Patent fees are usually treated as a dull administrative detail—a line item in a legal budget or a necessary friction for maintaining a patent portfolio. But in 2025, a reported U.S. proposal to charge patent owners a fee equal to 1–5% of the assessed value of their patents briefly turned these seemingly mundane charges into front-page innovation policy.

The proposal suggested that blockbuster patents worth billions should contribute more to the system than niche consumer inventions. By February 2026, the idea appeared dead, with Commerce Secretary Howard Lutnick telling a Senate panel that the USPTO would not pursue a valuation-based fee. Yet, even if the “value tax” was transient, the controversy revealed a deeper truth: patent fees are not merely a way to fund an office. They are a regulatory technology. They determine who enters the system, what gets examined, which patents survive, how patent offices behave, and, ultimately, who bears the cost of exclusive rights.

The applicant’s calculation: when fees change strategy

To understand how fees function as regulation, it is helpful to start by examining how inventors respond to price signals. In the economic literature, this response is measured by the price elasticity of demand.

A useful empirical rule of thumb is that a 10% increase in patent fees reduces patenting by roughly 3–5% [1,2]. For context, such a figure is closer to the price sensitivity observed for essential inputs such as oil or steel than for discretionary services such as leisure air travel, where a 10% price increase may reduce demand by more than 10%. In economic terms, patent demand is, therefore, relatively price-inelastic. Most applicants with a promising invention will not abandon patenting simply because official fees rise: the patent may be too important to the firm’s business strategy, fundraising prospects, licensing plans, or competitive position.

Low sensitivity to fees does not mean that fees are irrelevant. However, it implies that fee changes may need to be substantial before they produce visible effects. Historical evidence from Britain points in this direction. Tom Nicholas examines the Patents Act of 1883, which reduced the cost of patenting by a whopping 84%. The reform produced a large increase in patenting, but not a commensurate increase in measured innovation. The lesson is important: lower fees may induce greater use of the patent system without necessarily creating more inventions.

Furthermore, patent-office fees are not a single price, and applicants often respond along margins other than a simple yes-or-no decision on patenting. Fees are a schedule of charges for different choices, including, but not limited to, filing, examination, claims, continuations, disclosures, opposition, validation, and renewal. Instead of filing fewer applications, sophisticated applicants may adjust their filing strategies in response to a fee increase. The USPTO’s fee schedule, which saw major updates take effect on January 19, 2025, illustrates this kind of behavioral nudging. By introducing tiered surcharges for large Information Disclosure Statements (IDS), with fees rising as the cumulative number of submitted references crosses specified thresholds, the Office is seeking to discourage applicants from flooding examiners with prior art—not to discourage patent filings.

The IDS surcharge is an example of a “complexity tax”: a fee designed to make applicants internalize some of the administrative burden they create. Another example is the EPO’s 2008 introduction of a €200 claim-based fee for applications exceeding 15 claims, aimed at limiting excessive claim proliferation. More broadly, as Dietmar Harhoff and colleagues show, fees determine not only whether and how an invention is patented, but also where it is patented. In the fragmented European landscape, validation and translation costs force firms to make cold calculations about which national markets truly justify the “entry fee.”

Attorney fees set the gate; official fees steer the path

There is, however, an elephant in the room. Patent-office fees are only the visible, public part of the cost of patenting. For many applicants, they are not even the largest part. As innovators know, patenting costs include internal management costs, translation fees, patent attorney or agent fees, and administrative fees, with professional fees often dwarfing the fees charged by patent offices.

At first glance, the high patenting costs seem to weaken the case for the fee policy. If administrative fees are only a small share of the total cost of patenting, why should they affect behavior at all? The answer is that patenting decisions are not made once and for all. They are made sequentially at each procedural step, and many of those steps are governed directly by official fees.

The intuition is simple. Imagine that a firm has already paid a patent attorney to understand the invention, draft the specification, formulate the claims, and file the application. Much of that expense is now sunk: it cannot be recovered by abandoning the application. The relevant decision is no longer whether to incur the entire cost of patenting from scratch. It is whether the next step is worth paying for.

That next step may be filing a continuation, validating the patent in Greece or Belgium, responding to another office action, or paying the next renewal fee. These are marginal choices. They are often triggered directly by official fee schedules, even though the original attorney cost was much larger.

This distinction also varies by applicant type. Large firms with internal IP departments may draft applications in-house, standardize prosecution workflows, and spread fixed legal costs over large portfolios. For them, patenting may look more like an internal production process, with official fees becoming a more sizeable cost of portfolio management. By contrast, startups, universities, and independent inventors often rely on external counsel. For them, the first attorney bill may be the true barrier to entry.

Fees as a screening device: separating high value from low

If complexity fees shape how applicants draft and prosecute patents, application fees raise a more basic question: can the patent office use price to screen what enters the system in the first place? In principle, yes. A filing fee forces applicants to compare the expected value of protection with the cost of obtaining it.

Theoretical models by Bernard Caillaud and Anne Duchêne [3] and Florian Schuett [4] show how fees can induce self-selection. If applying is cheap, firms with marginal inventions may enter the system in the hope of obtaining protection. If application fees are higher, some low-value or obvious inventions no longer justify the cost of filing. Combined with meaningful examination, fees can therefore help create a separating equilibrium: stronger inventions are filed, while weaker ones are screened out before reaching the examiner’s desk.

Empirical evidence supports this screening logic, but with nuance. Studying the 1982 U.S. fee increase, one of my papers with Adam Jaffe finds that higher fees disproportionately weeded out patents in the lowest quality and value deciles. The effect, however, was concentrated among firms with medium-to-large portfolios, suggesting that fees screen most effectively where applicants already patent strategically.

Evidence from Japan reaches a similar conclusion from the opposite direction. Masayo Kani and Yoichiro Nishimura study Japan’s 2011 reduction in patent examination fees and find only a limited negative effect on the quality of patent applications, with no detectable decline in the quality of granted patents. Fee reductions may, therefore, widen access without necessarily flooding the system with low-quality granted patents, especially when examination remains meaningful. In this sense, the best way to ensure high-quality patents is to conduct high-quality examinations.

Access, inequality, and the small entity

The screening logic has an obvious distributional limit. Fees may discipline overuse, but they can also exclude applicants who lack cash, experience, or access to patent counsel. If fees are a regulatory technology, they must be calibrated not only to discourage weak filings, but also to preserve entry. For small firms, a discount may lower the toll charged by the patent office, but it does not eliminate the cost of crossing the bridge.

Recent policy shifts acknowledge that tiered pricing is part of innovation policy. The USPTO recently expanded discounts for small and micro entities under the Unleashing American Innovators Act framework. Similarly, in April 2024, the EPO introduced a support scheme giving qualifying micro-entities a 30% reduction in main procedural fees.

But the evidence also warns against expecting too much from fee reductions alone. In another paper with Adam Jaffe, we study whether lower application fees increase entry into patenting by small and micro entities. Their conclusion is sobering: fee reductions alone appear insufficient to significantly increase participation in the patent system among these applicants. The reason is precisely the broader cost structure discussed above. Official fees are only one part of the total cost of patenting; attorney fees, uncertainty, prosecution complexity, and limited familiarity with the patent system may matter just as much, and sometimes more.

This does not mean that small- and micro-entity discounts are useless. It means they are incomplete. They lower one barrier, but they do not transform the broader economics of access to the patent system.

The renewal test: which patents survive?

At the heart of fee design lies a basic tension: the patent system should be open enough to encourage entry and disclosure, but not so cheap that exclusion rights remain in force when their private value no longer justifies their social cost. Many major patent systems address this tension through back-loaded fee schedules: initial official fees are kept relatively modest, while renewal or maintenance fees rise over the life of the patent [5].

The economic logic of renewal fees is that they force patentees to reveal, periodically, whether a patent is still worth keeping alive. Ariel Pakes showed that the renewal decision can be understood as an option-like choice: each year, the owner decides whether the expected future value of the patent justifies paying the next renewal fee. Mark Schankerman and Ariel Pakes used this logic to infer the private value of patent rights from observed renewal behavior. A patent that is repeatedly renewed is not necessarily socially valuable, but it is valuable enough to its owner to justify the cost of continuation.

Renewal fees limit the duration of exclusion rights whose private value no longer justifies their cost. This public-domain function of renewal fees is well established, and several economists have modeled their optimal structure. Francesca Cornelli and Mark Schankerma, for example, show that optimal renewal fees should rise steeply with patent age, so that owners of highly valuable patents can buy longer protection while owners of lower-value patents are induced to let their rights lapse earlier. Marc Baudry and Béatrice Dumont (2006) make a related point: renewal fees can be designed as a mechanism for weeding out low-value patents. In this sense, the “renewal clerk” is as important to the public domain as the patent examiner is to the private one.

This logic also underlies more aggressive proposals for back-loaded fees. Jim Bessen and Brian Love propose “Pigouvian” renewal fees designed to induce the expiration of unused patents before they are acquired for nuisance-value litigation. A Pigouvian tax is a fee imposed to make private actors internalize the social costs they impose on others, as in the classic case of pollution. Applied to patents, the idea is to keep early fees low while raising late-stage maintenance fees steeply, forcing owners of older, unused patents either to justify their continued exclusionary rights or to let them fall into the public domain.

The same framework also highlights an institutional complication: the socially optimal fee schedule that economists devise need not be the revenue-maximizing one that the patent office may implement. Josh Gans and colleagues show that a self-funding patent office may have an incentive to flatten the fee schedule: raising application fees to collect revenue upfront while keeping renewal fees lower than socially optimal to encourage more patents to be kept alive. The result is a double distortion: some inventions may be deterred at entry, while too many exclusion rights may persist for too long. Fee design is, therefore, not only about steering applicants; it is also about ensuring that the institution setting the fees does not face incentives that push it away from social welfare.

The frontier: From cost recovery to behavioral pricing

Once fees are understood as behavioral instruments, the policy frontier becomes clearer. Existing fee schedules already use claim fees, IDS surcharges, late-continuation fees, and validation fees to discipline procedural complexity. More radical proposals push the same logic further, using fees not only to recover administrative costs but to reshape filing incentives, portfolio accumulation, and the stock of granted rights.

Neel Sukhatme, for example, notes that successful applicants typically pay more than unsuccessful ones and proposes higher fees for rejected applications to deter “lottery-ticket” filings. Another family targets the stock of granted patents. David Olson suggests scaling maintenance fees with the size of the owner’s portfolio, making very large portfolios more expensive to maintain and thereby discouraging the accumulation of anti-competitive patent thickets.

These ideas are not all equally feasible, and some move beyond ordinary patent-office cost recovery into quasi-taxation or litigation policy. But cost recovery should not be mistaken for social optimality. A fee schedule that balances the patent office’s books may still be too high at entry, too low at renewal, or poorly targeted across applicants and behaviors. Elsa Martin and Hubert Stahn push this point further, arguing that patent fees can be used as an innovation-policy instrument, not merely as a way to fund examination costs: in their model, fee revenues can be reallocated to universities to support research that expands the pool of freely available knowledge on which future patents build.

Conclusion: you are what you charge

Patent fees look like administrative charges, but they are really design choices. They decide who enters, who stays, who exits, and who pays for the system. The 2025 value-tax episode was a reminder that the price of patent protection is never neutral. A smart fee policy should not simply balance the books. It should ask what kind of patent system the fee schedule is quietly building: one that promotes entry, screens weak claims, releases unused rights, and preserves access—or one that merely maximizes revenue.

Refer a friend

Leave a comment

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. (2026). When patent fees become innovation policy. The Patentist Living Literature Review 12: 1–6. DOI: 10.2139/ssrn.6658140.

Standards solve a coordination problem—and then create a new one. Once an industry converges on a technical interface, compatibility becomes a requirement, not a preference. If implementing that interface requires patented technology, the patents become standard-essential, and licensing shifts into a high-stakes environment shaped by lock-in and complements: missing one license can block an entire product. Part I used this logic to explain hold-up, a recurring dispute in SEP markets. Part II asks what keeps this system workable in practice: how FRAND tries to pin royalties to ex ante value, why injunction leverage is the pressure point, and how thickets, stacking, and pools complicate the governance problem. The literature is quite complex and a bit jargon-heavy, so buckle up.

FRAND commitments: the governance response to SEP bargaining power

SSOs’ “FRAND” policies can be understood as an attempt to manage this transformed bargaining problem. FRAND stands for “Fair, Reasonable, and Non-Discriminatory.” The goal of FRAND is to create a neutral zone in which royalties reflect the technology’s ex-ante incremental value, rather than the value created by lock-in. Daniel Swanson and William Baumol explain that FRAND does this by pushing the negotiation back toward an ex-ante benchmark—the value of the technology before the standard was chosen and before implementers became locked in. The relevant comparison is not “how valuable is this feature inside today’s ubiquitous standard?”, but rather “how much better was this technology than the next-best alternative that could have been standardized?” This idea aims to strip out the portion of value stemming from the industry’s coordination on the standard itself (and from implementers’ sunk investments), leaving the patent holder rewarded for genuine technical merit.

If that sounds like a clean solution, the complication is that FRAND is intentionally incomplete. SSOs typically require a commitment to license on FRAND terms, but they generally avoid specifying numbers: they do not set a royalty rate, define the precise royalty base, or provide a formula that mechanically maps a patent portfolio into a fee. In practice, FRAND functions less like a posted price and more like a contractual framework—a set of constraints on bargaining behavior—within which firms negotiate bilaterally (and, when negotiations fail, ask courts or arbitrators to fill in the missing terms). Mark Lemley tells us that this incompleteness is not accidental. If SSOs tried to dictate prices centrally, they would risk becoming de facto price-setting bodies, raising governance and competition-law concerns, and would still face severe information problems: the value of a technology depends on context, alternatives, and portfolio interactions that are hard to observe ex-ante.

The most concrete way to see why FRAND matters is through injunction leverage. An injunction means that a court can order an implementer to stop making, using, or selling the standard-compliant product, so the SEP holder can credibly threaten to shut down sales unless a license is agreed. Mark Lemley and Carl Shapiro explain that the essence of hold-up is not merely that implementers are locked in; it is that a patent holder may credibly threaten exclusion—an injunction—unless the implementer agrees to pay. That threat can shift bargaining outcomes even when the patent’s contribution is modest, because the implementer’s downside from disruption is so large. FRAND commitments are meant to attenuate this dynamic: they signal that licensing should be available on reasonable terms, and, in many settings, they are invoked to argue that exclusionary threats should be limited when an implementer is willing to take a license.

At the same time, limiting exclusion can create a different concern: when enforcement is slow, and outcomes are uncertain, some implementers may have incentives to delay taking a license if the discounted expected cost of litigation is lower than the cost of reaching an agreement early—feeding the “hold-out” narrative. [1] However, evidence that post-judgment ongoing royalties are often higher than pre-judgment rates implies that delay can be costly when the implementer ultimately loses, strengthening incentives to settle in many cases. [2]

More generally, FRAND is a compromise between two risks: too much exclusionary leverage can let licensors capture value created by lock-in, but too little leverage can weaken payment discipline and reduce incentives to contribute technology to open standards. A vivid illustration is IEEE Standards Association’s 2015 patent-policy revision discussed by Michela Bonani, which sought to reduce FRAND uncertainty by offering more guidance on “reasonable” royalties (including a recommendation to anchor royalties to the smallest salable compliant implementation) and by tightening the conditions under which SEP holders could seek prohibitive orders (injunctions). In the aftermath, several major contributors—including Qualcomm, Ericsson, and InterDigital—reportedly refused to submit Letters of Assurance under the revised policy.

FRAND is therefore best understood as institutional plumbing rather than a magic number. It is an attempt to preserve the gains from standardization—interoperability, network effects, scale—while preventing the bargaining environment created by essentiality from turning into either opportunistic extraction by SEP holders or strategic delay by implementers.

From one SEP to many: patent thickets and royalty stacking

So far, the discussion has treated licensing mainly as if there were a single essential patent holder and a single implementer. In reality, major standards are typically covered by many patents owned by many entities. This creates what the literature calls a patent thicket: a dense landscape of overlapping and potentially blocking rights that an implementer must clear to bring a standard-compliant product to market. The thicket is not just “a lot of patents.” It is fragmented ownership plus overlap, which turns licensing from a bilateral negotiation into a multi-party coordination problem. In a previous post, we discussed one consequence of this problem: the tragedy of the anti-commons.

A visible symptom of this fragmentation is “royalty stacking.” When each SEP holder sets its royalty (or negotiates its share) independently, the sum of royalties can become large—even if each individual request appears “reasonable” when viewed in isolation. A simple illustration makes the point. Suppose a product needs licenses from ten SEP holders, and each asks for a 1–2% royalty. None of these numbers sounds outrageous on its own. But in aggregate, the implied licensing burden can quickly reach double digits, before even accounting for non-SEP IP, compliance costs, or manufacturing margins. The implementer experiences the total, not the components.

