Measurement and Causal Inference in Patent Strategy
- Author(s): Kuhn, Jeffrey Michael
- Advisor(s): de Figueiredo, Rui
- et al.
Determining the effect of patents on firms has long presented significant challenges to innovation scholars due to the specific and idiosyncratic nature of patent protection. This thesis develops new techniques for measuring the impact, technological relatedness, and extent of patents. It then describes two new approaches to performing causal inference on patent protection. Finally, it applies these new measures and causal instruments to investigate the effect of receiving a broader patent on patent sales and the effect of winning a patent race on follow-on innovation.
Patent scope is central to the sale of ideas, which can spur economic growth and provide significant gains from trade. Awarding an inventor a patent on a new idea partially solves a commitment problem that would otherwise prevent the inventor from selling the idea. (Arrow, 1962). In the absence of a patent, a prospective buyer cannot credibly promise not to steal the idea should the inventor reveal it, while the inventor cannot credibly promise to reveal the idea should the prospective buyer pay for it. A firm's ability to use a particular patent to overcome this transactional hurdle derives from two factors: (1) the scope of the patent's legal right to exclude and (2) the effectiveness of that legal right in providing market exclusivity. This thesis first shows that a broader patent is more likely to be sold by employing a causal instrument that provides a plausibly exogenous shock to the scope of a patent's legal right to exclude, holding fixed the underlying idea. It then examines variation in the effectiveness of the right by interacting the instrument with endogenous firm, industry, and market characteristics. These results shed light on how firms profit from innovation and also connect the important but understudied market for patents, widely believed to be illiquid and inefficient, with fundamental research about how markets function in other contexts.
Competition between firms to invent and patent an idea, or "patent racing,"' has been much discussed in theory, but seldom analyzed empirically. This thesis introduces an empirical way to identify patent races, and provides the first broad-based view of them in the real world. It reveals that patent races are common, particularly in information-technology fields. The analysis is then extended to identify the causal impact of winning a patent race, using a regression-discontinuity approach. It shows that both the winners and losers of patent races typically receive patent protection, but that the winners receive much broader patent scope. It also shows that patent race winners do significantly more follow-on innovation, and the follow-on research that they do is more similar to what was covered by the patent.
Underlying both of these analysis is a new measure of patent-to-patent similarity. Current measures of patent similarity rely on the manual classification of patents into taxonomies. This thesis describes and validates a machine-automated measure of patent-to-patent similarity developed by leveraging information retrieval theory and Big Data methods. It also demonstrates that the measure significantly improves upon existing patent classification systems.