TIIP Newsletter

Introduction
Why do only 1 percent of patent owners ever enforce their intellectual property rights in court? Prior research presented explanations such as alternative licensing arrangements, trading of thick patent portfolios, and sheer irrationality. Allison et al argue that these answers are unsatisfactory because they cover only a small fraction of patents. Instead, the authors suggest that some patents are intrinsically more valuable than others, and only this minority is actually worth enforcing. In this paper they seek to determine what makes patents valuable, and how valuable patents can be identified.

This notion of patent value is, of course, one of private value—a firm may be willing to bring suit because it blocks a competitor, but this does not necessarily mean that the patents involved in the suit are beneficial to society.

Methodology
This paper begins with the assumption that litigated patents are valuable by virtue of the fact that their owners believed they were worth enforcing. Hence, the best way to identify valuable patents is to compare the characteristics of litigated patents to those of patents issued but not enforced in court. In a two-pronged approach, the authors conducted the largest, most comprehensive study of the patent system ever attempted.

First, in what they call their “large population study,” the authors examine every patent issued from 1963 to 1999 (a total of 2,925,537 patents) and every patent lawsuit terminated between 1999 and 2000 (4,249 cases covering 6,863 patents) at an aggregate level. This large-scale study allows us to learn a few basic facts about a large number of patents. In their more intensive “sample study,” the authors randomly selected a subset of 1,000 patents issued between 1996 and 1998, and 300 patents issued during the same time period and litigated between 1999 and 2000. The smaller study presents a deeper investigation. Using multivariate statistical inference, the authors attempt to predict characteristics of the larger population of issued and litigated patents. They find that litigated patents differ from non-litigated patents in substantial, statistically significant ways. Therefore, in aggregate, if litigation is a measure of value—the authors’ central assumption—then valuable patents can be identified ex ante.

Results
The authors identify eight key characteristics that distinguish litigated patents from non-litigated ones. Compared to the average issued patent, litigated patents:

1. Tend to be new (litigated on average at three-years old). This implies that value is known early in the life of a patent.

2. Tend to be owned by domestic rather than foreign companies.

3. Tend to be issued to individuals or small companies as opposed to large corporations. This result may reflect what the authors refer to as the “genius of the small inventor,” issues of asymmetric stakes between small and large firms, or the existence of a market for patents. Small firms may have little to lose from entering patent litigation if they are only the inventors and not the manufacturers of products related to their patents. In addition, while large firms may have large patent portfolios to protect themselves against lawsuits, small firms generally do not. Finally, recent research indicates that a large percentage of patents that are litigated changed hands after the patent was issued but before litigation. The authors ask whether there is something about the “transfer” itself from a small firm to a large one that increases the likelihood of litigation.

4. Cite more prior art and are more likely to be cited by other patents.

5. Spend longer in prosecution at the U.S. Patent and Trademark Office (USPTO).

6. Are more likely to belong to a group of patents derived from the same original patent application (patent procedures permit applications to be divided into multiple patents).

7. Contain more claims. On average, litigated patents make 19.6 claims, while non-litigated patents make only 13.0. Because claims are costly to draft and prosecute, a willingness to invest in more claims suggests that a patent is valuable to its owners. The authors note that this could be an ex ante attempt to strengthen a patent in expectation of litigation.

8. Come disproportionately from specific industries. Patents from the mechanical, computer, and medical device industries are more likely to be litigated. Patents from the chemical and semiconductor industries are less likely to be litigated.

Allison et al also note that the number of US PTO or IPC classifications per patent shows either no statistical relationship or a surprisingly negative correlation with litigation. However, after creating their own fourteen areas of technology and assigning each of the sample patents to one or more areas, the authors did find that the average number of these technology areas per patent was a significant predictor of litigation, and, presumably, value. This suggests that the idea behind using the average number of classifications as evidence of breadth is a good one, but that the original patent classifications do not serve the purpose. US PTO or IPC classifications do not properly identify technology areas because they were designed as aids for prior art searching and identify particular functions at low levels of abstraction, and do not identify technology areas conceptually.

Implications
Allison et al argue that their findings suggest interesting implications for patent theory, policy, and practice. First, these data may shed light on what has often been called the “black art” of patent valuation. Second, since “valuable” patents are subject to a more extensive prosecution process, the PTO may be doing a better job than expected in evaluating the patents that really matter, but the authors note that the examiners’ reward system does not provide an incentive to spend more time on the hard cases. Third, with such distinctive differences across industries, these data suggest that an industry-specific design of the patent system may be worth considering.

Finally, it is noteworthy that most of the characteristics identified are either within the control of the applicant, or at least known to the applicant before or during litigation, such that they can be predicted and therefore influenced. As the authors explain, the fact that some patentees voluntarily enter into a longer and more costly prosecution process suggests that they expect their patent to be valuable and worth the investment.


