Robust empirical research confirms.

Why do we need patents? The usual argument, which we shall call the “second best” theory of patents, argues that patents are needed to solve a double problem: a problem of incentives, on the one hand, and a problem of knowledge diffusion, on the other hand (the so-called Arrow dilemma, 1962, which highlights the difficulty to achieve both optimal incentives to produce knowledge, and optimal level of dissemination of the knowledge produced [3]

[3]

In his seminal paper, Arrow (1962, p. 619) concludes: “To sum…). First, patents induce firms to invest in R&D and in the economic valorization of inventions (Nordhaus, 1969; Kitch, 1977); second, they encourage firms to disclose their inventions, i.e. disseminate technical knowledge within the economy and fight secrecy (Encaoua et al., 2006). Yet, in the short run, patents give market power to inventors, thus creating monopoly deadweight loss and static inefficiency. Consequently, patents are not a “first best”, but only a “second best” option. As reminded by Schumpeter (1942), static inefficiency is the price to pay in order to induce innovation and dynamic efficiency.

2If this vision of patents has the merit of consistency and simplicity, a growing number of researchers tend to oppose it more or less head-on (Kingston, 2001; Bessen, Meurer, 2008; Boldrin, Levine, 2008; Hilaire-Pérez et al., 2013). First, this standard interpretation relies on an unrealistic view of the innovation process. It is based on the traditional Arrovian framework, which considers innovation as an individual and isolated act to produce a public good (knowledge is reduced to information, i.e. is easy to reproduce). This view emphasizes incentives and the importance of patents to exclude imitators. It largely neglects the properties of knowledge (its tacit dimension), and of the innovation process (which is largely a collective process requiring collaborations and interactions). As suggested by Cohendet and Pénin (2011), when these properties are taken into account, the main source of market failure might not be a problem of incentives but of coordination, and the main role of the patent system might not be to exclude others but to include all the different stakeholders in the innovation process.

Second, the standard view is not confirmed by most empirical studies, which put forward three important insights (Levin et al., 1987; Cohen et al., 2000; Sakakibara, Branstetter, 2001; Arundel, 2001). First, patents are not the most efficient instrument to protect an invention (secrecy, lead time, complementarity with other assets, are usually preferred). Second, except for the pharmaceutical industry, patents are not necessary to increase incentives to innovate (most firms would invest a similar amount in R&D without the existence of the patent system). Third, the use of patents is strategic (defensive use, signaling, knowledge management, etc.). In conclusion, most empirical studies question the global positive effect of patents on incentives to innovate (see Bessen, Meurer, 2008, for a survey) [4]

[4]

As summarized by Lerner (2009, p. 347): “the lack of a positive…. This conclusion is robust according to the different methodologies used, such as a questionnaire-based inquiry (Mansfield, 1986; Levin et al., 1987; Cohen et al., 2000), natural experiments (Sakakibara, Branstetter, 2001; Lerner, 2002), or, more recently, lab experiments (Meloso et al., 2009).

4It is rare to observe such unanimity among empirical studies conducted over such a large period of time, on such a wide range of countries, and based on such different methodologies. Combined with the weak theoretical basis of the Arrow model, our conclusion is therefore that the second-best theory might not be the most appropriate way to justify the existence of the patent system. We therefore turn towards another possible argument, namely the link between patents and open innovation (Chesbrough, 2003; Pénin, 2017).