The Impact of Open Source Software Commercialization on Firm Value

Abstract

Vendors of proprietary software products are increasingly moving to business models inspired by open source software (OSS). This study investigates sources of heterogeneity in value appropriation associated with commercializing OSS. Specifically, I suggest that the relationship between a firm's OSS releases and its value depends critically on its stocks of protection mechanisms for intellectual property rights, such as software patents and software trademarks. I find that while software patent stocks positively affect the relationship between a firm's OSS product portfolio and its value, software trademark stocks have a negative effect on this relationship.

Keywords::

• Open source software
• intellectual property rights
• firm value

JEL Classification::

• O31
• O34
• L86

Acknowledgements

The author acknowledges helpful comments from Marco Giarratana, Andrea Fosfuri, Lars Frederiksen, Francesco Rullani, Eduardo Melero, Benjamin Engelstatter and the two anonymous reviewers. Prior versions of this research were presented at the Druid Academy Winter Conference 2011, Druid Summer Conference 2011, and 9th ZEW Conference on the Economics of ICT. The comments and suggestions of participants at these conferences were valuable in improving the article. Any remaining errors are my own.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ These industries include Electronics & Electrical Equipment, Computers & Office Equipment, Computer Software, Telecommunications and Semiconductors.

² The excluded firms are those that did not report financial information in some of the years during the period of analysis. This could have been due to a corporate event, such as an acquisition, filing bankruptcy or spinoff.

³ Of the 70 firms in the sample, 28 introduced at least one OSS product to the market during the 7-year period. Considering that commercialization of OSS is a rising trend but not yet a common practice, the sample size is within an acceptable range. I identified all the firms from the Fortune Global 500 list that operate in the software industry at some level, even if the principal sector in which they operate differed. In this respect, the sample is comprehensive.

⁴ The Bureau of Industry Economics (1994) adopts a discount rate of 15 per cent when assessing private returns to patent holders because the value of patents tends to fall over time. I follow a similar logic for assessing private returns from OSS commericalization because the value associated with older products will diminish over time as technology evolves and enhanced products enter the market.

⁵ The groups included are as follows: G06F 3, 5, 7, 9, 11, 12, 13, 15; G06K 9, 15 and H04L 9.

⁶ See Fosfuri, Giarratana, and Luzzi (2008) for details on the accuracy of the algorithm.

⁷ This study focuses on the moderating effect of software trademarks and software patents- and not on the moderating effect of hardware trademarks and hardware patents. As has been discussed in Hypotheses 1 and 2, I expect the relationship between OSS commercialization and Tobin's q to be affected by trademarks/patents through certain mechanisms that are mostly strategic and that can take place exclusively if these trademarks/patents pertain to software domain.

⁸ I split the data into five sub-samples and use the same model to estimate coefficients on the interaction terms. The sub-samples are created using Standard Industry Classification codes for the industries in which firms operate. The five categories through which I aim to proxy for firms' technical expertise in developing software are as follows: hardware, software, electronics, semiconductors and telecommunications. Although the significance of the coefficients on the interaction terms diminishes at some level, the effects remain the same in terms of directionality. The results suggest that the findings are robust to firms' technical expertise in developing software.

⁹ In Figures 1 and 2, the minimum value for OSS portfolio is below zero because the representative values were obtained on the basis of the mean-centered values of the explanatory variables that I used to compute the interaction terms.

¹⁰ Gross profit margin, also known as profit margin, is a commonly used accounting measure of performance that can be calculated by the following simple formula: Gross profit/Net sales.