This aggregation problem has a clean economic structure. Carl Shapiro emphasizes that when multiple firms control complementary (blocking) rights, they face a “Cournot complements” problem: each licensor sets its terms to maximize its own return while ignoring the effect on total output (and thus on other licensors’ royalty revenues)—but higher aggregate royalties depress implementation and shrink the pie. The result is an inefficiently high total royalty burden relative to what a coordinated set of licensors would choose. The thicket problem is compounded by uncertainty about what is truly essential. As noted earlier, many SSOs rely on self-declaration, and essentiality is rarely verified at scale. Mathias Dewatripont and Patrick Legros describe a related strategic response as “padding”: firms have incentives to declare (or contribute) patents that are not strictly essential in order to increase their perceived share of the licensing revenue associated with the standard. Padding makes the thicket denser and increases the number of claims that must be assessed, negotiated, or litigated—further intensifying the Cournot complements dynamic.

Finally, thickets impose transaction costs that go beyond the royalty rate itself. Each additional licensor means additional search and verification, negotiation, and enforcement risk. These frictions create delays and uncertainty that can be especially burdensome for smaller entrants, who find it harder to amortize legal and licensing costs. Consistent with this view, Iain Cockburn and Megan MacGarvie find that software start-ups in thicketed markets experienced delayed initial VC funding and that the negative effects of thickets were largely driven by their impact on new entrants (small, specialized firms) rather than established incumbents. Thus, patent thickets can function as an entry tax on standards-based innovation—one that is paid not only in money, but also in time and managerial attention.

Patent pools: when bundling solves (some) coordination failures

If the problem with SEP licensing is “too many doors to knock on,” patent pools are the canonical attempt to create a one-stop shop. A patent pool is a licensing arrangement in which multiple patent holders aggregate (some of) their patents—often those claimed to be essential to a standard—and offer a single portfolio license through a common administrator. Instead of negotiating coverage separately with each SEP owner, an implementer can obtain coverage for a bundle in a single transaction.

Economically, pools can address two distinct frictions created by patent thickets. First, they reduce transaction costs: fewer counterparties to identify, fewer bilateral negotiations, and a clearer licensing pathway for entrants who lack the scale to negotiate dozens (or hundreds) of agreements. Second, pools can mitigate royalty stacking by internalizing the complements pricing externality discussed above: when complementary inputs are priced independently, each supplier adds its own markup without accounting for its effect on total adoption, leading to an overall price that is too high and output that is too low. Bundling complementary rights can reduce that inefficiency by aligning incentives around the total royalty burden rather than each individual slice, consistent with Josh Lerner and Jean Tirole’s finding that pools enhance welfare by lowering prices when they aggregate complementary technologies. Historical evidence by Ryan Lampe and Petra Moser on the 19th-century sewing machine industry also suggests that pools can function as a coordination technology in their own right—sometimes facilitating diffusion by clearing fragmented rights—though the welfare effect depends on whether the pool aggregates complements or suppresses rivalry.

But pools are not a free lunch, and the controversies around them track the same economic logic. Pools are most defensible when they bundle complementary patents—rights that are jointly needed to implement the standard. If a pool starts bundling substitute patents (covering alternative ways of doing the same thing), it can become a vehicle for suppressing competition between technologies or for raising rivals’ costs, as Richard Gilbert discusses extensively. This is why essentiality screening is central. In principle, if a pool license covers only truly essential patents, the bundle behaves like a set of complements. In practice, essentiality is hard to verify, and “padding” incentives do not disappear simply because a pool exists. The credibility and rigor of the pool’s inclusion rules—what qualifies, how it is checked, and how disputes are handled—therefore matter greatly for whether the pool reduces frictions or exacerbates them.

A final wrinkle is that markets do not always converge on one pool, and not all major SEP holders join. Sometimes there are competing pools for the same standard, offering different coverage, different rates, or different governance. For example, licensing for the HEVC/H.265 video-compression standard has been offered through multiple rival pools (including MPEG LA, Access Advance/HEVC Advance, and Velos Media). This situation can preserve some competitive pressure among licensing platforms, but it can also recreate coordination costs: implementers must compare bundles, assess gaps in coverage, and potentially take multiple pool licenses plus additional bilateral licenses. In other words, pools can simplify the “too many doors” problem—but when there are multiple pools (and not all major SEP holders join), some of the fragmentation simply reappears in a new form.

Regarding the incentives of SEP holders to join a pool, Anne Layne-Farrar and Josh Lerner show that participation is higher when portfolios are more symmetric, while pools with more founders can deter later participation because rents are shared among more claimants. Antonio Tesoriere offers a complementary mechanism: when a pool shares revenue proportional to patent counts, it is stable against opportunistic reshuffling of patents (a SEP holder strategically splitting its patent portfolio into smaller portfolios). But this stability can come at the cost of participation, because small portfolio holders may prefer to remain outside and compete—highlighting a trade-off between stable sharing rules and broad membership.

Seen through this lens, patent pools are best understood as a market-design response to the same underlying issue: once standardization turns patents into complementary bottlenecks, the licensing problem becomes as much about coordination as about pricing. Pools can improve coordination, but their effectiveness depends on governance—especially on whether the bundle truly consists of complements, and whether inclusion is disciplined by credible essentiality checks—and on its coverage.

Declared vs. true essentiality: a measurement and governance problem

At this point, a subtle but crucial distinction comes back into focus: declared essentiality is not the same thing as true essentiality. This measurement-and-verification problem is not a footnote: pool credibility, stacking estimates, and even FRAND disputes depend on knowing which patents are truly essential. In most SSOs, firms are asked (or required) to declare patents that may be essential to a standard. [3] As explained earlier, the declaration is typically based on the firm’s own assessment and is rarely verified ex-ante by the SSO. The resulting SEP landscape is therefore noisy: it contains patents that are truly indispensable, patents that are plausibly relevant but not strictly required, and patents that are strategically “in the mix” even though they are inessential. A manual check by David Goodman and Robert Myers of patents declared essential to 3G cellular standards (3GPP and 3GPP2) found that only about 21% of these patents were actually essential. Lorenz Brachtendorf and colleagues confirm this noise using automated text analysis, finding that while truly essential patents show high semantic similarity to standards, a significant portion of declared SEPs exhibit low similarity, indicative of over-declaration.

This gap matters because it directly amplifies the coordination failures discussed above. Over-declaration—whether framed as caution, strategic positioning, or “padding”—inflates the apparent thicket by increasing the number of claimed rights that implementers must evaluate. That, in turn, raises transaction costs (search, legal review, and negotiation) and can worsen royalty stacking dynamics by expanding the set of parties with a claim to a share of licensing revenue. It also complicates patent pools: the more inessential patents are presented as essential, the harder it becomes for a pool to credibly claim that its bundle consists of complements rather than a mixture that includes irrelevant or even substitute technologies. Put differently, the economics of pools depends heavily on bundling genuine complements; over-declaration makes that classification problem harder and makes outcomes more contested.

Because of these costs, essentiality checks have emerged as a partial institutional response. The basic idea is straightforward: rather than relying solely on self-declarations, an independent technical review assesses whether a given patent claim is actually required to implement the standard. A credible checking mechanism can improve market functioning in three ways. First, it reduces uncertainty for implementers by clarifying what truly needs to be licensed. Second, it strengthens the informational foundations of pooling by helping ensure that pools bundle complements (and by making inclusion rules more defensible). Third, it can discipline incentives to “pad” declarations, since the expected payoff from declaring weakly related patents falls when those patents are likely to be screened out. Florian Schuett and Chayanin Wipusanawan formally model this dynamic, showing that essentiality checks reduce wasteful litigation by eliminating information asymmetries between patent holders and implementers.

A final implication is that SSO design matters. As Benjamin Chiao and colleagues empirically show, SSOs vary widely in their disclosure requirements (e.g., the specificity of required declarations) and licensing rules, often reflecting the organization’s orientation toward technology sponsors rather than users. Those governance choices shape not only the size of the declared SEP universe, but also the severity of the downstream coordination problems—thickets, stacking, and the effectiveness (or limits) of patent pools as a remedy.

Strategic behavior around standardization: positioning IP when the stakes are high

Standardization not only coordinates technologies; it also reshapes incentives. Once a standard becomes the focal point for an industry, being “inside” the standard can create a durable revenue stream through SEP licensing. It is, therefore, unsurprising that firms invest not only in R&D but also in strategic IP positioning around the standardization process.

One recurring pattern is “just-in-time patenting”: Byeongwoo Kang and Rudi Bekkers document the opportunistic filing of patent applications shortly before or during key standardization milestones, when the technical direction of the standard is becoming clearer and when contributions can be aligned with the specification. A related tactic involves using continuations (and other claim-drafting strategies) to adjust patent claims over time so that they “read on” the evolving standard. Cesare Righi and Tim Simcoe (2023) provide strong evidence of this practice, finding that standardization leads to an 80–121% increase in continuation filings, with the vast majority of SEP continuations filed after the standard is published. The point is not necessarily that the underlying engineering is trivial; rather, patent prosecution offers flexibility in how an invention is claimed, and that flexibility becomes more valuable when a particular technical pathway is about to be widely adopted. Consistent with this narrative, Florian Berger and colleagues provide empirical evidence showing that essential patents have more claims and amendments than a set of control patents.

This strategic behavior has two economic implications. First, it reinforces why declared essentiality can be noisy: as the standard crystallizes, the set of patents that plausibly cover it can expand through claim amendments and late-stage filings, increasing uncertainty for implementers. Second, it highlights why SSO governance choices—timing rules, disclosure expectations, and treatment of late disclosures—can matter for market outcomes. In short, once standards shape rents, firms respond by investing in both the technology and the IP boundary around it.

So what? Innovation, diffusion, and policy trade-offs

Standards are a powerful institution for diffusing innovation. By creating interoperability, they expand markets, lower adoption frictions, and enable cumulative innovation: firms can build new products and complementary technologies on a shared platform rather than reinventing interfaces from scratch. In that sense, standardization is often a force multiplier—turning dispersed inventions into an ecosystem.

SEPs sit at the center of this system in an ambivalent way: they can reward genuine technical contributions that enable interoperability, but they can also create bottlenecks when licensing becomes fragmented, noisy, or used as bargaining leverage. When standardization makes certain patents unavoidable, licensing shifts from a competitive setting to a bargaining environment shaped by lock-in, imperfect information, and multi-party coordination problems. Patent thickets, royalty stacking, and litigation risk can act as a tax on entry—paid not only in money, but also in time, uncertainty, and managerial attention. The result is a genuine trade-off: the industry benefits of interoperability can coexist with frictions that slow implementation and skew participation.

The policy question, therefore, is how to design institutions so that rewards for innovation do not choke diffusion. The broader lesson is simple. Standards create economic value by coordinating markets. When patents sit at the core of that coordination, the challenge is to keep the system in the neutral zone: innovators are compensated for genuine technical contributions, and implementers can bring standard-compliant products to market without paying for lock-in. Getting there is less about slogans than about institutional details—and those details, as this literature shows, are where the real economics lives.

Refer a friend

Leave a comment

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. (2026). When standards meet patents: The economics of SEPs and FRAND. The Patentist Living Literature Review 11: 1–11. DOI: 10.2139/ssrn.6306998.

A small museum could be built from the world’s electrical plugs: three prongs, two prongs, angled pins… and the adapter industry that exists only because standards differ. The punchline is that none of these designs is intrinsically “better” in a technological sense. They are better in the only sense that matters once coordination happens: they match what everyone else uses. When a society converges on a plug type, that choice quietly dictates what appliances can be sold, what adapters get bought, and how expensive it is to switch later.

That is the economics of standards in miniature. A standard is a shared specification that makes independent components work together. When it succeeds, it creates network effects (compatibility becomes more valuable as adoption grows) and reduces transaction costs (fewer tests, less customization, and fewer negotiations). But it also creates lock-in: once the ecosystem coordinates, deviating from the standard becomes costly, even if an alternative looks superior on paper. Standards, therefore, deliver large efficiency gains—and they also redistribute bargaining power toward whoever controls the critical interfaces.

Now replace plugs with technical standards (Wi-Fi, 5G, video codecs) and replace interfaces with patents. In many industries, implementing a standard requires the use of specific patented technologies. Those patents become standard-essential, and the same coordination that makes standards valuable can turn licensing into a high-stakes bargaining problem. That is the entry point for a jargon-heavy literature: SEPs, FRAND, patent thickets, and patent pools. The post delves into this rich and complex topic, covering the institutional plumbing that keeps interoperability from turning into gridlock. The argument unfolds in four steps: why standards coordinate markets; why SEPs transform bargaining; how FRAND and pools try to manage the resulting frictions; and why noisy essentiality claims complicate everything. Part I addresses the first two topics, and Part II covers the last two.

Standards and SSOs: coordination with rules

As the opening example illustrates, a compatibility standard is a shared specification that enables products and services to interoperate. Its economic value is often less about the intrinsic technical merit of a particular design and more about the fact that many independent actors agree to build to the same interface. In markets with strong network effects, such as the telecommunications industry, that agreement can be decisive: the value of a product rises as more users adopt compatible versions and more complementary products plug into the same ecosystem.

As Joseph Farrell and Garth Saloner explain, standardization can solve a basic coordination problem. Without a focal point, industries may face excess inertia—firms delay adoption while waiting to see what others will do—or costly standards wars, where incompatible technologies split the market and force duplication of effort. By converging on a common interface, standards reduce transaction costs, support economies of scale, and make it easier for innovators to build compatible complements.

Standard-setting organizations (SSOs), such as ETSI or 3GPP, are one way industries create that focal point. They provide a process for developing, selecting, and maintaining standards—effectively turning a coordination problem into a governed, repeatable procedure. Crucially, once standards become widely adopted, the interface they define can become economically powerful: it shapes what can be sold, what must be compatible, and how costly it is to deviate later.

Interestingly, the issue of excess inertia may also occur once a standard exists, if a new, superior alternative becomes available. Farrell and Saloner also analyze whether the benefits of standardization can “trap” an industry in an obsolete or inferior standard when a superior alternative is available. They find that when there is incomplete information regarding firms’ preferences for one standard over another, “excess inertia” can occur, causing an industry to fail to adopt a new standard even when doing so would benefit all parties. This inertia arises because firms act as “fencesitters,” willing to join a “bandwagon” once it begins but unwilling to be the first to switch, for fear of being stranded with an incompatible technology if others do not follow.

From standards to patents: what makes a patent “essential”?

Once a standard is adopted, “compatibility” ceases to be a preference and becomes a requirement: products must implement the specified interface to participate in the ecosystem. This is where patents enter the story. A standard-essential patent (SEP) is a patent that must be used to comply with a standard. In the strict technical sense, it is not possible to implement the standard without infringing the patent.

Essentiality matters because it changes the market structure of licensing. Before standardization, technologies can compete: multiple approaches may perform the same function, and implementers can select among substitutes. After standardization, that competitive constraint largely disappears. If a patented technology is baked into the standard, implementers cannot “design around” it while remaining compliant. In economic terms, standardization can produce what Nobel Prize winner Oliver Williamson calls a “fundamental transformation”: an environment with alternatives ex-ante becomes one with no effective substitutes ex-post.

Two additional features amplify this shift. First, Carl Shapiro notes that essential patents are perfect complements: to implement the standard legally, an implementer may need licenses from many SEP holders. If the standard requires N separate essential patents, holding N−1 licenses is not “almost compliant”—the product still infringes. Second, essentiality is often not a clear, court-validated label. In many SSOs, patents are self-declared as potentially essential, and declarations are rarely verified. Because SSOs do not verify essentiality claims, and because firms can face significant legal and contractual consequences for failing to disclose relevant patents, firms have an incentive to declare more patents than necessary.[1]

Put these pieces together, and the bargaining environment looks very different from ordinary patent licensing. Joseph Farrell and colleagues frame standard setting as shifting bargaining from ex-ante competition to ex-post leverage. They explain that implementers make sunk, standard-specific investments (engineering, testing, certification, supply chains). Once those investments are made and the market coordinates on the standard, SEP holders can gain bargaining leverage they did not necessarily have beforehand. This is the classic backdrop for hold up: SEP owners try to charge for the value created by lock-in rather than for the ex-ante incremental value of the invention. After an implementer has sunk millions into designing a smartphone or a car to comply with 5G, switching away from the standard if the patent holder demands a high royalty is rarely a realistic option, creating lock-in.

