1.2.1. The key question of appropriability of returns for innovation

What economists call “club goods” are the result of the combination between non-rivalry and the possibility to preclude usage. The economic model of club goods consists of requiring payment in exchange for access to the good. It is therefore possible for the individual or company who produces a club good to receive a part of their return from the use by each individual to whom access to the good has been granted. The compensation for the producer of this kind of good is therefore proportionate to the number of individuals who are willing to pay the asking price, and when applicable, the number of units that each individual wants. A live performance is generally a club good. Provided that the number of spectators does not damage the perception that each individual has of the performance, there is a non-rivalry. It is also possible to exclude individuals from accessing the location where the performance is occurring. On the other hand, when it is not possible to exclude usage, the individual or company who produces the good finds themself unable to derive returns from individual beneficiaries through an appropriate price structure. It is therefore not a viable economic model in the context of private production. An air show is an example of this. Even if, like the live performance, there is indeed non-rivalry as long as the number of spectators remains limited so that no one interferes with one another while watching the air show, it is generally impossible to prevent individuals outside of the airfield from watching the show. In the absence of market incentives to produce them, goods that have both properties of non-rivalry and non-excludability are produced either directly by public authorities or indirectly by public commission of private actors. What about innovations? To answer this, it is necessary to distinguish different steps in the process of creating innovations and, at the same time, clarify the term “innovation”.

Innovations are a link in a larger process of knowledge creation in which we can identify three major steps:

• – the domain in which they are implemented;
• – the primary activity that characterizes them;
• – the result that they produce.

The first step falls under the domain of science and is characterized by research that can be either purely speculative or oriented toward a concrete goal but its result is always a kind of scientific discovery. The application perspectives are generally too distant in time to be correctly predicted or even suspected. Their production therefore cannot rely on a market process. As scientific discoveries are a base for the following steps, it is nonetheless necessary to ensure their production with adequate incentives. The economics of science tends to highlight the role of peer evaluation in the careers of scientists, even simply the personal satisfaction derived from moral recognition by others or from the very fact of having solved a problem². The absence of rivalry in the use of scientific knowledge is obvious. There is a possibility of direct excludability, for instance by controlling access to conferences or to publications that present discoveries. In contrast, there is generally an impossibility for indirect excludability, in the sense that we cannot often prevent a person who has had access to the discovery from passing on the discovery to others. Consequently, scientific production is essentially supported by public funds.

The second step falls under the domain of technology and is characterized by the production of inventions that make it possible to solve a practical problem. The concrete character of the problem solved makes it possible to consider a more or less short-term application for the invention and therefore a potential market value. There is no rivalry in the use, as one application can be developed without preventing other applications. The possibility to exclude is, however, subject to discussion. It can be imposed by force³. It also partly depends on the technological field under consideration and the possibility for reverse engineering in that field. This consists of finding the concept for the invention by examining what is produced thanks to the invention. This is a common practice in the engineering industries. It is more difficult in the tire industry, which relies on tacit knowledge, appropriable with difficulty, as opposed to explicit knowledge. Where the possibility of excludability is not imposed by the very nature of the inventions, notably by the difficulty to proceed with reverse engineering, it can be imposed by law. This can happen through prohibitions. Over time, the law has moved more toward the attribution of rights. This is how the issue of patents, the subject of this book, emerged. Before discussing this issue in more detail, we will present the third step of the process.

The third step falls under the domain of the economy, and is characterized by the creation of value within a society. The creation of value can be achieved in a marketable form, such as through the successful commercialization of a new product, or a non-marketable form, through the dissemination of best practices regarding how to produce a good. It is only when the creation of value becomes effective that we are talking about innovation. It then becomes important to distinguish innovations, which are inventions that have created value, from inventions in general that have not, or not yet, led to the creation of value. In addition, innovations are not exclusively technological. A typology by object, directly inspired by the definition given by Schumpeter [SCH 11], led to the present distinction between four types of innovations (Oslo Manual [OEC 05]):

• – Process innovation (implementation of a new production technique). Thanks to this kind of innovation, a company can produce an existing good for a lower cost than its competitors. This type of innovation follows a logic of vertical differentiation in production tools. A classic example is the process invented and patented by the English engineer and inventor Henry Bessemer in 1855 to manufacture steel in a more efficient way.
• – Product innovation (commercialization of a product that offers new features or responds to needs that were hitherto not satisfied or poorly satisfied). Thanks to this kind of innovation, a company can be the only one to supply the new good on the market. This type of innovation follows a logic of differentiation that is at least as much horizontal as vertical. A relatively recent example is the development of mobile phones in the 1990s and then smartphones in the 2000s. The iPhone from Apple Inc., commercialized as of June 2007 in the United States, was the first smartphone with a touch screen interface available at that time.
• – Organizational innovation (rethinking the organization of tasks, human resources, decision procedures, and client and supplier relations). Thanks to this kind of innovation, a company can reduce its production inefficiencies or informational inefficiencies. This type of innovation is the responsibility of management and can be paired with an engineering logic. One widely documented example is the Taylor system adopted for the assembly-line production of the Model T by the Ford Motor Company from 1908 to 1927.
• – Marketing innovation (modification to the design of a product or the way of selling it). Thanks to this kind of innovation, a company can succeed in attracting new customers. This type of innovation follows a marketing logic. The site Amazon.com introduced the idea of an online library in 1994.

The different types of innovation are not incompatible. It is possible to combine a product or process innovation, an organizational or marketing innovation. For instance, it is with the batteries developed by its subsidiary Batscap that the Bolloré group produces the Blue Car, the iconic model of the car-sharing service known as Autolib’ implemented by the city of Paris. In this example, there is a product innovation based on batteries. However, to create value from this product, rather than selling the batteries to automobile manufacturers, the group decided to offer a turnkey solution to the city of Paris that included a network of charging stations and the supply and maintenance of the vehicles for that network. This is an organizational innovation that makes it possible to circumvent the issue of the critical mass of users beyond which a network becomes profitable. This organization breaks with the strategy of automobile manufacturers who are often content to offer electric car models without getting involved in setting up charging stations networks, with the noticeable exception of Tesla Motors. The organizational innovation led to a marketing innovation because the firm generates its income not through vehicle sales but through automated, short-term rentals to customers.

The problem of the impossibility of excludability raised by the second step of the development of applications takes on its full meaning when inventions are likely to result in innovations, meaning the creation of value. However, it should be emphasized that the impossibility of excluding is contingent on the application; for instance, it is stronger where reverse engineering is simple. When this impossibility prevails, it is very difficult for inventors to get financial compensation through the commercialization of their invention even if it has a high value for the community. This difficulty, referred to as the problem of appropriability of returns from the innovation, comes from the early entrance of competitors copying the invention at low cost, so that the rent of the inventor is very quickly eroded⁴. The competition forces the selling price down to a level that covers the production or reproduction costs of the invention, but does not often allow the inventor to cover her own costs for research and development. Since the inventor is the only one to be subjected to the cost of R&D, she can find herself in a paradoxical situation where imitators earn profits, even small ones, while she is losing money. If this problem is anticipated by the inventor, she may refrain from engaging in the activity of R&D. The community will then suffer from a lack of incentive because innovations whose social value exceeds the cost of R&D will not be created. The problem of the appropriability of returns from the innovation by the inventor is at the heart of the economic analysis of patents⁵. In order to best understand the importance of patents to respond to this, it is important to highlight two points. The first point is, as noted by Schankerman and Schuett [SCH 16], that a good patent system is able to only target innovations that are appropriable with difficulty, and which would not be created without this system. Innovations that are not subject to this difficulty not only do not need a patent system, but also should not solicit one in order to avoid it serving to claim undue rents. The second point is that there are alternatives to the patent system to remedy the flaw of appropriability. It is essentially information imperfections that create a context where the patent system appears a priori to be more specifically appropriate.

1.2.2. Patents as a solution for the lack of appropriability

In a utopian world where information would be perfect, public authorities would be able to identify the innovations that suffer as a result of this lack of incentive and to plan for their realization. They would also be able to identify which inventors are capable of realizing a given innovation at the lowest cost of R&D and to precisely evaluate that cost. In such a world, the public authorities could thus overcome the lack of incentives with the planning of inventions and the efficient allocation of funds for R&D. But the real world is made of imperfections, especially in the quality of information and its distribution among economic agents. In many ways, it is these information flaws that are at the basis of innovation policy and that underlie the patent system.

The first fundamental flaw comes from the inherently random character of any R&D activity. It is difficult for any person other than the researchers themselves to determine if an R&D program fails because the researcher encountered insurmountable difficulties or whether the absence of results reflects a lack of effort or skill on the part of the researcher. This problem is one variation among many in economics of the so-called “moral hazard” phenomenon. To encourage an appropriate level of effort on the researchers’ part, it is essential to implement incentives, to make their compensation and career somewhat dependent on obtaining results. In the case where the research in question occurs upstream in the process of innovation, and therefore more in the domain of science, the incentive is necessarily largely disconnected from the value created, which is itself highly uncertain. It therefore generally takes the form of a job promotion and an increase in autonomy at work. In the case where the research in question occurs downstream and where its impact in terms of value creation is more explicit, the incentives are generally more tied to profit-sharing⁶. The exact forms of this profit-sharing can a priori be very diverse, and notably, nothing makes it possible to say at this stage whether an innovation prize is more or less appropriate than a patent. The arguments to choose between one or the other are based on a second informational problem.

