Discover more from Risk & Progress
Rethinking how we approach intellectual property
Risk & Progress| A hub for essays that explore risk, human progress, and your potential. My mission is to educate, inspire, and invest in concepts that promote a better future for all. Subscriptions and new essays are free and always will be. Paid subscribers gain access to the full archives.
The search for the proper balance between incentivizing innovation and the dispersion of new ideas has vexed many. How we design our “innovation infrastructure” is crucial to expanding and enriching human advancement, with implications for technological progress, economic development, and human well-being. However, existing IP law is inflexible and flawed. New ways of managing IP may be on the horizon.
Why Do We Have IP?
A patent or copyright is a temporary government-sanctioned monopoly on an idea or product. For patents, this monopoly is typically 20 years, providing an inventor a temporary exclusive right of profit, after which the idea is placed into the public domain for anyone to use and improve upon. The intent is to allow the inventor to capture the positive externalities of their work and internalize them, for a time.
Patents exist because markets fail to properly incentivize innovation on their own. Typically, the cost of developing a new idea or product is high, while the cost of copying it is comparatively low. In a world without IP, copycats can reproduce an idea near marginal cost, leaving the inventor unable to recoup their initial expense. Absent patents, most will rationally choose not to innovate or jealously guard their IP behind trade secrets, preventing its diffusion.
This is especially a problem with pharmaceuticals. The cost to develop a new drug is extremely high, often over $1 Billion, but the marginal cost to produce it is usually rather low. Absent patent protection, the smart entrepreneur will rationally wait for the competition to develop a new drug and copy it, undercutting the inventor. Copying, not invention, is rewarded. Patents and copyright protections attempt to solve this by using the legal system to temporarily protect an idea from being copied.
The Failures of Patents
Critics contend that patents create almost as many problems as they solve. They protect innovators but then prevent humanity from fully reaping the benefits of their new ideas. As one example, the patent pricing of AZT, a drug used to treat AIDS, priced many people of modest means out of treatment and left millions of children with preventable illnesses.
Patents can also perversely incentivize duplicative research, as competitors attempt to “patent around” existing patents to avoid paying royalties. This creates immense waste in research and development dollars. Worse still, patent “trolls” hold patents, not in the name of innovation, but by exploiting ambiguity and the high cost of litigation to extract money from the innovators themselves.
Additionally, patents can lockout follow-on innovation. With monopoly pricing, the cost to buy or license a patent is such that most cannot afford to do so. As a consequence, would-be innovators, ready to improve upon an idea, remain on the sidelines for years, if not decades, until the idea enters the public domain.
What are the Alternatives?
There are alternatives to patents that have some success. Direct Investment from the government, as I discussed here, could be targeted at the bridge between basic and applied research through a DARPA-like organization, avoiding direct competition with the private sector. Direct investment can have a complementary role by investing in areas that are deemed too risky, and/or returns too distant.
Prizes, where the government or private entities offer a cash prize for inventors to demonstrate new technology, also have a history of limited success. Cash prizes could be offered for socially beneficial innovations like life-extension medications, reusable rocketry., or battery breakthroughs.
The problem with Direct Investment and Prizes is that they require the government to foresee what ideas society needs and accurately determine a value for it. As history illustrates, it’s not clear that the government can do this reliably. In late 1970s America, for example, garage computer hobbyists invented the first desktop computers, kickstarting the PC era and an entirely new industry.
Around that time, the Soviet Union also had garage tinkerers, but in the command economy, were unable to tap VC funding or start a business; they had to seek bureaucratic support instead. Like many at the time, bureaucrats could not see the value in a personal computer and predictably refused to provide funding. As a consequence, the computing industry took off in America and stagnated in the USSR.
Such is a cautionary tale for those assuming that the government can successfully pick and choose which innovations to nurture. Thus, Prizes and Direct Investment would likely still leave significant gaps in the innovation infrastructure. These gaps require us to explore creative means of improving intellectual property, namely patents.