Hold-out is a distinct concern, mostly relating to bargaining under enforcement. The basic idea is simple: in some settings, a firm can keep selling a standard-compliant product while a licensing dispute plays out, because patent cases take time. Hold-out arises when implementers delay taking a license because enforcement is costly and slow, and legal remedies may arrive only after the fact—so using the technology without a license can be a profitable bargaining strategy. In this view, the dispute is not primarily about lock-in, but about the expected cost of refusing (or postponing) agreement.

From the patent holder’s perspective, essentiality also shifts enforcement economics: once a standard is widely adopted and the patented technology is unavoidable, potential infringement can be widespread and costly to police across many implementers. FRAND commitments can therefore be seen not only as a constraint on leverage, but also as a mechanism that facilitates broad licensing and more predictable compensation.

At this point, the puzzle is clear: the very coordination that makes standards valuable can also hand SEP holders (and, in a different way, implementers) bargaining leverage that has little to do with the invention’s intrinsic merit. So what can standard setters do—without becoming price regulators—to keep licensing in a neutral zone? Part 2 covers the core responses: FRAND as an ex-ante commitment, why injunction leverage matters, and how thickets, stacking, and pools complicate the picture.

Refer a friend

Leave a comment

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. (2026). When standards meet patents: The economics of SEPs and FRAND. The Patentist Living Literature Review 11: 1–11. DOI: 10.2139/ssrn.6306998.

The previous post on The Patentist reviewed the theoretical literature on patent rights and cumulative innovation. We discovered that there is no one-size-fits-all patent system. Because technologies and industry structures differ widely across sectors, a policy choice that works well in one setting may be counterproductive in another. While economic models are helpful for clarifying the mechanisms at work and mapping out the trade-offs, they cannot, on their own, tell us how large the effects are in the real world. For that, we need data. This post moves from theory to evidence, delving into the empirical literature.

Ideally, we would randomly assign inventions to a stronger or weaker patent protection (or no protection at all), let time pass, and study differences in follow-on inventions. This experiment is, of course, not possible, and scholars have used a variety of creative methods to explore the question.

Plausibly random patent invalidation

A first line of enquiry has exploited randomness within patent offices or courts. Specifically, Alberto Galasso and Mark Schankerman have exploited the random assignment of judges to three-judge panels at the U.S. Court of Appeals for the Federal Circuit (CAFC), knowing that some judges are more lenient than others. In other words, some patents were upheld because they had a lenient panel, while others weren’t because of a tougher panel, producing random variation in outcomes that enables the study of patent rights (or the lack thereof) on cumulative innovation. The authors find that, on average, the judicial removal of patent rights leads to about a 50% increase in follow-on innovation by other parties (as measured by patent citations), providing causal evidence that patents block cumulative innovation. This blocking effect is highly heterogeneous, concentrated in complex technology fields (like computers, communications, and medical instruments).

In a similar vein, Fabian Gaessler and colleagues have used post-grant opposition at the EPO, exploiting random variation in whether the original granting examiner participates in the opposition proceedings. (They argue and find evidence that the opposed patent is more likely to be upheld if the granting examiner participates.) According to the authors, patent invalidation increases follow-on innovation by 16% on average. Crucially, the invalidation effect is heterogeneous with respect to the value of the original innovation. For low-value patents, invalidation predominantly increases the number of low-value follow-on inventions outside the patentee’s product market. For high-value patents, invalidation primarily increases the number of high-value follow-on inventions in the patentee’s product market. This finding aligns with the prediction that the blocking effect is caused by licensing failure due to excessive transaction costs for low-value patents and rent dissipation for high-value patents.

These two studies focus on a set of patents that were “important” enough for third parties to challenge the patent’s validity—precisely those for which we expect large effects on follow-on innovation. However, only a fraction of patents are challenged, and other research approaches complete the picture.

Looking beyond patentable inventions

A second line of enquiry has focused on the broader blocking potential of patents, looking beyond technological innovation. For instance, Fiona Murray and Scott Stern study whether patent rights hinder the free flow of scientific knowledge. They focus on patent-paper pairs—inventions documented in a scientific article and protected by a patent—and exploit the substantial delay between the scientific publication and the eventual patent grant. Focusing on publications in Nature Biotechnology, they compare the citation rate of the scientific paper in the pre-grant period with that in the post-grant period, using similar “non-patented” papers as a control. They find that the scientific article citation rate declines by approximately 10% to 20% after the patent grant, relative to control papers that are not subject to patent rights.

In follow-up work, Fiona Murray, Scott Stern, and coauthors examined a natural experiment in which NIH agreements reduced IP restrictions and high access costs associated with genetically engineered mice, a key research tool in life sciences. Access was facilitated, among other measures, through a simple, standardized one-page Material Transfer Agreement and an institution-wide license. The mice were further made available through a nonprofit research mouse repository, on an open-access basis. The sudden increase in openness led to a significant boost in follow-on scientific research, as evidenced by a 30% average increase in annual citation rates to the original mouse articles. The authors further find that the increase in cumulative innovation was driven by greater exploration and the entry of new researchers into the field, and suggest that IP restrictions had stifled research diversity.

Bhaven Sampat and Heidi Williams further broadened the analysis, focusing not only on scientific research but also on commercial investment. They compared follow-on research associated with genes claimed in accepted versus rejected patent applications. On average, gene patents have had no quantitatively important effects on follow-on scientific research (as measured by scientific publications) or commercial investments (as captured by clinical trials and diagnostic tests). The authors suggest that cross-firm licensing contracts in this market seem to have operated at least somewhat efficiently to mitigate the blocking effect of patents on downstream research.

Cumulative innovation… by the incumbent

The research and innovation capacity of a single firm is necessarily inferior to the combined capacity of all other firms. This is perhaps why most of the literature has focused on cumulative innovation at large. However, some studies have also examined the effect of patent rights on the patent-owning firm’s cumulative innovation. In a sense, if patent rights are important for protecting a market niche, not securing patent rights means a more competitive market… and less revenue to invest in follow-on research by the incumbent.

Deepak Hegde and colleagues look at exactly that question in the context of startups. They find that receiving a broader patent scope (because a more lenient examiner examined the patent) significantly spurs cumulative innovation for the patent holder. Specifically, each additional granted claim increases subsequent patent applications and grants by about 6% and boosts the quality of follow-on inventions, increasing citations to subsequent patents by 6% per patent.

A study by Ashish Arora and colleagues estimated the causal effect of patent protection on a firm’s commitment to further internal scientific research by analyzing sudden, unanticipated reductions in the effective strength of granted patents, termed ‘priority disclosures.’ This occurred when a competitor’s patent application, filed earlier but disclosed later, unexpectedly became prior art, weakening the focal patent’s protection. The main finding was that firms responded to this decrease in protection by reducing follow-on investments in that specific research trajectory, observing an average 16% drop in internal citations to the associated scientific work.

The role of invention disclosure

So far, we have focused on how the sudden invalidation, restriction, or narrowing of patent rights affects subsequent innovation. An equally important dimension of the patent system is disclosure: simply putting an invention’s technical details in the public domain may itself spur follow-on research. As we explained in a previous post, patent documents must disclose enough detail for examiners to assess novelty, for inventors to define their claims, and for others to avoid infringing the protected technology. But the technical information they contain can also inspire new ideas. A growing body of empirical work shows that policies that enhance disclosure or lower the cost of accessing patent documents tend to stimulate subsequent innovation.

One major quasi-experimental approach leverages the American Inventors Protection Act (AIPA) of 1999, which required U.S. patent applications to be published 18 months after filing. Before AIPA, U.S. patents became public only at grant, so the reform shortened the secrecy period by roughly 1.5 to 2 years. Deepak Hegde and co-authors compared U.S. patents affected by AIPA with their ‘twin’ European patents (which were already published after 18 months). They find that earlier disclosure significantly increased knowledge diffusion: U.S. patents received more and faster follow-on citations, with citations rising by about 15% within ten years and the average delay to first citation falling by roughly 25%. The reform also appears to have reduced duplicative R&D—overlap declined between highly similar patents, and applications were less likely to be abandoned. Finally, firms that were more exposed to lengthy pre-AIPA grant delays increased their R&D spending by almost 4% after the reform. Overall, the benefits of cheaper access to knowledge seem to have outweighed any losses from free-riding, leading to more patenting and higher R&D investment.

Stefano Baruffaldi and Markus Simeth also use AIPA to study how disclosure timing affects knowledge diffusion. They find that earlier disclosure led to a 12% increase in forward citations, indicating stronger knowledge flows, but that this effect was concentrated in technologically proximate domains rather than in geographically close ones. Their conclusion is that faster disclosure does help diffusion, but its benefits depend on inventors’ ability to screen relevant information and on their absorptive capacity.

A second research design, used by Jeff Furman, Markus Nagler, and Martin Watzinger, exploits the gradual expansion of the U.S. Patent and Trademark Depository Library system between 1975 and 1997. Before the internet, these libraries were the only places outside the USPTO where the full technical content of patents was available, so their arrival sharply lowered the cost of accessing patent information in a given region. The authors show that after a patent library opened, local patenting rose by 8–20% relative to similar regions without a library. This increase was driven by knowledge transfer: new patents in those regions increasingly used technical terms that were new locally but already common elsewhere, indicating that the libraries—and, therefore, the patent literature—broadened the knowledge base of local inventors.

A third line of evidence comes from settings where disclosure is suppressed rather than accelerated. In work with Gabriele Pellegrino and Emilio Raiteri, I examine the impact of the U.S. Invention Secrecy Act of 1951, which allows the USPTO to withhold the publication of patent applications deemed detrimental to national security. Using a matching approach, we compare ‘treated’ prior-art patents (cited by an application subject to a secrecy order) with similar ‘control’ patents not linked to secrecy. We find that when an invention is kept secret for at least three years, related patents receive about 41–45% fewer forward citations than the controls—a negative effect that persists even after the secrecy order is lifted. These results complement Dan Gross’s analysis of World War II secrecy orders, which likewise shows that longer secrecy periods depress future citations. Together, these studies provide strong evidence that the patent system’s disclosure requirement actively supports cumulative innovation and that forced secrecy has lasting costs for the development of related knowledge.

Flying half-blind

Despite all the progress in theory and empirics, we are still seeing only part of the picture. Most work treats the decision to patent as given and then studies what happens when patent rights are stronger or weaker, or when disclosure occurs earlier or later. But in practice, firms also choose whether to patent at all or to rely on secrecy. Because the main empirical strategies all condition on inventions that were patented in the first place, they tell us how tweaks to rights affect follow-on innovation for that subset of inventions, not how the patent system compares with a world in which more (or fewer) inventions stay out of the patent record altogether.

A second blind spot concerns what ‘disclosure’ really means. Existing studies mostly treat it as a binary variable: whether the information is public or not, and how early others can see it. In reality, the quality and usability of disclosure vary enormously. Claims can be clear or deliberately vague, descriptions can be technically deep or superficial, and classification and drafting practices can make it easier or harder for others to find and understand relevant patents. We know that greater access to patent documents supports cumulative innovation, but we know far less about what ‘good’ disclosure looks like from the perspective of follow-on R&D across sectors.

Finally, the empirical literature leans heavily on patent citations and other noisy proxies to measure cumulative innovation. Citation practices have changed over time, search tools and examination procedures have evolved, and firms sometimes advise inventors to avoid reading patents altogether to limit legal exposure. This means that simple counts of citations can give a distorted picture of who is building on what, and by how much. Alternative outcome measures—such as products, clinical trials, software releases, or scientific outputs—exist only in a few domains. A key task for future research is, therefore, methodological: building richer, more robust indicators of cumulative innovation that combine multiple data sources and better capture the real economic and scientific value of follow-on works.

Refer a friend

Leave a comment

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: 1–10. DOI: 10.2139/ssrn.5983416.

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.

Refer a friend

Leave a comment

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: 1–10. DOI: 10.2139/ssrn.5983416.

National treatment requires that foreign applicants receive the same legal treatment as domestic applicants. This principle is embedded in Article 2 of the 1883 Paris Convention for the Protection of Industrial Property and is reaffirmed under the 1995 TRIPS Agreement. The national treatment principle is a cornerstone of international patent law and binds the vast majority, if not all, of the world’s patent systems. Yet, despite its prevalence today, national treatment has not always been the norm. The U.S. Patent Act of 1836 set application fees of $30 for national applicants, $500 for any “subject of the King of Great Britain,” and $300 for all other persons.

Why discriminating?

Economists have studied governments’ incentives to treat foreign patentees less favorably than domestic ones. A core reason for discrimination is that governments seek to maximize national welfare rather than global welfare. A welfare-maximizing government cares about domestic producers’ profits and consumer surplus but not foreign profits [1]. Weaker protection for foreigners can raise national welfare by facilitating imitation and faster diffusion of imported technologies. In this view, stronger protection for a foreign company looks like a transfer abroad with limited domestic upside.

Difei Geng and Kamal Saggi show that trade frictions strengthen this logic [2, 3]. With tariffs, transport costs, or other wedges, export-market protection is less valuable to firms than home-market protection. If national treatment forces a single level of protection for locals and foreigners, it prevents governments from tailoring policy to those wedges and can depress innovation incentives.

Yasukazu Ichino notes another efficiency argument, related to optimal discretion. Governments would optimally give stronger protection to the firms that are more responsive to it and weaker protection to the firms that are less responsive. Requiring national treatment forces the government to provide uniform protection, thereby reducing global welfare under certain conditions.

Finally, beyond pure welfare calculus, states face strategic motives. Discrimination can operate as industrial policy—tilting the field toward domestic R&D, encouraging leapfrogging in “strategic” sectors, or selectively weakening foreign rights where local absorptive capacity is low [4].

Is there a case for not discriminating?

Discrimination is often justified as industrial policy, but the effect on domestic R&D is not uniformly positive. Reiko Aoki and Thomas Prusa show that the impact of discrimination hinges on the industrial context. In infant industries, favoring the domestic firms can raise their efforts, yet in mature industries with lucrative incumbent products, protection induces a replacement effect—current rents dull incentives to invest in the next generation. In short, the same lever that can accelerate catch-up in nascent sectors can entrench complacency where established innovative products already earn rents.

The welfare calculus also flips once we relax the frictions that motivate discrimination. Whereas Geng and Saggi show that trade wedges make asymmetric protection appealing, in low-friction or liberalized trade settings, the authors show that national treatment can raise world welfare by increasing the value of foreign markets and reducing the static consumer cost of foreign patent rights.

Strategic considerations cut both ways. Just as discrimination can serve industrial policy, open favoritism invites retaliation—tit-for-tat fees, enforcement hurdles, or litigation strategies—that may dissipate any short-run domestic gains. National treatment also advances fairness and legal parity, thereby lowering political risk premia for multinational R&D.

Finally, national treatment addresses the global failure arising from governments maximizing their own country’s welfare. It acts as a coordination device: by committing to non-discrimination, countries reduce the scope for beggar-thy-neighbor policies, raise the expected value of foreign markets ex ante, and shift private incentives toward socially efficient (global) R&D.

The many facets of discrimination

We could be forgiven for thinking that discrimination belongs to the past. On paper, it mostly does: de jure discrimination—rules that openly treat foreigners differently—has largely disappeared. For instance, Tajikistan maintained distinct fee bases for foreigners until it adhered to the national treatment principle in 2011, in line with WTO standards [5]. But in light of rising trade tensions and talk of “technology protectionism,” this issue is emerging in the empirical literature.

Outcome gaps can persist de facto. Even with identical rules on paper, foreigners may face heavier effective burdens. Sometimes, the mechanism is intentional but operates through discretion rather than statute. For example, governments can pursue “second-degree” discrimination by quietly tightening practice in government-designated strategic sectors while keeping the statute facially neutral. At other times, the mechanism is unintentional, driven, for instance, by language barriers or procedures.

Discrimination in the patent prosecution process has been measured along two margins. At the extensive margin, scholars have asked whether a right is obtained at all; at the intensive margins, they have asked how costly, how slow, and how narrow that right becomes.

A rich literature documenting differences in outcome

Early cross-office comparisons reported systematic differences by applicant residence. Masaaki Kotabe showed that the Japan Patent Office (JPO) appears to discriminate against foreign applicants with longer pendency periods than for domestic applicants. In contrast, Western patent offices appear to discriminate against foreign applicants with lower patent grant rates than for domestic applicants.

Outcomes in China have been subject to the most scrutiny. Overall, the literature finds that Chinese applicants receive grants faster than foreigners, and the bias is largest in strategic fields [4, 6, 7]. The picture concerning grant rates is mixed. While some analyses found bias against foreigners in grants, others found that aggregate grant rates were higher for foreigners due to better-quality patents compared to locals [4, 6, 8, 9].

