Patent disclosure and R&D competition in pharmaceuticals

Abstract

The prominent role played by patents within the pharmaceutical domain is unquestionable. In this paper, we focus on a relatively neglected implication of patents: the effect of patent-induced information disclosure on the dynamics of R&D and market competition. The study builds upon the combination of two large datasets, linking the information about patents to firm-level data on R&D projects and their outcome. Two case studies in the fields of anti-inflammatory compounds and cancer research complement our analysis. We argue that patent disclosure induces R&D competition and shapes firms’ technological trajectories. In fact, we show that under conditions of uncertainty, patent disclosure can contribute to generate knowledge spillovers, promoting multiple parallel research efforts on plausible targets and stimulating private investment and competition.

Keywords:

• patent disclosure
• innovation
• R&D competition

JEL Codes :

• D23
• D83
• O34

Acknowledgements

This paper is the result of the ‘R&D competition’ research project carried out jointly with Adrian Towse and Martina Garau of the Office of Health Economics, London, UK. A preliminary draft of the paper has been presented at the DRUID Summer Conference 2006 (Copenhagen), and at the 11th ISS Conference (Sophia-Antipolis). Comments and suggestions by Fabio Montobbio, Mark Supekar, Cristina Tinti, conference discussants and participants, and two anonymous referees are gratefully acknowledged.

Notes

See also David, Mowery, and Steinmuller (1992), Pammolli (1996), Orsenigo, Pammolli, and Riccaboni (2001).

Art. 29(1) of the TRIPs agreement states that ‘[m]embers shall require that an applicant for a patent shall disclose the invention in a manner sufficiently clear and complete for the invention to be carried out by a person skilled in the art…’. Provisions similar to the one set forth in the TRIPs agreement appear, for instance, in art. 83 of the European Patent Convention and in 35 U.S.C. §112, para. 1.

This suggests the existence of some correlation between the scope of disclosure and the scope of claims, although it seems that a consensus is emerging on both sides of the Atlantic on the fact that this correspondence should not be considered excessively strict (Janis 2000).

A rather uncontroversial account of the success rate of innovative efforts in the pharmaceutical industry reveals that less than 1% of the new compounds object of preclinical investigation reach the human testing stage and that only around 20% of compounds going through the clinical trials ultimately gains FDA approval (DiMasi 1995). As for market uncertainty, a hint of its relevance comes from observing the distribution of returns to R&D for new drug introductions, which is highly skewed, with few blockbuster drugs accounting for most of the returns (Grabowski, Vernon, and DiMasi 2002). Moreover, while the degree of cumulativeness is overall high in the sector, cumulativeness at the firm level is low. Indeed, once a firm introduces a new product of major therapeutic value, its rivals explore the new line of research trying to develop similar or related drugs, and, in this race, the discovering firm seems not to have an advantage in discovering chemically related drugs. Sutton (1998) analyzed the top 50 selling drugs in 1960, 1973, and 1986 and provided evidence of high instability in the leadership within therapeutic classes.

The evidence presented in this paper looks at this link but highlights the reverse direction of causation, the one running from patent disclosures to the structure of innovative activity.

US patents in selected IPC and US classes are included in the database.

For a subset of projects, the database lists the patents protecting the compound under study. Old molecules and/or natural products, which do not have any associated patent, have been omitted. For projects listing a patent granted by a patent office other than the USPTO, we considered the US patent in the same family as the one listed in the database. In case no US patent is identified, the record is not considered in the analysis.

We refer to marketed/discontinued patents as the patents associated to marketed/discontinued R&D projects.

See Jaffe and Trajtenberg (2002) and the literature referenced therein.

We are aware that patent citation count is only a noisy proxy of the relevance of the knowledge disclosed, since citations might be included for strategic purposes or added by firm's lawyers or by patent examiners (Alcacer and Gittelman 2006). However, survey evidence shows that, even if noisy, patent citations are indicative of knowledge spillovers and communication among inventors (Jaffe, Trajtenberg, and Fogarty 2000).

See Jaffe and Trajtenberg (1996) for methodological details.

Trajtenberg (1990); Lanjouw and Schankerman (1999); Harhoff et al. (1999); Jaffe, Trajtenberg, and Fogarty (2000); Trajtenberg, Henderson, and Jaffe (1997); Jaffe and Trajtenberg (2002).

We compare the date of termination of the project (either marketed or discontinued) with the application year of the citing patents. Need it here to mention that average time to market or to discontinuation is not substantially different, being equal to 7.8 years for discontinued R&D projects and to 8.3 years for marketed R&D projects.

The share is computed over the total number of observed citations (respectively, self-citations and citations by other companies/institutions) for each patent. For example, 19.9% of the self-citations made to discontinued patents take place after the outcome of the project becomes known. Put it differently, 80.1% of the self-citations made to discontinued patents take place before the project is terminated.

Further research is needed to properly address this issue. Moreover, in order to distinguish ‘real’ knowledge spillovers, it would be useful to distinguish the citations added by the patent examiner.

We further restrict our sample and discard patents applied for before the year 1980, and patents assigned to individual inventors, leaving us with a sample of 2057 patents.

The instrumental variables selected for pt_succ and pt_fail are a measure of importance of the backward citations (pt_importb), the number of previous failed patents (Nfailure), and number of previous successful patents (Nsuccess).

Patents US4175127, WO8801169.

Oxazole patents: WO9513067, US5559137. Pyrazole patents: WO9705877, WO9705878. Substituted pyrazole patents: WO9852937, WO9852940, WO9852941, WO9852558.

One of the more exciting developments in oncology has been the identification of Topo I as a molecular target of a variety of anticancer drugs. First discovered in 1971, the enzymes are referred to as topoisomerases because they are able to change the topology of DNA molecules without changing the underlying chemical structure of the DNA.

Check the similarity in the molecular structures reported in Figure 4.

See Section 801.