Patent rights and cumulative innovation: Part I
Balancing the first breakthrough against the next thousand improvements
In the first post of The Patentist, we explained that patents should only be granted for inventions that would not have been created in the absence of the patent system. Granting exclusivity to inventions that would have been developed anyway simply creates monopolies without delivering the intended incentive effect. The goal of the patent system is, therefore, to maximize the production of these “additional” inventions.
Innovation, however, is a cumulative process: new ideas are built on top of existing ones. As Suzanne Scotchmer emphasized, this creates a core trade-off in patent system design. Strong patent rights can encourage the creation of “first-generation” inventions, but they may also discourage follow-on work that builds on them. Weaker protection can make it easier to develop follow-on inventions, yet may not offer enough reward to justify the initial R&D investment. This post explores that delicate balancing act.
Before going further, it is helpful to distinguish two related terms. Cumulative innovation refers to the overall process in which innovations build upon one another over time. Follow-on innovation (or subsequent innovation) refers to the individual steps, discoveries, or products that appear later in that cumulative chain.
Two competing views
Theory offers two opposing views on how patent rights affect cumulative innovation. On one side is Edmund Kitch’s “prospect theory” of patents. In this view, the patent system grants exclusive and publicly recorded ownership of a technological “prospect” soon after discovery—much like a mining claim. The pioneer inventor can then coordinate subsequent development of that prospect. Stronger patent rights are seen as helpful because they give the initial inventor the authority to organize follow-on research, reduce wasteful duplication of R&D, and limit free-riding and coordination failures among downstream innovators.
On the other side is Michael Heller and Rebecca Eisenberg’s “tragedy of the anti-commons,” which emphasizes the blocking effect of patents. The classic tragedy of the commons arises when a shared resource is overused because many users have the privilege to use it but no one has an incentive to conserve it (think of unregulated fishing in the ocean). The anti-commons flips this logic: when too many parties hold rights to exclude others from using a resource—such as multiple fragmented patent rights over components of a single product—the resource tends to be underused. Uncoordinated bargaining among many right holders can generate high transaction costs and royalty stacking, making it difficult or impossible for follow-on innovators to secure all necessary licenses. In this view, stronger patent rights reduce cumulative innovation because broad upstream patents can block valuable subsequent improvements.
Jerry Green and Suzanne Scotchmer formalize this tension in a model of sequential innovation. What matters most, they argue, is not only how strong the initial patent is, but also how the economic gains from later inventions are shared between the pioneer and follow-on innovators through licensing. If the first inventor can capture almost all of the surplus from subsequent improvements, potential followers have little reason to invest. Conversely, if followers keep most of the surplus, incentives to undertake the initial invention are too weak. In this perspective, the patent system and licensing contracts must work together to divide rewards along the innovation chain so that both first-generation and follow-on research remain attractive.
Economists have built on these ideas to study the conditions under which “stronger” patent rights are more or less likely to stimulate cumulative innovation. To do so, they first need a way to formalize what “strength” means. The next section explains how economists model the strength of patents in the context of cumulative innovation.
What economists mean by “patent strength”
In a previous post, we distinguished between law on the books and law in practice. The former concerns written, official patent rules; the latter concerns how those rules are applied in real life. Economists studying cumulative innovation have built both aspects into their models.
On the law-on-the-books side, patent strength is typically captured by three design choices: the length of patent protection, the size of the inventive step, and the breadth of protection. On the law-in-practice side, the main element modeled is the validity of a granted patent.
Patent duration (or length) can be understood in two ways. The statutory term is the legal life of the patent (typically 20 years from the filing date). The effective term is the period until either the patent expires or a non-infringing substitute appears and erodes the patentee’s market power, whichever comes first.
The inventive step is a second key element of patent design. It sets the minimum threshold an invention must clear to be patentable. A higher inventive-step requirement means that only more novel and less obvious inventions qualify for protection.
Patent breadth (or scope) refers to the technological territory covered by the claims—that is, how different rival products must be to avoid infringement. Ted O’Donoghue and colleagues decompose breadth into lagging breadth and leading breadth. Lagging breadth offers protection against competition from imitators and is linked in patent law to the doctrines of disclosure and enablement. Leading breadth offers protection against future innovators by determining which follow-on innovations infringe the original patent. It depends on how courts interpret “use of a technology” (i.e., how broadly they define what counts as making, using, or selling the patented invention), as well as on the doctrine of equivalents and the reverse doctrine of equivalents.
Finally, the validity (or enforceability) of a granted patent is the likelihood that it will be upheld in court. A model by Silvana Krasteva and colleagues shows that strengthening patent validity creates a trade-off. On the one hand, it protects the innovator’s profits against imitators, which can encourage initial R&D. On the other hand, it makes lawsuits more credible and more threatening, which can discourage follow-on innovators from working in nearby technological space.
Why patent-strength dimensions can’t be set one by one
The different dimensions of patent strength are policy instruments that interact and can sometimes be traded off against each other to achieve better innovation incentives.