The second informational flaw that it is important to consider in matters of innovation is linked to the social returns created. Unless a specific scientific or technical problem has been clearly identified as an obstacle to be overcome in order to implement one or several applications that create value, the returns generated by an innovation are most often only correctly understood, and assessed, after the invention has been developed. Similarly, different competing solutions are often studied simultaneously until one of them manages to take the lead. When the automobile was being developed at the end of the 19th Century, it was not clear that the internal combustion engine that ran on liquid fuel would win out. Natural gas, electricity and steam were all considered as options for power before proving less interesting⁷. It is also the case that inventors often have a better idea of the advantages and disadvantages of their inventions, if only through the prototype testing that they carry out, and consequently of their chances of success. There are three consequences to this. First, innovators may be more inclined to take on the risks inherent to an invention, based on a more favorable belief in its success. Second, it is generally more difficult for the community to evaluate what is worth supporting or not before the invention is available and begins to spread. Third, inventions can have belated success, unforeseen applications, or a context that is favorable for their development may not emerge until later⁸. It is therefore often admitted that the incentive mechanism must operate ex post, must moderate compensation as a function of the observed value and not only the predicted value, and must be generic, in the sense that the compensation need not be negotiated for every invention. This perspective advocates for a technological neutrality for incentives, in the sense where they must support innovations but without making an a priori choice between the various options initially considered. The patent system responds to these different criteria and also offers a solution for the problem of temporal inconsistency that is likely to affect support for the innovation.

Like the moral hazard mentioned above, the problem of temporal inconsistency applies in economics to many more questions than simply innovation. Applied to innovation, it refers to the idea that it is in the interest of public authorities to incentivize innovation ex ante, before an innovation has emerged, by committing to compensate inventors when they are successful, but not to respect this commitment ex post. Indeed, once an innovation is available, the collective interest would be to spread it as widely as possible at the lowest cost, and therefore promote its free imitation by the inventor’s competitors. The problem is that if they anticipate this reversal, inventors will not trust commitments formulated ex ante and the incentive will be ineffective. To avoid the problem of temporal inconsistency in incentives, it is essential that commitments made ex ante be binding. For this, they must pass through mechanisms in which the discretionary power of public authorities is low. To do this, the support system must be established at a high level in the hierarchy of standards, which is the case for patents, which rely on a system of intellectual property rights (IPRs). IPRs are basically a bundle of legal standards used as a means of protection, compensation, conservation, and valuation of rights related to an “intellectual creation”. As will be detailed later in this text, because this type of right has major particularities, it is a public institution or independent administrative authority, namely the patent office, that grants and regulates intellectual property titles⁹. It is this relative independence vis-à-vis the legislative and executive powers¹⁰ that limits the issue of temporal inconsistency and guarantees a certain efficiency of patents as an incentivizing tool¹¹.

A third information issue pertains to the apparent contradiction of designing a support system that is generic, but should only address innovations that, without such a system, would not be created. It is important to remember that a “good” patent system is a system which is only solicited for this type of innovation. However, a patent office cannot immediately identify innovations that it should target and rule out the others. For this, a mechanism encouraging inventors to self-select has to be put in place. One strategy to achieve this is to make the inventor pay to obtain a patent, through filing¹² and processing¹³ fees, for example. The appealing feature of a patent is to allow for the appropriability of returns from the invention in case of success. When appropriability is already high, the gain supplement provided by a patent is relatively low and there is a strong chance that it will be canceled out by the filing fee if that fee is expensive enough. Conversely, when the appropriability is initially low, the gain supplement provided by a patent is relatively high and it has a good chance of staying positive once the filing and processing fees are subtracted. These fees therefore serve to discourage filing patents on inventions for which the appropriability is initially high, while encouraging it for inventions whose appropriability is initially low. This view assumes, however, that appropriability is an exogenous factor in every invention. In practice, appropriability is constructed and, more specifically, the ability to exclude competitors is not totally independent from the inventor. A protection strategy based on secrecy, for example, can consist in dividing the activity of R&D between different centers, even different countries, in order to prevent a competitor from easily copying the invention by poaching researchers. Such a strategy is costly and, in the presence of a generic patent protection mechanism, the trade-off could be made in favor of the latter. This means that patents tend to lessen the excludability effort by other means, such as secrecy or lead time, and are therefore subject to the moral hazard phenomenon which fuels the belief that patents would be obtained for inventions that do not require them. Patents also have another quirk that has long attracted attention: they confer a market power to their holder, and this power is exercised to the detriment of the community and must be balanced with the incentives created. From this point of view, patents (and beyond that, intellectual property rights) contrast strongly with more classic property rights, such as land ownership rights.

1.2.3. Patents and their design

The controversy over the merits of patents is old, as it struck most European countries between 1830 and 1870. Some authors refer back to the origin of patents to highlight the ambiguities to which they are subject. In this vein Frumkin [FRU 45], recalled that in different countries of Western Europe, kings granted privileges (“Litterae Patentes”) in the form of rights to economic exclusivity in very varied domains¹⁴. An opposition developed against these rights, particularly in England, due to the obstacles that they presented to the freedom to pursue economic activities. Rights protecting inventions, or more generally the development of new industries, were a notable exception to this opposition. The Statute of Monopolies adopted in 1624 in England is often considered to be a founding act for the modern patent system (Holdsworth [HOL 45]), especially because, as highlighted by Machlup and Penrose [MAC 50], it proposed the idea of granting exclusive rights to the first inventor. France, in 1791, followed closely by the United States in 1793¹⁵, also instituted the patent system for inventors and would be imitated by most European countries over the course of the 19th Century. Two important elements appeared in the historical analysis conducted by Machlup and Penrose [MAC 50]. The first element involves the way in which patents were perceived. Although for the English, patents emerged as an exception to the rule of the limitation of monopolies, for the French, inspired by the Revolution of 1789, they were considered to be the “natural”¹⁶ right of every individual to freely dispose of their person and their work¹⁷. This brings up the question of the nature of these property rights which, as we will explain later on, cannot be considered “standard” rights even if some authors like Demsetz [DEM 69] make that argument. The second element concerns the seniority of an anti-patent movement based on the absurdity of their presence in economies that promote free trade¹⁸. Thus, citing costs to obtain patents, and the uncertainty of obtaining them, demand grew for laws that were more favorable to inventors in England. This provoked a hostile counter-attack to patents that almost led to their abolition in the second half of the 19th Century, based on the argument of Chevalier [CHE 62], according to which an invention is a collective phenomenon, and whatever one inventor may have found, another person could have discovered¹⁹. This contestation failed because of a return to protectionist ideas, starting from 1873. The adoption of patents was delayed during the 19th Century in Germany. Conversely, the Netherlands voted to abolish patents in 1869²⁰ and joined countries like Denmark, Switzerland²¹, Mecklenburg, Turkey, and Greece, who would do without invention patents²². It is therefore false to think that questioning patents is a recent phenomenon. On the contrary, patents have long been considered ambivalent and economists have questioned their relevance, including during the 20th Century²³. As Machlup [MAC 58, p. 80] noted, “on the basis of present knowledge, [no economist] could possible state with certainty that the system [of IP] as it now operates, confers a net benefit or a net loss upon society. […] If we did not have a patent system, it would be irresponsible, on the basis of our present knowledge of its economic consequences, to recommend instituting one. But since we have had a patent system for a long time, it would be irresponsible, on the basis of our present knowledge, to recommend abolishing it”.

From the modern perspective, patents can be defined as exclusive rights to exploit a patented invention. This means that a patent confers to its holder the possibility of forbidding others to use, produce, import or sell within the territory where the patented invention is protected without the consent of the patent holder. From a legal point of view, the patent is an industrial property title that provides its holder not with the right to exploit, but the right to forbid third parties from exploiting a patented invention. Patents confer the right to use (usus), profit from (fructus) and to dispose (abusus) of the patented invention for a certain duration and in one (or more) determined territories, within the limits of the claims contained in the technical description of the invention, in exchange for disclosing the invention. Among other things, the usus implies that the patent holder can choose not to develop or attempt to commercialize the invention, but simply make use of a patent to prevent others from doing so²⁴. Combined with the fructus, it also allows the patent holder to share a market while controlling the product price. In practice, the company can grant the use of licenses²⁵, which is to say they can allow third parties to develop and commercialize the invention in exchange for financial compensation. Finally, the abusus allows the patent holder to sell or decide to terminate the patent by not paying the renewal fees. Patents are not de facto rights but de jure rights and, aside from different parameters detailed later which make it possible to delimit them, the conditions required for their issue play a fundamental role in assessing the social advantages they can offer and their disadvantages.

In order to have a real incentivizing role for innovation and preventing them from resembling undue privileges to the detriment of the community, patents are not granted unless they satisfy three criteria²⁶:

• – novelty, which means that an invention must not have been achieved before²⁷;
• – non-obviousness (United States) or inventiveness (Europe), which means that the invention must not be obvious for a person skilled in the art;
• – utility (in the United States, Australia, and Canada, among others) or industrial application (Europe)²⁸, which means that the invention must be complete, so that it can be used and applied.

The first criterion requires that what is patented does not already exist before it is patented or that its existence was not known to the applicant, the patent office, or more generally, specialists in that field. The first major implication is that this excludes anything that exists naturally from being patented. In this sense, the applications of scientific discoveries can be patented but the scientific discoveries themselves cannot be because they only reveal the laws of nature that already existed, even if they were not known. Similarly, living organisms have long been kept outside of the field of what is patentable²⁹. This principle was initially illustrated in the United States by the Ex Parte Latimer decision of 1889³⁰. The patent filed in 1873 by Louis Pasteur on a yeast was an exception, but at that time, yeast was not considered a living organism. On the other hand, since the 1980 decision in Diamond v. Chakravorty, that which is living but has been transformed by a person or obtained through the intervention of a person can be patented in the United States (Darr [DAR 81]). Although this rule was applied in some European countries in the past, the European Patent Convention (EPC) in 1973, which is still in effect, excludes it, despite several attempts to find loopholes (Llewelyn and Adcock [LLE 06]). The implementation of the criterion of novelty requires defining the state of the art when the request is filed, which is partly the responsibility of the applicant, who must indicate relevant pieces of information in the application, and partly the responsibility of the patent office, which must conduct anteriority searches in order to establish the state of the art in the technical domain of the invention. Today, in Europe as well as the United States, the principle of first to file applies in case of a dispute over two patents on the same invention. Prior to 2013, the principle of first to invent applied in the United States. This meant that a patent filed at a later date by an inventor who could provide proof of the prior existence of the invention could prevail over a patent filed earlier by a competitor. It should be noted that the principle of first to file is only valid if the invention was not disclosed by people other than the applicant before the filing, even without filing for a patent, in which case the criterion of novelty would not be satisfied.