Patent Buyout Auctions
One solution to the above-noted patent problem was proposed by Michael Kremer in 1997: Patent Buyouts. In his proposed system, an auction is used to determine the value of a patent, whereupon the government buys and open-sources it. As opposed to Direct Investment, the government purchases the output of research, rather than directing money at the input.
This concept has historical precedent with Daguerreotype photography, invented in 1837, the first means of taking photographs. The government of France recognized the monumental importance of this new technology and made the decision to purchase the patent and place it in the public domain.
Now free to use and improve by anyone, the technology rapidly spread across Europe. Within months, the process of taking photographs was translated into dozens of languages. Chemists all across Europe quickly improved upon the technology, far faster than they would have had the patent not been in the public domain.
Kremer’s proposal works like this: When a patent is filed, it would automatically go into a public auction. The auction would be a Sealed Bid Second Price Auction (SBSPA). In an SBSPA, bids are submitted in written form without knowing the other bids in the auction. The highest bid wins, but only pays the second highest bidder’s price.
Unlike traditional first-price auctions where bidders are trying to guess what other parties are offering, the SBSPA incentivizes participants to bid the patent’s true value only. An honest and true valuation is crucial to the functioning of the system. The key is that the government would also submit a bid, but the government’s bid would take the private value of the patent, determined by the auction, and add a multiplier.
Why? New ideas create positive externalities that cannot be fully captured privately. If they could, inventors would be better incentivized to innovate even more. The data suggests that the social return on R&D is at least twice the private return. Thus, Kremer suggests a multiplier of 2x that would roughly account for the positive externalities that the private sector cannot capture.
By default, the government’s bid would win the auction. The patent holder could elect to sell or retain the patent. If sold, it would be placed into the public domain for the benefit of humankind. The beauty of patent buyouts is that inventors will be rewarded for their work, while society gets to immediately enjoy the fruits of new ideas.
Kremer’s System is Impractical
The problem with Kremer’s system is that it may be vulnerable to manipulation. Since the government is buying patents at a significant markup, there is an immense risk that bidders and inventors will collude, either explicitly or implicitly, driving up patent valuations and forcing the government into excessive and wasteful spending.
Kremer recognizes this and suggests that to keep bidders honest, the government steps aside on random auctions, forcing the highest bidder to take on the patent at the price they had bid, creating risk for colluders. As an additional measure, the government might occasionally base its valuation on the third-highest bid rather than the second, forcing the cooperation of three parties rather than two, making manipulation more difficult and risky.
With that said, even with safeguards, it is too easy for three CEOs to meet on a golf course and arrange a “patent pumping” scheme that would net billions in profit, even if they occasionally would take losses. There is simply too much upside to manipulation and collusion, and little corresponding downside.
Additionally, there is an open question as to whether such a system could ever be affordable. In 2019, the United States spent $667 Billion on R&D, both privately and publically funded. Even if we assume that the output of this R&D is only valuable at face value, a 2x buyout markup brings total spending to ~$1.3 Trillion annually. But in a world with patent buyouts, we should expect more patent generation. Even assuming no collusion, a patent buyout system could cost trillions of dollars every year.
Patents as Probabilities
Ian Ayres and Paul Klemperer propose a more radical solution to the patent problem: Probabilistic Patents. They argue that an efficient patent policy should provide “constrained” market power as opposed to an absolute one. This is, the government should allow a limited amount of patent infringement that balances returns to inventors with social costs. One means of achieving “constrained” market power is to combine uncertainty with delay of IP enforcement.
Why is this? Probabilistic determination means that some “infringers” will make the calculation that the risk of violating a patent is worth the potential profits. A few “infringers” will enter the marketplace and reduce the monopoly power of the patent holder. Only a limited number of players would step in because the market would naturally reach a point where the risk of “infringement” would outweigh the potential for profit. Thus, competition would be limited and would not wipe out the patentholder’s profits or eliminate their incentive to innovate.
