In spring quarter, I took “Market Design” from Eric Budish. This was my final paper:
Markets for Ideas
Despite their central importance for economic growth, markets for ideas remain woefully understudied. This paper aims to describe extant markets for ideas and subject them to analysis using ideas from the market design literature.
Briefly, a market for ideas involves two sides – buyers and sellers. For simplicity, we can assume that sellers are inventors, and buyers are firms (although additional trading can occur between firms alone, without the inventor). The purpose of a market for ideas is to match an idea (or invention) with the entity best-suited to developing, producing, marketing, and selling them (for simplicity, we will refer to this bundle of activities as developing). Any number of simple stories can account for this need to trade, i.e., why the original inventor may not be best-suited to develop an idea: there may be complementarities between ideas, advantages from economies of scale, or merely advantages to a division of labor.
Although both the structure of the problem as well as the salience of market design tools are both very apparent, the current literature context here is thin. There is a long and robust market design literature, but almost none of it has tackled this particular market. Likewise, there is a long economics literature on markets for technology, but its analyses have yet to embrace tools from market design. One notable attempt has been made to marry these two literatures, and attack the markets for ideas problem – Gans and Stern (2010) identify three unique features of markets for ideas from the economics literature (idea complementarity, user reproducibility, and value rivalry), and use Roth’s (2008) three market design criteria (thickness, lack of congestion, and safety) to analyze these features’ effect on the market in question. Though fairly cited six years later, few if any authors have followed Gans and Stern in this particular vein of work.
This paper seeks to build upon Gans and Stern (2010) by describing the dominant set of extant markets for ideas, and subjecting them to analysis using Roth’s three criteria described above. Our contribution is twofold: an initial roadmap as to fruitful directions for additional markets for ideas research, and some preliminary implications for policy/practice.
The paper is organized as follows. Sections II and III examine the role of intellectual property rights (IPRs) in a market for ideas, and in particular examine a world absent IPR, and the U.S. patent system, respectively. Sections IV and V assume an IPR regime, and explore two recent and novel market designs provided by private companies – an auction design operated by Ocean Tomo and ICAP, and a consulting arrangement operated by InnoCentive. Section VI concludes with suggestions for future work, and implications for policy.
- Absent IPR
The vast majority of human history has operated without any formal intellectual property rights. Indeed, the first patent system did not arrive until the 1400s, and the first U.S. patent was not granted until 1790. Throughout this time, lawyers, economists, and policymakers have debated whether patents are even necessary: for example, Jaffe and Lerner (2004) hone in on a notable episode of debate in the Dutch system at the end of the 19th century. Even now, well-respected scholars make serious arguments for abolishing the patent system. Thus our analysis begins not with a moot point, but an object of both historical and present-day interest.
The primary issue with such a system is the lack of market safety, which breeds a thin market. As Gans and Stern (2010) point out, there is a major user reproducibility problem with ideas, differentiating them from physical goods. More specifically, without intellectual property rights, it is nearly impossible for an inventor to exclude others from copying the idea. Consequently, inventors have a difficult time appropriating the full value of their invention, and are less willing to spend their time doing the costly work of inventing in the first place.
This would be a difficult enough problem if inventors were also always developers, since secrecy is difficult to maintain; the problem is exacerbated when we want there to be trading/selling of ideas. Because an inherent property of a new idea is that it lacks comparables, potential buyers have no way of valuing a given idea without seeing it in action. But this very act compromises the secrecy of the idea, and the inventor has no recourse against potential buyers “stealing” it, and paying nothing for the privilege. Nor can we remedy the situation by establishing prices before such an idea-demonstration, ex ante – in such a situation, the inventor has every incentive to misrepresent the true value of the idea, and charge exorbitant prices for what s/he knows to be a dud.