The criterion of inventiveness means that the patented invention must not be obvious to a person in the field or, more generally, for a “person having ordinary skill in the art” (shortened to the acronym PHOSITA). The idea is that a substantial effort of reflection and codification should be necessary to create something patentable. In this way, we avoid patents on inventions that would have had a very strong chance of emerging without needing to be generated by a new patent. Eisenberg [EIS 04] highlights a subtle difference between simply ensuring that the invention is not obvious to a PHOSITA and the criterion of inventiveness. A PHOSITA can never be an inventor, which suggests that the degree of effort that underlies the concept of inventiveness is more demanding than the notion of non-obviousness for a PHOSITA.

The third criterion can be interpreted differently depending on whether the concept of utility or industrial application is applied. At first glance, the American concept of utility refers to the fact that the invention responds to a need. As noted by Machin [MAC 99], however, it is only the broad understanding of this criterion (general utility) alongside which other meanings coexist, such as specific utility, practical utility and moral utility. The concept of general utility can pose a problem, notably because it requires that what the invention contributes to society be proven from the time of the patent application. Machin [MAC 99] argues for an interpretation that is more in line with the highly uncertain nature of the innovation through what he calls prospective utility. This consists of assessing the probability and scope of any advantages provided to the community by the invention. For Erstling, Salmela and Woo [ERS 12], who discuss the different interpretations of the utility of an invention by different patent offices, the European approach is very similar to Machin’s prospective utility (Machin [MAC 99]). One notable difference is that, if it does indeed suggest that the invention must have the potential for application, then the European approach limits these applications to the industrial domain, and consequently to only product or process innovations. This does not mean that, in the European approach, there is no other kind of innovation than industrial innovation, but rather that the patent system is only adapted for this type of innovation. In Europe, software and business methods are not patentable as such but can, at least for software, be protected by copyright. In both Europe and the United States, the applications considered by the patent applicant are included in the patent description document in the form of what are called the patent claims. By describing and detailing a specific foreseeable application, each claim implicitly identifies a potential market for the invention. In this sense, the claims contribute to delimiting the patent, but they are only one component of the delimitation.

Establishing criteria that make it possible to say whether an invention can be patented is not enough to define patents. What is covered by the patent must also be delimited. The delimitation takes place in three dimensions: i) length, ii) width, and iii) height.

1. i) The length of patents corresponds to delimiting the duration of patents. As already noted, patents have ambivalent effects. On the one hand, they have an incentivizing role and make it possible, at least in theory, for the community to benefit from innovations that would not have been created in their absence. On the other hand, because they convey the exclusive right to exploit an invention, they are the source of a market power that is exercised to the detriment of the community. The case of product innovation demonstrates this negative effect most clearly. Because of the exclusive rights of sale on a new product, the patent holder can increase his price above the marginal production cost, at the risk of seeing some consumers refrain from purchasing it, but without worrying that consumers will head to other suppliers. The relatively low losses in volume compared to the increase in the margin for each unit encourages the patent holder to increase their price above the marginal production cost³¹. Compared to a situation where the incentive of competition pushes a company to maintain a price close to the marginal cost, this results in smaller quantities at higher prices for consumers. The decrease in quantities signifies that globally, producer and consumers combined, the value created for the community is lower. The producer increased their profit by capturing a more substantial part of the value created for consumers but did so by reducing the total amount of this value³². By limiting the duration of patents, the public authorities make a temporal trade-off between the incentive to innovate and the limitation of market power. There is therefore a sequential solution to the problem. Nordhaus [NOR 69, 72], taken up by Scherer [SCH 72], demonstrated that a socially optimal trade-off leads to a patent lifespan that is not zero, but also is not infinite. Today, the maximum legal duration of patents granted in both Europe and the United States is 20 years from the date of filing the application. It is the same regardless of the technological domain. This principle of undifferentiated lengths (called the “one size fits all” principle in the literature) is, however, lessened by the patent renewal mechanism³³. To keep a patent in force, the holder must pay renewal fees at regular intervals³⁴. A default on the payment will lead to the patent falling into the public domain. It is therefore possible that a patent will last less than the legal maximum of 20 years. The data about renewals used by Schankerman and Pakes [SCH 86] show that a significant proportion of German patents (from the 1952 to 1978 cohort), French patents (from the 1951 to 1979 cohort)³⁵ and English patents (from the 1950 to 1976 cohort) were abandoned well before the 20-year term expired. Cornelli and Schankerman [COR 99] also demonstrated that this renewal system made it possible to solve a problem of information asymmetry, related to the inventiveness criterion, that the patent office encountered. The goal of the criterion of inventiveness as described above is to promote inventions that require an effort to be obtained. However, it is very difficult for a patent office to immediately identify which inventors have the strongest innovative capacities. By proposing different combinations of duration and cost for patents, with the cost increasing more than proportionally with the duration, the patent office drives inventors to reveal their performance in R&D. Well-performing inventors know that they will obtain a high-value innovation relatively more easily than poorly performing inventors and will accept paying more to keep their patent longer than them. Therefore, renewal fees that increase more than proportionally with the duration of the patent make it possible to offer a greater financial incentive to inventors who perform better and produce inventions that require a greater degree of inventiveness. Gans, King and Lampe [GAN 04] demonstrated that the renewal system is more appropriate, because it is more flexible, than a system of fixed filing fees whose price would be dependent on a firm time commitment as soon as the dynamics of returns earned on the patent are stochastic rather than deterministic.
2. ii) The breadth of patents is less clearly defined than their length³⁶. At least, it is represented rather differently in models presented in the theoretical literature of economics, depending on whether these models address it in terms of process innovations or product innovations and, in the latter case, whether they place emphasis on horizontal or vertical differentiation. For a process innovation, the breadth measures to what point competitors can be inspired by the patented innovation to reduce their production costs. In the model presented in Nordhaus [NOR 69], taken up again in 1972 [NOR 72] and by Scherer [SCH 72] and Dore, Kushner and Masse [DOR 93], when the marginal production cost is constant, the breadth de facto sets the marginal cost gap between innovators and their competitors, and therefore the cost benefit that innovators can exploit. For a product innovation, Klemperer [KLE 90] was one of the first to define the breadth based on the concept of horizontal differentiation. For this, generally, it must be considered that products are distinguished from one another by characteristics, with the optimal combination of these characteristics varying from one consumer to the next. The question that is posed to companies offering or wishing to offer the good revolves around positioning themselves in the space of these characteristics. The breadth will set the distance, in the space of the characteristics, that competitors must respect in relation to the new patented product. This presumes that the characteristics are objectively measurable. For instance, in the case of a phonographic device, we can consider at least two characteristics: the capacity to render high frequency sounds and the capacity to render low frequency sounds. A new phonograph might innovate through its capacity to render low frequency sounds even if it is revealed to be less effective at rendering high frequency sounds. It is therefore directed at a particular segment of the customer base that prefers low sounds over high ones. The breadth of a patent on such a device is supposed to set up to what point, in terms of improving the rendering of low frequency sounds, the competitors can approach without infringing on the patent, but it would not apply any restriction on competitors concerning improving the rendering of high frequency sounds. However, there is also a vertical differentiation of products. This is operative when a new product is distinguished from previous ones by a certain degree of improvement to all of its characteristics. At an identical price, it would be preferred by all consumers and not only one segment of the consumers. This would be the case of a phonograph that increased its capacity to render both high and low frequency sounds. The breadth of a patent on such a device is supposed to set up to what point a competing product can improve the rendering of low and/or high frequency sounds without infringing on the patent. This improvement would also need to be obtained in the same way as described in the patent. Yet, it is possible for a competitor to get close to the same performances by circumventing the patent if the way of obtaining the results does not occur through exactly the same technical solutions. This requires an effort on the part of the competitor to develop a solution that is not necessarily obvious and could therefore justify a competing patent. This is why some authors like Gallini [GAL 92] argue directly that the breadth is reflected in the cost of circumventing a patent in order to imitate without infringing. It is also the case that competitors can, without circumventing the patent technical solution, improve and surpass the product. The degree to which they must surpass the performances in order to consider that the patent has not been infringed corresponds more to the height of the patent.
3. iii) The height of a patent is often likened to its breadth. However, we can define it differently from the breadth based on the idea that it measures the degree of inventiveness for a patent to be granted that was described above. It takes on its full meaning as soon as the question of sequential innovation is addressed, as in Green and Scotchmer [GRE 95], Moshini and Yerokhin [MOS 08] or van Dijk [VAN 96], even though the latter authors discuss the breadth of a patent rather than the height. The height is then a key tool that public authorities can use to balance between, on the one hand, incentivizing the development of a first-generation innovation that opens up the field to a whole series of subsequent inventions at the risk of curbing incremental innovations and, on the other hand, incentivizing the acceleration of the appearance of second generation inventions at the risk of discouraging the initial invention on which they are based. It is often incremental inventions that really make it possible to move on to commercialization and industrialization, or in other words to progress from invention to innovation. That is why it can be so important to promote them and, in doing so, limit the height of patents. It is important to ensure, in the context of sequential innovation, that the very system of patents remains optimal, a point that Bessen and Maskin [BES 09b] emphasize. The sequential character of innovation also lends meaning to the obligation to disclose the content of the patent. This obligation appeared very early with the idea of protecting inventions. Frumkin [FRU 45] attributes it to King Henry II of France based on the publication of an invention dating from 1555. Matutes et al. [MAT 96] examine how the inclusion of the disclosure of content affects the optimal design of a patent, but by focusing on its length and scope, with the later concept referring more to the number of claims than the height. The intent of the obligation to disclose is to stimulate second generation innovations by giving competitors access to knowledge which could otherwise be hidden from them. From the point of view of the inventor of the initial innovation, this obligation constitutes a disadvantage to using a patent as opposed to a secret to protect the innovation. It plays a key role, as discussed by Hall et al. [HAL 14], in the decision for inventors concerning whether or not to use a patent. These same authors also insist, in the same vein as Lemley and Shapiro [LEM 05], on the probabilistic nature of the issue of a legal dispute if a patent is infringed, which calls into question the effectiveness of delimiting the rights conferred by a patent.

