Geradin (2007) - Royalty Stacking In High Tech Industries Separating Myth From Reality

• This page is referenced in the Patent Thicket Literature Review
• This page is listed on the PTLR Core Papers page

Reference

• Geradin, D., Layne-Farrar, A. and Padilla Blanco, A. (2007), "Royalty stacking in high tech industries: separating myth from reality", Working Paper

@article{geradin2007royalty,
  title={Royalty stacking in high tech industries: separating myth from reality},
  author={Geradin, D. and Layne-Farrar, A. and Padilla Blanco, A.},
  year={2007},
  abstract={A few recent contributions to the literature have claimed that in high-tech industries -- where innovation is often cumulative and products include many components protected by patents held by many different patent holders – the cost of obtaining all necessary licenses is too high. Some have even requested sweeping policy reforms to deal with the so-called royalty stacking problem. In this Essay we find that the empirical evidence – including new evidence for 3G telecom – does not corroborate the gloomy predictions of the proponents of the royalty stacking hypothesis. A careful look at the theoretical underpinnings of this hypothesis explains the lack of empirical support. First, three necessary conditions must be satisfied for a royalty stacking problem to exist: (a) innovation must be cumulative, so that the patents are complementary; (b) there must be many patents for a given product; and (c) the many patents must be held by numerous, distinct rights holders. Buy royalty stacking may not be a problem even if the three necessary conditions are met; i.e., the three conditions are necessary but not sufficient. Moreover, several market mechanisms, such as cross licensing or voluntary patent pools, can be used to mitigate the costs of multiple concurrent patent negotiations. We conclude that the so-called royalty stacking problem is more myth than reality and that there is no reason to adopt the dramatic reforms in antitrust and patent law that have been recently proposed.},
  discipline={Econ},
  research_type={Empirical},
  industry={},
  thicket_stance={},
  thicket_stance_extract={},
  thicket_def={},
  thicket_def_extract={},  
  tags={},
  filename={Geradin (2007) - Royalty Stacking In High Tech Industries Separating Myth From Reality.pdf}
}

File(s)

• Download the PDF
• Repository record

Abstract

A few recent contributions to the literature have claimed that in high-tech industries -- where innovation is often cumulative and products include many components protected by patents held by many different patent holders – the cost of obtaining all necessary licenses is too high. Some have even requested sweeping policy reforms to deal with the so-called royalty stacking problem. In this Essay we find that the empirical evidence – including new evidence for 3G telecom – does not corroborate the gloomy predictions of the proponents of the royalty stacking hypothesis. A careful look at the theoretical underpinnings of this hypothesis explains the lack of empirical support. First, three necessary conditions must be satisfied for a royalty stacking problem to exist: (a) innovation must be cumulative, so that the patents are complementary; (b) there must be many patents for a given product; and (c) the many patents must be held by numerous, distinct rights holders. Buy royalty stacking may not be a problem even if the three necessary conditions are met; i.e., the three conditions are necessary but not sufficient. Moreover, several market mechanisms, such as cross licensing or voluntary patent pools, can be used to mitigate the costs of multiple concurrent patent negotiations. We conclude that the so-called royalty stacking problem is more myth than reality and that there is no reason to adopt the dramatic reforms in antitrust and patent law that have been recently proposed.

Review

Theoretical Model

TBD

Measures of Patent Thicket

Quantitative measures of patent thicket or royalty stacking effects is:

• Citation concentration of the top 4 firms in a technology area:
  • The measure sums citation shares of each of the top 4 firm's backward patent citations to the other firm's patents at a point in time;
  • As a robustness check, the authors also considered an 8 firm concentration measure and a traditional HHI measures based on sum of squared citations, and obtained qualitatively similar effects.

Sample

The data analyzes 63 firms with at least one 3G patent over the period from 1996-2005:

• The universe of 60 firms contributing patents to the 3G telecomm standard, plus 82 firms in SIC code 3663 (Radio and Television Broadcasting and Communications Equipment) and 4812 (Radiophone and Communications) was restricted to those with at least one U.S. 3G patent.
• Firm financial data (sales, assets, employees) for 1980-2005 is drawn from Compustat.
• Patent data (grant dates, technology classes, citations and prior art listing) from USPTO.

Results

"Since concentration at this level has nearly no statistically significant effect on firm market value, we conclude that four firms translate into concentration for this industry and fragmentation is low among public cellular telecom firms, especially among vertically integrated firms. Thus, we find that the predicted royalty stacking problem is not an issue in this industry."

• R&D spillover effects are positively related to market value for upstream firms, but negatively related for vertically integrated and Operator firms, although estimates are not significant;
• Patenting propensity is negatively related to market value for upstream and vertically integrated firms, but positivey among Operator firms, and estimates are significant for vertically integrated firms.
• The patent thicket measure, citation concentation, is positively related to market value for upstream and vertically integrated firms, negative for operators, although estimates are not significant.
• Insignificant results may reflect the low levels of fragmentation in this data - the concentration measure for public cellular telecom firms ranges from 53% to 100% with an average of 83%, indicating that most patent citations are concentrated in the top 4 firms.

Social Welfare Consequences

"The royalty stacking theory is not robust and, what is most important, even in such industries where innovation is cumulative and products integrate many components each of which may be subject to one or more patents, there is no real evidence that royalty stacking is a problem in practice. In contrast to recent claims by Lemley and Shapiro (2006), we find no evidence of a significant royalty stacking problem in the 3G telecoms industry...The empirical and theoretical results of this Essay seriously question the underpinnings of recent proposals advocating various changes in antitrust enforcement and patent law, which are meant to address the purported tragedy of the anti-commons and the misfortunes caused by ever growing patent thickets."

Dependent Variable and Model

The dependent variable is Tobin's Q (ratio of firm's market value to tangible assets).

• The model specification is OLS with robust standard errors.
• Patenting control variables include:
  • Citation Concentration, meant to capture patent thickets or royalty stacking effects (see above for definition);
  • Patenting propensity, as measured by the firm's patent stock and R&D stock in a given year;
  • Spillover, as measured by the sum of R&D spending of close competitors weighted by the citation concentation, which is the uncentered correlation between the two firms patent citations (excluding self-citations) across technology classes.
  • Each of these patenting variables is interacted with indicators of whether the firm has an Operator or an upstream/vertically integrated structure.
• Firm and industry controls are:
  • Shadow cost of physical capital, measured by ratio off R&D spending to tangible assets (net stock of property, plant and equipment), and higher order terms;
  • Shadow cost of intellectual capital, measured by ratio of the number of a firm's patents to tangible assets (net stock of property, plant and equipment), and higher order terms;
  • Indicators of whether firms are Operators, Upstream (design firms such as Qualcomm) or Vertically Integrated firms (>10% of revenue from manufacturing), with vertically integrated firms as the omitted group;
  • log Net sales and its lag;
  • Technological opportunity in a technology area, as measured by the logarithm of the weighted sum of the patent stock of a company's technologically close competitors for each patent technology class, and a lag;
  • Year and industry indicators.