Empirical Data on “Comparable Licenses” in Patent Infringement Suits

Economists Ink: A Brief Analysis of Policy and Litigation



EI Vice President Thomas R. Varner has experience in complex litigation involving intellectual property and other issues. His research in licensing is described in “An Economic Perspective on Patent Licensing Structure and Provisions,” Business Economics, (Vol. 46, No. 4, 2011), reprinted in les Nouvelles (March, 2012).

Economic damages experts in patent infringement suits are often called upon to determine a reasonable royalty arising from a hypothetical negotiation for a license for the patents-in-suit. Determining a reasonable royalty often involves evaluating, among other factors, the financial terms of “comparable licenses.” A number of recent court decisions in patent infringement suits have highlighted the importance of correctly determining which licenses are comparable for purposes of determining a reasonable royalty. For example, in Wordtech Systems v. Integrated Networks Solutions (2010), the Federal Circuit stated, “Comparisons of past patent licenses to the infringement must account for ‘the technological and economic differences’ between them.” Other recent cases addressing this issue include Lucent v. Gateway (Fed. Cir. 2009), ResQNet.com v. Lansa (Fed. Cir. 2010), Uniloc v. Microsoft (Fed. Cir. 2011), and Oracle v. Google (N.D. Ca. 2012).

What should experts consider as relevant “economic differences” when deciding which licenses are comparable and which are not? Courts have described several economic differences as relevant to the patent infringement cases before them. These differences include whether an agreement had a lump-sum payment versus a running royalty payment, whether the license was exclusive or nonexclusive, whether the parties were competitors, and the number and nature of patents licensed and products covered.

Research conducted on thousands of technology licenses sheds light on the economic differences between license agreements and their royalty rates. Information from a dataset of over 4,000 complete technology licenses collected from material agreements submitted to the U.S. Securities and Exchange Commission (SEC) over the past 20 years indicates that a number of factors are relevant to the financial consideration specified in patent licenses.

One of the first steps in analyzing technology licenses is to identify the general nature of the agreement, e.g., is it a product license, a development agreement, a research agreement, a joint venture agreement, a bare patent license, a patent plus know-how license, or some other type of agreement? The median and average royalty rates observed in technology licenses can vary significantly across agreement types. Not surprisingly, product licenses generally have among the very highest royalty rates, whereas bare patent licenses generally have among the very lowest royalty rates. Courts have found that software rebundling or repackaging agreements are not comparable to software bare patent licenses. The research supports this finding: the median royalty rate of software product licenses in the dataset is approximately 10 percent, whereas the median royalty rate of software bare patent licenses is approximately 3 percent. Moreover, bare patent licenses have among the lowest median royalty rates compared to other agreement types regardless of the nature of the licensed technology (e.g., software, hardware, pharmaceutical, medical, and telecommunications).

In analyzing the comparability of different licenses, the question often arises of how to apportion the value of the patented technology from other technologies included in the licenses, such as technology “know-how.” The research shows a statistically significant difference between royalty rates found in patent plus know-how licenses and royalty rates found in bare patent licenses. Although the premium for technology know-how varies across different industries, patent plus know-how licenses generally have median royalty rates 30 percent to 50 percent higher than median royalty rates of bare patent licenses.

Courts also have found that the nature of the licensor is relevant in assessing whether agreements are comparable. The research only partially supports that view. The median royalty rates of patent licenses are generally higher when the licensor is a commercial entity rather than a non-commercial entity. However, there was little or no statistically significant difference in patent license royalty rates across different types of non-commercial licensors (i.e., universities, nonprofits, government agencies, and individuals).

While many experts focus on a single royalty rate expressed as a percentage of sales, just one-third of patent licenses in the dataset specify only a running royalty rate. The remainder of these agreements specify a running royalty plus a fixed fee, a fixed fee only, or are royalty-free. Additionally, many agreements that specify a running royalty in terms of a percentage of sales specify not just one royalty rate but a tiered structure of royalty rates (e.g., a royalty rate of 2 percent on the first $10 million dollars of sales then a royalty rate of 1 percent thereafter). This declining structure of tiered royalty rates (e.g., 2 percent to 1 percent as sales increase) is by far the most common form of tiered royalties in the high-tech and medical device industries. Patent licenses in the pharmaceutical industry, however, generally specify increasing tiered royalty rates (e.g., 1 percent to 2 percent as sales increase).

Experts also need to consider the terms in license agreements that modify base royalty rates. Licenses can contain terms that specify conditions in which a base royalty rate will be reduced. Approximately 30 percent of all patent licenses observed in the dataset included such royalty reduction provisions. For example, a “royalty stacking” provision lowers the base royalty rate if the licensee subsequently needs to take a license from a third-party and is required to pay royalties to that third-party. Almost two-thirds of the patent licenses in the dataset that included a royalty stacking provision specified a maximum royalty reduction of 50 percent of the base royalty rate.

Basic economic principles suggest that an exclusive license would command a higher royalty rate than the rate for a nonexclusive license. Nonetheless, little significant difference in royalty rates was observed between exclusive and nonexclusive patent licenses in the dataset, even accounting for other factors such as the nature of the licensor, the nature of the licensed technology, and the inclusion of know-how in the agreement. Other researchers who have studied royalty rates in technology licenses have made similar observations as well. Determining how exclusivity affects royalty rates is complicated by the presence of other factors that are difficult to observe, such as the established commercial success of products incorporating the licensed technology or the availability of alternative technologies. However, a number of license agreements in the dataset include a royalty reduction provision giving the licensor the option to convert the license from an exclusive basis to a semi-exclusive or nonexclusive basis. In these agreements, because the technology and the parties remain constant, the effect of the conversion from exclusive to nonexclusive can be isolated. In such agreements, the typical royalty reduction for converting an exclusive license to a nonexclusive license is 50 percent of the base royalty rate.

Research involving thousands of technology licenses filed with the SEC provides empirical support for recent patent infringement decisions that addressed the economic differences between licenses. Courts, and damages experts, will likely give added scrutiny to these differences in the future.