Ayres and Klemperer developed a simple linear model that illustrates how a small amount of uncertainty reduces deadweight loss (the aforementioned social costs of patents) by much more than it reduces the patentees’ expected profit. For example, if the probability of enforcement is 95%, the market price of the patent is cut by 10% and the social cost is reduced by over 18%.
But won’t this reduce the incentive to innovate? It certainly does by a small amount, but we can compensate for this by lengthening the term of the patent. For example, reducing the probability of enforcement to 90 percent would require lengthening the patent term by just 3.4 percent to achieve the same profits for the patent holder, while still reducing the social cost of the patent by roughly 30 percent. In short, consumers of IP benefit from oligopolistic pricing for a longer period than they do under monopoly pricing for a shorter one.
While a probabilistic patent system is likely to face significant legal hurdles, a similar outcome could be achieved through a Duopoly Auction. In that system, a patent would confer not only the right to an idea but also a mandate to auction off that patent to a second owner. The inventor would still receive profit from its own sales alongside a hefty lump sum payment at the patent auction. Consumers would benefit from access to lower-cost duopoly-priced patents rather than monopoly-priced ones.
Ayres and Klemperer indirectly identified what I call the “Dark Side of Property,” or what is known as the Tragedy of the Anticommons. There is a kind of “spectrum” of optimal property rights. No ownership results in high allocative efficiency (anyone can use the property) but low investment incentives. When ownership rights are too strong, however, investment incentives are high but allocative efficiency is low, so the property is not used in the most efficient ways. At both binary extremes, the end result is not optimal.
The optimal pathway for property is to recognize that property rights are not binary and exist across a continuum. That is, the best property is not commonly owned or privately owned, but rather a hybrid of both. We need not roll the dice on patent enforcement, however, we can introduce the same partial ownership directly through the tax system.
The Harberger Tax
One way to do this is via Harbeger Taxation. Patents are legal monopolies and the monopoly protection enjoyed by the inventor is a pure creation of the sovereign state. As such, it may be socially optimal to reframe patents as partially publically owned. The owner has the right to profit from his idea but must pay a portion to the sovereign authority that made this possible.
A small annual tax, almost a kind of “lease” from the sovereign state, can transform patents from a monopoly into partially publicly owned property. As in Kremer’s system, the challenge lies in accurately assessing the value of the IP. That’s where Harberger taxes come in. To determine that valuation, we turn to the IP owners themselves. They self-assess the value, and the government charges an annual tax based on that valuation.
Now, won’t IP owners asses the value low so as to reduce their tax burden? They certainly will, which is why anyone can buy the IP at the self-assessed value at any time. The combined constraints of self-assessment and a recurring tax force a broadly honest valuation.
A patent holder for a widget, as a hypothetical example, might want to value their patent at $10 billion to prevent anyone from buying it. But with a 2.5 percent annual tax, for example, they would have to pay $250 million a year to keep their monopoly protection. Instead, they will rationally choose a realistic $100 million valuation, incurring a reasonable $250 thousand annual levy that is worth paying for patent protection.
Like probabilistic patents, a small degree of uncertainty has a big impact. A tax of just 2.5% percent on property would significantly improve social welfare. Under this regime, IP can be purchased at a fair value without the “holdout problem” of monopoly pricing, thus allocative efficiency is dramatically improved. At the same time, the tax is not too high so as to completely deter investment/innovation. Crucially, this recurring tax would discourage rent-seeking behavior, including patent-trolling, by imposing a cost on sitting and holding patents unproductively.
Modifying Kremer’s Proposal
For all of its systemic flaws, Kremer had identified that, if one can accurately value a patent, copyright, or idea, we have a basis upon which to either open-source it or make it liquid and tradeable. The question becomes 1) How best to value the IP and 2) Which IP, if any, should be open-sourced? Can we refine his concept and make it practical?