1.2.4. Are patents a property right like any other?

If patents are part of property rights, it must be said that rarely are other property rights so contested, even within societies where economic activity relies directly on property and the market exchanges that it permits. The contrast with property rights on more tangible assets is important. A systematic comparison with land-based property is offered below in order to stress the specificity of intellectual property rights, and patents in particular.

A first explanation proposed by Stiglitz [STI 08] and detailed by Bessen and Meurer [BES 09c] is that, contrary to many other property rights, the boundaries of what is covered by a patent can be unclear. For instance, it is easy to delimit the property of land-based goods and to establish a land registry in the form of a detailed map that indicates very precisely where the boundary between what belongs to different owners is located. It is often much more difficult to set the boundary of what is covered by a patent. One famous example is the patent US Patent 821393 obtained in 1906 by the Wright brothers for the control system of a flying machine that used a wing torsion system, a patent that mentioned the possibility of using other methods to obtain the same result but did not detail them (Freiwald [FRE 13]). In 1908, Glenn Curtiss filed a patent on a means to obtain the same effect using ailerons, a system that is still used today. A legal dispute followed between the Wright brothers and Curtiss, with the former accusing the latter of having infringed their patent. Notwithstanding the vague formulation about “other means to produce the same effect” in the text of the Wright brothers’ patent, they won the case in 1913. Today, historical studies tend to consider that this issue put aeronautical engineering in the United States at a disadvantage compared to other countries, especially France, which applied a much narrower interpretation and in doing so, did not discourage competitors from pursuing innovation. Freiwald [FRE 13] also explains that at the time when the Wright brothers won their case, their solution was already technically outdated. It was not until the implementation of a patent pool in 1917 that the American aeronautics industry succeeded in surpassing the inherent difficulty of a broad interpretation of the delimitation of patents (Bittlingmayer [BIT 88], Szalkalski [SZA 11])³⁷. The difficulty of establishing the boundaries of what is patentable is even stronger when the technology is complex and hard to codify. Hall et al. [HAL 14] see this as an explanation for the industry differences in the propensity of innovative companies to patent. Inventions in chemistry are often more easily codifiable (in the form of chemical formulas), which would explain why innovative companies in this sector patent on average more easily than in other industries.

A second explanation for the contestation of patents as property rights involves the non-rivalrous nature of what is patented. This characteristic, intrinsic to the very function of patents, contrasts with the features of property rights on more tangible assets, especially land ownership rights. Notably, it translates into the fact that an invention is only protected by a patent to the full extent of what the patent permits within the territory covered by the protection. It is important to specify that the criterion of novelty applied by most patent offices prevents a priori that a patent on the same invention be filed by a competitor in another office³⁸. On the other hand, it does not allow the holder to exclude competitors from using what is patented outside of the territory covered by the protection of the patent. To do so would involve requesting an extension of the patent from other offices, provided that they accept. Moreover, the interpretation of what is covered by a patent can differ substantially from one country to another. This explains why the Wright brothers’ patent did not allow them to exclude their European competitors, since the patent offices of different European countries maintain, as indicated above, a narrower interpretation of what is patented than the US patent office. Similarly, inventions are not divisible. It follows that, at least in the jurisdiction of the office that granted the patent, intellectual property is concentrated in the hands of only the patent holder. This is the source of market power, partly intended to create financial incentives where they are lacking, but which also creates an imbalance in the economy along the way. Property on more tangible assets also aims to create incentives. It is well-known that it can avoid the over-exploitation of certain natural resources by eliminating the problem of the tragedy of the commons popularized by Hardin [HAR 68]. Applied to free access to pastures, this problem means that the villagers decide the size of their herd without considering that every additional animal enters into competition for food with the other villager’s animals and lowers the production of milk and meat. Dividing the grazing land into parcels that can be sectioned off (known as enclosures) solves this problem. Unlike inventions, each parcel can be allocated to a different proprietor, which maintains the pressure of competition and thereby avoids the adverse effect of an increase in market power. In a dynamic context, the property of natural resources also allows the owner to plan for the future and define the optimal pace for exploitation, thus transforming the resource into an asset (Hotelling [HOT 31]). Like for patents, land property can inspire a race, in the form of exploration to discover deposits of a mining resource or a source of fossil fuels, for example. There again, however, the winners of this race will only gain the property of the deposit they discovered, leaving their competitors the opportunity to find other deposits of the same resource and compete with them on the resource market.

A third explanation relies on the concept of the tragedy of the anti-commons. Heller [HEL 98a] defined the anti-commons as a property regime in which multiple owners have exclusive rights to the same resource. As the name indicates, this case is the opposite of the tragedy of the commons described above. Although Heller [HEL 98a] applied this concept to land property, which is not in principle immune from subsequent problems, economists refer chiefly to the article by Heller and Eisenberg [HEL 98b] to underscore the importance that the concept of the anti-commons takes on when it comes to intellectual property. The tragedy of the anti-commons occurs when the user of a technology needs to negotiate ex ante the licenses of different holders of multiple source patents on the basis of which the user intends to develop ex post his own activity. Each holder of a source patent recognizes that the total licensing demand for his patent depends on the accumulated value of the licenses required for the activity downstream, so he is aware that they are all dependent on each others when it comes to setting the licensing fee. On the other hand, the different holders of source patents have no incentive to incorporate the impact that this will have on others’ profits in their licensing pricing structure. In other words, rate setting for licenses occurs in a non-cooperative manner, which leads to a greater sum to pay for a set of licenses than if these licenses were each held by a single agent. This results in a lesser propensity to adopt the technology or, in a more dynamic perspective, this represents an obstacle for the development of incremental innovations. The sub-optimal character of this non-cooperative equilibrium between two upstream companies holding monopoly power is known as the double marginalization problem. This problem has been analyzed for a long time, now with the literature referring especially to Cournot [COU 38] who demonstrated it using the example of the fabrication of brass using copper and zinc. Buchanan and Yoon [BUC 00] formalized this idea by highlighting its symmetry with the problem of the tragedy of the commons. More precisely, these authors demonstrated that the tragedy of the commons results from the multiplicity of access rights without exclusive power while the tragedy of the anti-commons results from the multiplicity of exclusion rights without access power. In both cases, the lack of coordination between the rights holders leads to an inefficient use of the resource, in the form of its over-use in the case of the tragedy of the commons and its under-use in the case of the tragedy of the anti-commons. Although Buchanan and Yoon [BUC 00] do not develop their analysis much in terms of patents, the applications are nevertheless immediate. Shapiro [SHA 01a] develops a similar idea making explicit reference to patent thickets. Notably, in the appendix of his chapter, he proposes a model where the companies of a downstream industry, in pure and perfect competition between themselves, must buy different and complementary components from upstream companies who each hold a patent on the component that they manufacture. Shapiro [SHA 01a] shows that the margin rate, which measures the market power ultimately exerted to the detriment of consumers of the downstream product, exceeds the margin rate that would occur if all of the components were produced internally by a single integrated company acting as a monopoly. In other words, the fragmentation of intellectual property between multiple patent holders reinforces the distortion of competition inherent in the patent system and curbs the adoption of applications that could be made; this is known as a patent thicket. Upstream, there is a risk that the obstacles to adopting innovations that represent downstream patent thickets will in turn discourage companies from innovating³⁹.

1.3.1. From factory to fabless: the growing role of the obligation to disclose the content of patents

Intellectual property is increasingly becoming an independent source of revenue. In the industrial sector, for Western companies, we are seeing an increase in revenue drawn from granting licenses compensating for a dip in revenue from industrial production⁴¹. The most representative example of this transformation is without a doubt that of the Technicolor company. Since the recovery, in 1999, of the patent rights of RCA (the sector of television bought by General Electric in 1987) which held a global portfolio of more than 55,000 patents related to digital technologies for image and sound, Technicolor, previously Thomson Multimedia (TMM) effectively raised the research and development of intellectual property to the status of activities in their own right⁴².

This transformation is all the more remarkable because, threatened by a sale at a symbolic one French Franc in 1996 and filing for bankruptcy in 2009, today Technicolor draws the majority of its revenue from its “patents and licenses” unit on analog technologies. The group, specialized in services related to the entertainment, image and sound processing industries, holds an intellectual property portfolio that includes more than 30,000 patents and applications, in the areas of video compression, image processing, telecommunications, user experience, security, and screens. Technicolor is the leading French patent holder. In 2015, it filed 769 patents with the European Patent Office (the leading French applicant in terms of filings) and 745 in 2014⁴³. The 30,000 patents held by Technicolor can be broken down into 7,000 families and about 1,100 related license agreements. Technicolor is also one of the European companies that draws the most revenue from intellectual property. In 2015, the “Technology” division of Technicolor generated 511 million euros from the value of its patents (up 4.2% from 2014), which represents 67% of its profits for only 14% of its total sales⁴⁴. This activity with low operating expenses (about 20% of the sales figures) is a providential source of revenue for the profitability of the group. However, a large part (288 million euros) is generated by the MPEG LA consortium in the MPEG 2, MPEG 4 and HEVC video compression technologies⁴⁵. The other part (202 million euros) comes from the transfer of licenses managed internally.

For instance, since it integrated its patents into the MPEG LA consortium in 2002, the group collects two dollars for every decoder using this standard. Another example is the MP3. Since it combined its patents with those of the German research institute Fraunhofer in 1995, the firm Technicolor has been able to capitalize on the commercial success of iPods and iPhones by the American company Apple. Still, we can question the capacity of Technicolor’s patent portfolio to maintain its profitability in the future. In fact, the short-term outlook is bleak, with the MPEG LA consortium licensetransferring agreement ending by 2017 for the MPEG 2 video compression technology (representing 55% of their revenue from licenses) for which all patents have fallen into the public domain. There is a risk of a lack of visibility after the loss of the MPEG LA patent because the business model founded on royalties will henceforth be replaced by signing contracts with consumer electronics manufacturers due to the shortening lifespan of video compression technologies. Technicolor is counting on new internal programs for granting licenses in digital television or computer monitor screens to take over.

