measure and should rather be able to capture the change in returns between periods. The probability of observing a decision to let the patent expire is monotonically decreasing with each additional citation which is in accordance withHypothesis 8. Furthermore, the probability is clearly lower for patents which have successfully survived opposition and increasing with the number of inventors.
Corporations, especially small ones, are less likely to let their patents expire compared to non-profit organizations and universities as well as individual inventors. The latter often lack the means to appropriate value from the patented inventions. Small corporations usually rely more heavily on the ability to protect their business interests through patents compared to large corporations. Patents in discrete technologies are also less likely to lapse compared to patents in complex technologies. Furthermore, applicants with larger patent portfolios seem to be less willing to let their patents expire. The effect remains if we control for patent value and increases if we additionally control for market characteristics (see Appendix 1.7.3). This can be ascribed to the effect that the private value of a patent within a patent portfolio is often higher than its stand alone value. Expiration is also less likely if the patent protects a core technology. The importance of the technology within the applicant’s patent portfolio has a significant negative effect on the probability to let the patent expire prematurely. IPC4_SHARE has a negative and significant effect for European patent applications but not for patents granted by the DPMA. Technological competition significantly reduces the probability to observe an expiration decision once we control for the technology’s importance to its owner.
states of the European Patent Organization.
Using data on German patents we show that for almost 6% of all patents willingness to license is declared. For patents granted by the German Patent and Trademark Office (DPMA) the number goes up to even more than 11%. The decision to declare LOR is significantly affected not only by the characteristics of the patent but also by the characteristics of its owner and the technological area it belongs to. The likelihood to observe a declaration is considerably larger in complex technical areas such as electrical and mechanical engineering compared to discrete ones. Moreover, patents which are owned by large corporations are more likely to be endorsed LOR. They can use the provision to reduce the costs of their patent portfolios or to supplement cross-licensing agreements. Once we control for the patent portfolio size, patents owned by individual inventors become equally likely to be endorsed LOR. Somewhat surprisingly, for patents filed by non-profit organizations and universities the probability is significantly reduced.
Moreover, it is less likely to observe a declaration if the patent belongs to the applicant’s core technology or if the level of competition in the market is high.
Within our theoretical framework we show that the effects of returns from patent pro-tection and the age of the patent on the probability to declare the willingness to license heavily depend on the motive for the declaration. If pure cost-saving considerations are relevant, the probability increases with the returns from patent protection and decreases with a patent’s maturity. If licensing (commitment) considerations prevail, the effects are completely reversed. However, we are not able to determine empirically what motive is the relevant one in each period of a patent’s life. Using static measures for correlates of patent value, like the number of received patent citations within a given period, the patent family size and survival of an opposition procedure, we show that patents of presumably higher total value have a lower probability to be endorsed LOR. However, if we use a dynamic measure for the returns from patent protection, which should be better capable of capturing changes in returns from patent protection, the effect on probability vanishes.
For younger patents the probability is increasing with a patent’s age and decreasing for older ones. However, for presumably very valuable patents, for which the (licensing) commitment motive should prevail in all periods, the probability is almost monotonically increasing in a patent’s age.
Although the willingness to license is declared for a significant number of patents in Germany, we lack evidence on whether the inventions are indeed taken up and exploited by third parties. Data on licensing agreements is rarely available. Nevertheless, the prevalent opinion by practitioners and also researchers is that the market for technology licensing is underdeveloped (see Gambardella et al. 2007). This might be due to a lack of awareness but also due to the nature of the German technology market. According
to the practitioners we have interviewed, firms usually do not actively search for existing technology but try to develop the technology themselves. As a consequence of this, licensing agreements are often only the outcome of litigation settlements. Thus, as already proposed by Schovsbo (2009) and Gowers (2006), in order to promote licensing of patents endorsed LOR it is important to create a desirable environment and provide support for active use of the system. A first step would be to increase the visibility of the patents endorsed LOR by creating an innovation pool, which would enable third parties to easily identify what patents relevant to their area of R&D are available for licensing. Here, Singapore is a leading example. The Singapore government has established a special web base for all patents endorsed license of right which anyone can easily access. A second step would be to introduce a model license agreement provided by the patent office to further reduce the transaction costs involved in the conclusion of licensing contracts.
The German patent system incentivizes applicants to declare the willingness to license and open up their inventions by reducing their maintenance fees for patents. Shovsbo (2009) argues that this mechanism may entail substantial social costs. Patentees may use the reduction in maintenance fees to apply for patents of little value or “artificially”
extend patent protection. The first objection is easily invalidated since in almost all patent systems the maintenance fees, especially in the first 5 years, are marginal compared to the costs that have to be involved to get a patent granted. The possible reduction in maintenance fees should not play a major role for the decision to apply for a patent and request a costly examination. The second argument, however, may be important for the general welfare assessment of the LOR system. From the theoretical model we were able to determine two motives for the declaration, the cost-saving and the commitment motive. The first one is unambiguously welfare decreasing, since without the LOR option the patentee would have let his patent lapse. The commitment motive, however, can be welfare increasing, since the patentee voluntarily eliminates exclusivity from an otherwise valuable patent. Hence, the welfare assessment will heavily depend on the ratio of the two motives. Unfortunately, we do not have a reliable measure for the returns from patent protection in a given period, which would allow us to categorize the declarations.
However, the problem can potentially be solved using structural estimation techniques by using the theoretical model developed in this paper and applying a simulated GMM estimator. A comparison of simulation results of one patent system with and one without the LOR option should then allow for explicit welfare statements.