How can patents pay inventors without blocking an industry?

The bargain in early patent statutes

Venice’s statute of 19 March 1474 is often described as the first general patent statute. It offered ten years of protection for “any new and ingenious device, not previously made,” provided it was useful. The Republic logged well over a thousand petitions across the following century. The inventor received a limited period of exclusivity in exchange for disclosing the device.

England addressed a different abuse about 150 years later. The Crown had issued monopoly grants for ordinary commodities such as salt and starch, whether or not the recipient had invented anything. Parliament’s Statute of Monopolies (1624) voided those grants. Among the exceptions it kept, the one that mattered for invention allowed a patent of fourteen years or less for the “true and first Inventor” of a new manufacture. That exception came with a condition. The grant could not be “mischievous to the State, by raisinge prices of Commodities at home, or hurt of Trade, or generallie inconvenient.” The founding statute asked what a patent would do to prices and trade. The modern U.S. tests for a patent (35 U.S.C. §§ 101–103 and 112) ask whether an invention is eligible subject matter, new, useful, non-obvious, and adequately described, and they ask nothing about the market position of the firm that will hold it.

In the League’s reading, both statutes were anti-Moloch instruments. Venice paid for disclosure with a limited monopoly, and England took back grants that had nothing to do with invention, then wrote price effects into the condition for new ones. Neither document supplies a universal design for modern patent law.

Where Moloch gets in

The failure modes are game structures, and most of them map onto the League’s trap taxonomy. Two historical cases show how collective licensing changed the terms of a fight over patents.

A modern patent thicket contains enough overlapping rights that a manufacturer may need several licenses before it can sell a product. Bessen and Meurer estimated the direct cost of disputes with non-practicing entities, firms that earn mainly by asserting patents rather than selling related products, at $29 billion in 2011. Their estimate covered 5,842 defendants, most of them small and medium firms, and they found that little of the money reached small inventors. The estimate is contested because it relies on a survey by RPX, a company that sells patent-risk services, and it excludes indirect costs. In the taxonomy this is an extraction problem: firms that make nothing collect from firms that do, and the cost of defending a lawsuit is how they collect.

Standards create a different problem. ETSI, which develops cellular standards used in 4G and 5G, asks patent holders to declare patents that might be essential. Studies have found that fewer than half of declared-essential patents are essential, with published estimates for 5G between 10% and 20%. A payment system based on declared patent counts gives holders a reason to over-declare. Standards bodies use FRAND commitments, promises to license on “fair, reasonable and non-discriminatory” terms, to limit that problem. Because “reasonable” was not fixed in advance, courts have had to define the negotiation process and rates. The Court of Justice of the European Union set out negotiation steps in Huawei v. ZTE (2015). UK courts set a global royalty rate in Unwired Planet v. Huawei, a decision upheld by the UK Supreme Court in 2020. Unwired Planet made no phones. Ericsson had transferred 2,185 patents and applications to it in 2013 in exchange for a share of Unwired Planet’s gross licensing revenue: 20% of the first $100 million, 50% of the next $400 million, and 70% of everything above $500 million. The practice is called privateering: an operating company hands patents to a firm with no products, which can assert them without risking a countersuit over products of its own.

The Bayh-Dole Act, the 1980 law that lets universities and contractors patent federally funded research, includes “march-in rights.” The government may require additional licenses when statutory conditions are met. For more than four decades no agency used them. The NIH denied every march-in petition it received, including petitions concerning the prices of Norvir and Xtandi. In August 2025 Commerce Secretary Howard Lutnick told Harvard that the department was starting the march-in process, a step no agency had taken before. His letter cited late invention disclosures, licenses that did not require substantial U.S. manufacturing, and failure to bring inventions into practical use. As of early 2026 Commerce had announced no determination. No agency has ever completed a march-in, over price or on any other ground, and a lever that never fires disciplines no one.

When the term runs out

Every case above concerns the price of access while patents are in force. A harder problem starts when they expire. A patent can end on schedule while the monopoly it built keeps going, because the holder spent the exclusive years gaining control of whatever a customer would need in order to switch.

Humira is the clearest recent case. AbbVie’s base patent on adalimumab, the drug’s active ingredient, expired at the end of 2016. A further 132 U.S. patents from 20 patent families, together with settlements with would-be competitors, kept U.S. biosimilars (near-copies of a biologic drug) off the market until January 2023, more than four years after they reached Europe. When they arrived, pharmacy benefit managers, the intermediaries that decide which drugs a health plan covers, kept Humira on their lists in exchange for rebates. Biosimilars held about 1% of U.S. adalimumab use in November 2023, according to IQVIA. Their share barely moved until CVS Caremark dropped Humira from its major national commercial formularies on 1 April 2024. By that month one biosimilar, Sandoz’s Hyrimoz, held 14% of U.S. adalimumab claims. For ordinary pills, where pharmacists substitute generics automatically, Hemphill and Sampat found that effective market life hardly changed from 2001 to 2010, averaging about 12 years, and their evidence suggests generic challengers held it there by targeting weak, late-expiring patents. The League reads the difference as a question of who controls the switch.