This example is representative of companies that manage to get through a rough patch thanks to a strategy built on patents and can inspire others. The American company IBM, who transferred all of its computer production units to the Chinese company Lenovo in 2005, holds a global patent portfolio of more than 40,000 patents and generates more than a billion and a half dollars in revenue from licenses and the sale of intellectual property assets. Its patent portfolio is also a means of exerting pressure in legal proceedings with competitors. The patents it holds focus on online reputation, natural language processing, the identification of malicious applications, the ability to offer virtualization services in the cloud, and the coordination of information sharing between mobile apps, among others. The group filed 8,088 patents in 2016, which is equivalent to 20 patents per day⁴⁶.

If the production can be disconnected from the invention, it is in large part because the patent responds not only to a lack of appropriability but also an informational problem that has been known for a long time as well. Arrow [ARR 62] discusses this fundamental information paradox. Analyzing the request for information, he highlights that its value is only revealed to the user once he possesses the information and no longer needs to pay to get it. There is therefore a market failure concerning the exchange of information. This applies particularly well to inventions once the inventor engages in an interaction with actors to develop it, such as an investor who could fund its development or another company who has the appropriate material means to more efficiently implement it (a distribution network, for instance). These actors would not be inclined to engage unless they are informed about the content of the invention, to assess its technical and economic viability. However, in the absence of a patent, once informed, these actors could freely imitate the invention without needing to compensate the inventor. Anticipating this behavior, the inventor will not be very inclined to disclose the information and will also be less inclined to develop it, and even will not feel prompted to innovate if we go back further in the decision. The obligation to disclose the patented invention, coupled with the right to exclusion, solves this paradox, at least in principle. It renders the information public while restricting its use thanks to the right of excludability. Potential partners of the inventor thus have access to the content of the invention but must enter into negotiations with the inventor if they want to participate in its development. Therefore, without the obligation to disclose, there would be no “fabless”. Several questions remain open, however. First of all, the information disclosed is technical. Can it take a fiscal and/or financial form, and if yes, how and under what conditions? That is the question of the recognition of the role of patents as intangible assets. Second, a piece of information can only produce all its economic effects if its credibility is well-assured. What are the mechanisms that govern this credibility? This question, addressed from the perspective of signal theory, makes it possible to better understand the role of patents in access to external funding for innovation. Finally, once the answers to these questions have been determined, what chances do we have of seeing a knowledge economy based on a technology market emerge, and what risks would that present?

1.3.2. The emergence of patents as intangible assets

The example of IBM, like that of Technicolor, shows how the performance of a company can be constructed around intangible assets. In this area, the role of patents and more generally of intellectual property and R&D have long been highlighted in economic and financial literature.

As noted by Lindenberg and Ross [LIN 81], the performance of a company is measured by the difference between its value on the market and its value in liquidation. The first value, which we could call the financial value, revealed by the market as soon as the company is listed, is guided by the results obtained by the firm. The second value is guided by the logic of resources mobilized and corresponds to the book value. Tobin and Brainard [TOB 68] introduced the idea, which was then developed by Tobin [TOB 69, TOB 78] and ample other literature that we will only briefly touch on here in relation to innovation and patents, that in a world where resources are perfectly mobile, the two values would tend to balance out. A favorable gap between the market value of a company and its book value indicates that one more euro invested in resources will generate more than one euro in revenue, and it is therefore desirable to increase the resources mobilized by the company. An unfavorable gap indicates that it is preferable to abandon a euro in that resource by selling it rather than keeping it within that company. In a world of mobile resources, there can only be transitional gaps between these two values, the time that investors become aware of this gap and consequently adjust their decisions. On the contrary, if a lasting gap is observed between the two values, it means that there are frictions in the acquisition or sale of resources by the company.

Lustgarten and Thomadakis [LUS 87] argue that these frictions can be associated with the specificities of intangible capital. There is a first series of frictions in the transfer of resources, which Lustgarten and Thomadakis [LUS 87] call barriers to exit. The specificities of mobilized intangible assets are the first source of friction. By specificity of assets, we mean that they are suited for a unique purpose and cannot be redeployed in another activity. If the company wants to resell them, it will be inherently difficult to find a buyer. The assets in question need only be specific to the company’s industry, but not necessarily specific to the company. The conjunctural causes likely to push a company to dispose of their assets will be shared by all companies in that industry and therefore, there will be no buyer. The specificities of intangible assets apply particularly well to knowledge, whether it is codified or tacit. That is why the literature mentioned very early on that the result of R&D expenditures belongs in this category. The asymmetries on the secondary market of assets are a second source of friction in the transfer of resources. A classic situation of information asymmetry is when a potential buyer is unsure of the reasons prompting a company to part with certain assets. These reasons may not be related to the assets themselves but to the situation of the company, which should not in principle affect their resale value in a situation of perfect information. However, the reasons may also be related to the assets, such as a disparity with respect to expected performance. In this case, the buyer can only accept to buy them with a duly proportioned discount. In an information asymmetry, the doubt about the real reasons for the transfer will lead to demanding a discount which the selling firm will not consent to if it knows that the causes are not related to the asset. So, only “bad” assets will remain on the secondary market, which justifies the discount. This is the adverse selection phenomenon first developed by Akerlof [AKE 70]. This type of friction can typically concern the transfer of patents on product or process inventions for which the divesting company has not yet moved on to the industrialization or commercialization phase. Aside from this first series of frictions involving the sale of resources, there is a second series of frictions involving the acquisition of resources. A favorable gap between the market value and book value of a company can result from the time necessary to provide resources for the firm’s development. Knowledge, intangible capital accumulated over time thanks to R&D efforts and training within the company, is representative of this type of resource. This is especially the case for tacit knowledge, which is only transferred through the training of personnel, which takes time, and not for codified knowledge, which is easily transferable from one company to another in a short time. Whether it involves barriers to the acquisition or transfer of resources, resources generated by the activity of R&D are primarily concerned.

Thus, an empirical literature developed that sought to evaluate what part of the value of the company could be attributed to R&D expenditures and/or the patent portfolio. Griliches [GRI 81] provides, to our knowledge, the first estimation of an average value for patents based on Tobin’s Q ratio⁴⁷. The author evaluated $2 at the time as the value generated over the long term by $1 of R&D expenditure and $200,000 as the value generated by a non-anticipated patent, which is to say above the number of expected patents based on the amount spent on R&D. Pakes and Griliches [PAK 84] examined the extent to which the number of patents can be used as an appropriate measurement of accumulated knowledge in the form of intangible capital by a firm. They consider the fact that patents are only one facet of knowledge stock, but show that they constitute an adequate proxy variable. They base this on an analysis of the variance in a time series model. Pakes [PAK 85] partly resumes this discussion by explaining that R&D and patents result from the inter-temporal maximization of the value of the firm. Although the number of patents is an error-ridden measure of knowledge stock, the author finds an average value of $810,000 for a non-anticipated patent. This pioneering work clearly established that patents are intangible assets whose value can be substantial.

Lindenberg and Ross [LIN 81] discuss the link between, on the one hand, a lasting difference separating the market value and the book value of a firm and, on the other hand, the exercise of market power. This difference is perceived not only as the consequence, but also as an indicator, of the firm’s competitive advantage. The connection to the concept of market power is intuitive once we note that the equalization between market value and book value is a declination of the well-known principle in economics that production factors are paid their marginal product. According to this principle, in the absence of market power the firm will only take into account, in the profit resulting from the growth of a resource unit, the increase of physical production valued at a market price considered to be constant because the firm does not control it. The firm will therefore find it profitable to increase by one unit the resource as long as the increase in production valued in this way exceeds the unit value of the resource⁴⁸. If, on the other hand, the firm does have market power, it will consider the fact that the increase in production resulting from the expansion of a resource unit is generally accompanied by a decrease in a buyer’s willingness to pay for an additional unit of production, and therefore a drop in the production value. This additional negative effect of resource expansion in the presence of market power leads to rationally limiting their expansion. However, R&D expenditures can typically confer a competitive advantage to a firm in the form of lead time over its competitors and allow it to exploit opportunities that others do not have access to, an advantage that translates into market power and the associated rents. Through the right to exclude that they accord to a firm, in principle, patents allow the firm to secure its present or future rents. The idea that secured rents are not necessarily observed in the present but are based on future opportunities emerged fairly early on in the literature. Thomadakis [THO 77] noted that the difference between the market value of a firm and the book value has an option value dimension by explaining that it could correspond to a “set of options which the firm holds for future investment”. The importance of this option value was later confirmed by Bloom and van Reenen [BLO 02] who emphasized that a firm’s patent portfolio contributes to explaining a firm’s market value in a more statistically significant way that its sales or added value. These authors explained this by emphasizing that the future opportunities of a company are well-integrated in investor expectations, even though they do not contribute to its present results. Bloom and van Reenen [BLO 02] also proposed modeling firm value in terms of the option value of its patents, but we owe it to the contribution by Abel et al. [ABE 96] to have made explicit the link with Tobin’s Q ratio.

1.3.3. The delicate question of assessing patents as intangible assets

As soon as the idea of patents as intangible assets is raised, the question of the investor information concerning their value follows. A few of the empirical works already cited here, such as Griliches [GRI 81], Pakes and Griliches [PAK 84] and Pakes [PAK 85] proposed estimates for the average value of patents. One of the most complete studies on this subject, because it includes the qualitative aspects of patents by considering the number of citations received, is that of Hall et al. [HAL 05b]. First and foremost, these works provided empirical proof that R&D expenditures are a source of intangible asset creation, eventually validated by obtaining a patent, and that investors are aware of this. However, they do not provide evidence that the information provided by these estimates is pertinent for directing investor’s choices. Indeed, estimates produced this way reveal what investors think intangible assets associated with R&D are worth, but not necessarily what the value that they actually produce is. This step can only be taken if the financial markets function efficiently, in the sense that all information useful for forecasting a firm’s performance would be instantly taken into account by market actors (Fama [FAM 98]).