The literature on post-grant outcomes is considerably thinner, but the existing empirical findings present a mixed picture of home-country bias. In Canada, Joseph Mai and Andrey Stoyanov found that foreign firms are about 25 percentage points less likely to prevail in court when asserting IP. In China, Christian Helmers and colleagues found that foreign companies perform as well, if not better, than local firms in patent suits. Furthermore, foreign patentees received remedies commensurate with those given to domestic patentees. However, the remedies during the study period were extremely low—probably too low to act as an effective deterrent.

Differences in outcome do not “prove” discrimination

Differences in outcome do not, by themselves, suffice to establish discrimination. Many reasons could explain discrepancies in grant lag or grant rates between foreigners and locals. For instance, foreigners may take longer to respond to the patent office’s communications, thereby increasing grant lags. Regarding the grant probability, there may be intrinsic differences in the quality of inventions between foreigners and locals. Correlation is not causation, and establishing the causal impact of being a foreigner on the outcome of the patent prosecution process is key to affirming discrimination.

The economic literature offers many studies of discrimination. Marianne Bertrand and Sendhil Mullainathan have proposed a rather neat way of establishing discrimination in the labor market. They sent fictitious resumes to help-wanted ads in U.S. newspapers. To manipulate the perceived race of the job applicant, the authors randomly assigned resumes to African-American- or White-sounding names. They found that White names received 50 percent more callbacks for interviews.

Adopting such an approach to studying discrimination in the patent system is, however, not feasible. One cannot just send the same fictitious patent applications several times, but simply changing the applicant’s origin; patent applications are only granted to inventions that are new to the world, and duplicate inventions would be quickly spotted and rejected on the grounds of novelty.

Elizabeth Webster and colleagues have proposed a clever test of discrimination. They have used a matched sample of patent applications granted by the USPTO and filed at the JPO and the European Patent Office (EPO). In other words, they observe patent applications for the same invention in three jurisdictions (so-called ‘twin patents’). They found that Japanese firms were less likely than European firms to have their patent applications granted in Europe (and vice versa at the JPO). Because this method compares the fate of the same inventions across different jurisdictions, it alleviates concerns about systematic differences in the quality of inventions by foreigners and locals. A follow-on study using more recent data and a larger number of patent offices found similar results.

However, this method has its limits. Although it compares the same invention by the same applicant in different jurisdictions, the applicant may behave differently across jurisdictions. For example, it may hire more qualified patent attorneys in its domestic market or know best how to navigate the intricacies of its national patent system.

The story does not end here

To address these concerns, empirical research has recently delved into the details of the examination process. Bruno van Pottelsberghe and colleagues have tracked several metrics related to the intensity of the examination process performed by patent examiners at the USPTO, EPO, and JPO. They confirm that international applicants face a lower grant probability than locals. However, when analyzing the underlying internal processes, researchers found no convincing evidence of discrimination. This leads to the conclusion that the disparity in grant outcomes is likely driven by economic forces or heterogeneous applicant behavior, such as a reduced willingness or capacity of applicants to invest time and resources to maintain their patent application active “at all costs” in foreign jurisdictions.

Some work with a former student of mine focused precisely on this question, seeking to measure the applicant’s level of effort at the USPTO. Our model included the number of final and non-final rejections received, as well as the total count of transactions initiated by the applicant during the prosecution phase. A higher count of rejections or a greater number of applicant transactions was taken as evidence of greater effort by the assignee, reflecting higher costs and a greater willingness to fight for the patent. These transaction data revealed that foreigners generated fewer rejections and fewer examiner and applicant transactions than locals, consistently suggesting that they put less effort into the prosecution process. In other words, the lower grant rates and greater scope reductions for foreigners are partly explained by the fact that they concede more or abandon earlier than domestic applicants.

A clear case of discrimination?

Together with Emilio Raiteri, we have exploited the twin-patent approach within a statistical framework that allows controlling for a large number of variables that could affect the grant rate, such as the “quality” of the patent attorney firm. Interestingly, as we start controlling for more and more variables, traces of foreign discrimination in China vanish. They vanish for all but one type of patent application: those in technological areas the Chinese government considers strategic. We also find that these applications are subject to more amendments during the examination phase.

In a follow-up study, joined by Rudi Bekkers, we provide additional evidence of strategic discrimination against foreign innovators in China. We utilized a novel identification strategy leveraging standard-essential patents (SEPs): we tracked whether foreign SEP applications were publicly disclosed as essential to a standard (such as 3G/4G mobile communication standards) before or after substantive examination began. We find that foreign SEPs declared as such before examination took place were approximately nine percentage points less likely to be granted in China than those disclosed after substantive examination. Furthermore, these early-disclosed foreign applications faced a one-year prosecution delay and experienced greater scope reduction than later-disclosed applications.

Looking forward

The emerging picture is nuanced. Outcome differences between foreign and local applicants are real in many settings, but they do not by themselves establish discrimination; composition and behavior matter, and credible identification requires designs that hold constant invention quality and other “unobservables.” Where stronger designs exist, evidence of systematic bias often weakens, yet targeted disadvantages can remain—especially in government-prioritized fields or when procedural frictions (such as translation and timeliness) bite unevenly.

Collectively, these studies have only scratched the surface of the discriminatory treatment of foreigners. We need to dig more into the data to understand the root causes of the apparent discrimination. Scholars have proposed new statistical methods and examined various stages of the patent prosecution process. We need more such methods (or ‘identification strategies’ in economists’ jargon) and more metrics to identify the channels through which discrimination might occur. Finally, the literature has focused primarily on the largest offices because they provide high-quality data. But the more opaque offices might also be those discriminating the most, calling for more scrutiny.

Refer a friend

Leave a comment

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). National Treatment: Principle or Practice? The Patentist Living Literature Review 9: 1–6. DOI: 10.2139/ssrn.5702262.

Few things matter more for innovation and economic growth than the way ideas spread. Economists have long emphasized that when new ideas spill across organizational or regional boundaries, they fuel the progress of others and amplify the returns to discovery. But while the importance of these “spillovers” is widely accepted, measuring them is far from straightforward.

What are knowledge spillovers?

In an earlier post, we explained that knowledge is a special kind of good: it is non-rival (many can use it simultaneously) and often non-excludable (making it difficult to prevent others from using it). Unlike a barrel of oil, knowledge doesn’t run out when someone uses it. Once created, it can spread widely—across universities, firms, researchers, and inventors.

Economists generally refer to “knowledge flows” as the movement of ideas from one actor to another. These flows can be paid for (through a licensing contract, for instance) or not. Knowledge spillovers are the subset of flows for which the originator does not capture the full value. Spillovers can happen deliberately, as when scientists share results at a conference, or indirectly, for instance, when competitors learn from a patent disclosure. In practice, it is nearly impossible for innovators to appropriate all the value of what they create; part of the value “spills” beyond the original creator.

Why do we care about knowledge spillovers?

Knowledge spillovers matter for two main reasons. First, they imply that the social benefits of innovation exceed the private ones. A company may invent a new tool, material, or method, but suppliers, rivals, or even customers can benefit far beyond what the firm itself earns. Take semiconductors: when Intel develops a new chip design or manufacturing process, competitors study the patents, reverse-engineer the products, and adapt some of the underlying techniques. Even if Intel fiercely guards its IP, part of the knowledge inevitably spills over to the rest of the industry. The result is faster diffusion of better technology into laptops, smartphones, and data centers—benefits to society that Intel cannot fully capture.

Second, spillovers don’t just spread ideas more widely; they make innovation cumulative. Each generation of ideas becomes a platform for the next. The smartphone, for example, combined breakthroughs in semiconductors, wireless communication, touchscreens, and software—each of which built upon earlier advances that had spread across firms, universities, and industries. This cumulative process is what economists mean when they say innovation is “standing on the shoulders of giants.” But the metaphor can be sharpened: every time knowledge spills over, it does not just let others climb higher—it makes the giants taller, raising the platform for all future innovators.

In search of knowledge spillovers

Paul Krugman remarked in his 1992 book that knowledge flows “are invisible; they leave no paper trail by which they may be measured and tracked.” A frustrating observation, given their central role in economic growth. Just a year later, Adam Jaffe, Manuel Trajtenberg, and Rebecca Henderson countered: “knowledge flows do sometimes leave a paper trail, in the form of citations in patents.”

The central question of their paper was simple: are knowledge flows more likely to occur locally? To find out, they compared the location of citing and cited patents against a neutral baseline. Each citation was matched with a “control patent” from the same technology class and year but without a citation link. If citing inventors were unusually likely to be in the same place, it would suggest that geography matters for knowledge flows.

The results were striking. Citing patents were far more likely to be located near the cited ones than were the controls—three times more likely at the state level, and up to ten times more likely within a metropolitan area. The effect was strongest soon after the original patent was filed and diminished over time, consistent with the idea that knowledge starts local before spreading more widely.

Because spillovers are a subset of knowledge flows, much of the literature has focused on testing whether citations track flows in general. Only then can we return to the sharper question: do they capture the special case of spillovers?

Is the trail reliable?

The finding that patent citations capture knowledge flows became a cornerstone of the literature, sparking decades of debate, replication, and refinement. Criticism, however, has been twofold: methodological and interpretative.

On the methodology, the sharpest critique came from Peter Thompson and Melanie Fox-Kean, who argued that Jaffe and colleagues’ matching procedure was too coarse. Using finer technology subclass controls, the localization effect largely disappeared. Jaffe and colleagues countered that such fine matching reduces sample size and risks excluding real spillovers that cross technological boundaries. Later work by Yasusada Murata and co-authors, using actual distance measures, again found evidence of localization even under strict controls.

In other work, Peter Thompson sidestepped the matching problem entirely by exploiting newly released USPTO data that distinguish citations added by applicants from those added by examiners. He showed that applicant-added citations are far more localized than examiner-added ones—a result that Juan Alcácer and Michelle Gittelman somewhat tempered, but that was also found by Paola Criscuolo and Bart Verspagen at the European Patent Office.

Overall, these studies strengthened the case that citations can capture knowledge flows, but they also revealed a key weakness of citation data: because examiners (and patent attorneys) add so many references, the overall citation record dilutes the true signal of knowledge flows.

Errors of commission: citations without knowledge flows

The results above suggest the presence of “false positives,” namely citations that do not reflect knowledge flows. The prevalence of these “errors of commission” was quantified by Adam Jaffe, Manuel Trajtenberg, and Michael Fogarty using a survey of inventors. Each inventor was asked about two patents their own patent had cited, plus a placebo patent from the same year and technological class that had not been cited. A parallel survey asked cited inventors whether they had communicated with the citing inventor and whether the later invention appeared to build on their work.

The results painted a mixed picture. On the positive side, inventors were far more familiar with the patents they had cited than with the placebos, and around 40 percent reported learning from the cited invention during their work. On the negative side, roughly one-third of inventors were entirely unaware of the patents they had cited, and perhaps half of all citations did not correspond to any real knowledge flow. Many of these were likely inserted by attorneys or examiners. (The survey was done at a time when the USPTO did not release examiner vs. application citations.) The conclusion: citations can indicate knowledge flows, but many also reflect something else.

Errors of omission: knowledge flows without citations

If some citations occur without knowledge flows, the opposite is also true: many genuine flows leave no trace in the citation record. A striking example comes from Andrew Nelson’s study of the Cohen–Boyer recombinant DNA patents. Stanford University’s licensing office signed 464 agreements for the technology, yet only 135 organizations held patents citing the original rDNA patents, and just 55 overlapped with the licensees. In other words, citations missed 88 percent of the organizations that were actively building on the technology.

Michael Roach and Wes Cohen reached a similar conclusion by comparing citation data with survey reports from R&D managers. They found that citations systematically miss knowledge flows occurring through contract-based channels (consulting, sponsored research, collaborative projects) that rely on tacit, face-to-face communication. Citations also understate the role of public science in basic research, where outputs may not be patentable in the first place. The conclusion: knowledge flows can be captured by citations, but many also leave no citation trail.

Spillovers… or knowledge flows?

Up to this point, we have seen that citations can capture knowledge flows—but the original ambition of the literature went further. From the start, Jaffe, Trajtenberg, and Henderson interpreted local citation patterns not just as flows, but as evidence of knowledge spillovers. Their logic was straightforward: by excluding self-citations, they focused solely on links between independent organizations. The excess local clustering of cross-firm citations was taken as proof that citations do not just record knowledge flows, but specifically reflect spillovers—knowledge benefits that spill beyond the boundaries of the firm.

However, these localized citations may have been driven by market transactions. As a case in point, David Mowery and Arvids Ziedonis show that technology licenses—clear market-mediated transfers—are more geographically localized than patent citations. This suggests that proximity may matter just as much, if not more, for contracting and collaboration than for unpriced spillovers. Thus, localization in citations alone does not prove the existence of spillovers.

Other scholars argued that localized citations might simply reflect shared local environments: inventors clustered in the same region naturally work on similar problems and therefore cite one another. Ashish Arora, Sharon Belenzon, and Honggi Lee formalized this critique using “citation reversals”—cases where a citing patent has an earlier priority date than the cited patent, making sequential knowledge transmission impossible. If citations really captured spillovers, then standard citations (where transmission is possible) should be more geographically concentrated than reversals (where it is not). Yet both types of citations were similarly localized, suggesting that proximity in the citation record may reflect the geography of related inventive activity rather than knowledge spillovers.

So, do localized citations capture spillovers or just flows?

On the surface, these critiques suggest that localization in citations may have little to do with true spillovers. Yet some research points in the opposite direction: mobility, social ties, and personal presence show that genuine spillovers do exist—and that citations, read carefully, can reveal them.

Paul Almeida and Bruce Kogut showed that regions with high intra-regional labor mobility, like Silicon Valley, exhibit stronger localized knowledge flows, suggesting that ideas literally “move” with people. Ajay Agrawal, Iain Cockburn, and John McHale demonstrated that mobile inventors continue to generate citations back to their prior location, evidence that enduring ties facilitate diffusion even after departure. Jasjit Singh showed that inventors with direct collaborative ties are four times more likely to cite each other than unconnected inventors, and Stefano Breschi and Francesco Lissoni confirmed that once co-inventor networks and mobility are controlled for, the residual role of geography shrinks dramatically. Finally, Benjamin Balsmeier, Lee Fleming, and Sonja Lück provided quasi-experimental evidence on the importance of inventors as vectors of spillovers. When an inventor of a co-invented patent died between application and grant, local citations to that patent fell by about 25 percent within 20 miles of the deceased inventor’s location relative to the surviving co-inventors.

Together, these studies show that geography matters not because knowledge is “in the air,” but because it travels through people, relationships, and tacit interactions. That is exactly what we should expect if localized citations capture genuine spillovers rather than just parallel invention and shared local environment.

Concluding remarks

In the end, patent citations are neither a perfect lens on knowledge flows (and spillovers) nor an empty signal. They capture part of the picture, but with both noise and blind spots: many citations do not reflect true knowledge transfer, while many genuine flows leave no trace. What the literature now makes clear is that spillovers exist, they are localized, and they travel through people, networks, and institutions. Citations provide a valuable—if imperfect—paper trail, one that must be read critically and complemented with other evidence.

Refer a friend

Leave a comment

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 citations: Tracing spillovers or chasing shadows? The Patentist Living Literature Review 8: 1–6. DOI: TBC.

Patent documents cite earlier patents, much like scientific articles reference prior research. Scholars and practitioners routinely use the number of citations a patent receives as a proxy for invention importance, based on the assumption that patents attracting more citations are more valuable. This post reviews what the evidence says about that assumption.

An intuitive idea, but…

The intuition is simple and powerful: valuable ideas attract attention. Nobel Prize laureates, for instance, almost invariably have highly cited papers to their name. Citation counts, in this view, provide a tidy signal of impact.

However, while scientific citations are widely understood as a way to give credit—recognizing a researcher’s contribution to the advancement of knowledge—patent citations serve an entirely different purpose, rooted in legal procedure rather than scholarly acknowledgment. Citations in patents are part of the legal assessment of patentability, disclosing prior art that helps examiners determine whether an invention is novel and non-obvious. This legal function blurs the use of patent citations as signals of value—at least in the way we interpret citations in science.

And then there’s the question of value itself, as discussed in a post on patent quality. What exactly do we mean by a “valuable” patent? Are we talking about the value of the underlying invention, or the legal right it confers? Are we referring to technological value or economic value? And if economic, is it private value to the firm, or social value to the world?

Not all patent citations are the same

Before diving into the empirical literature, a word of caution about patent citations is warranted. Virtually all studies rely on so-called front-page citations—the references printed on the first page of granted patent documents. These citations are standardized and widely available, making them a convenient data source.

