Patenting Behaviour and the Survival of Newly Listed European Software Firms

Abstract

We test whether patenting activity impacts on software companies' likelihood of survival after going public in the UK, Germany, France, Sweden, Italy and Spain. Our database covers all software companies undertaking IPOs in these six countries between 1 January 1997 and 31 December 2005, and includes data from various sources (the Bureau van Dijk Zephyr database, the Questel-Orbit QPAT patent database, financial documents available on the company websites and specialised websites). Survival analysis follows a semi-parametric approach, based on the stratified Cox competing risk model, controlling for other determinants of survival. We find that, after controlling for the firms' main entry characteristics (experience, size, sales, profitability and solvency, together with market conditions), the influence of the size and the quality of the firms' patent portfolios is different according to the type of exit and the type of software firm. In particular, the number of patents reduces the risk of failure and acquisition for software developers, while quality increases their attractiveness as an acquisition target.

Keywords::

• European Software Firms
• patent metrics
• initial public offering (IPO)
• survival
• start-ups

JEL Classification::

• O34 Intellectual Property Rights
• O32 Management of Technological Innovation and R&D

Acknowledgements

I would like to thank Valerio Sterzi for his econometric support and for his valuable comments. Special thanks also go to Laurent Berger, Philippe Gorry, Yann Ménière, Julien Pénin, Ammon Salter, Sébastien Rouillon and the anonymous reviewers of Industry and Innovation for their helpful comments and suggestions to improve previous drafts of this paper. All errors are mine.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ Most of the newly listed software companies in Europe are by definition “New Technology Based Firms” which are newborn Young Innovative Companies (YICs) operating in high-tech sectors.

² On 19 February 2013, 25 Member States of the European Union agreed to implement a “European patent with unitary effect”. The European “unitary patent” will co-exist with national patents and with classical European patents. The aim of the unitary patent protection will be to make access to the patent system easier, less costly and legally secure. However, this institutional innovation is too recent to be considered in this paper.

³ In Europe, a computer-implemented invention must also be “technical” to qualify for patent protection (See European Patent Convention Rule 27). In consequence, “a computer program claimed by itself is not excluded from patentability if the program, when running on a computer or loaded into a computer, brings about, or is capable of bringing about, a technical effect which goes beyond the ‘normal’ physical interactions between the program (software) and the computer (hardware) on which it is run” (see: T935/97 and T1173/97, Official Journal of the EPO 1999, 609).

⁴http://www.questel.com/Prodsandservices/FamPat.htm

⁵ In our sample, companies that were acquired while they were in the process of liquidation were coded as acquired.

⁶ A firm can also be delisted from a particular public stock exchange if it is taken privately or if the company decides to change stock market. This was the case for some European companies in this sample. In such cases, companies in our sample which changed stock market but continued to operate in the new stock market until 31 December 2011 were coded as censored.

⁷ We did not evaluate “software patents”, the definition of which can be arbitrary (Mann 2005), but we did evaluate all the innovative input that might emerge in a complex innovation process with other firms, providers and clients in the different business segments in which software companies operate.

⁸ We are aware that this window in the number of forward citations is not easily observable by potential firm acquirers but they may reflect a better indicator of the quality of technologies patented prior to IPO.

⁹ The Bureau van Dijk Zephyr database defined shareholders' funds as the sum of the target company's capital and other shareholders' funds including reserves.

¹⁰ In most cases, for companies developing software, we found “software developer” but there was also “video games developer”, “image technology developer”, “systems developer”, “platforms developer” or “website developer”. For Internet-related companies, we also use the key words “e-commerce”, “online”, “website” or “web”.

¹¹ Results are available upon request.

¹² In robustness checks, we compared the Cox model regression with the stratified Cox regression (See Appendix 4, Table 10 in Supplemental data). Results reflect that the coefficients are very similar among both models. However, the stratified Cox regression provides more efficient results for all the parameters.

¹³ We also test the number of PCT applications, as an alternative measure of patent quality, which are not statistically significant for the hazard of exit in a short duration through the different mechanisms. For the sake of brevity, these regressions were not included.

¹⁴ It should be pointed out that in our simple test, on average, software companies were acquired in 5.3 years while companies with cited patents were acquired in 7 years. Only two companies that have patents with forward citations were acquired within three years of their IPO. For only one company, two forward citations were obtained after the firm's acquisition and these citations were not made by the acquirer.

¹⁵ The quadratic terms of revenues, assets and age at IPO are not statistically significant. For the sake of brevity, the regressions have not been included but are available upon request.

¹⁶ As an alternative measure of market conditions, we also included a dummy variable (bubble period) coded one if the company was introduced between 1999 and 2000, 0 otherwise. Surprisingly, results reflect that the market conditions were not statistically significant for the hazard of exit through the different mechanisms in Europe. We do not control by SOFT_ENTRY and bubble period at the same time because they are strongly collinear and introduce problems of non-proportionality. Results are available upon request.

Previous drafts of this paper have benefited from helpful discussion and comments of colleagues at MINES PARISTECH, University of Bordeaux, DRUID 2013 Conference and CODE (ZEW) 2014 conference.

Additional information

Funding

Financial support from GREThA-CNRS Bordeaux is gratefully acknowledged.