We can fix ideas with a simple example. Suppose that you are deciding whether to spend $40,000 researching and inventing a new energy-efficient lightbulb. You are not a manufacturer, so you plan to sell the lightbulb “blueprints” to a manufacturer once invented. Suppose further that you know the true value of this invention to be $1 million; that is what any competent manufacturer could expect in profits by producing this lightbulb. After inventing the lightbulb, you face a dilemma – you shop around the prospect of buying the blueprints to various manufacturers and ask your fair price, $1 million, but no one believes you! Hucksters come around all the time, making similar claims, and then making off with the money in exchange for non-working blueprints. Furthermore, you know that if you show your blueprints to the manufacturers, establishing their authenticity and value, they are smart enough to go produce it on their own without paying you a dime. With no sellers willing to buy sight-unseen, and you unwilling to have your invention stolen, you decide not to spend $40,000 inventing the new lightbulb in the first place (despite a potential $1 million payout, and the associated social welfare effects).
This is what economists refer to as a market unraveling, with similar phenomena posited by classic papers like Akerlof’s market for lemons (1970). Clearly we can play with the numbers in the above example, and relax various assumptions, but the central point still holds – we have far too few sellers, and far too few buyers, since there is no mechanism for establishing value without introducing the chance for theft.
There are two primary ways around this problem. The first is by establishing some other form of monetary reward, provided not by the market but by government. Since we cannot prevent people from reproducing an idea freely, and we want to properly incentivize the production of the idea, we might think that central governments can simply offer the “true value” as a reward to the first inventor who comes up with a given invention. Indeed, there is wide historical precedent for such a system – proponents often point to the invention of a method for determining longitude at sea, induced by prizes from the British government, or more recently, DARPA’s Grand Challenges (which arguably kickstarted driverless car technologies). Unfortunately, economists have pointed out that such a system is potentially riddled with problems. First is the measurement problem: it is very difficult for government, even a very informed and technically adept one, to place the proper value on an invention ex ante. Too low, and inventors will not be induced to invent; too high, and soon the funding agency will run out of money. Ex post solutions, like awarding the inventor the equivalent of a licensing fee, get far closer to an equilibrium outcome but can be costly to calculate. Most damningly, however, governments are simply ill-equipped to predict all of the inventions that will be most socially useful, and it is infeasible to offer tens of thousands of prizes per year. So, while this may work for special cases (and DARPA’s success indicates as much), it is unlikely to solve the market as a whole.
Instead, we might solve the unraveling problem via a trusted third-party intermediary that both buyers and sellers can rely upon to a) fairly value the invention, and b) prevent unauthorized “theft” of the idea, or reproduction absent sale. We find just such a solution in the patent system.
- S. Patent System
The first U.S. patent was granted in 1790, to Samuel Hopkins. Since that time, the patent system has evolved in minor and significant ways, and indeed has hardly ever remained static for long. Furthermore, almost every country has its own patent law, each with its own idiosyncratic granting and enforcement process (e.g., China has a notoriously weak IPR regime). Exploring the full range of variations, over time and across countries, is far beyond the scope of this paper. Instead, we will focus on the current incarnation of the U.S. patent system, as of time of writing, and try to identify the most salient features from a market design perspective.
First, we provide a brief sketch of the U.S. patent system as it stands today. Any individual, or team of individuals, may apply for a patent on any product or process. The three requirements for granting are that the idea must be novel, non-obvious, and useful. Once a patent is granted, the duration is twenty years, during which time the patent holder has exclusive rights to manufacture, produce, or sell the invention. These rights may be transferred in any fashion, to any number of other parties, so long as those rights are not overlapping (e.g., you cannot sell the exclusive rights to manufacture in California to two different parties). Critically, the entire patent need not be sold – the patent holder may sell pieces of exclusivity by geography, time, etc., or do the same by licensing. The limits are essentially only in the creativity of the contract.
Should any party infringe upon these exclusive rights, the patent holder has the right to sue. Relief typically comes in two forms – injunctive relief, by which a judge can force the infringer to stop the manufacture/sale of the product, and damages, which are a monetary payment that may either cover what a license would have provided and/or may be further punitive (e.g., due to lost sales, or changed product strategies).