Working his system backward, we must first ask if it is truly necessary for the government to pay a 2x markup on a patent. I question the value of striving to permit IP owners to fully capture the positive externalities of their work. As discussed in Stanford Lawyers magazine, we do not permit the full internalization of social benefits in any other property realm. Thus, it is somewhat disingenuous to place such an extreme demand on IP.
Take the example of planting a flower on your front lawn. As the landowner, existing property law does not permit you to hunt down a passerby and charge them for the beautification that the flowers provided. The market need not provide an absolutely perfect capture of social benefits, only enough to cover the fixed and marginal costs of producing IP.
Realizing this, we may tweak Kremer’s system and lower the markup paid to <20 percent above market value instead of 100. Alternatively, it could be abolished entirely. Doing this has two effects. First, it significantly lowers the cost of the system. Second, it reduces the upside incentive for participants to manipulate that system, further lowering the cost.
Working backward further, a second reform might be to invert the proportion of patents/IP that the government purchases. That is, instead of purchasing >80 percent of patents, as Kremer advocated, a minority of around 20 percent are placed into the public domain. This eliminates concerns about manipulation as colluders are more likely to lose than win, and further reduces the financial cost.
Third, inventors that refuse the sale of their IP, must pay a Harberger tax on the value as determined by either the offers of purchase or self-valuation. From that point on, the IP remains available for purchase by parties who feel that they can make better use of it. If it is purchased by another party, the new owner pays the Harberger levy instead.
To the extent that the tax may dull inventors’ profits, we can borrow from Ayres and Klemperer and extend the length of the typical patent term.
A Practical Alternative
So how would such a system work in practice? Inventors can self-assess the value of their IP or register for auction. Either way, the moment that the IP is valued, it is open for purchase by the government or the private sector. If the government chooses to purchase the IP, it will pay the markup value (if any) above the valuation, and open source it for all to use.
This combines the best of Kremer’s Patent Auctions, without fear of manipulation and at an affordable cost. Because the government’s purchases could engender a markup, inventors will strive to create socially beneficial inventions that are more likely to be purchased.
Of course, private entities and non-profits can also bid for IP. This gives private players, who may be able to make more productive use of that IP, a chance at acquiring it. In other words, the IP market will become somewhat liquid, improving the allocative efficiency that is quite limited under the current regime with monopoly pricing. The tax would likely incentivize IP owners to more readily license their IP at a lower cost.
One might imagine that private players could also band together into Associations to purchase IP that is beneficial for all of its members. For example, airlines might partner with aircraft manufacturers to purchase a new patent for a fuel-saving jet engine turbine blade. By spreading the cost across multiple firms, all parties benefit financially from a relatively small investment.
In the end, as in Kremer’s system, the IP owner has the right to refuse the sale. The caveat is that refusing the sale of the IP engenders a Harbeger levy of 2.5 percent on the higher of the self-assessed value or the auction value. This levy generates some revenue to fund patent buyouts, charging inventors proportionally to the cost of the maintenance of their IP.
At any point thereafter, the IP remains available for purchase but the valuation can always change. For instance, should a higher bid be received and the owner again refuses to sell, the new bid value becomes the baseline of the Harberger levy. The higher the valuation, the higher the levy will become. IP owners can avoid paying the Harberger levy at any time, all they need to do is open-source the patent or copyright for all to use.
This Solves IP’s Challenges
This system would address most of the issues facing IP. Because IP would be subject to valuation and taxation, it would not be rational for orphaned works to exist. Conceivably, most would immediately be put into the public domain. Similarly, the existence of a recurring tax on IP would discourage patent trolls as they would now face potential losses.
Additionally, it would also greatly simplify and reduce the cost of IP litigation. In infringement cases, courts’ roles would be limited only to adjudicating whether or not an infringement took place. They would be freed from the arbitrary task of determining the damages suffered, as this number would be publically available.
Finally, it would make the IP market more liquid, giving the government and private players the ability to purchase and/or open-source key innovations much more quickly and easily than they can under the current system, greatly accelerating the diffusion of new ideas and technology into society for improved social welfare and faster progress.