A classic debate in the patent-design literature contrasts “long and narrow” patent rights with “short and broad” ones. In models of a single, stand-alone innovation (ignoring cumulative research), Richard Gilbert and Carl Shapiro show that the optimal patent can, in theory, be infinitely long but very narrow—just broad enough to let the inventor recover R&D costs—thereby minimizing the social cost of monopoly power. By contrast, Nancy Gallini finds that when imitation is costly, short and broad patents can be preferable. Paul Klemperer further demonstrates that the optimal design depends critically on consumer preferences, in particular, how costly it is for consumers to switch between product varieties and how much they are willing to pay for their preferred variety.
When innovation is cumulative, the relationship between length and breadth becomes more complex. The statutory patent term may become almost irrelevant if market power ends as soon as a non-infringing substitute displaces the protected product. In plain English, extending the patent term to, say, 30 years after grant may have little effect if lagging breadth is narrow enough that competitors can easily commercialize non-infringing substitutes.
O’Donoghue and colleagues obtain a striking result in a model of sequential innovation that analyzes the trade-off between patent length and leading breadth along a long sequence of improvements (a “quality ladder”). They show that a given rate of innovation can be supported by two very different patent designs. One design combines infinite leading breadth with a finite statutory length. Every improvement infringes the original patent until it expires, so the effective patent life equals the statutory term. This arrangement minimizes the cost of delayed diffusion because monopoly pricing ends relatively quickly. The alternative design combines finite leading breadth with an infinite statutory length. Here, the effective patent life ends when a sufficiently large, non-infringing improvement is discovered. This second policy often does better at minimizing R&D costs.
The model by Silvana Krasteva and colleagues highlights a different interaction, this time between patent length and patent validity. When validity is low, infinitely long patents are preferred to maximize the positive competition effect (protecting innovators against imitators). When validity is high, however, long patents amplify the negative litigation effect on followers, so finite patent length becomes necessary to keep innovation incentives in balance.
The broader lesson is that any sweeping claim about cumulative innovation and “patent strength” should be treated with caution. Patent strength is a multi-faceted concept, and its various dimensions interact in subtle ways.
When stronger patents help—and when they hurt
Whether stronger patents help or hinder cumulative innovation depends on a few broad factors: how well licensing works, the nature of the technology, and the cost and frequency of R&D.
Patent strength is more likely to encourage cumulative innovation when licensing is relatively frictionless. In this ideal world, the upstream patentee can coordinate subsequent development and make sure that valuable follow-on projects go ahead. Problems arise when follow-on innovators cannot obtain the necessary licenses, for example in settings that resemble an anti-commons or a patent thicket (many overlapping patents that must all be licensed). In such contexts, licensing can fail because of high transaction costs, asymmetric information (leading to holdup or bargaining failure), or because incumbents simply prefer not to license potential competitors to protect their product-market profits (known as “rent dissipation”).
Stronger patents are also more likely to have a positive effect in discrete technology areas (for example, some chemicals or drugs), where products rely on a small number of patents and the space for potential improvements is relatively clear and bounded. By contrast, blocking effects are more likely in complex technologies (such as computers, electronics, and medical instruments), where new products embody many patentable elements and innovation is highly cumulative and complementary. In these fields, fragmented patent ownership makes bargaining harder and increases the risk that negotiations break down.
When innovation is relatively cheap or frequent, weaker or shorter protection is often preferred. Very strong, long patents can encourage incumbents to delay their own improvements—because new products would cannibalize existing profits (the replacement effect)—and can also discourage entry by followers. A model by James Bessen and Eric Maskin even shows that in highly cumulative fields such as software, a regime with no patents can generate more innovation than one with strong patents. By contrast, stronger protection is optimal when initial, pioneering innovation is difficult or costly, and thus requires high ex ante incentives—as in many pharmaceutical markets. The overall prediction is non-monotonic: to maximize innovation, patent strength should typically be set at some intermediate level [1,2].
From theory to evidence
As the discussion suggests, there is no one-size-fits-all patent system. Technologies and industry structures differ widely across sectors, and a policy choice that works well in one setting may be counterproductive in another. The challenge for policymakers is, therefore, multi-dimensional: they must not only balance the needs of different sectors, but also calibrate the various facets of patent strength—length, breadth, inventive step, validity, and the institutions that govern licensing and enforcement.
Economic models are valuable because they clarify the mechanisms at work and map out the trade-offs. But they also rely on assumptions, and they cannot tell us on their own how large the effects are in the real world. For that, we need data. In the next post of The Patentist, we will move from theory to evidence and delve into this empirical literature. Stay tuned.
If you enjoy evidence-based takes on patents and innovation, join hundreds of readers who receive The Patentist directly in their inbox.
Please cite this post as follows:
de Rassenfosse, G. (2025). Patent rights and cumulative innovation. The Patentist Living Literature Review 10: 1–X. DOI: TBC.