This system is a very deep and complex web of idiosyncrasies, and an entire research literature could be devoted to its study with relation to market design; therefore, we will confine our analysis here only to the most prominent features described above. With regards to the granting of patents as a property right for ideas, the system (ostensibly) acts as a first-line verifier of quality for potential sellers. That is, if someone tries to sell you a patent, there must be some floor of quality that it has already cleared. More importantly, the system’s remedy for infringement via litigation effectively solves the unraveling problem, giving inventors a recourse if potential buyers attempt to “steal” or reproduce the idea without paying. Thus, even if the patent office literally granted every patent application coming through the door, and served as no indicator of quality whatsoever, market participants could still safely put up an idea for valuation and sale without risk of going uncompensated (although the litigation system would strain under the weight). Consequently, the patent system is an obvious improvement over the no-IPR scenario with regards to two aspects of market design – the market is far safer, and so naturally thickens. Sellers are induced to invent in the first place, and then feel comfortable putting the invention on the open market without appropriability concerns, and buyers can then freely assess the quality of a given idea and offer that value to the seller.
With this added market thickness, however, comes new problems of congestion. In particular, the addition of more sellers (or ideas) into the market makes it progressively more difficult for buyers to find what they are looking for – since there is no good clearinghouse for patents, as the USPTO listings make little attempt to organize or simplify them (and patent language is typically dominated by legalese), search costs quickly become very high. In addition, although patents are in principle supposed to be completely novel, in reality many patents are granted which are overlapping. As the number of sellers rise, the number of overlapping patents also rises (and indeed may do so nonlinearly), which introduces additional difficulty for buyers – they do not know if they will require a bundle of patents, or if one relevant patent will suffice, and there is no way to check prior to litigation if the buyer is 100% in the clear. These additional costs prevent the market from growing as large as it should on the buy-side, and skew the distribution of patent-holders to larger firms which can more easily bear those costs.
This tension between thickness (stemming from safety) and congestion is at the core of our analysis, because it suggests that the patent system must make a tradeoff between favoring patent holders and favoring patent buyers. In a no-IPR scenario, the market completely favors patent buyers, because there is no fear of litigation and no safety – therefore the market is thin, since sellers do not want to participate. But moving to a patent system is not merely a binary decision. The easier it is for patents to be granted, coupled with ease of litigation (and probability of winning damages), increases the market safety for patent holders but drives up costs of doing business for buyers (through congestion-related search costs, overlapping-patent costs, and risk tolerance for potential litigation), in turn thinning that side of the market. We posit therefore that a given patent system lies on a spectrum, that that position on the spectrum forces a tradeoff between desirable market design features, and that the likely welfare curve takes on a U-shape, with an optimal system balancing safety, thickness, and congestion.
While it is incumbent upon the USPTO and Congress to adequately address the litigation issues raised above (and explored in depth by other authors), private actors have had some luck in resolving the problem of search costs, even as the market explodes in size.
- Ocean Tomo and ICAP
In 2006, the private firm Ocean Tomo claims to have held the first open auction for patents. It operated the auction as a first-price English auction with ascending prices. For sale were only patents which had previously been presented to and vetted by Ocean Tomo’s internal patent review team, and this limited set of patents had to meet high (undisclosed) standards of value. Similarly for the buy-side, Ocean Tomo required registration from potential buyers, including fixed fees in the thousands of dollars for the privilege of bidding on a given patent or patent lot.
In 2009, ICAP (the world’s largest broker) acquired Ocean Tomo’s Transactions division, which ran the auctions. ICAP took over these activities, switching over to a sealed-bid auction, but detailed information on these new auctions appears unexplored by researchers. The exclusivity on both the seller and buyer side persisted, though – clients have included prominent IP players such as NASA and Foxconn.
The market design literature is rich in describing the advantages of auction systems, broadly construed. First, we should not have strong preferences between the types of auction offered by Ocean Tomo, then ICAP; each should lead, theoretically, to equivalent outcomes. In particular, the strength of auction systems comes down to market safety – representing one’s true value in bidding for an object is a dominant strategy, with all other “gaming” being inefficient and not in one’s best interest.