		Valuable Patents By John R. Allison (Texas), Mark A. Lemley (Berkeley), Kimberly A. Moore (George Mason), and R. Derek Trunkey (George Mason) FULL TEXT
		--Summary by Leslie Schafer (Wharton)
		Introduction Why do only 1 percent of patent owners ever enforce their intellectual property rights in court? Prior research presented explanations such as alternative licensing arrangements, trading of thick patent portfolios, and sheer irrationality. Allison et al argue that these answers are unsatisfactory because they cover only a small fraction of patents. Instead, the authors suggest that some patents are intrinsically more valuable than others, and only this minority is actually worth enforcing. In this paper they seek to determine what makes patents valuable, and how valuable patents can be identified. This notion of patent value is, of course, one of private value—a firm may be willing to bring suit because it blocks a competitor, but this does not necessarily mean that the patents involved in the suit are beneficial to society. Methodology This paper begins with the assumption that litigated patents are valuable by virtue of the fact that their owners believed they were worth enforcing. Hence, the best way to identify valuable patents is to compare the characteristics of litigated patents to those of patents issued but not enforced in court. In a two-pronged approach, the authors conducted the largest, most comprehensive study of the patent system ever attempted. First, in what they call their “large population study,” the authors examine every patent issued from 1963 to 1999 (a total of 2,925,537 patents) and every patent lawsuit terminated between 1999 and 2000 (4,249 cases covering 6,863 patents) at an aggregate level. This large-scale study allows us to learn a few basic facts about a large number of patents. In their more intensive “sample study,” the authors randomly selected a subset of 1,000 patents issued between 1996 and 1998, and 300 patents issued during the same time period and litigated between 1999 and 2000. The smaller study presents a deeper investigation. Using multivariate statistical inference, the authors attempt to predict characteristics of the larger population of issued and litigated patents. They find that litigated patents differ from non-litigated patents in substantial, statistically significant ways. Therefore, in aggregate, if litigation is a measure of value—the authors’ central assumption—then valuable patents can be identified ex ante. Results The authors identify eight key characteristics that distinguish litigated patents from non-litigated ones. Compared to the average issued patent, litigated patents: 1. Tend to be new (litigated on average at three-years old). This implies that value is known early in the life of a patent. 2. Tend to be owned by domestic rather than foreign companies. 3. Tend to be issued to individuals or small companies as opposed to large corporations. This result may reflect what the authors refer to as the “genius of the small inventor,” issues of asymmetric stakes between small and large firms, or the existence of a market for patents. Small firms may have little to lose from entering patent litigation if they are only the inventors and not the manufacturers of products related to their patents. In addition, while large firms may have large patent portfolios to protect themselves against lawsuits, small firms generally do not. Finally, recent research indicates that a large percentage of patents that are litigated changed hands after the patent was issued but before litigation. The authors ask whether there is something about the “transfer” itself from a small firm to a large one that increases the likelihood of litigation. 4. Cite more prior art and are more likely to be cited by other patents. 5. Spend longer in prosecution at the U.S. Patent and Trademark Office (USPTO). 6. Are more likely to belong to a group of patents derived from the same original patent application (patent procedures permit applications to be divided into multiple patents). 7. Contain more claims. On average, litigated patents make 19.6 claims, while non-litigated patents make only 13.0. Because claims are costly to draft and prosecute, a willingness to invest in more claims suggests that a patent is valuable to its owners. The authors note that this could be an ex ante attempt to strengthen a patent in expectation of litigation. 8. Come disproportionately from specific industries. Patents from the mechanical, computer, and medical device industries are more likely to be litigated. Patents from the chemical and semiconductor industries are less likely to be litigated. Allison et al also note that the number of US PTO or IPC classifications per patent shows either no statistical relationship or a surprisingly negative correlation with litigation. However, after creating their own fourteen areas of technology and assigning each of the sample patents to one or more areas, the authors did find that the average number of these technology areas per patent was a significant predictor of litigation, and, presumably, value. This suggests that the idea behind using the average number of classifications as evidence of breadth is a good one, but that the original patent classifications do not serve the purpose. US PTO or IPC classifications do not properly identify technology areas because they were designed as aids for prior art searching and identify particular functions at low levels of abstraction, and do not identify technology areas conceptually. Implications Allison et al argue that their findings suggest interesting implications for patent theory, policy, and practice. First, these data may shed light on what has often been called the “black art” of patent valuation. Second, since “valuable” patents are subject to a more extensive prosecution process, the PTO may be doing a better job than expected in evaluating the patents that really matter, but the authors note that the examiners’ reward system does not provide an incentive to spend more time on the hard cases. Third, with such distinctive differences across industries, these data suggest that an industry-specific design of the patent system may be worth considering. Finally, it is noteworthy that most of the characteristics identified are either within the control of the applicant, or at least known to the applicant before or during litigation, such that they can be predicted and therefore influenced. As the authors explain, the fact that some patentees voluntarily enter into a longer and more costly prosecution process suggests that they expect their patent to be valuable and worth the investment. © 2003. Verbatim copying and distribution of this entire article for noncommerical use are permitted provided this notice is preserved.