model was introduced by Hausman et al. (Citation1984). For further discussion, see Cameron and Trivedi (Citation1986), Winkelmann and Zimmermann (Citation1995), and Cincer (Citation1997).
10 As with the random effects model, we also looked at fixed effects Poisson and negative binomial models. The results from the Poisson model were also very similar to the results for the Logit model in terms of signs and statistical significance. In particular, the coefficients on the IPR variables were positive and statistically significant. As with the random effects negative binomial model, however, the fixed effect negative binomial model failed to converge (i.e. the likelihood function become nonconcave).
11 Evidence shows that government policies on intellectual property rights protection have large effects on foreign technology spillovers in countries (Xu and Chiang, Citation2005).