Sunday, January 10, 2010

An Anlysis of IP Licensing: The Empirical Challenge

Research on IP licensing tends to be of two types--either anecdotal or empirical. The strength of each is also its weakness. Anecdotal research, while it drills down into the specifics of a given licensing situation, lacks the ability to support broader generalisation; empirical research, while it provides a broader overview of licensing, usually lacks any guidepost for the practical implications of the results.

With that in mind, I finally had an opportunity to review a study published last year as a Working Paper of the OECD Directorate for Science, Technology and Innovation (DSTI/DOC(2009)5)). Entitled "Who Licenses Out Patents and Why? Lessons from a Business Survey," the study was carried out by Maria Pluvia Zuniga and Dominque Guellec. As described in the Abstract, "[t]he goal was to investigate the intensity of licensing to affiliated and non-affiliated companies, its evolution, the characteristics, motivations and obstacles met by companies doing or willing to license." To this end, responses were studied from 600 European, and 1700 Japanese, corporate patent holders.

The first point is the unique nature of technology transactions. The authors mention a number of factors that make IP licensing a challenging activity: (i) markets for technologies are seen as less efficient economically than product markets; (ii) the incorporeal nature of IP rights, whether registered and disclosed, as with patents, or tacit, as with trade secrets, makes both valuation and deal structure uncertain; (iii) a license arrangement requires the licensor to find a satisfactory way to share the fruits of its innovation with a third party, with both commercial and psychological implications; (iv) the mixed nature of an IP license, having both proprietary and contractual aspects, makes the structure and drafting a license agreement a non-trival and often times costly activity; (v) the licensor runs the risk that the licensee may become a competitor, lawfully or otherwise; and (vi) the characteristics of a license are industry-specific with significant differences across industries.

Let's go now straight to the statistical findings:
1. About 20% of companies in Europe and about 27% of their Japanese counterparts license-out at least one patent to an unrelated party.

2. The probability of a company licensing-out patents is U-shaped, namely, small and large companies are more likely to license-out than mid-size companies.

3. In Europe, UK companies, followed by Nordic companies, are most likely to license-out.

4. Revenue is the major driver of licensing-out, followed by the sharing of technology (i.e., cross-licenses) and "constrained licensing" (an alleged infringer is compelled to accept a license).

5. About 24% of European companies holding patents, and about 53% of their Japanese counterparts, expressed the unrequited desire to license-out, due to the failure to locate a partner.

6. The use of patents in connection with fund-raising is most notable for young firms, i.e., venture capital (31%) and private equity (40%).
The authors then state as follows:
"The survey shows that licensing markets are less developed than they could be, in view of the willingness of patent holding companies to license more of their portfolio. Helping suppliers to find partners would substantially increase transactions in patent markets. Both market and government solutions exist which could alleviate obstacles and reduce transaction costs. Market-based mechanisms have recently emerged (technology brokers, internet platforms, patent funds, auction houses, IP consulting companies, etc.)."
I have a number of comments about this.
1. It has always seemed to me that the greatest threshold obstacle to getting our empirical arms around IP licensing is the quality of the data. Most IP licenses are private arrangements and the kind of (self-) sampling described in this study seems dicey as a statistical matter. Until we solve this problem, whatever the results are, they must viewed with a healthy degree of skepticism.

2. It would have been helpful if the authors had suggested the reason for the U-shaped probability curve for who licenses-out. Is this a robust result or an artifact of inadequate data? If the former, it strikes me as having significant implications for policy and practice, but only after a plausible explanation is offered for it.

3. I remain wholly puzzled by the geographical focus of licensing-out on the UK and the Nordic countries. Once again, is this simply the result of inadequate data, or does it represent a significant difference in the way that IP is commercialized in these countries, as contrasted with such jurisdictions as Germany, France and Benelux? I think that there the empirical jury is still out on this point.

4. As for the search for better licensee identification and reduced transaction costs through market mechanisms, there is a "quest for the Holy Grail" aspect here. More ink seems to have been spilled at positing the possibility of promoting such markets, as contrasted with identifying such markets on the ground. I really want to be proved wrong here. If so, research should begin with evaluating the effectiveness of the market alternatives suggested by the authors.
More generally, I wonder just how useful empirical studies of IP licenses can really be. This unease about the entire enterprise has not been allayed by this study, despite some tantalizing findings and suggestions.
It May be Easier to Find Transaction Costs After All