This is done by some authors, such as Hirschey et al. [HIR 01] and Hirschey [HIR 82]. The former also explicitly integrate information about patents through, in addition to counting the patents obtained, different metrics like an impact index calculated on the basis of citations received at age five by patents, an index of connection to science based on references to the academic literature in patent documents and an indicator for the technological cycle length assessing to what extent a firm’s patents are based on recently patented inventions. Hirschey, Richardson and Scholz [HIR 01] conclude that information about patents adds statistically significant information to R&D expenditures that accounts for a company’s market value. They explain this through the fact that patents reflect the result of R&D activity and that, coupled with the R&D effort, information about patents makes it possible to approximate the performance of a company’s R&D process. This type of work falls under a larger trend in private accounting which considers that accounting information and the market valuation of firms must be consistent and that this can serve to guide accounting standards. Lev and Zarowin [LEV 99] argued that the discrepancy between accounting documents and the market value of the firm grew during the last decades of the 20th Century. They also determine a link between this increased discrepancy and the greater instability of the relative position of firms in terms of performance, an instability that has been fostered by new regulatory policies and/or the development of new innovations, notably in sectors like telecommunications⁴⁹. As a result, Lev and Zarowin [LEV 99] propose making changes to accounting standards so that they better incorporate the prominent place occupied by intangible assets in firm dynamics. This trend, defended by Barth et al. [BAR 01] among others, is strongly contested by others, like Holthausen and Watts [HOL 01] who argue that accounting standards aim to provide the most objective description possible of a company’s situation and financial health. Holthausen and Watts [HOL 01] reject any practice that tries to incorporate elements that are too speculative into accounting standards and recommend sticking to what is proven. More in line with this second approach, Sougiannis [SOU 94] attempts to evaluate intangible assets resulting from R&D through the connection between the cumulative R&D expenditures and the revenue of a firm. In the process, Sougiannis [SOU 94] proposes a method for estimating the rate of depreciation of R&D expenditures⁵⁰. This idea is taken up by Lev and Sougiannis [LEV 96] who also study the link between the revenue projected by such a model and the market value of firms. For the moment, there do not seem to be any contributions that incorporate information about patents into this approach. Even if such work existed, they would not be exempt from criticism regarding the consistency of the value of patents produced in this way with accounting standards⁵¹.

Since the beginning of the 2000s, the accounting standards in effect in developed countries have converged downwards with international standards. The movement was more delayed in the United States than in Europe but has been initiated⁵². The adoption of International Accounting Standards/International Financial Reporting Standards (IAS/IFRS) for all listed companies across the European Union starting from January 2005 led to rules that were often more restrictive than before for member countries concerning the possibility of capitalizing intangible assets associated to R&D. Their capitalization is based on a more general standard, the standard IAS 38 about intangible assets⁵³. As presented in the Official Journal of the European Union, it makes explicit reference to patents in paragraph 9 refining the concept of intangible assets. In paragraph 10, it is indicated that:

“Not all the items described in paragraph 9 meet the definition of an intangible asset, i.e. identifiability, control over a resource and existence of future economic benefits. If an item within the scope of this standard does not meet the definition of an intangible asset, expenditure to acquire it or generate it internally is recognized as an expense when it is incurred. However, if the item is acquired in a business combination, it forms part of the goodwill recognized at the acquisition date (see paragraph 68)”.

The last sentence paves the way for the capitalization of patents as assets when they are acquired externally. Nevertheless, in the economic logic of fabless companies for instance, what matters here is more knowing to what extent a patent obtained internally can be capitalized as an asset. In the United States, the US GAAP (Generally Accepted Accounting Principles) precludes capitalization of an internally developed asset. To do so in Europe, the patent, or more generally the intangible asset, must be scrutinized for its identifiable character, management of the resource and the existence of future economic benefits. The fact that a patent is a clearly recorded and transferable property right makes it possible to respond unambiguously to the first and second criteria. The third criterion, presented in paragraph 17, states that:

“The future economic benefits flowing from an intangible asset may include revenue from the sale of products or services, cost savings, or other benefits resulting from the use of the asset by the entity. For example, the use of intellectual property in a production process may reduce future production costs rather than increase future revenues”.

The reference to intellectual property is also explicit here. However, the details about the accounting and evaluation of intangible assets, notably in paragraphs 21–24, can greatly hamper the possibility of capitalizing patents. These paragraphs specify that:

“An intangible asset shall be recognized if, and only if: (a) it is probable that the expected future economic benefits that are attributable to the asset will flow to the entity; and (b) the cost of the asset can be measured reliably. An entity shall assess the probability of expected future economic benefits using reasonable and supportable assumptions that represent management's best estimate of the set of economic conditions that will exist over the useful life of the asset. An entity uses judgment to assess the degree of certainty attached to the flow of future economic benefits that are attributable to the use of the asset on the basis of the evidence available at the time of initial recognition, giving greater weight to external evidence. An intangible asset shall be measured initially at cost”.

The success of an invention is generally subject to strong uncertainty, often because of a technological gamble. The only patents that respond easily to these details are patents actually exploited by a company, or in other words, patents for which the holder has moved on to the industrialization and commercialization phase, or has transferred a license to another agent. In addition, the cost related to a patent can be difficult to determine, especially for complex technologies. To do so, it must be possible to isolate the development costs associated with one patent in particular⁵⁴. This is only possible if the process or product innovation is based on only one patent, and not several at the same time. An industry like pharmaceuticals where a medication is associated with a limited number of patents can submit to this requirement, but not an industry like mobile telephones where a single device can involve several thousands of patents today. Finally, even when the development costs associated with a patent are identified, paragraph 57 further restricts the possibility of accounting for these costs as assets. This paragraph states that:

“An intangible asset arising from development (or from the development phase of an internal project) shall be recognized if, and only if, an entity can demonstrate all of the following:

1. a) the technical feasibility of completing the intangible asset so that it will be available for use or sale;
2. b) its intention to complete the intangible asset and use or sell it;
3. c) its ability to use or sell the intangible asset;
4. d) how the intangible asset will generate probable future economic benefits. Among other things, the entity can demonstrate the existence of a market for the output of the intangible asset or the intangible asset itself or, if it is to be used internally, the usefulness of the intangible asset;
5. e) the availability of adequate technical, financial and other resources to complete the development and to use or sell the intangible asset;
6. f) its ability to measure reliably the expenditure attributable to the intangible asset during its development”.

As noted by Morricone [MOR 11], companies are rarely able to demonstrate that all of these criteria are satisfied. Barring situations in which an entity has effectively moved on to the phase of industrialization or commercialization or transferred a license, the economic value of a patent essentially consists of its real option value, or in other words, the value accorded to future but uncertain opportunities that will make it possible to exploit the patent. To use the terminology of Bloom and van Reenen [BLO 02], only “embodied” patents, not “disembodied” ones, seem able to be really capitalized as assets in a company.

It seems that, generally and for reasons of consistency unique to them, accounting standards tend to minimize the value that they allocate to patents compared to their economic value. If, as we saw previously, investors still incorporate the value of patents in their estimate of the market value of a firm, it is because they have information other than purely accounting information. Several trends in the literature have taken an interest in the information that firm behaviors reveal. We can distinguish two categories. The first, which is losing interest as IFRS standards become more widespread, exploits the fact that some countries like Australia and Canada allow companies to choose between expenditure accounting or accounting for their intangibles as assets, when they respond to a series of criteria quite similar to those described above for the IFRS standards. Focusing on R&D expenditures, Chan et al. [CHA 07] argue that the choice can prove complex to analyze, because both capitalizing R&D and accounting R&D as expenditures convey a message to external financers of the company. Capitalizing as an asset requires that an external auditor of the company’s accounts gives approval, which is only done after being assured of conformity with accounting rules. From this perspective, voluntarily capitalizing sends a positive signal regarding the probability of future revenue. At the same time, accounting R&D as expenditures is penalizing because it minimizes the company assets, and with it the compensation of managers. It can thus be seen by investors as a sign of the company’s financial soundness. These effects can cancel each other out, which would explain the lack of a clear trend regarding the effects of the movement from old national standards, often more flexible, to IFRS standards for European countries⁵⁵. The second category of publications that attempted to exploit what patent holder behaviors reveal about the value of their intellectual property assets focused on renewal behaviors. A patent only remains valid if its holder pays renewal fees at regular intervals (1 year in most European national offices, three and a half years in the United States). The failure to pay causes the patent to fall into the public domain permanently. Schankerman and Pakes [SCH 85, SCH 86] were the first to propose a model that made it possible to estimate the distribution of the value of a cohort of patents, based on renewal data. They demonstrated that, even if the fees⁵⁶ are a modest amount, the method can reveal significant patent values. Their model paved the way for a series of works that will be discussed more in-depth further on in this book. A first critique that can be addressed to all of these works is that they overlook all strategic behavior in the decision to renew. Inversely Crampes and Langinier [CRA 98] developed a theoretical model demonstrating that a firm may deliberately not renew a patent in order to make a potential competitor believe that the market associated with the invention is not profitable and to dissuade him from entering it. At least from a theoretical point of view, the message conveyed by decisions to renew or abandon a patent can be ambiguous. Generally, the limit to be considered when assuming that investors rely on the behaviors of inventors to infer the value of their patents is that the messages conveyed by these behaviors are too ambiguous. By focusing on the way in which patents convey credible information, exploitable by investors to externally assess the value of the patents, the literature highlights the role of the examination and granting processes by the patent office itself. This literature relies on the signal theory developed by Spence [SPE 73] to demonstrate how patents can play a key role in the access to external funding by innovative firms. On a more empirical level, Baum and Silverman [BAU 04] were among the first to validate the idea that the existence of a steady stream of patents is interpreted by investors, and more significantly by venture capitalists, as a signal that a company is maintaining its place in the innovation race.