Furthermore, Ocean Tomo and ICAP provide an additional value-add service by vetting the patents for quality, bundling them into relevant portfolios, and providing brief descriptions as to their purpose. In this fashion, apart from any aspects inherent to an auction design, these private firms reduce search costs for buyers. Even as the market thickens, a trusted-third party acting to bundle the goods and simplify the valuation mitigates issues stemming from congestion. From this perspective, the auctions run by Ocean Tomo and ICAP are ideal next steps in designing effective markets for ideas, since they both guarantee market safety beyond litigation-remedies and make strides toward reducing congestion.
Some secondary problems arise, however, from these designs. Specifically, there should be some cases in which neither buyers nor the broker-as-intermediary are best placed to value an idea – indeed, there must be some cases in which the seller has a better idea of the true value, and auction systems provide no incentive for that information to be made public.
More pressingly, an auction system like one of these will inevitably run into the same thickness/congestion constraints described above, in the more general market for patents. Ocean Tomo and ICAP have artificially constrained the market by setting an undisclosed floor on value – only sellers with patent portfolios exceeding that floor, and buyers with capital sufficient to bid past that floor, can participate in the market. If the firms were to reduce these constraints, however, and thicken the market, we should expect some of the regular congestion problems to return – it would remain difficult for buyers to assess the true value of a given patent, and bid accordingly. In particular, it may be the case that Ocean Tomo / ICAP can perform their bundling and valuation services only because their fee is high enough from these high-value patents – there is no guarantee that such an incentive will exist for a private actor across the entire spectrum of patent quality, and theoretically there must be some patents of sufficiently low value as to not merit such services at all.
Although it is clear that an auction design, like that run by Ocean Tomo or ICAP, is both feasible in markets for ideas and desirable from a market safety standpoint, it nonetheless does not do enough to solve the tension between market thickness and congestion beyond the scope of very high-value patents or patent portfolios.
Spun off from Eli Lilly in 2005, InnoCentive is an innovation consulting firm which helps individual companies find solutions to pre-defined problems. Roughly, the process works as follows: a client company will provide InnoCentive with a pressing problem that it wishes solved, InnoCentive will reach out to its network of scientists and inventors with details of the problem, and finally InnoCentive will identify the “winning” idea and facilitate any necessary transfer or licensing of intellectual property.
Since InnoCentive only works with one client firm at a time on a given problem, we have termed this model a monopsony matching market – there is one buyer for ideas, but multiple sellers, and InnoCentive’s service is to match the best seller with the buyer as given. The economics literature has engaged with this model to a small extent, but because this is a consulting arrangement and not a commonly applicable design, the market design literature has not engaged with markets of a similar type as of yet.
From the perspective of Roth’s criteria, this design looks very similar to the auctions explored above. The principle advantage of this system is market safety – because this is a repeated game for the sellers, and InnoCentive acts as a gatekeeper for who gets to participate, the sell-side has very strong incentives to be honest about the true value of their inventions. Very simply, if a successful seller has misrepresented the value of their idea, and that misrepresentation is assessable any time after the sale, that seller will be excluded from any further possible matches by having fallen out of favor with InnoCentive. Unfortunately, the market is similarly thin as the auctions run by Ocean Tomo and ICAP – the buy-side is necessarily a monopsony simply because of InnoCentive’s business model, and the size and efficacy of the sell-side is limited by InnoCentive’s personal and likely idiosyncratic network. The thinness likely avoids congestion problems, but there is not even the bundling and assessment mechanism in place from Ocean Tomo’s and ICAP’s model to compensate for increasing complexity if InnoCentive were to engage a larger portion of the potential market.
Overall, while InnoCentive’s design may indeed be beneficial for a given firm and a given problem, the ex ante limiting of the buy-side as well as the absence of a mechanism for reducing congestion concerns severely constrains the ability of this design to be effective, or more importantly, scalable.