Bell shows the same pattern with a different lock. A 1956 consent decree required Bell to license its existing patents royalty-free. Follow-on invention building on those patents rose, mostly among young and small firms, but the lasting increase came only outside telecommunications equipment (Watzinger, Fackler, Nagler, and Schnitzer). Inside that market, Bell’s operating companies were the customers, and they bought from Bell’s own manufacturer, Western Electric. Free patents opened every field except the one where Bell also controlled the buyers.

A pool, a flat rate, a buyout, or a prize sets the price of using a patent, and none of them decides who controls substitution when the patent ends. Where the holder or an ally controls it, a license has to come with a second rule that opens that channel: automatic substitution at expiry, access for rival products to the holder’s buyers, or release of the data the holder gathered as the only supplier.

What patents can provide

A granted U.S. patent publishes claims and a technical specification. Under 35 U.S.C. § 112, the specification must describe the invention in enough detail for a person skilled in the field to make and use it. Individual patents still vary in clarity and practical usefulness, so “published recipe” is too strong a description. McLean’s container patent, U.S. 2,853,968, is available to anyone with a browser. Unlike a trade secret, its claims and specification remain public after the exclusion period ends. That public record is also what lets a court or a government find a patent, read it, and open it, as the Bell decree did. Control held through contracts, data, or secrecy has no term and no public text to act on.

Temporary exclusion can also help a young company obtain financing. Patent applications are assigned to examiners in a process that is close to random, and examiners differ in leniency. Farre-Mensa, Hegde, and Ljungqvist tracked 34,215 first-time startup applications through those assignments. Startups whose first patent was granted showed 55% higher employment growth and 80% higher sales growth over the following five years. The authors attribute much of the difference to improved access to venture and bank funding. For a small entrant facing an incumbent, the patent works as collateral and as a shield against being copied at scale.

The same asset can pull a startup toward the incumbents it might have displaced. In a study of 118 startup projects, startups holding at least one patent were about 23 percentage points more likely than those without patents to commercialize by licensing their technology or being acquired, instead of competing on their own (Gans, Hsu, and Stern). That is a correlation, and cooperation can be efficient, but a patent that finances an entrant does not guarantee a new competitor.

Patent holders can also license broadly to encourage adoption. McLean made his container patents available royalty-free during ISO standardization so competitors could build compatible equipment; the profile documents that history. The 1856 sewing-machine pool and the 1917 aircraft pool also placed multiple patents under a common licensing arrangement, although they arose under different pressures and used different terms.

Available policy and licensing tools

Between full exclusion and abolition sits a shelf of instruments. Each has run somewhere, and each solves a different problem.

• Statutory flat rate. In 1909, concerned that piano-roll reproduction could consolidate under one company, Congress created a compulsory mechanical license for published songs. A new recording could be made under the statute if the user followed its conditions and paid the set rate. The rate began at two cents per copy and did not change until 1978. A statutory rate therefore needs a credible review process.
• Private flat rate at the interface. Avanci licenses cellular patents from dozens of holders to carmakers at one per-vehicle price: $15 when it published its first 4G rate in December 2017, $20 for new 4G licenses signed from September 2022, and $32 for its 5G program ($29 for early signers). Avanci reported that its licenses covered more than 325 million connected vehicles as of September 2026. The carmaker pays once per vehicle regardless of how many covered patents or components it uses. Because the total charge is tied to the vehicle rather than the number of declared patents, adding another patent does not increase the bill. The vehicle program licenses only carmakers. Suppliers that build the connectivity modules cannot take a license of their own, although Avanci says the carmaker’s license covers them. Continental, one of those suppliers, sued in 2019, arguing that the patent holders refused to license component makers on FRAND terms; the courts dismissed its antitrust claims (Fifth Circuit, 2022, unpublished).
• Buyouts. France bought the daguerreotype patent in 1839 and put photography in the public domain. Kremer’s 1998 mechanism generalizes it: auction the patent to discover its private value, pay the inventor that value times a social markup, and release most purchases into the public domain.
• Prizes and commitments. An advance market commitment promises, before a product exists, to buy it at a set price if it meets defined conditions. Donors committed $1.5 billion to a pneumococcal vaccine program in February 2007. The program launched in 2009 and contracted with GSK and Pfizer at $10.50 per three-dose course. Gavi, the vaccine alliance that administered the program, reported that more than 215 million children had been immunized by the end of 2019. The contracts tied payment to qualifying vaccine supply rather than control of downstream research.
• Proposed self-assessed pricing. Posner and Weyl propose in Radical Markets (2018) that a patent holder name a license price, pay a recurring fee proportional to that price, and license at the stated amount. Under the proposal, setting a high price to deter use would also increase the holder’s recurring fee.

The League’s recommended approach

Match the instrument to where the invention sits.