However, the patenting process gives rise to other types of citations. Depending on the jurisdiction, these include references listed in Information Disclosure Statements (IDS) submitted by applicants, prior art discussed in Office actions issued by examiners, and citations that may appear in the body of the patent text. Additional citations may also emerge in post-grant proceedings, such as oppositions. These alternative sources are less frequently used in research but may offer different insights into how patents engage with prior art.

The first validation study

We owe the first validation study of patent citations as a signal of value to information scientist Francis Narin and colleagues at CHI Research, one of the earliest firms in patent analytics. Observing the use of citation-based metrics in science, they set out to test whether patent citations could similarly be used to identify important patents.

Their approach was elegantly simple. They identified patents behind products that received the IR-100 award between 1968 and 1974. This award, granted by the Industrial Research and Development magazine, recognizes the most significant new technical products of the year. For comparison, they selected a set of 100 “control” patents randomly drawn to match the time distribution of the award-winning group. The result? The award-related patents were cited more than twice as often as the controls. The authors concluded that “citation analysis will be useful in identifying important patents.”

However, the study was vague about what exactly was being measured. The IR-100 award focused on “products of certified industrial or commercial importance,” which may reflect the underlying technical merit or the economic potential of the invention, or a combination of these two dimensions.

Citations and the “technical merit” of the invention

A decade later, a follow-up study by the same group looked specifically at technological importance. The authors examined 77 patents related to silver halide technology granted to Eastman Kodak between 1982 and 1983. They asked 20 senior Kodak scientists to rate how much each patent had advanced the state of the art in the field. The results showed a significant relationship between expert ratings and citation counts.

Interestingly, the signal was strongest at the top of the distribution: highly cited patents were more likely to be rated as “important.” At the lower end—between patents with zero or only a few citations—there was little discernible difference in expert ratings.

A handful of other studies

Only a handful of additional validation studies have examined patent citations as a proxy for the technological importance of an invention.

One recent example is a study by Arianna Martinelli and colleagues, which follows the tradition of linking citations to external markers of innovation quality. The authors focus on patents protecting products that received a UK Queen’s Award for Innovation. Specifically, they identified 1,468 patent families filed between 1976 and 2013, covering 401 award-winning innovations. The number of citations these patents received was a strong predictor of award status.

Another notable example comes from Dirk Czarnitzki and colleagues, who also rely on expert judgment—but with a twist. They focus on 188 “wacky” patents, collected by an employee of the World Intellectual Property Organization. These patents were selected for their futile or absurd nature, typically involving only a marginal inventive step or barely satisfying the non-obviousness criterion. The authors then compared the citation rates of the wacky patents to those of a control group filed in the same year and technology class. The result: wacky patents received significantly fewer citations.

Finally, Petra Moser and colleagues adopted a distinct approach. Their study links patents to objective measures of improvement in the quality of the patented invention, using hybrid corn as their empirical setting. Unlike studies based on expert opinion, improvements in agricultural performance can be directly observed in field trial data, by comparing a patented hybrid’s yield to that of the best-performing comparison hybrids. They analyzed 315 patent–hybrid corn pairs between 1985 and 2002 and found that citations strongly correlate with improvements in yield, although this relationship was driven primarily by applicant-added citations.

Citations and the “economic merit” of the invention

In the early 1990s, Manuel Trajtenberg offered a different kind of validation—one grounded in economics. Using the rise of Computed Tomography (CT) scanners in the U.S. market between 1973 and 1982, he documented a strong link between citation-weighted patent counts and the social value of innovation, measured in terms of both consumer surplus and producer profits.

The intuition behind his approach is as follows: imagine that all available CT scanners can be described using a handful of key attributes, say, resolution, scan time, and safety features. Over time, innovation leads to the introduction of new scanner models and improvements in existing ones. These product changes expand consumer choice and generate value, either by improving quality or lowering prices. Using tools from industrial organization, Trajtenberg quantifies that value based on detailed hospital purchase data and scanner characteristics. He then shows that yearly improvements in social value metrics correlate with a citation-based count of CT scanner patents. While the statistical analysis is impressive, the reliance on yearly aggregated data means that the evidence ultimately rests on just ten data points, calling for more validation studies.

A flurry of studies has since examined the economic value of patents. These studies draw on a range of data sources to estimate value. Still, their focus is primarily on the private economic value of an invention—that is, the monetary benefits that an invention generates for its owner.

Survey-based measures of value

One group of studies has collected direct, subjective estimates of patent value from inventors or patent-holding firms through surveys.

Dietmar Harhoff and colleagues regressed the number of citations received by 192 U.S. patents filed in the late 1970s on patent owners’ retrospective estimates of private value [1]. The key survey question asked inventors to estimate how much their patent would have been worth—in hindsight—knowing what they now knew about the profits the invention ultimately generated. The study finds a significant positive correlation between citation counts and these retrospective value estimates.

A similar approach was used in a large-scale survey of inventors by Alfonso Gambardella and colleagues. Drawing on responses from the PatVal-EU survey—covering more than 9,000 European patents with priority dates from 1993 to 1997—the authors tested the relationship between citations and inventor-assessed value. They find that citations are the strongest predictor of value among several indicators (including patent family size). However, their statistical model explains only 2.7% of the variation in private value. This low number suggests that while citations carry some signal, they are a noisy indicator of value.

Exploiting auction price data

While survey-based measures offer a rare window into how inventors themselves assess patent value, they suffer from important limitations—including recall bias, selective responses, and inconsistent interpretations. These concerns have motivated researchers to explore alternative, more objective measures.

Some scholars have investigated the determinants of patent value in an auction setting [2]. Auction prices provide a directly observed measure of private economic value, based on what buyers are actually willing to pay.

The authors analyzed a dataset of 402 patents grouped into 99 lots from Ocean Tomo’s live patent auctions held in 2006, including 49 sold and 50 unsold lots. They found a positive relationship between citation counts and auction value. Each additional citation was associated with approximately $10,600 in additional price per patent. That said, patent auctions remain a relatively thin and selective market—not all patents are auctioned, and those that are may differ systematically from the broader population. As such, while auction prices offer a cleaner measure of value than surveys, they come with their own selection and generalizability concerns.

Note that data from Ocean Tomo’s auctions have also been used by other scholars, who similarly find a positive correlation between the auction price and the count of citations [3, 4, 5].

Stock market data

Another approach to estimating patent value draws on stock market data. A classic study in this stream is that of Bronwyn Hall, Adam Jaffe, and Manuel Trajtenberg, who exploit Tobin’s q, a widely used measure in finance that compares a firm’s market value to the replacement cost of its assets (that is, what it would cost to rebuild the firm from scratch). A high Tobin’s q suggests that investors believe the firm owns valuable intangible assets—like knowledge, brand, or intellectual property.

The authors link more than half a million U.S. patents to publicly-traded firms, constructing three measures of firms’ “knowledge stock”: R&D spending, patent counts, and citation-weighted patents. They find that all three indicators explain variation in Tobin’s q, but citations per patent turn out to be particularly informative. Firms with more highly cited patents tend to be valued more highly by the market, even after accounting for their R&D spending and number of patents. In other words, citations help distinguish high-value patents from run-of-the-mill ones, and this distinction shows up in firms’ stock prices.

Leonid Kogan and colleagues offer a more direct use of stock price data. Their method estimates the private economic value of individual patents by analyzing how the market reacts in a narrow window around each patent’s grant date. They find a strong and positive relationship between this market-based value estimate and citation counts.

A note of caution is warranted, however. While stock market data offer a powerful lens for assessing patent value, they come with limitations. Most importantly, stock prices reflect investor expectations, which may be influenced by hype, noise, or factors unrelated to the patent’s actual economic returns. Furthermore, this assessment occurs early in the patent’s life, and the value may not be realized.

Patent renewal data

Another empirical strategy for validating patent citations as a proxy for private value draws on patent renewal records. Maintaining a patent over its full term requires periodic fee payments, and inventors are expected to make these payments only when the expected benefits outweigh the costs. Renewal behavior thus offers a revealed-preference indicator of a patent’s private value. James Bessen applies this method to estimate the value of U.S. patents and finds that highly cited patents tend to be more valuable. However, patent citations explain little variance in value.

Renewal-based value estimates capture only the lower bound of expected value—enough to justify paying the fee—but not the full distribution of returns. Moreover, some patents may be renewed for strategic reasons unrelated to immediate commercial value, such as blocking competitors or signaling capability.

Administrative records

Finally, Jesse Giummo explores a unique source of data for patent valuation: inventor compensation records mandated by the German Employee Invention Act of 1957 [6]. This law requires firms to compensate employees for patented inventions they create in the course of their work, based on the economic benefit derived from the invention’s use.

Using these records, Giummo assembles a dataset of over 1,100 U.S. patents of German origin filed between 1977 and 1982. He finds a positive correlation between citation counts and inventor compensation. This finding supports the idea that citations capture the economic relevance of patented inventions, even when value is measured in terms of use. The study thus adds further evidence that citation data reflect meaningful differences in patent value.

Concluding remarks

The evidence reviewed here suggests that patent citations can serve as a meaningful proxy for patent importance. Their appeal lies in their accessibility and scalability: they are publicly available, quantifiable, and cover the entire patent universe. Yet, as with any proxy, they should be used with care.

Validation studies span three conceptions of value: technical merit, private economic value, and social value. The evidence is uneven across these categories. For technical merit, the number of studies is limited and often based on small expert-rated samples, yet they consistently find a positive correlation with citations. For private economic value, the literature is more extensive and methodologically diverse, consistently finding that citations correlate with private value. However, these correlations explain only a modest share of the overall variation in patent value. For social value—the contribution of inventions to societal welfare—there is only one major study, highlighting a clear need for more research.

Importantly, not all citations are equal. Most analyses rely on front-page citations, but other sources—such as citations from Information Disclosure Statements or examiner Office actions—may provide distinct insights. Citation practices also vary across jurisdictions, complicating cross-country comparisons. And citations may reflect legal tactics or spillovers rather than patent value per se.

Refer a friend

Leave a comment

Please cite this post as follows:

de Rassenfosse, G. (2025). Do more citations mean better patents? The Patentist Living Literature Review 7: 1–6. DOI: 10.48550/arXiv.2508.20503.

International trade rarely leaves the headlines. Politicians fret over bilateral deficits, pundits debate tariff hikes and supply-chain “re-shoring,” while analysts track shipping costs and trade volumes. Yet trade costs, including tariffs, are only part of the story. Trade flows also pivot on structural fundamentals—geography and distance, relative factor endowments, technological capabilities, consumer preferences, and firms’ competitiveness—as well as on historical and cultural ties such as common languages or colonial legacies. Overlaying these fundamentals is an often-invisible architecture of rules and institutions that determine who can sell what to whom, and on what terms.

A central piece of that architecture is intellectual property (IP) protection. The rise of knowledge-intensive manufactures—from biotech reagents and advanced semiconductors to design-rich consumer electronics—and the explosive growth of digitally delivered products and services have made legal rights governing copying, licensing, and data access critical determinants of market access. Since the World Trade Organization’s 1995 Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), and its “TRIPS-plus” upgrades in mega-regional agreements, IP provisions have moved from the margins of trade negotiations to their core. But how exactly does IP affect global commerce?

Countervailing forces

The overall theoretical framework for studying the effect of intellectual property rights (IPRs) on international trade primarily revolves around two countervailing forces, as explained by academics Keith Maskus and Mohan Penubarti in a widely-cited 1995 paper.

A first force is the market power effect, which lowers the quantity exported. Stronger IPRs mean that it is harder for rivals in the destination market to imitate the technology sold by the foreign firm. In economic terms, the supply of close substitutes shrinks, and the foreign firm can exercise increased monopolistic power. This leads the firm to charge a higher price and ship fewer units, lowering export volume even though per-unit profits rise.

A second, opposing force is the market expansion effect, which boosts exports. Tighter enforcement lowers the risk that local imitators will free-ride on the exporter’s technology or brand. This increases the demand for the original products in several ways: by limiting the number of substitute products available and by increasing the value of the original product. With imitation less likely, the producer finds it worthwhile to offer robust warranties, after-sales service, and complementary inputs (such as software updates and spare parts), thereby raising customers’ willingness to pay. In economic terms, this result is an outward shift of the residual demand curve—at every possible price, the exporter can sell a larger quantity.

The net impact on trade flows depends on which force dominates, but some broad theoretical generalizations can be made. For example, the market power effect is likely to be most visible in situations with strong parallel-trade restrictions, so arbitrage—where products sold at lower prices in one market are resold in another where prices are higher—cannot undo the price increase. The market expansion effect may dominate in sectors where brand integrity and after-sales service are crucial, such as the medical device and luxury apparel industries.

The effects described above are “static,” in the sense that they consider how strengthening IPRs affects the existing set of products. However, there are also “dynamic” effects, which lead firms to improve their products (quality upgrading) or introduce new ones (product differentiation). The mechanism works as follows. Stronger IPRs reduce the risk of imitation by domestic firms in importing countries, which in turn increases the expected returns on innovations. This increased security incentivizes exporting firms to invest more in R&D to create new products and improve existing ones. With better protection for their intellectual assets, firms are more willing to bring differentiated and higher-quality products to market. This can involve climbing the quality ladder (vertical differentiation) or creating new varieties that set them apart from rivals (horizontal differentiation).

Measuring the effect is challenging

Quantifying the effect of IP on international trade is no small matter, for scholars must address several empirical challenges. First, measuring the level of IP protection is notoriously difficult. We have discussed in a previous post the various indices that have been proposed, capturing both the law on the books and the law in practice. These indices, despite providing significant structure, may still lack the nuances necessary to fully grasp the various facets of the IP regime that affect trade. Many indices focus solely on patents, overlooking other forms of IPRs, such as trademarks, copyrights, and trade secrets. There is also a lack of comprehensive, time-varying data on certain IP-related measures (e.g., parallel trade, pricing regulations), making fine-grained analyses challenging.

Second, and at least as challenging, is establishing causality. Do stronger IPRs cause trade increases, or do expanding trade flows and rising economic development (or the prospect thereof) persuade governments to tighten IP protection? Trade flows are shaped by a host of hard-to-observe forces (e.g., macroeconomic reforms, institutional quality, geopolitical risks) that often correlate with changes in IPRs, and teasing out the independent effect of IP on international trade is difficult. The problem is compounded by policy bundling: IP upgrades rarely arrive in isolation. Modern “deep” trade agreements bundle IP chapters with liberalisation of services, investment, and public procurement. It is thus challenging to separate the effect of IP provisions from other trade-related issues within the same trade agreements. Unless researchers can disentangle these concurrent reforms, any econometric coefficient labelled “IP strength” is liable to absorb the impact of the whole package rather than the IP provisions themselves.

An evolving empirical literature

Despite these hurdles, a sizeable body of empirical work has emerged over the past three decades that quantifies the impact of IP on trade. These studies differ in several ways, resulting in distinct strands of evidence. A first dividing line runs between studies that examine general shifts in IP strength—typically proxied by the Ginarte-Park index—and those that exploit specific reforms, such as TRIPS. The early literature primarily belongs to the former camp, where causal claims are more difficult to sustain for the reasons outlined above. More recent research treats policy events like TRIPS as quasi-natural experiments: by inducing many developing countries to tighten their IP regimes, the agreement provides an exogenous shock that helps isolate the effect of stronger IP from other confounding factors. A newer strand goes further, matching trade flows to the actual patenting behaviour of exporting firms, thereby complementing country-level IP indicators with actual use of IP by firms.

A second line of demarcation is the unit of analysis. Work has progressed from aggregate trade flows to increasingly granular data—industry-, firm-, and even product-class level. Greater disaggregation reveals how IP effects differ across sectors (high-tech vs. low-tech, pharmaceuticals vs. agriculture) and across destinations with varying imitation risk. It also allows researchers to unpack the mechanisms at play, distinguishing, for example, between changes at the extensive margin (new product varieties or quality upgrades) and the intensive margin (export volumes of existing products).

Yes, IP matters. But how depends on context

Early studies using aggregate data yielded mixed results, sometimes positive, negative, or insignificant, depending on factors such as the level of patent protection, country types, or goods traded. Some seminal works, like Maskus and Penubarti’s 1995 paper, have found that stronger patent protection generally has a positive impact on bilateral manufacturing imports, particularly in developing economies, indicating a market expansion effect. Economist Pamela Smith introduced the concept that the impact depends on the importing country’s imitative ability: market expansion effects dominated for countries with high imitation threat, while market power effects were more prevalent for those with weak imitation threat [1].