This paper has explored four forms of market design for markets for ideas, and their predicted properties with regards to Roth’s three criteria – market thickness, lack of congestion, and market safety. We conclude with suggestions for areas of future work, and implications for policy and practice.
Although we have covered four principle designs for markets for ideas, one of the main questions left unanswered is whether other forms of markets for ideas exist, and how and why they might be superior (or inferior) to the types described herein.
Two particularly fruitful areas for study unexplored by this paper include Nathan Myhrvold’s suggested market for ideas, which approximates a VC-like structure, and considering patents as securities, as recently suggested in the legal literature. Furthermore, scholars would be well-served by studying what small tweaks in markets for ideas might lead to especially large effects in effective market design – e.g., the CAFC reforms of the 1990s appeared administratively small, but ended up having a large impact on the efficacy of the market. Researchers would do well to be able to predict the effect of such changes ex ante.
With regards to market design research veins more specifically, we note most emphatically that our focus on Roth’s three primary criteria need not be the central focus going forward. Although this provides a useful starting place, these criteria are by no means the be-all and end-all, and scholars should ask what features of a successful marketplace are particularly salient in the context of markets for ideas (e.g., Gans and Stern focus on repugnance, which we have left untouched in this paper). Critically, this literature would be best served by careful empirical analysis of the presence of these criteria, and their probable effects on outcomes (whether that be forward-citation weighted patents, product outcomes, or something else). We can tweak theory and assumptions endlessly, but often hard data can lead us in the most fruitful directions, rather than trying “every bottle on the shelf.”
Finally, this analysis holds some preliminary implications for policy and practice. First, it is apparent that we are not anywhere close to either the perfect market for ideas, nor having exhausted all ideas about the possible forms for that market. As for the patent system itself, it is apparent that the government has a lot of room for small changes that might yield large improvements in market design, as the market appears right now to be too tilted toward patent holders. As for building systems atop intellectual property rights, we would be well-served by more experimentation like Ocean Tomo and InnoCentive have done, and especially if these experiments are run by a social welfare maximizing central planner, i.e. the government.
This paper has been an early attempt at a large, largely unexplored, but highly important task – applying the tools from market design toward designing an optimal market for ideas. Tremendous work remains to be done, but we hope that this paper serves at least as a spur for others in pursuing this critical line of inquiry.
 Jaffe, Adam and Lerner, Josh, Innovation and its Discontents (2004). Princeton University Press.
 Akerlof, George A. “The Market for” Lemons”: Quality Uncertainty and the Market Mechanism.” The Quarterly Journal of Economics 84.3 (1970): 488-500.
 Jaffe and Lerner, ibid.
 Note that we provide only a very brief sketch here. Thorough treatments of the U.S. patent system, from granting to litigation, are the subject of entire books.
 It is of course the case that information asymmetries can induce bidding away from true values. If the seller has more information, s/he can ask for and receive a higher price than the idea is worth. If the buyer has more information about the quality of the idea, which we might expect to happen especially in the case of idea complementarities (i.e., the buyer owns complementary inventions, which the seller has no knowledge about), then the buyer can bid for and receive a lower price than the idea is worth. Remedying this asymmetry should be a key task for market designers going forward.
 E.g., it is not uncommon for patent applicants to hire specialists to conduct searches of prior art, and it is further not uncommon for these searches to fail to yield all relevant material
 Jaffe and Lerner, ibid.
 Fischer, Timo, and Jan Leidinger. “Testing patent value indicators on directly observed patent value—An empirical analysis of Ocean Tomo patent auctions.” Research Policy 43.3 (2014): 519-529.
 Lakhani, Karim R., et al. “The Value of Openness in Scientific Problem Solving.”
 Myhrvold, Nathan. “The big idea: funding eureka!.” Harvard Business Review88.2 (2010): 40-50.
 Risch, Michael, Patent Portfolios as Securities (September 19, 2013). Duke Law Journal, Vol. 63, p. 89, 2013; Villanova Law/Public Policy Research Paper No. 2013-3022. Available at SSRN:http://ssrn.com/abstract=2227103