1.3.4. Patents as funding leverage

The granting of a patent on an invention by a patent office requires that an invention responds to the three criteria of novelty, inventiveness (Europe) or non-obviousness (United States) and industrial applicability (Europe) or utility (United States, Australia). Even if, in the details, the verification of these criteria occurs through practices that can vary substantially from one patent office to another (see Chapter 3, which is dedicated to institutional aspects), the patent office consistently appears as an administrative authority that filters patent requests according to relatively objective criteria. By publishing a patent, a patent office accomplishes two things at once. First, they contribute to resolving the information paradox by disclosing the content while granting a right of excludability on the exploitation of the content. The patented invention can then be known by all and examined in detail while offering a certain guarantee against being stolen. Then, because it applies some selection criteria and filters the requests independently and objectively, the patent office provides the guarantee of a certain quality of the inventions on which they grant the patent. From this point of view, obtaining a patent resolves much of the adverse selection problem for the external funding of young companies.

The problem of adverse selection comes from the fundamental asymmetry that exists between, on the one hand, young innovative companies and, on the other hand, their potential financers. The companies know what their invention provides that is really new and what its limitations are. They are also able to appreciate its chances of success, and of becoming an innovation. Potential financers do not often have a sufficient grasp on the technical subject and/or the time necessary to examine in detail the many inventions that need funding, but they are well aware of the heterogeneity, in terms of the chances of success and the scope of that success, of the projects that are proposed to them. Without a credible signal allowing them to discriminate between good and bad projects, they would incorporate a substantial risk premium on the funding conditions⁵⁷. Companies proposing good projects would consider these conditions to be prohibitive because they place a high probability on their success and therefore on the realization of the considerations required by the financers. On the other hand, companies proposing bad projects would associate a small probability with the realization of these considerations and be more inclined to accept the terms of the contracts. This is the adverse selection problem that is well-known from the work of Akerlof [AKE 70]: without a credible signal about the quality of the projects, the “bad” projects would drive away the “good” ones in the sense that it would be above all the “bad” projects that would accept to deal with the investors.

This phenomenon is not specific to the funding of innovative projects and a classic solution is to require a partial self-financing, so that the people proposing the project also assume a part of the risks that they ask the financers to incur. If the people proposing the project are aware that the risks of failure are considerable, they will anticipate that committing to self-finance a substantial part of the project will be very costly for themselves and will not do it, and vice versa for people who propose a project that has a high likelihood of succeeding. However, an important characteristic of innovative companies that request external funding is their youth and low credit standing. Their ability to self-finance is therefore often inherently limited, unlike more established firms. On the other hand, expenses to file a patent application are generally within their reach. Starting from this observation, Conti et al. [CON 13a] developed models highlighting the role of patents as a signal intended for investors. Conti et al. [CON 13a] validated their model with an empirical study on a sample of Israeli start-ups while Conti et al. [CON 13b] worked with start-ups from the Advanced Technology Development Center (ATDC) of the Georgia Institute of Technology to test their model. Using a sample of Belgian companies, Czarnitzki et al. [CZA 16] also demonstrated that patents allow small companies to lessen R&D financial constraints, an effect that is specific to them and does not occur for larger companies⁵⁸. By filing a patent, innovative companies take the risk of having their request refused and having it be disclosed by the patent office that their invention does not respond to at least one of the three criteria applied by the office, and therefore that it would also likely not satisfy the requirements of investors, either. On the other hand, if the patent is accepted and published, companies can leverage this with investors and it saves investors from having to do their own costly and demanding technical examination of the invention. Only companies that are sufficiently confident in the outcome of the examination process accept the risk of filing the application; the others know that the effect will be counter-productive and will not go through with it. This means that companies self-select and in doing so, reveal what they think of their own capacity to develop meaningful innovations. The patent system therefore plays, even if partially, the role of a revealing mechanism. The centralization of the examination activity also allows the office to have the largest frame ofreference for comparing different inventions, and therefore to produce more reliable information than a decentralized examination system.

The signaling role of patents and the responsibility that lies on the patent office is sometimes also considered just as important, or more so, than their incentivizing role. This has led some authors like Long [LON 02] to completely revise the economic importance of patents. From this perspective, two elements are essential to ensure the credibility of the signal conveyed by the patent office. The first is that the patent requests must be public. If this was not the case, a company could hide the fact that they had difficulty validating their requests, hiding the rate of rejections it has suffered if it makes multiple requests, for instance. Second, an office must be sufficiently selective in granting patents so they are adequately considered by financers, at least in their first approach to projects that are submitted to them. The next chapter focuses on institutional aspects and reforms and shows that this is far from being the case for all offices. The offices do not all make the same trade-off between, on the one hand, bearing the costs related to a comprehensive and selective examination of patents and, on the other hand, relying on the legal system to correct any imperfections in the delimitation of patents; imperfections which can result from too superficial of an examination. From this point of view, the European Patents Office (EPO) is often regarded as the best of what a patent office can be, whereas the US Patent and Trademark Office (USPTO) is often considered to rely more on the legal system to compensate for any imperfections⁵⁹. This comparison between offices is confirmed by Useche [USE 14] who showed that, during their initial public offering (or in other words, their introduction to the stock market), all things being equal besides, European companies in the software industry enjoyed a better capitalization on the initial price of their shares for patents that they obtained from the EPO than their American counterparts regarding patents that they obtained from the USPTO⁶⁰.

More generally, several recent empirical contributions have sought to validate the idea that patents are indeed used as signals by investors, either during an initial public offering like in the article by Useche [USE 14] or in the first steps toward financing by venture capital with results that validate this signal role (Hoenen et al. [HOE 14]) or not (Hoenig and Henkel [HOE 15]). Hsu and Ziedonis [HSU 13] argue that the signal effect of patents can only be correctly understood by focusing, on the one hand, on the very first steps of a venture capital operation (where the asymmetry of information is greater) and, on the other hand, by focusing on one particular industry. Indeed, as soon as only one industry is considered, the degree of appropriability of returns associated with the exploitation of patents is similar for all companies and therefore cannot account for the probability of reaching an agreement for venture capital or the time necessary to reach it. Although these works are limited to examining the influence of the number of patents obtained, Haeussler et al. [HAE 14] go further by demonstrating, using a sample of British and German companies seeking venture capital, that the characteristics of patents (such as references cited by the European Patent Office having a classification in X or Y to flag the references that challenge the novelty of the patent) as well as certain events affecting the patents (especially oppositions) have a significant influence on the time it takes for firms to reach a venture capital agreement. Thus developed the idea that the recognition of the role of patents as a signal accompanied the development of new ways of funding innovation. Kortum and Lerner [KOR 00] documented the way the large increase in the number of patents published by the USPTO in the last two decades of the 20th Century is concurrent with the development of venture capital.

1.3.5. The commoditization of patents

Although the literature has mostly concentrated on the role of patents as a signal that facilitates access to venture capital, patents can also be used to find other funding strategies⁶¹. For instance, Munari et al. [MUN 11] cite the case of patents used as collateral for loans, or even the more “exotic” sell and lease back mechanism⁶². In both cases, the effectiveness of the system relies on the prospect of securitizing the patent, in the sense that the financer hopes to be able to easily resell the patent to a third party. This is only conceivable if a technologies market exists, and if the practice is sufficiently anchored with economic actors, on the one hand, to turn to at least partial outsourcing of their R&D and, on the other hand, to not seek to exploit the results of their R&D themselves and to make selling technological solutions the core of their business. It is worth specifying the ruptures that accompany the development of a technologies market and then examining the difficulties and limitations of this development.

Arora et al. [ARO 01] and then Arora and Gambardella [ARO 10] propose both a state of affairs and an analysis of the function of a technologies market. Arora et al. [ARO 01] focuses on the resource-based view of the firm theory. According to this theory, to be a lasting source of competitive advantage, a resource must be rare and imperfectly mobile. In other words, there must not be an efficiently functioning market for this resource, so that competitors of the firm cannot compete on equal terms. A firm must attempt to develop its control over these types of resources and above all to play on the complementarity between them. To be able to harness the entirety of the value associated with its intellectual property assets, a firm must seek to ensure control over all complementary assets. If, for example, an innovation is based on exploiting the properties of a rare natural resource, the firm must seek to control the deposits of this resource. If an innovation has applications in the form of different products whose success depends on the quality of the distribution network, the firm must invest in and control these networks. This results in the firms which are the most effective at exploiting their R&D being large firms, potentially closely integrated both vertically or horizontally. This paradigm prevailed for a large part of the 20th Century but has been strongly challenged in the last few decades. The new regulatory policies implemented first in the United States and later in Europe, for network industries, and more generally for opening markets to competition, have contributed to limiting the barriers to entry for competitors. In network industries, the separation between network managers and operators has promoted a wider variety of operators that have been able to develop thanks to equal treatment for access to the essential resource that forms the network. The greater variety of operators has in turn facilitated the development of an equipments market, limiting the complementary assets to be controlled to promote R&D by equipment manufacturers. The separation between the production and transport of electricity, for instance, contributed to the development of innovations in renewable energies. Generally, the existence of a large number of actors upstream and downstream from a firm limits the risks inherent to specializing in the exploitation of specific resources by diversifying the portfolio of clients and suppliers⁶³. This way, some firms can specialize in the production of technological solutions for which they grant licenses or sell their patents, while others can instead outsource their R&D in the hopes of being less confined to routines.