With the implementation of TRIPS, studies increasingly treated this as an exogenous policy shock, allowing for stronger causal inferences. Research has found that the implementation of TRIPS is associated with increased trade in knowledge-intensive goods, particularly exports by developed countries to developing countries, and in specific sectors such as ICT and biopharmaceuticals [2]. Another study similarly found that stronger patent rights in developing countries, due to TRIPS, increased developed countries’ exports, especially in patent-sensitive industries such as medicinal and pharmaceutical products. This increase was driven by an expansion in the quantity of exports, rather than higher prices, suggesting that it did not limit access to innovative products [3].

The more recent literature disaggregates the quantity effect along the extensive margin (the variety of goods) and the intensive margin (the volume of existing goods). There appears to be a consensus that stronger IPRs primarily boost trade through the extensive margin, by increasing the variety of products traded, rather than the intensive margin [4, 5]. This implies that IPRs encourage firms to introduce new products into foreign markets.

Beyond shipments at the border

The effect of IP on international trade also manifests through different channels that may take a longer time to materialize but can have a significant impact on trade flows. Stronger IPRs can influence a firm’s choice of how to serve a foreign market, potentially shifting from exporting to licensing or foreign direct investments (FDI), as discussed by Lee Branstetter and colleagues in a 2006 paper.

If patents and trade secrets are well enforced in the destination country, the foreign innovator may choose to license its know-how to a local producer rather than exporting the finished good. Exports of physical goods can fall while cross-border royalty flows rise. Similarly, stronger IP protection lowers the risk that proprietary know-how leaks out of a foreign factory. Firms are, therefore, more willing to build or buy production facilities abroad, especially for R&D-intensive or design-heavy stages of the supply chain. Trade in intermediate parts (i.e., components sent to the affiliate) can increase, while finished-goods exports can rise if final assembly now takes place locally.

Put simply, strong IP can persuade a company to “make or license there” rather than “ship from here.” Customs data alone might show smaller export volumes, but that needn’t signal lost business: the economic activity has merely shifted into royalties, intra-firm trade, and local production.

Looking ahead

Digital and other remotely-deliverable services are a growing component of global trade. Unlike containerized goods, these flows depend almost entirely on a web of intangible rights, including copyrights in software and creative content, patents on algorithms, database rights, and trade-secret protection for proprietary code and know-how. Yet the empirical base lags far behind their economic weight. Streaming royalties collected through app stores, SaaS subscriptions paid to cloud providers, in-app micro-payments, and cross-border data-processing fees rarely appear—at least not cleanly—in customs or balance-of-payments statistics. Until those statistical blind spots are filled, our understanding of how IPRs shape twenty-first-century trade will remain fragmentary. Strengthening measurement, therefore, appears as one of the most pressing tasks for researchers and policymakers seeking evidence-based guidance on the next generation of IP and digital trade agreements.

Refer a friend

Leave a comment

Please cite this post as follows:

de Rassenfosse, G. (2025). International trade and intellectual property. The Patentist Living Literature Review 6: 1–5. DOI: 10.48550/arXiv.2506.18929.

As discussed in a previous post, patent “quality” has many facets. One is compliance with statutory requirements. Do patent offices grant and refuse applications in line with their own rules, or do they make mistakes? Under this lens, a “low-quality” patent is one unlikely to survive a court challenge. But perfect compliance is not the whole story. A patent system could have a bar so low that the office issues patents on every trivial invention. The office would never make a “mistake,” yet such a low threshold is hardly a mark of true quality.

That discussion brings us to a second, equally important dimension: rigor. How high is the bar? Are patents awarded only for genuine advances—novel, non-obvious contributions—or is the threshold so low that almost anything under the sun qualifies as “inventive”? To capture these differences, scholars have devised indices measuring the strength of patent systems. This post reviews some of the major indices and explores what they reveal about international variation in patent systems.

Ginarte and Park’s index

The Ginarte and Park index was first released in 1997 and remains the most widely used measure of patent-system strength. Drawing on the formal laws “on the books,” it assigns each country a composite score based on five legislative and institutional dimensions:

1. Coverage. Assesses which categories of inventions are eligible for patent protection (e.g., pharmaceuticals, chemicals, plant varieties, software, etc.). Fewer exclusions imply a more expansive (“stronger”) system.

2. Membership in international treaties. Awards points for participation in key agreements such as the Paris Convention, the Patent Cooperation Treaty, and the International Convention for the Protection of New Varieties of Plants. Treaty obligations typically raise minimum standards and promote harmonization.

3. Duration of protection. Measures the statutory patent term. The international norm of 20 years from filing serves as the benchmark; shorter terms reduce a country’s score.

4. Enforcement mechanisms. Captures whether national law provides tools for patent holders to defend their rights, such as preliminary injunctions, contributory‐infringement claims, reversals of the burden of proof, and similar remedies.

5. Restrictions on patent rights. Penalizes systems that impose compulsory licensing, domestic‐working requirements, or other legal limits that weaken patent exclusivity.

By summing the scores across these five categories, the Ginarte and Park index delivers a single number reflecting each country’s formal patent-law strength. Higher scores indicate more robust regimes—that is, systems with broader coverage, stronger enforcement, and fewer built-in limitations. This index is widely employed in economic research to examine how variations in patent-law strength affect indicators such as innovation, trade flows, and foreign direct investment (FDI).

Yet, despite its broad adoption, scholars have identified notable limitations of the Ginarte and Park index. First, as a general measure, it may obscure sector-specific features of patent regimes. This issue prompted the creation of targeted indices, such as those for pharmaceuticals and plant varieties. Second, by relying solely on “law on the books,” it overlooks the real-world effectiveness of those laws—what scholars call “law in practice.” Third, because it focuses on the generosity of formal rules and remedies, calling high scores “strong” can be misleading: these values often reflect applicant friendliness more than patent system robustness. In the sections that follow, we explore the challenges of measuring enforcement in practice and consider alternative approaches to assessing patent quality.

Law in practice

The Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), administered by the World Trade Organization (WTO), entered into force on January 1, 1995. It is the most comprehensive multilateral treaty on intellectual property (IP), setting minimum standards for the availability, scope, and use of IP rights. TRIPS spurred rapid convergence in “law on the books” across WTO members, but significant disparities remain in enforcement and administration. Patent owners’ real ability to exercise their rights depends not only on statutory rules but on how those rules are applied in practice. Hence, the need for a complementary perspective: law in practice, which captures the effectiveness of IP-law enforcement by courts, police, customs authorities, and other institutions.

Because “law in practice” involves diffuse enforcement routines, nuanced procedural norms, and informal institutional behaviors that resist simple codification, it is far harder to measure than “law on the books.” Direct data on enforcement outcomes—such as court filings, injunctions granted, or damages awarded—are often unavailable or incomplete. Even where records exist, compiling comparable indicators of judicial independence, administrative efficiency, or enforcement-official competence across dozens of countries is no small feat. Layered on top of this are culture-bound conventions and unwritten protocols—how judges, prosecutors, and customs officers actually interpret and apply statutes day to day—which defy straightforward quantification.

Despite these challenges, researchers have devised several proxies for “law in practice.” Political scientist Robert Ostergard, for example, incorporated enforcement evidence drawn from the U.S. State Department’s Country Reports on Economic and Trade Practices. Economist Beata Javorcik used assessments from the International Intellectual Property Alliance’s Special 301 recommendations. Management scholar Nick Papageorgiadis and colleagues offer perhaps the most comprehensive analysis, with an index that quantifies enforcement based on three categories of transaction costs faced by rights holders:

1. Servicing costs. Quality of patent administration.

2. Property-rights protection costs. Judicial efficiency, corruption, and related factors that affect court-based enforcement.

3. Monitoring costs. Resources required to detect infringement, including police engagement and public commitment to IP protection.

“Strength,” or applicant-friendliness?

The indices we have reviewed so far capture the strength of patent laws and enforcement, but they don’t tell us whether the patents actually granted would survive a court challenge. In other words, they measure how generous or friendly a system is to patent owners, not how robust its patents are.

Recognizing this limitation, economist Bruno van Pottelsberghe produced a series of studies aimed at measuring the quality of patent systems. He developed a “quality index” built around nine operational design features. Seven of these relate to procedural safeguards—such as the window for requesting examination, the availability of post-grant opposition, and grace-period length—while two indicators capture patent office resourcing: personnel expenses per staff member and the average number of claims examined per examiner.

Building on this procedural focus, he also exploited procedural data published by patent offices, seeking to convert them into quantifiable examiner-centred metrics, including search completeness through classification and citation practices, the upfront certainty of information delivery, the speed of information delivery, and the relative stringency assessed via grant rates. This research line admittedly offers a more direct gauge of patent strength. However, it does not tell us how high or low the bar is for patent applications to be granted.

Insights from “twin patent” studies

A complementary approach to gauging patent‐system rigor uses twin patents—matched filings for the same invention made at different offices—and compares grant outcomes across jurisdictions. A team of Australian economists, for example, examined triadic patent applications (filed in the United States, Japan, and Europe) and documented significant “disharmony”: decisions on the same invention often diverge, with some offices granting patents that others reject.

Such inconsistencies are not inherently problematic, for different countries may legitimately apply distinct patentability standards. However, they can also signal examiner errors. In a follow-up study, Gaétan de Rassenfosse and colleagues decompose these discrepancies into two sources: variation in patentability thresholds and outright mistakes. Analyzing data from the five largest offices (the triadic offices plus China and South Korea), they find that systemic differences in examination standards explain most disharmony, while true errors play a more limited role.

Crucially, their model also quantifies each office’s effective “bar height.” They show that the JPO applies the strictest standard, followed by the EPO, the KIPO, the USPTO, and finally CNIPA—clearly documenting differences in patent offices’ rigor.

What explains cross-country differences in patent protection?

Recognizing the vast differences in the level of patent protection across countries, scholars have sought to understand the factors that explain these variations. They find that countries tend to calibrate their IP regimes dynamically as they develop. At early stages of industrialization, when domestic firms lack significant innovative capacity, patent systems are often weak or narrow. Weak systems allow local actors to imitate established technologies without stringent legal barriers. As a nation’s technological base and R&D capabilities grow, however, pressure mounts to strengthen IP rights, both to reward nascent innovators and to secure access to foreign markets. China exemplifies this trajectory: its patent laws and enforcement apparatus have been progressively tightened over the last three decades, in tandem with the rise of home-grown “national champions” that compete globally and demand robust protection at home.

However, whether weak IP regimes truly benefit developing economies remains a topic of hot debate. Critics of lax IP argue that multinational firms will hesitate to establish R&D centers or high-value manufacturing in countries where their core technologies can be pirated. Empirical studies have found that stronger patent protection is correlated with greater inflows of technology-intensive FDI, suggesting that weak regimes can deter knowledge transfer and thereby limit learning-by-doing opportunities.

However, proponents of early-stage flexibility counter that imitation is a key engine of catch-up. By allowing domestic firms to reverse-engineer and adapt foreign innovations, weak IP rights can accelerate capability building, foster competitive industries, and lower entry barriers.

Ultimately, the political economy of IP reform—shaped by domestic interest groups, trade negotiations, and institutional capacity—determines each country’s unique path along this dynamic spectrum.

Refer a friend

Leave a comment

Please cite this post as follows:

de Rassenfosse, G. (2025). The strength of patent systems. The Patentist Living Literature Review 5: 1–5. DOI: 10.48550/arXiv.2505.07121.

Patent quality is a central concern for all stakeholders in the patent system. Inventors, patent attorneys, scholars, and policymakers consistently emphasize its importance, recognizing that the value and effectiveness of the patent system hinge on the issuance of high-quality patents.

But what exactly is patent quality? As we will see, the term is used to describe a wide range of distinct constructs. Conversely, the same construct may be described using different terminology, such as strength, importance, or value, adding another layer of confusion. This post explores the many facets of patent quality.

Clarifying the foundation: What is invention quality?

Since patents are meant to protect inventions, defining what constitutes invention quality seems like a natural starting point. In the economics literature, invention quality is a stylized and abstract concept that captures ‘how much better’ a product or technology is compared to previous versions. It refers to the productivity-enhancing or utility-enhancing characteristics of that product or technology [1, 2].

While it is intuitive to conceive invention quality as a measure of an invention’s technological superiority, this definition is not satisfactory for practical purposes: the notion of technical superiority is just as vague as the concept of quality itself. An alternative perspective emerges from the innovation literature, which distinguishes between types of technological advances, namely incremental vs. radical inventions. The former refers to improvements or refinements in known technologies, whereas the latter represents significant breakthroughs with strong disruptive potential, such as the invention of the first 3D printer. In addition, a restricted number of inventions have a widespread, transformative impact on the entire economy, often leading to long-term changes in how we live and work—for instance, the steam engine, electricity, and the internet. We call these inventions general-purpose technologies.

While this perspective offers a gradation of inventions’ importance, it does not provide a concrete way to assess quality improvements across successive generations of inventions. Such an assessment requires acknowledging the multidimensional and context-dependent nature of quality.

Consider the case of a piezoelectric sensor, which converts mechanical pressure into an electrical signal. This single invention can be embedded in a wide array of products, each placing different demands on its performance. In electronic drum pads, for instance, the sensor is used to detect strikes and trigger digital sounds. Relevant dimensions of quality include responsiveness (how accurately the sensor captures input), latency (how quickly the signal is processed), and durability (its ability to withstand repeated physical stress). In contrast, when embedded in a car’s airbag system, the same sensor must detect sudden shocks to trigger deployment. Here, while responsiveness and latency still matter, other dimensions—such as resistance to temperature fluctuations and humidity—become equally critical. Piezoelectric sensors are also found in a variety of everyday products, including electric lighters, digital bathroom scales, and inkjet printers. In each case, quality is assessed over multiple dimensions, and improvements along one dimension may be essential in one application but irrelevant in another.

This example illustrates the challenge of measuring invention quality with a single, unified metric. Yet, at least in principle, it is possible to assess an invention’s quality ex-ante by identifying its relevant technical dimensions—such as efficiency, durability, or precision—and evaluating them under controlled conditions, even in the absence of a concrete use case. (This is particularly true for incremental inventions, as radical inventions may introduce entirely new quality dimensions, leaving no baseline for measuring improvements over previous generations of the technology.)

What about invention value?

While invention quality can be assessed in the absence of concrete use, invention value is inherently tied to use. An invention has economic value only insofar as it is, or can be, embedded in a product or process that generates impact in the real world. Unlike quality, however, value lends itself—at least theoretically—to a single, comparable measure: its monetary worth. While measuring quality requires navigating multiple, often incommensurable dimensions, value can be expressed in real money.

An important corollary to this statement is that different uses for an invention will lead to different valuations. Consider the case of the ring-pull can. The inventor allegedly licensed the system to Coca-Cola at 0.1 pence per can and obtained £148,000 a day in royalties. Had the inventor granted an exclusive license for his invention to a company specializing only in canned green beans, he would have amassed far less money—highlighting that the actual use of an invention directly affects its value.

This example focuses on what economists call the ‘private’ value of the invention, namely, the returns accruing to the innovator. The private value forms only one component of an invention’s social value, which also encompasses the consumer surplus (capturing the fact that some inventors may have paid less than what they were willing to pay), spillovers to competitors and other actors such as suppliers (who may capture some of the value created by the invention), and wider societal effects (such as environmental benefits).

While the private value of an invention is tied to how a particular firm chooses to use it, the social value of an invention reflects its broader potential—across users, applications, and contexts. The inventor of the ring-pull can may have initially licensed it for use in canned beverages, but as other companies adopt the invention for canned vegetables, soups, or even pet food, its benefits diffuse throughout the economy. Consumers gain from improved convenience, competitors are nudged to innovate, and suppliers may benefit from increased demand or efficiency gains. These spillovers, along with broader societal impacts—such as improvements in accessibility—contribute to the social value of the invention.

The distinction between private and social value highlights a central insight: while private value depends on specific market conditions and the innovator’s approach to appropriate returns, social value emerges more diffusely as others adopt and adapt the invention. As a result, the realized private value of an invention may fall well short of its potential, whereas the realized social value may come closer to being fully realized—precisely because it is not restricted to a single use or actor.

Finally, note that the term ‘value’ is not confined to the ‘economic merit’ of the invention. A previous post introduced the notions of ‘use value’ and ‘exchange value.’ Use value refers to an object’s utility or practical usefulness, whereas exchange value refers to what it can be traded for in the market. The former concept more closely relates to the notion of invention quality, while the latter refers to the invention’s economic value (capturing one dimension of the social value).

So, where does patent quality fit in?

We have taken a long detour in our effort to understand what makes a patent ‘high quality.’ A natural temptation is to equate patent quality with the quality of the underlying invention. While some scholarly works adopt this definition, patent quality can, in fact, be understood in several distinct ways. It is entirely possible for a high-quality patent to protect a low-quality invention, just as a low-quality patent may cover a groundbreaking one. There are at least four broad definitions of patent quality.

