Are two better than one? Modelling the complementarity between patents and trademarks across industries

ABSTRACT

Intellectual property (IP) rights are a major component of firms’ strategies to appropriate the benefits of their innovations. This paper aims at assessing the interactions between two types of IP rights, namely patents and trademarks. We first model the effect of these two types of IP rights on the returns of innovations for firms. Based on a supermodularity analysis, we then show that the complementarity between trademarks and patents varies according to the characteristics of the market. Depending on the levels of advertising’s spillovers and depreciation rate, trademarks are found to be complementary or not to patents. Finally, based on a data set encompassing the IP activity of a sample of publicly traded firms among the top corporate R&D investors worldwide, we find that patents and trademarks are complementary in chemical and pharmaceutical sectors, but not in Information and Communication Technologies (ICT) sectors.

KEYWORDS:

• Patents
• trademarks
• IPRs
• complementarity
• supermodularity
• innovation
• brands

JEL:

• O32
• O34
• L10

Acknowledgments

The opinions expressed in this paper are the sole responsibility of the authors and do not necessarily reflect those of the OECD or of the governments of its member countries. Authors are grateful to Tim Folta, Georg von Graevenitz, Mariagrazia Squicciarini, Hélène Dernis, Betrand Koebel, Isabelle Maret, Christian Martinez-Diaz, the editors and the referees for their helpful suggestions and valuable comments. All mistakes remain those of the authors alone.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ This constitutes a simplifying assumption as, in practice, firms in some industries may protect their innovation with several patents or several trademarks..

² Cassiman and Veugelers (2006) use the supermodularity framework for the analysis of the complementarity between internal R&D and external knowledge acquisition; Guidetti, Mancinelli, and Mazzanti (2009) for the complementarity in training practices; Cavaco and Crifo (2014) for the complementarity between various dimensions of corporate social responsibility and financial performance; Ballot et al. (2015) for complementarities in performance between product, process and organisational innovation; and Fares, Raza, and Thomas (2018) for the complementarity between labels and brands.

³ For the sake of readability, we do not report all the calculation steps here. The full demonstration with detailed calculation steps is available from the authors on request.

⁴ The cost associated to filing a trademark or a patent are not negligible. Indeed recent research shows that patent and trademark fees have significant effect on the demand for patent and trademark applications (Rassenfosse 2019; Herz and Mejer 2016). Moreover, firms often use other services such as legal advice for IP filings which are likely to generate additional costs..

⁵ In other competition frameworks, we might observe that followers have lower advertising expenditure than leaders, and compensate by a significantly lower selling price of the same product, which corresponds for example to the situation of firms selling generic drugs.

⁶ Assuming an annual interest rate of r₀, and a patent period of T years, the discount rate between the two periods in the model can be approximated by $r\equiv \frac{{\sum }_{t=0}^T{\left(\frac{1}{1+r_0}\right)}^t}{{\sum }_{t=T+1}^{\mathrm{\infty }}{\left(\frac{1}{1+r_0}\right)}^t}=1-{\left(\frac{1}{1+r_0}\right)}^T$. Taking r₀ = 0.05, and considering that the patent period lasts twenty years, we obtain a value of $\cong 0.6$.

⁷ The conditions of the model require that${\left(\frac{4\beta \left(1-r\frac{1-\delta }{1-\bar{s}s}\right)-{\tau }^2}{6\beta -{\tau }^2}\right)}^2>\frac{1-\delta }{1-\bar{s}}$, so that with β = τ = 1 and $\bar{s}=1/2$, the model does not admit an equilibrium for δ < 0.85.

⁸ Szegedy-Maszak (2004) for example found that the advertising for a new drug for adult attention deficit disorder, Strattera, increased demand for all drugs targeted to this disorder.

⁹ According to Bilir (2014), electronics products and computers have some of the shortest product life cycle length.

¹⁰ As noted by Ben Elhadj, Lahmandi-Ayed, and Laussel (2014), advertising spillovers can even be negative in some cases, as for example in the mobile phone market.

¹¹ The fact that we look at the evolution of the ratio of market value to assets over a limited period of time reduces the possibility that our empirical model captures the effect of innate differences in the nature of the firms, in addition to the effect of IP strategy choices.

¹² The previous threshold is ± 1.282 at 10% level, and ± 2.326 at 1% level.

¹³ A number of other initiatives have aimed at linking firm-level data and IP data, notably the NBER patent data project, linking USPTO patents to the Standard and Poor’s Compustat database on US firms (Hall et al., 2000) and the Oxford Firm-Level Intellectual Property (OFLIP) database, linking UK firm data from the FAME database and UKIPO patents and trade marks data (Helmers, Rogers and Schautschick, 2011). For an overview of other initiatives, see for example Helmers, Rogers and Schautschick (2011)..

¹⁴ To be consistent with the theoretical model, the sample should ideally be restricted to innovative firms. Otherwise, we cannot know if firms have no IP right activity because they do not innovate (which would have a negative impact on market value compared to other firms) or because they innovate but do not protect their innovations with IP rights. One possibility to have information on innovating behaviour would be to match the dataset with innovation survey data. However, because of the small size of innovation survey samples, this would reduce our sample size drastically. Moreover our sample, which focuses on the top corporate R&D investors worldwide, is by nature likely to be composed of innovative firms..

¹⁵ This corresponds to our theoretical framework, where the IP choice is considered as discrete: the firm either files a trademark or not, and either files a patent or not..

¹⁶ These results are robust to including additional controls on the firm characteristics such as detailed information on the sector of activity and information on size based on the number of employees. The corresponding results are available upon request.