In this way, a technologies market supported by patent transactions develops (Kline and Rivette [KLI 00]). Not only does this market allow certain firms to specialize in R&D, it also allows all firms whose creation of value relies on holding patents to specialize their patent portfolio. This specialization occurs through the transfer of patents for which a company does not have adequate resources to derive all of their value and also through the acquisition of patents from which they have more potential to profit. This perspective is particularly important for so-called complex technologies, or in other words, technologies that require the mobilization of several patents to develop an invention. Odasso et al. [ODA 14] cite the example of the portfolio of the bankrupt firm Nortel Networks Inc. that included more than 6,000 patents which were purchased in a block in 2011 for nearly 4.5 billion dollars (3.3 billion euros) by the Rockstar Bidco consortium associating Apple with companies that are in principle rivals, like EMC, Ericsson, Microsoft, Research in Motion and Sony⁶⁴. The firms of the CPTN⁶⁵ consortium seem to have wished to protect themselves against usage at their expense by other buyers, notably in the context of legal disputes over patent violations. The consortium led by Microsoft also purchased 882 patents by Novell in 2011 for 450 million euros. Similarly, Google paid more than 12.5 billion dollars (8.75 billion euros) in 2011 to acquire 17,000 patents from Motorola Mobility⁶⁶. Reflected in these cases, we can see that companies in an industry can identify patents held by others that they need to develop an innovation, and inversely, can identify patents within their own portfolio that they cannot effectively exploit, either because they lack too many complementary patents held by third parties or because they do not have the complementary assets required to do so, such as an adapted distribution network in the case of a product innovation. The driving force behind this exchange is the search for complementarity between patents, widely recognized in the business world but whose analysis in the academic literature is still in its infancy. An exception to this can be found in Gambardella et al. [GAM 12], taken up and summarized by Gambardella [GAM 13], which proposes a model that puts forward the search for complementarities and validates their importance using survey data. The logic of patent portfolios leads to a form of patent securitization, in the sense that the actors can specialize in building patent portfolios with strong complementarity, without being themselves inventors, applicants, or exploiters of patents, but by carrying out patent acquisition operations to reassemble portfolios and resell them. These actors take advantage of an over-additivity of the economic value of patents; the value of a portfolio of complementary patents being more than the sum of the value of all of the patent parts that constitute it. Faced with the need to improve the flow of technology transactions, other new actors like Innocentive, Yet2.com and Ocean Tomo⁶⁷ act as technology brokers and in doing so participate in improving the flow of the invention market (McDonough [MCD 06]). The economic models are not yet fixed, but an attempt at a typology gives the following broad categories:

• – Patent investment funds that intend to preserve the freedom to operate for companies whose interest they are acting in. This category notably includes France Brevets. Such investment funds can be found at the limit of legality to the extent that, notably in the European context, if they are implemented by public authorities and act in the interest of the companies of a member-state, they can conflict with regulations on state aids for the distortion of competition and with WTO rules⁶⁸.
• – Technology brokers who introduce technology suppliers and those who seek them. One example is the British company BTG which has operated in this domain since 1995. Similarly, online exchange platforms like Yet2.com, pl-x.com and innovationsmarkt.de were created.
• – Technology aggregators, such as Intellectual Ventures, a company created in 2000, which holds more than 70,000 patents and in which firms like Microsoft, Intel, Sony, Nokia, Apple, Google and eBay have invested. These are investment funds that identify essential patents and buy them to group them together to license them in blocks⁶⁹. These aggregators are defensive patent pools, such as Constellation Capital, RPX, Allied Security Trust and Open Innovation Network. These pools are designed to minimize the risks of patent assertion.
• – Intermediaries who develop and commercialize the intellectual property of third parties. Some like IP Group specialize in the transfer of technologies derived from public research.
• – Companies that organize patent auction sales, notably in Taiwan (through the ITRI Institute) and in Japan since 2003, in the United States (with the world leader Ocean Tomo Auctions since 2006) and in Germany (with IP Auctions since 2007).
• – Banks who accept lease-back operations or securitization on patents. In Germany, for example, the pioneers in this area are Credit Suisse Deutschland since 2005 and the Deutsche Bank since 2006. Similarly, some banks – especially in Japan (which is the case for Intellectual Property Bank Corporation: IPB) and in Germany – will now consider a patent or a patent portfolio as liability coverage for granting a loan. In this way, patents become equivalent to mortgage-backed securities. More generally, the patent acquires the properties of a common stock equivalent through these operations. As indicated by Yanagisawa and Guellec [YAN 09], the basic principle is that owners of patent assets can exchange the revenue that they will likely obtain in the future for funds obtained in cash.

A significant limitation to the phenomenon of securitization is that the actors implementing it cannot have as comprehensive a view of the complementarities to be exploited as companies in the different and heterogeneous industries. It is tempting to implement such a securitization to proceed to a treatment of patents according to a relatively standardized, even automated, procedure. However, the literature about the value of patents tends to cast doubt on the possibility of estimating the value of patents on a large scale, and a fortiori those in a patent portfolio. The first movement in empirical literature based this on survey data (Harhoff et al. [HAR 99], Harhoff et al. [HAR 03a], Giuri et al. [GIU 07], Gambardella et al. [GAM 08]). Using micro-economic data, these publications not only confirmed a very strong asymmetry in the distribution of the value of patents already demonstrated through the analysis of data concerning renewal by patent cohorts (Schankerman and Pakes [SCH 85, SCH 86]), they also showed that it is very difficult to predict the value of a patent based on metrics only. Patent metrics are features of patents that are objectively measurable through information contained in the patent document published by the patent office. For instance, they include the number of backward and forward citations, the number of claims, the size of the family, even composite indicators combining the data about citations and the technological fields designated by the citations, such as indicators of generality and originality⁷⁰. A second movement in the literature sought to use patent renewal data for an approach based on renewal behaviors (Barney [BAR 02], Bessen [BES 08], Baudry and Dumont [BAU 12]). These works confirmed the results from the survey data, namely that even if certain metrics have a statistically significant influence on the value of patents, their observation is not sufficient to predict this value reliably⁷¹. As noted by Fischer and Leidinger [FIS 14], the share of the variance in the values explained in these models, whether they rely upon survey data or renewal data, is in the order of 12%.

The difficulty of estimating the value of patents is also due to the fact that there are no, or very few, observations of transactions. For the most part, the sale of patents occurs over-the-counter and, even if, as specified above, the IFRS accounting norms authorize explicitly capitalizing patents acquired externally, it is still difficult and laborious to populate a database with these transactions. Even if such a database is available, nothing ensures that its treatment allows for improving predictions about the value of patents. Patents have two features that significantly complicate this prediction: uniqueness and indivisibility. Uniqueness refers to the fact that every patent protects a different invention and that its value is in essence very difficult to approach by comparison. Patents are not the only type of goods that have this particularity. Real property and land assets, to take up a comparison made above about the nature of property rights, are also unique, if only because of their localization. In the case of real property, this difficulty is circumvented by making recourse to estimating hedonic prices, or in other words, by considering a functional relationship between the value of observed transactions and objectively measurable characteristics of the goods involved in these transactions. However, this approach requires determining characteristics that are not only objectively measurable but that also cover the heterogeneity of goods in a sufficiently exhaustive way. Although information about the features of real property goods is relatively exhaustive today, notably thanks to the use of the Geographic Information System (a digital system of maps that can be overlaid) that describes the environment of land-goods in detail, it is doubtful that the few metrics of patents will allow us to understand all of the heterogeneity of patents. Indeed, two patents from the same technology sectors with the same metrics continue to protect inventions that, according to the criteria of inventiveness required to obtain the patent, are necessarily different. The indivisibility corresponds to the fact that it is not possible to own a fraction of a patent. At best, in the case of multiple owners of the same patent (which occurs in the context of R&D cooperation between companies or between public research organizations and companies), decisions about the patent are made according to conditions that substantially limit the discretionary power of each coholder⁷². The result is that, unlike for many financial assets such as shares, the value of a patent cannot be inferred from recurring transactions that are carried out on parts of the patent. In other words, statistical information is inherently rarer. The only cases of transactions on patents for which we have data gathered in a sufficiently systematic way are cases of patent auctioning. Platforms have been developed to connect vendors and purchasers of patents and respond to two types of alignment problems. First, potential buyers do not often want their desire to acquire a patent to be known, out of fear of revealing a competitive or legal weakness in their patent portfolios. Second, bilateral negotiations can take a lot of time and be costly for both vendors and purchasers. Auctions reduce some of the costs associated with bilateral negotiations and are a faster alternative. By screening potential buyers and the patents offered for sale through the calculation of an objective measurement of the value of each lot, the platform offers vendors serious contenders who can honor their financial offers, all while helping them to establish a reserve price. In this way, auctions support the disclosure of some private information while preserving the anonymity of the process. For example, Jarosz et al. [JAR 10] described in detail the platform implemented by the American company Ocean Tomo. These authors showed that the volume and magnitude of the patent transfers were limited due to a lack of flexibility in the structure of auctions, combined with the inherently complex nature of patents. This meant that, although the lots were pre-screened, they could still be subject to adverse selection. Indeed, it is possible that a patent be put to auction precisely because its vendors are having difficulty developing it. They also demonstrated that a large part of the revenue obtained from auctions was generated by a small number of transactions and that many lots did not receive any offers. The lots that did receive an offer were the subject of limited competition and were sold at a substantially reduced rate compared to their estimated value. It is therefore difficult to confirm that patent auctions are a success⁷³.

Odasso et al. [ODA 14] and Fischer and Leidinger [FIS 14] used data from the same platform to re-test value prediction models using real transactions with the help of patent metrics. The methodological difficulty consists of controlling the selectivity bias, which is potentially strong. As a matter of fact, patents put to auction are not necessarily representative of all patents; only certain types of patents are put to auction for specific reasons. For example, the two publications mentioned above demonstrate that the information technologies and communications sectors are over-represented. In addition, not all patents put up for auction find buyers (Jarosz et al. [JAR 10]). The information about the transaction value only concerns patents that actually found buyers. This particularity must be treated correctly in econometric models. By accounting for these particularities, the two studies manage to explain a much greater portion in the value variance, in the order of 60–80%, than studies about survey or renewal data. The risk is that the value resulting from the auction itself results from a prediction model and is somewhat self-fulfilling. One element that could cause concern about this phenomenon is that the platform developed by Ocean Tomo proposes ex ante an indicative estimate and that Ocean Tomo simultaneously developed an automated scoring model for patents… for which the company obtained a patent from the USPTO⁷⁴. However, a reading of the patent document suggests that the scoring is largely based on the metrics and data for renewal. Even if the evaluation of indicators provided by Ocean Tomo also involves expert opinion, it is susceptible to being strongly guided by the scoring model and influencing the outcome of auctions. If this is actually the case, the securitization of patents would be confronted with the same misuse as some financial instruments, with major and brutal risks of correction when economic agents become aware of the mismatch between the market value of these titles and their actual value.