As just mentioned, the first definition conflates the quality of the patent with the quality of the invention it protects. In this context, invention quality is measured, for instance, by its novelty, inventiveness, or the size of the ‘inventive step’ separating the invention from existing technology. This abstract, standardized metric of quality is useful for stylized economic modeling but fails to recognize the multidimensional and context-dependent nature of quality, as the example of the piezoelectric sensor illustrates.

A second perspective links patent quality to the economic value or impact of the patented invention. From this standpoint, a good patent is one that fulfills the key objectives of the patent system: to reward and incentivize innovation while promoting knowledge diffusion and follow-on innovation. Accordingly, high-quality patents are those that generate significant social value.

A third definition focuses on the drafting quality of the patent. In this view, a high-quality patent is a well-written document, especially in terms of claim clarity and appropriate scope. It provides a clear and comprehensive description of the invention, and its claims are precise and broad enough to offer strong protection.

A fourth perspective defines quality in terms of compliance with statutory requirements. A high-quality patent is one that meets the criteria of patentability: subject matter eligibility, utility, novelty, non-obviousness, enablement, and written description. Here, a low-quality patent is one that should not have been granted in the first place. Conversely, a high-quality patent is one that is likely to be legally valid if challenged. The USPTO largely adopts this legalistic definition, focusing its quality initiatives on identifying and minimizing the issuance of potentially invalid patents.

It is worth emphasizing that no single, universally accepted definition of patent quality exists. Indeed, each profession may have its favorite definition: patent attorneys may prioritize clarity and breadth of claims, engineers may value technological sophistication, and economists may focus on the invention’s impact.

What about patent value?

Patent value refers to a patent’s potential to deliver economic returns or strategic benefits to its owner. Value is shaped by the existence and size of the relevant market—the actual use of inventions. Selling the ring-pull can patent to Coca-Cola, for example, clearly yields more value than licensing it to a niche soup manufacturer.

It would be erroneous to conflate patent value with the value of the underlying invention. Indeed, patent protection creates additional value due to the exclusionary rights over the invention. The mere existence of a patent can enhance the private value of an invention. Take the ring-pull can again: without a patent, selling the invention to Coca-Cola would yield limited benefits—rivals could freely copy the invention, eroding any first-mover advantage and driving the private value of the invention to zero. By contrast, the patent allows the would-be buyer to obtain exclusive rights, increasing the invention’s marketability. This added value is known in the literature as the ‘patent premium.’ One estimate from a team of Australian economists using data from a survey of inventors puts it at 40 to 50 percent.

As noted in a previous post, firms file patents for many reasons and employ them in varied ways—whether to block competitors, attract investment, or strengthen bargaining positions. These different uses represent as many value-creation strategies that go beyond the invention’s worth, further explaining why the value of the invention differs from the value of the associated patent right.

Crucially, while the existence of the patent right adds value, the actual value realized depends on the quality of the patent—particularly its legal strength. Because patentees typically need to prove infringement and withstand invalidity challenges in court, legal validity is a core component of patent value. In that sense, quality and value are tightly intertwined.

However, economists Joseph Farrell and Carl Shapiro explain in a 2008 paper that patents with dubious validity (‘weak’ patents) can still command substantial royalty payments. Even if a downstream firm suspects that a patent is weak, the cost of proving its invalidity in court can be substantial, making a settlement with the patent holder (through royalty payments) a more economically rational option. Thus, even weak patents can shape—or distort—markets, making high-quality patent examination essential to a well-functioning innovation system.

Refer a friend

Leave a comment

Please cite this post as follows:

de Rassenfosse, G. (2025). What is patent quality? The Patentist Living Literature Review 4: 1–5. DOI: 10.48550/arXiv.2504.08785.

Lire la version française.

A previous post highlighted that not all patentable inventions are patented, prompting the question of why firms choose to patent or not. The present post explores the factors that affect the propensity to patent. Examining these factors helps clarify the conditions under which the patent system operates effectively.

Is patent protection worth it?

Every firm considering patenting an invention asks itself this question, seeking to assess the costs and benefits of patent protection. The costs are generally known, incurred upfront, and encompassing both monetary and non-monetary factors. In contrast, the benefits are more challenging to identify and quantify and remain uncertain. Moreover, not all patent applications are approved, meaning that firms could incur the costs of patent protection without realizing the benefits if their application is rejected. Let us analyze the cost-benefit tradeoff of patent protection.

On the cost side, the monetary expenses are straightforward to quantify. They encompass the patent attorney fees and the administrative charges, with total costs in the United States typically ranging between $5,000 and $30,000—although there is obviously no strict maximum limit. However, because patent rights are jurisdictional, having legal effect only in the jurisdiction that issues them, firms seeking international patent protection must file patents—and bear the associated costs—in every country where they seek protection. These costs can escalate quickly and be prohibitively expensive for small and medium enterprises.

The non-monetary cost primarily relates to the disclosure requirements. To secure a patent, an inventor must disclose the workings of the invention, which become public information. This transparency may assist competitors in designing around the patent or advancing alternative technologies. Furthermore, once the patent expires, competitors can copy the disclosed invention. The disclosure requirement is one strong reason why some firms opt for secrecy instead of patenting. Stories of inventions kept secret abound, although examples commonly found primarily relate to the food and beverage industry, including the Coca-Cola formula, KFC Original Recipe of ‘11 herbs and spices,’ and the Pastéis de Belém. However, survey data consistently indicate that a larger share of companies valued secrecy over patents across many industries—a tendency especially pronounced among smaller firms [1,2].

What are the benefits of patent protection?

A patent grants its holder the right to exclude others from making, using, selling, or importing the patented invention. Patents help firms ‘appropriate the returns’ or ‘capture the value’ of their innovations—terms coined by economists and management scholars, respectively, to indicate that patents shield inventions from imitation. However, as we will discuss, there are many ways to leverage patents, resulting in diverse motives for filing—and a broad range of associated benefits. To explore these benefits, researchers have surveyed patent applicants about their reasons for filing. These studies cover, among others, established U.S. firms, U.S. high-tech startups, European small and medium enterprises, and firms in Switzerland, France, Germany, and Spain.

The surveys reveal a variety of overlapping and sometimes context-specific reasons why firms seek patents. Alongside the traditional goal of protecting their inventions, companies also file patents strategically to block competitors, restricting access to patented content even when they have no intention of using it themselves. An ‘offensive’ patent blocking strategy might involve disrupting rivals’ R&D or product pipelines, while a ‘defensive’ blocking approach focuses on preserving a firm’s freedom to operate and avoiding infringement lawsuits. Tactics such as building ‘patent fences’ or ‘patent walls,’ or contributing to ‘patent thickets,’ support these objectives.

Another key motive for filing patents is to use them as negotiation tools or ‘bargaining chips’ in inter-company agreements and collaborations, such as cross-licensing deals, joint ventures, and standard-setting activities. Holding patents can significantly strengthen a firm’s bargaining position and help deter litigation, with such strategy sometimes leading to what is known as a ‘patent arms race.’

Despite their variety, these motives for filing patents rest on a fundamental aspect: the exclusion right granted by patents. This legal power underpins every strategy discussed above—from protecting inventions and blocking competitors to facilitating negotiations—by ensuring that the patent holder can control how the invention is used. However, there is also a range of patent uses that do not rely on the exclusion right, as discussed below.

Benefits extend beyond the exclusion rights

Firms might file patent applications solely to create prior art. This approach stops others from patenting the same invention and ensures that the firm can continue to use its own technology freely. In these cases, the objective is to establish a documented record that limits future patent claims, rather than to secure patent protection.

Patents can also serve as signals of a firm’s quality and innovation. A well-known example is Audi’s 2006 A6 advertisement, which compared the number of patents held by NASA with those held by Audi to highlight the car’s advanced technology. The commercial ended with the line “To date, Nasa have filed 6,509 patents. To get to the A6, Audi have filed 9,621 patents.” Although Audi likely did not pursue patents specifically for the ad, the campaign illustrates that firms can use patents to enhance their reputation.

Firms, especially smaller ones and startups, may also file patents to bolster their attractiveness to investors, but patents serve two distinct roles in this regard. On one hand, the exclusion right directly protects a firm’s market position and enhances its chances of survival, something that investors value. On the other hand, the mere presence of patents can act as a signal to investors that a company is innovative and professionally managed, lending credibility to the firm’s technology novelty. This ‘signal’ increases investor confidence and operates irrespective of the way the exclusion right will be used.

Finally, patents are sometimes used as performance indicators for researchers and R&D teams, offering a measure of innovation and productivity. Furthermore, being named as an inventor on a patent not only can lead to additional compensation and career advancement but may also reinforce an employee’s value within the company. Some organizations showcase their patents in the form of ‘patent walls,’ underscoring their commitment to innovation and celebrating the creative contributions of their personnel.

Overall, protecting innovations from imitation and blocking competitors consistently emerge as key drivers across different firm types and countries. However, strategic uses such as negotiation and licensing, signaling value, and securing financing are also important motivations, with their significance varying depending on the specific characteristics of the firms and their environments.

The challenge of evaluating the benefits of patents

Although there may be numerous potential benefits, quantifying their value at the time of the patenting decision is highly speculative. This uncertainty complicates the choice of whether to patent an invention. First, there is no guarantee that the application will be granted. If it is rejected, the applicant incurs all costs—including the cost of disclosure, as the application is published—without gaining any of the anticipated benefits. Moreover, while the grant decision is based on an invention’s technical merits, its actual economic payoff hinges on market potential, creating a disconnect between the likelihood of grant and the value ultimately realized. Finally, because the rewards of patent protection materialize in the future, firms must predict how technology and competition will evolve, adding yet another layer of uncertainty.

Considerations about the ‘option value’ of patents further complicate the patenting decision. Even if a simple cost-benefit analysis may not justify the immediate expense of filing a patent, the patent’s option value can still make it a worthwhile investment. Once a patent is filed, the firm secures the right to benefit from its protection, much like holding a call option on future innovations. By contrast, choosing not to file means permanently relinquishing this option, even if market conditions later become more favorable for enforcing or monetizing the technology.

Given this uncertainty, patents are often likened to ‘lottery tickets’ that offer potentially significant, albeit speculative, rewards to their holders. Moreover, the ultimate value of a patent frequently depends on its ability to stand up in court, meaning that its benefits are sometimes conditional on successful enforcement. Several studies suggest that a considerable number of U.S. patents would be deemed ‘invalid’ if challenged in court (a topic we will explore in a separate post), further adding to the hypothetical nature of the rewards associated with patent protection.

Barriers to patenting

In addition to examining why firms pursue patents, scholars have investigated factors that discourage patenting. Unsurprisingly, high costs and the mandatory disclosure of key information stand out as major barriers. Beyond these concerns, research also points to the limited effectiveness of patents in preventing imitation, given that competitors can often ‘invent around’ existing patents or that enforcement can prove both difficult and costly.

Another set of barriers involves the complexity and time of the patenting process. Because it can be tedious, firms may be discouraged from pursuing patents altogether. Moreover, the delay between filing and issuance—often spanning several years—can be particularly problematic in rapidly evolving fields, where a patent might become obsolete before it is granted.

Lastly, some innovative firms, especially those with limited experience or resources, may not realize that their inventions are patentable in the first place. This knowledge gap can arise from inadequate internal processes for identifying intellectual property or from a general lack of familiarity with patenting strategies and best practices. As a result, potentially valuable innovations might go unprotected simply because firms remain unaware of their patentability or are unsure of how to navigate the system.

If firms don’t patent, how do they capture the value of their inventions?

While patents offer a recognized path to securing returns on an invention, firms often rely on a variety of other strategies that may be more suitable for specific industries, innovation types, or organizational circumstances. One obvious alternative is secrecy, as previously discussed, which works especially well for inventions shielded from public view, such as manufacturing processes. Moreover, if a product or process is inherently complex, firms can rely on that complexity to discourage imitation. Inventions that are costly and time-consuming to imitate form a natural barrier that deters potential competitors.

Another strategy is to gain a lead time advantage by introducing an innovation before competitors, allowing the firm to build market presence and cultivate brand loyalty. This approach is particularly valuable for smaller companies, which often emphasize speed to market as a core protection mechanism. Being first to market also enables a firm to refine production processes ahead of potential imitators, helping it move quickly down the learning curve, thereby reducing costs and boosting efficiency—advantages that can be challenging for competitors to replicate. Likewise, by regularly upgrading or improving its offerings, a firm can stay one step ahead, rendering existing products or processes obsolete before imitators can catch up.

Innovators can also capture value by offering complementary products or services, drawing on capabilities in manufacturing, marketing, sales, and service. Even if the core innovation is easily replicated, a firm’s established network, specialized expertise, and infrastructure can still provide a crucial competitive edge.

It is important to recognize that these strategies are not mutually exclusive and can often be combined for enhanced effect. For example, a strong patent portfolio can work in tandem with efforts to build brand loyalty, while secrecy and formal patent protection may coexist even within the same product line. The protection of a novel beauty cream might rely on patents for individual compounds and secrecy for the specific mixture.

To conclude, even amid questions of costs, validity and effectiveness, patents continue to offer strategic advantages that many firms find indispensable. Firms patent not only to deter imitators but also to create new opportunities—through collaboration, financing, and reputation-building—underscoring why patents remain central to modern innovation strategies.

Refer a friend

Leave a comment

Please cite this post as follows:

de Rassenfosse, G. (2025). Why do firms patent? The Patentist Living Literature Review 3: 1–5. DOI: 10.48550/arXiv.2503.11555.

Lire la version française

Knowledge is the lifeblood of innovation, fueling economic growth and improving living conditions. However, we don’t treat all knowledge the same way. Some is patented, while other knowledge remains freely available or kept secret. Understanding the share of knowledge that is patented provides important insights into the functioning of the knowledge economy and the contexts in which the patent system effectively incentivizes innovation. If certain types of knowledge persistently fall outside the realm of patenting, one can hardly argue that patents should be credited with their creation. Exploring these issues deepens our understanding of how intellectual property shapes the creation and dissemination of knowledge.

What is knowledge?

Before attempting to quantify the proportion of knowledge that is patented, we need a clear working definition of knowledge. Knowledge is a complex concept interpreted differently across fields like philosophy or sociology. Information science gives us a useful definition from the viewpoint of economics. It introduces three components: data, information, and knowledge. Data refers to raw, unorganized facts. Information is processed and organized data that carries meaning. Knowledge, in turn, is the application of information in a specific context to enable decision-making.

Knowledge encompasses a vast spectrum of human understanding and experience, from practical skills to scientific theorems to cultural traditions. For example, Indigenous communities have observed plant growth cycles, animal behavior, and weather patterns over generations (‘data’). They have come to recognize which plants have medicinal properties, understand animal migration routes, and predict seasonal changes (‘information’). They can then use this information to cultivate crops, hunt effectively, and prepare for seasonal challenges (‘knowledge’). Knowledge is embedded into all our actions, from riding a bike to cooking to applying a rule of thumb.

What kind of knowledge does the patent system aim to encourage?

The patent system provides an economic incentive to create new technical knowledge. The ‘economic incentive’ implies that the patent system channels inventive activities toward knowledge with significant economic value. However, it overlooks certain inventions that are highly useful but not economically valuable. To understand this, we need to think about the difference between use value and exchange value. Use value refers to the utility or practical usefulness of an object, whereas exchange value refers to what an object can be traded for in the market. Although goods with high use value may sometimes hold high monetary value, this is not always true. For instance, water is indispensable and possesses high use value, yet diamonds—despite being far less useful—command a much higher exchange value due to their rarity and demand. This phenomenon is known as Adam Smith’s ‘paradox of value.’

The term ‘technical knowledge’ must be understood in the context of the ‘patentable subject matter,’ which denotes inventions considered eligible for patent protection under the law. In the United States, four categories of invention are appropriate subject matter, including processes, machines, manufactures, and compositions of matter. The first category defines ‘actions’—inventions that consist of a series of steps or acts to be performed—while the latter three categories define ‘things’ or ‘products.’ The law also explicitly excludes abstract ideas, laws of nature (like physical laws or biological processes), and naturally occurring substances, organisms, or phenomena from patentable subject matters.

In sum, the patent system incentivizes the creation of new, economically valuable technical knowledge. This knowledge is undoubtedly key to advancing social and economic progress. However, an equally important aspect of progress relies on other types of knowledge.

• First, ‘old’ knowledge—knowledge that is not new to the world and, therefore, not patentable—plays a vital role in economic growth. Think about the wheel—it was invented a long time ago, but we still use it in cars, bikes, and airplanes. Economists have long recognized that growth depends not only on creating new inventions but also on disseminating and applying existing ones across firms and societies.