• Products with documented development costs. Keep exclusion where evidence shows that copying would otherwise prevent the inventor from financing development. Keep it short and enforced, and pair it with correction levers that fire on published criteria. March-in, which no agency has completed in more than four decades, shows what a lever without criteria does.
• Technology included in a shared standard. Convert automatically. Entering a standard should switch the patent from exclusion to a blanket license priced per product unit, with the rate on a fixed review schedule and open to every layer of the supply chain. Avanci shows one private version of this structure.
• A defined outcome known before research begins. Pay for the result with prizes, advance market commitments, or patent buyouts. The pneumococcal vaccine program shows that an advance commitment can operate at scale.
• An open interface supported by an operating business. Hold the patent briefly as a coordination instrument, open the interface, and earn from implementation, manufacturing, support, or services. McLean and Hock illustrate different versions of that approach, and it requires no legislation.
• A holder that controls the switch. Where the holder or an ally controls substitution through a formulary, captive buyers, an installed base, or data gathered as the only supplier, pair any license or remedy with a rule that opens that channel at expiry. Automatic generic substitution at the pharmacy is the working version for ordinary pills.
• Granted patents that are held without active use. Test self-assessed pricing as a backstop, so that holding a patent in reserve for litigation carries its own cost.

None of this abolishes the patent. Every piece already runs somewhere: a 1909 statute, a per-vehicle pool, a Gavi contract, state generic-substitution laws. The work is assembly, plus keeping the rate-setters awake.

Scope. These directions draw on cases from different industries and legal systems, and they have not been tested as one combined system. A complete proposal would need to name the jurisdiction, covered patents, rate setter, review schedule, appeals process, and evidence used to change the terms. The patent-assignment study supports an early patent for some young firms and says nothing about the right exclusion period. What would change the verdict: biosimilars substituted automatically at the pharmacy taking the market as slowly as adalimumab did under formulary control, or a remedy aimed only at patents ending a dominant position while the holder kept control of its buyers.

Sources: Venetian Patent Statute (1474); Primary Sources on Copyright: commentary on the 1474 statute; Statute of Monopolies (1624); Statute of Monopolies, section VI (legislation.gov.uk dates the Act 1623 because Acts then counted from the first day of their session, which opened in February 1624, still 1623 under the old calendar; royal assent came on 29 May 1624); 35 U.S.C. §§ 101–103; Wikipedia: Manufacturers Aircraft Association; Katznelson & Howells, “The myth of the early aviation patent hold-up” (Industrial and Corporate Change, 2015); Wikipedia: Sewing Machine Combination; Mossoff, “The Rise and Fall of the First American Patent Thicket” (Arizona Law Review, 2011); Lampe & Moser, “Do Patent Pools Encourage Innovation? Evidence from the Nineteenth-Century Sewing Machine Industry” (Journal of Economic History, 2010); Bessen & Meurer, “The Direct Costs from NPE Disputes” (Cornell Law Review, 2014); Light Reading: Ericsson sells 2,185 patents to Unwired Planet (2013); Unwired Planet Form 8-K on the Ericsson revenue share (January 2013); Lemus & Temnyalov, “Patent Privateering, Litigation, and R&D Incentives” (2015); 17 U.S.C. § 115; Avanci Vehicle; Avanci: more than 325 million connected vehicles covered (September 2026); IAM: Avanci publishes its first 4G vehicle rate (December 2017); Avanci 4G rate change (2022); Continental Automotive Systems v. Avanci, No. 20-11032 (5th Cir. 2022); CJEU, Huawei v. ZTE, C-170/13 (2015); Unwired Planet v. Huawei, [2020] UKSC 37 (summary); Brachtendorf, Gaessler & Harhoff on declared essentiality (JEMS, 2023); CRS, “March-In Rights Under the Bayh-Dole Act”; GAO-26-107885, “Intellectual Property: Information on Draft Guidance to Assert Government Rights Based on Price” (February 2026); IPWatchdog, “Lutnick Warns Harvard of Plan to March In on Patents” (August 2025); Commerce Department letter to Harvard on march-in (August 2025); Mayor & City Council of Baltimore v. AbbVie, 42 F.4th 709 (7th Cir. 2022); IQVIA, Humira biosimilar tracking summary for the Biosimilars Council (2024); GaBI: adalimumab biosimilar share after formulary changes (January 2025); KFF Health News on Humira rebates and pharmacy benefit managers (September 2023); Hemphill & Sampat, “Evergreening, patent challenges, and effective market life in pharmaceuticals” (Journal of Health Economics, 2012); Watzinger, Fackler, Nagler & Schnitzer, “How Antitrust Enforcement Can Spur Innovation: Bell Labs and the 1956 Consent Decree” (AEJ: Economic Policy, 2020); U.S. patent 2,853,968 (McLean); Farre-Mensa, Hegde & Ljungqvist, “What Is a Patent Worth?” (Journal of Finance, 2020); Gans, Hsu & Stern, “When Does Start-Up Innovation Spur the Gale of Creative Destruction?” (RAND Journal of Economics, 2002); Kremer, “Patent Buyouts” (QJE, 1998); Gavi: pneumococcal AMC; Gavi: pneumococcal vaccine support; Posner & Weyl, Radical Markets (Princeton, 2018).