ارزش گزینه شکایت های قانونی ثبت اختراع: نظریه و شواهد

In this paper, I use a real options approach to investigate patent litigation when enforcement is costly, winning is uncertain, and beliefs about validity are stochastic. I consider both finite horizon and infinite horizon models. The theoretical results demonstrate that patent value depends not only upon the underlying technology but also upon the degree of uncertainty over the property right. Additionally, imperfect enforceability creates an effective patent term that is less than the statutory term. Using simulation methods and patent data, I estimate the hazard rate of patent litigation. Contrary to previous studies, I find that the rate of forward citations is negatively associated with the litigation rate. The difference arises because (1) I use a dynamic model that exploits the information contained in the timing of litigation and citations, and (2) I control for truncation using a duration model. Using random effects models, I find that heterogeneity in patent litigation is embedded primarily in the heterogeneity in receiving patent citations. Because of this, the patent litigation decision can be modeled as a unilateral decision.

Patents are classic examples of real options: a patent holder has the option to develop certain types of products, to license the technology, or to use it as an input for further research. Pakes (1986) models the patent as an option to renew in patent regimes that require renewal fees. Reiss (1998) models the decisions of firms to develop and patent innovations when competition arrives stochastically. The paper explicitly examines the trade-off between patents and trade secrets as modes of protection, and it introduces some uncertainty in the patent right, in that competition may develop a substitute technology that does not infringe the patent. Takalo and Kanniainen (2000) and Weeds (2002) develop models investigating the patenting decision. Neither paper explicitly models the patent enforcement process, although Takalo and Kanniainen incorporate enforcement costs into the general cost of patenting. Most dynamic models treat the patent holder as having exclusive rights to develop a technology. However, any option on the patented technology presupposes an enforceable property right. The fundamental value of a patent right is the “right to exclude” others from using the technology. Because enforcement is imperfect and costly, the right to exclude becomes the right to sue with some probability of success. In this paper, I treat the patent as an option to bring a lawsuit against an alleged infringer. As with financial options, the option to sue need not be exercised in order for it to have value. Thus, the value of a patent is a function of the enforceability of the property right, the underlying technology, and the distribution of beliefs about those parameters. The model enables me to be precise about the sources of patent value and to examine the litigation decision. Section 2 presents the basic framework of patents as options in cases where there is uncertainty over the validity of a patent. I solve the model analytically for infinitely lived patents, and numerically for the finite horizon case. The model establishes that imperfect enforceability leads to an effectively shortened patent life, and that rewards to innovation involve a trade-off between the length, breadth, and strength of patent rights. Failure to recognize the consequences of uncertainty and costly enforcement inevitably leads to myopic patent policy. In Section 3, I investigate the litigation rate using simulations. The results lead to some testable implications about the timing and incidence of litigation, including: (1) that greater uncertainty over the probability of patent validity should delay litigation, and (2) that greater relative profit flow should decrease the litigation rate. These results are tested using a matched sample of litigated and non-litigated patents, described in Section 4. Section 5 estimates the hazard rate of patent litigation. I find that the rate at which patents receive citations is a good proxy for profit flow and that patent claims are a proxy for scope. In contrast to previous studies, I find that the patent citation rate – normalized by patent claims – has a negative impact on the litigation rate. This highlights the importance of using dynamic methods to investigate the enforcement decision in both theoretical and empirical models. Finally, I find that heterogeneity in patent litigation is primarily a function of heterogeneity in the rate of receiving citations. Consequently, patent litigation can be treated as a unilateral decision on the part of the patent holder, conditional on observable patent characteristics. Section 6 concludes with a discussion of policy implications.

A central theme in this paper is that it is important for policy-makers to explicitly understand the role of enforcement in patent value because it is implicitly a policy instrument. At the administrative level, some amount of legal uncertainty is inevitable; however, it can be influenced by the PTO. The PTO may choose to direct more resources to patent authentication. Alternatively, it may want to depend on individual firms to enforce their own patents. In fact, it may be more cost effective to accommodate some degree of uncertainty in the system. In this way, expenditure on each granted patent will be reduced, and only those that are in dispute will be investigated (in court) at further cost. The fact that the enforcement decision is deferrable makes a real options model appropriate. Modeling litigation in a dynamic setting leads to important theoretical and empirical results. From a theoretical standpoint, an important result of this paper is that the exercise boundary for litigation falls to zero prior to the expiration of the patent, which effectively reduces the patent term below the statutory term. This result has important consequences for patent protection, especially if policy-makers ignore enforcement problems. Differences in validity and uncertainty across technology classes will lead to arbitrary differences in patent terms. On the other hand, explicit recognition of this effect gives policy makers more flexibility in setting patent policy. With regard to appropriation, the strength of property rights enters symmetrically with patent scope from a theoretical perspective. Thus, the fundamental trade-off is not between breadth and length, but rather among breadth, length, and strength. Legal uncertainty is especially pervasive in emerging technology areas, and especially in emerging patenting areas like biotechnology in the 1980s and 1990s, software in the 1990s, and business methods more recently.31 In new patenting areas, it may be desirable to give more limited patent rights by issuing intentionally weak or uncertain patents, in terms of validity. Over time – as the courts and the PTO define clearer standards – beliefs for certain classes of patents will be revised upward or downward based on related cases. In this way, patents that the courts and the PTO wish to make stronger will be just those patents that experience an increase in confidence about validity. In fact, it is more feasible for the PTO to issue weak patents than to adjust the patent term on an ad hoc basis. Policy makers must also realize that the different components of patent value provide different incentives for enforcement. My simulation results demonstrate that increases in patent scope and validity increase the litigation rate, but increases in the relative profit flow x/pz decrease litigation. From an empirical standpoint, the use of dynamic methods provides insight into the litigation decision, and generates results that contradict previous studies. I find that the ratio of forward citations to claims is negatively correlated with the litigation rate. For comparison, I re-estimate the probability of litigation using a simple probit. In accord with Lanjouw and Schankerman (2001), the probit generates positive coefficients for forward citations and citations per claim.32 The contradictory results highlight the need for researchers to correctly model the decision-making processes that underlie econometric results. The dynamic approach exploits the timing of the litigation decision – which is important in an options framework – and accounts for truncation. Furthermore, the empirical results of this paper indicate that the forward citation rate is a good proxy for profit flow, and that most heterogeneity among patents – with regard to the litigation decision – can be tied to heterogeneity in the forward citation rate. This is important because it establishes that the litigation decision can be treated as unilateral on the part of the patent holder, conditional upon observable characteristics. Finally, I find there to be predictable differences across technologies. Patents in computers, electronics, and medical/biotechnology evidence higher levels of uncertainty than chemical and mechanical patents. As found in the simulation results, increased uncertainty delays litigation and increases duration dependence in the hazard rate. Current political attention on tort reform in the U.S. is evidence that policy-makers recognize the policy dimension of legal uncertainty on a broad scale. To the extent that the effects of uncertainty and enforcement are ignored in patent policy, incentives for innovation will be inconsistent and arbitrary and the economic costs of patent litigation will be exacerbated.