• Second, not all useful knowledge is economically valuable. For instance, drugs targeting rare diseases, while highly beneficial, lack commercial appeal due to their small market size. Consequently, the patent system alone provides insufficient incentives for firms to develop such drugs.

• Third, useful knowledge is not always ‘technical.’ Think of management practices or organizational strategies that enhance efficiency and productivity. Moreover, scientific knowledge and discoveries, which fall outside the patent system, are also of immense importance.

What proportion of patentable knowledge is patented?

Now that we have more narrowly defined the type of knowledge that can be subject to patent protection, it becomes evident that the overwhelming majority of knowledge lies outside the scope of the patent system. While quantifying the exact proportion of knowledge that is patentable is a vain task, economists have focused on the more modest endeavor of estimating the proportion of patentable inventions that are actually patented. At first glance, one might assume that all patentable inventions would be protected by patents—after all, why would inventors forgo patent protection if they are entitled to it? However, as research on the ‘propensity to patent’ has shown, this intuition is misleading.

Estimating the exact percentage of inventions that are patented is challenging due to a lack of comprehensive data on all inventions. However, academic studies tend to show that patents are not the primary means of protecting inventions in most industries. A researcher used data from nineteenth-century world fairs to estimate the proportion of inventions patented. She estimated that one in nine innovations presented at the 1851 Crystal Palace Exhibition in London were patented, and one in eight at the 1876 Centennial Exposition in Philadelphia. Patent law has strengthened significantly since then, increasing patenting incentives, and ‘knowledge’ has become a greater source of competitive advantage for firms, potentially increasing the urge to use patents.

Modern evidence about the ‘propensity to patent’ comes primarily from survey data. Another researcher surveyed 100 U.S. manufacturing firms in the early 1980s, asking them to estimate the percentage of their patentable inventions that were patented. He reported numbers ranging from 50 percent for the primary metal industry to 86 percent for the machinery industry. In a survey of 1478 labs in the U.S. manufacturing sector in the early 1990s, a team of researchers obtained significantly lower numbers. Their respondents applied for patents on 49 percent of their product innovations and 31 percent of their process innovations. Others surveyed 604 of Europe’s largest industrial firms in the early 1990s and reported that the average propensity to patent was about 36 percent for product innovations (but close to 80% in pharmaceuticals) and 25 percent for process innovations (and close to 50% in precision instruments).

Another research project adopted a different approach to investigate patent propensity. The authors analyzed a sample of 2,800 inventions that received the ‘R&D 100’ award, a competition organized by the journal Research and Development (R&D) to recognize the 100 most technologically significant new products available for sale or licensing in the year prior to evaluation. Their dataset, spanning the period from 1977 to 2004, reveals that a mere 10 percent of these inventions were patented.

These results have considerably enriched our understanding, but they are based on relatively small samples. A recent study addresses this limitation by providing estimates representative of the broader population of firms in the United States. Drawing on data from the U.S. Census Bureau’s Business R&D and Innovation Survey, collected between 2008 and 2015, the researchers estimate that about 80 percent of firms performing R&D activities—presumably leading to patentable inventions—do not apply for patents. In other words, patenting is not the norm but, rather, a privilege of a select few firms. However, the study also reveals that patenting firms account for the vast majority (above 90 percent) of reported R&D. Importantly, this finding does not imply that 90 percent of inventions are patented. Instead, it highlights that most inventions come from firms engaged in patenting, even if not every invention they create results in a patent application.

Why are not all patentable inventions patented?

The evidence presented above indicates that not all patentable inventions are, in fact, patented. While firms in some industries may seek patent protection for most or even all of their inventions, others patent selectively or choose not to patent at all. A body of literature in economics and management has explored the factors influencing firms’ propensity to patent from both theoretical and empirical perspectives.

A key insight from this literature is the significant heterogeneity in patenting behavior across invention types, firm characteristics, industries, and market structures. While a comprehensive discussion of these factors is beyond the scope of this post, it is worth noting that, in some cases, inventions eligible for patent protection are better kept secret or disclosed to the public without seeking IP protection.

To illustrate why firms may prefer secrecy over patenting, consider the ‘enablement requirement’ of patent law, which mandates that a patent application must disclose sufficient information to enable a person having ordinary skill in the art (PHOSITA) to replicate the invention. While patent protection would in practice prevent anyone from using the replicated inventions (at least in countries where patent protection is valid), the public disclosure of the invention’s technical details can discourage firms from patenting. Instead, they may favor secrecy, particularly for inventions that are difficult or costly to imitate. Conversely, patenting becomes more attractive for inventions that are easy to reverse-engineer, such as mechanical designs or chemical formulas. This distinction helps explain why surveys consistently find that product inventions are more likely to be patented than process inventions. While products are often exposed to public scrutiny, manufacturing processes conducted behind closed doors can often remain effective trade secrets.

Good news for the free use and sharing of knowledge?

This review has highlighted that the vast majority of knowledge is not patentable, and even when it is, only a fraction becomes patented. Consequently, most knowledge remains free to use and share. However, a critical distinction must be made between using knowledge and sharing it. While patents restrict the use of patented knowledge, they promote its sharing by requiring detailed public disclosure. Patent databases have thus become invaluable repositories of technical information, freely accessible to all—a stark contrast to trade secrets, which, while freely usable if independently discovered, are rarely shared (except under non-disclosure agreements).

That said, even if only a small fraction of the world’s knowledge is patented, the tensions arising from the exclusive rights of patents cannot be overlooked, particularly when they pertain to inventions of immense social importance. Pharmaceuticals exemplify these tensions, where patents often provide the financial incentive for innovation but can also delay access to life-saving treatments. Although patent protection is time-limited—typically 20 years from filing—this period can feel disproportionately long in critical situations, such as for life-saving treatment or during a global pandemic. The balance between incentivizing innovation and ensuring equitable access remains a pressing challenge for the IP system.

Refer a friend

Leave a comment

Please cite this post as follows:

de Rassenfosse, G. (2025). What proportion of knowledge is patented? The Patentist Living Literature Review 2: 1–4. DOI: 10.48550/arXiv.2501.18043.

Subscribe now

Lire la version française · Lesen Sie die deutsche Version.

Simply put, a patent is a legal document securing the right to exclude others from making, using, or selling an invention for several years. At first glance, the idea of granting exclusive rights to an invention might seem surprising—perhaps even unsettling—since we often think of knowledge as a shared resource. Wouldn’t the world benefit more if knowledge flowed freely for everyone to use?

While it is hard to argue against the idea that knowledge should be shared as widely as possible, it is important to ensure that new knowledge is being created in the first place. That’s where patent rights come into play, as we explore below.

Before diving into why the patent system exists, however, it is worth noting that most inventions are not subject to patent protection. The United States Patent and Trademark Office (USPTO) issued its twelfth millionth patent in mid-2024. Of these, around five million patents were granted in the last two decades and could still be in force. However, patents do not automatically stay valid for their full 20-year term—owners must pay renewal fees regularly to maintain them. Many choose not to, so a rough estimate is that only about three million inventions are currently protected by active patents. Furthermore, not all patentable inventions are patented—many are kept secret—and scientific discoveries are not eligible for patenting. According to OpenAlex, a comprehensive open dataset of scholarly work, approximately 120 million scientific articles have been produced over the past two decades—far exceeding the number of patents granted in the same period. In other words, only a tiny portion of the world’s knowledge is protected by patents.

What is all the fuss about patents, then?

If only a small fraction of knowledge is protected by patents, why are patents such a polarizing topic? One reason lies in the philosophical debate surrounding the nature of knowledge itself. Many believe that knowledge, as a universal and shared resource, transcends individual ownership and should remain free from barriers, serving the collective progress of humanity.

Beyond this philosophical divide, patents stir controversy because of their real-world implications. Critics argue that patents limit the use of knowledge, particularly in critical fields like medicine, where restricted access to patented drugs can have life-or-death consequences. The debate surrounding patent protection for antiretroviral medications during the AIDS crisis at the turn of the millennium is a stark example of these tensions. Patent protection helped keep prices high, effectively excluding patients in the least-developed countries from accessing life-saving treatments. More recently, similar debates resurfaced with COVID-19 vaccines, as many stakeholders advocated for waiving patent rights to ensure equitable global access.

These arguments, and others, are used by some to argue that the patent system should be abolished, pure and simple. However, economists studying the patent system generally hold a more nuanced view, as explained below.

Why do we have a patent system after all?

The answer to that question often lies in the economic rationales offered to justify the patent system. However, these rationales are ex-post explanations constructed long after the system’s inception. In reality, we have a patent system today largely because, on March 19, 1474, the Senate of Venice happened to enact the Venetian Patent Statute—a legal innovation shaped by the unique political and economic conditions of Renaissance Venice. The Statute recognized that creative minds thrived in Venice’s bustling and innovative environment, and it sought to incentivize these inventors by granting them exclusive rights to their creations for up to ten years. This protection ensured that others could not copy their work and “take their honor,” providing a strong motivation to invent and contribute to the state’s prosperity.

What began as a local measure to encourage invention became the foundation of a global system. Over the centuries, governments across Europe and beyond saw the potential of such systems to foster technological progress and economic development, leading to the evolution of the patent systems we use today. In the United States, the importance of patents was enshrined in the Constitution, which explicitly grants Congress the power to promote the progress of science and useful arts by issuing exclusive rights. From the 15th-century canals of Venice to the globalized innovation landscape of the 21st century, the patent system has evolved through countless amendments and adjustments, shaped by the changing needs of societies and economies.

The understanding of the patent system has evolved alongside advances in theory and analytical tools available to economists. The so-called utilitarian theory of patent rights began with the influential political economists of the 18th and 19th centuries. Pioneers like Adam Smith, Jeremy Bentham, and John Stuart Mill were among the first economists to argue for granting inventors limited “monopolies” to promote innovation. Building upon these insights, prominent economists such as Arthur Cecil Pigou, John Bates Clark, and Kenneth Arrow further refined economic thinking about intellectual property. Their collective work culminated in what is now considered the classical framework for modeling the patent system, formalized by William Nordhaus in 1969.

Nordhaus’ model focuses on the optimal duration of patent protection, recognizing that a longer patent term incentivizes innovation by letting inventors earn more profit but also harms society by extending the period of monopoly pricing. The optimal patent length, according to Nordhaus, balances these competing forces. The Nordhaus model highlights the fundamental trade-off in patent policy. On the one hand, longer patent terms provide greater incentives for innovation by allowing inventors to capture more of the social benefits through higher profits. On the other hand, longer patent terms also prolong the period of monopoly pricing, leading to reduced consumer welfare.

Why do we need to incentivize innovation?

The theory behind the patent system rests on the idea that we need to incentivize innovation. But does this mean society would cease inventing if the patent system were abolished? The answer is a resounding no. Ingenuity and creativity are fundamental aspects of human nature, and people would undoubtedly continue to invent. In fact, we already know that only a small fraction of inventions are ever patented, which suggests that many inventions emerge and thrive without the promise of exclusive rights. Inventions would likely continue to see the light of day, driven by inventors’ curiosity, necessity, and desire to solve problems.

However, the question is not whether society would stop inventing but rather how many inventions society would create without a patent system. A straightforward argument is that patents, by granting exclusive rights to inventors, make it harder for others to build on patented inventions, thereby limiting follow-on innovation. In this view, abolishing patent rights could potentially free millions of inventions for society to use and improve. However, this reasoning overlooks a critical point: many of these inventions were created in the first place because of the incentives provided by the patent system. Economists caution that, without the patent system, society might not innovate at the same pace or scale as it does today.

To understand why this is the case, we must introduce the concept of knowledge as a public good. Public goods have two key traits: non-excludability and non-rivalry. Non-excludability means that others cannot be prevented from benefiting from the good once it is produced, whereas non-rivalry means that one person’s use of the good does not diminish its availability to others. Knowledge is the quintessential example of a public good. Consider the Pythagorean theorem: “In a right triangle, the square of the hypotenuse is equal to the sum of the squares of the other two sides.” One architect’s use of the theorem does not reduce its availability for others, nor can anyone be prevented from using it.

Recognizing the public-good nature of knowledge, Kenneth Arrow concluded in 1962 that a free market economy often underinvests in innovation compared to the social optimum. From a firm’s perspective, investment decisions in R&D are driven by expected private profits, which fail to account for broader societal benefits, such as consumer surplus. Moreover, the non-excludability of knowledge means that innovators cannot fully capture the benefits of their creations, as knowledge can diffuse to competitors and other firms. Consequently, projects with high social value but insufficient private returns are frequently overlooked, depriving society of innovations that could have significantly improved collective welfare.

The gap between social and private returns, and the resulting underinvestment in R&D by the free market, is a primary justification for R&D support mechanisms such as grants and subsidies. The patent system is another such mechanism, specifically designed to strengthen the appropriability of inventions. By allowing inventors to exclude others from using their inventions, patents increase the private returns on R&D investments, thereby fostering greater overall R&D activity in the economy. The exclusivity conferred by patents is not a flaw but an intentional feature of the system. While patent protection enables firms to charge temporarily higher prices for their innovative products—limiting their immediate diffusion in society—this trade-off represents the (imperfect) solution to incentivize firms to invest in developing these products in the first place. Economies accept the ‘static inefficiency’ of higher prices because the ‘dynamic efficiency’—greater incentives to invest in innovation—presumably outweighs it.

From the theory of patents to the design of the patent system

The theoretical considerations thus far allow us to understand some important features of the patent system. Following the theory’s logic, patents would only be granted for inventions that would not have been created if the patent system did not exist. Granting exclusivity to inventions that would have been developed anyway creates monopolies without delivering the intended incentive effect. However, it is exceedingly difficult for patent offices to determine whether an invention was motivated by the prospect of higher profits delivered by the patent system. In practice, patent offices grant protection based on criteria such as worldwide novelty, non-obviousness, and utility. Worldwide novelty ensures that the invention is genuinely new, while the non-obviousness requirement—indicating an inventive step—can be seen as a proxy for identifying inventions that are sufficiently innovative to justify exclusivity.

From an economic efficiency perspective, following Arrow’s argument, knowledge should be freely distributed to ensure its optimal utilization. This is because the marginal cost of providing existing knowledge to an additional person is effectively zero, reflecting the non-rivalry aspect of knowledge. This principle aligns with the concept of marginal-cost pricing, a cornerstone of welfare economics, which dictates that goods and services should be priced at their marginal cost of production to achieve allocative efficiency. While patents limit the use of an invention by others, the patent document provides a detailed description of the invention, freely accessibly to everyone. Patent collections thus serve as publicly available repositories of knowledge from which inventors can learn.

The requirement for patent documents to disclose the details of inventions serves several practical purposes: enabling examiners to assess novelty, allowing inventors to clearly define their claims, and helping others avoid infringing on the protected technology. Beyond these practicalities, the temporary exclusivity granted in exchange for public disclosure is often described as a “bargain” or a “social contract” between inventors and society, or the “quid pro quo” of the patent system. Moreover, patent protection is both time-bound and geographically limited. Patents are jurisdictional rights, valid only in the specific country where protection is sought and granted. Obtaining worldwide protection is prohibitively expensive, so most inventors seek patents in only a select few countries. As a result, most inventions are freely available for use in many parts of the world.

The disclosure requirement, along with the temporal and geographic limitations of patent protection, reduces the scope of the exclusion rights conferred by patents, partially mitigating the economic inefficiencies associated with restricted utilization. However, some scholars have argued that patents can confer exclusion rights that are excessively strong, potentially rendering the patent system inefficient. Michele Boldrin and David Levine, for instance, suggest that the current intellectual property system, by granting excessive control over the use of ideas, often creates more problems than it resolves. They distinguish between two components of patent rights: the right to own and sell ideas—commonly referred to as the right of first sale—and the right to control the use of those ideas after the sale, known as the downstream licensing right. While they see the first component as essential, they view the second as economically harmful. They argue that allowing producers to dictate how consumers use purchased products containing intellectual property stifles competition from those consumers, weakening market competition overall.

Ultimately, the necessity of the patent system is for readers to judge. Perhaps the best way to conclude is with the words of economist Edith Penrose, who aptly observed in her book The Economics of the International Patent System:

“If national patent laws did not exist, it would be difficult to make a conclusive case for introducing them; but the fact that they do exist shifts the burden of proof and it is equally difficult to make a really conclusive case for abolishing them.”

Refer a friend

Leave a comment

Please cite this post as follows:

de Rassenfosse, G. (2024). The Role of Patents: Incentivizing Innovation or Hindering Progress? The Patentist Living Literature Review 1: 1–5. DOI: 10.48550/arXiv.2412.14370