WORKING PAPER NO. 03-17/R AN EMPIRICAL LOOK AT SOFTWARE PATENTS James Bessen* Research on Innovation and Boston University School of Law (Visiting Researcher) Robert M. Hunt** Federal Reserve Bank of Philadelphia First Draft: August 2003 This Draft: March 2004 * [email protected] ** Ten Independence Mall, Philadelphia, PA 19106. Phone: (215) 574-3806. Email: [email protected] Thanks to Peter Bessen of May8Software for providing a software agent to acquire our patent database and Annette Fratantaro for her work with the Compustat data set. Also thanks to John Allison, Tony Breitzman and CHI Research, Iain Cockburn, Mary Daly, Dan Elfenbein, Terry Fisher, Bronwyn Hall, Joachim Henkel, Brian Kahin, David Mowery, Leonard Nakamura, Cecil Quillen, Eric von Hippel, Rosemarie Ziedonis and seminar participants at APPAM, Berkeley, EPIP Munich, Federal Reserve Banks of Philadelphia and San Francisco, the Federal Reserve System Applied Micro meetings, Harvard, IDEI, MIT, NBER, and OECD. The views expressed here are those of the authors and do not necessarily represent the views of the Federal Reserve Bank of Philadelphia or the Federal Reserve System. 2004, Verbatim copying and distribution of this entire article for noncommercial use is permitted in any medium provided this notice is preserved.AN EMPIRICAL LOOK AT SOFTWARE PATENTS James Bessen Research on Innovation and Boston University School of Law (Visiting Researcher) Robert M. Hunt Federal Reserve Bank of Philadelphia March 2004 Abstract: U.S. legal changes have made it easier to obtain patents on inventions that use software. Software patents have grown rapidly and now comprise 15 percent of all patents. They are acquired primarily by large manufacturing firms in industries known for strategic patenting; only 5 percent belong to software publishers. The very large increase in software patent propensity over time is not adequately explained by changes in R&D investments, employment of computer programmers, or productivity growth. The residual increase in patent propensity is consistent with a sizeable rise in the cost effectiveness of software patents during the 1990s. We find evidence that software patents substitute for R&D at the firm level; they are associated with lower R&D intensity. This result occurs primarily in industries known for strategic patenting and is difficult to reconcile with the traditional incentive theory of patents. Keywords: Software, Patents, Innovation, Technological Change JEL classification: O34, D23, L863Introduction Federal courts, and to a lesser extent the U.S. Patent and Trademark Office (USPTO), dramatically changed standards for patenting software-related inventions over the last three decades. During the 1970s, federal court decisions typically described computer programs as mathematical algorithms, which are unpatentable subject matter under U.S. law.1 Systems using software could be patented, but only if the novel aspects of the invention did not reside entirely in the software.2 At this time, the U.S. Congress considered the question of patenting software and instead opted to protect computer programs under copyright law.3 But after the Supreme Court decision in Diamond v. Diehr in 1981,4 a series of court and administrative decisions gradually relaxed the subject matter exception that restricted the patenting of software-related inventions. The 1994 decision In re Alapat eliminated much of the remaining uncertainty over the patentability of computer programs.5 During this same period, new legislation and other court decisions lowered standards for obtaining patents in general, while strengthening aspects of patent enforcement. This paper explores the general characteristics of software patenting over the last two decades, paying particular attention to the rapid growth in software patenting and the effect of this growth on R&D. We construct our own definition of a software patent (there is no official definition) and assemble a comprehensive database of all such patents. In Section I we describe this process, and the process of matching these patents to firm data in the Compustat database. In Section II we summarize the general characteristics of this data. We find that over 20,000 software patents are now granted each year, comprising about 15 percent of all patents. Compared with other patents, 1 See, for example, the Supreme Court decision in Gottschalk v. Benson, 409 U.S. 63 (1972). 2 Parker v. Flook 437 U.S. 584 (1978). 3 U.S. Copyright law was amended in 1976, and more explicitly in 1980, to include computer programs. See H. Rpt. No. 94-1476 (1976) and P.L. 96-517 (94 Stat 3028). There is a voluminous literature on the merits of different forms of intellectual property protection for computer programs. See, for example, Dam (1995), Graham and Zerbe (1996), and Samuelson et al. (1994). 4 450 U.S. 175 (1981). 5 33 F.3d 1526 (Fed. Cir. 1994).4software patents are more likely to be assigned to firms, especially larger U.S. firms, than to individuals. They are also more likely to have U.S. inventors. Surprisingly, most software patents are assigned to manufacturing firms and relatively few are actually assigned to firms in the software publishing industry (SIC 7372). Most software patents are acquired by firms in industries that are known to accumulate large patent portfolios and to pursue patents for strategic reasons (computers, electrical equipment, and instruments). These large inter-industry differences remain even after we control for R&D, software development effort, and other factors. In Section III we perform regressions that explore the “propensity to patent” software inventions. This builds on the model of Hall and Ziedonis (2001) which, in turn, builds on the empirical literature of “patent production functions” (including Scherer 1965, Bound et al. 1984, Pakes and Griliches 1984, and Griliches, Hall, and Hausman 1986). We find a dramatic growth in software patent propensity even after controlling for R&D, employment of computer programmers, and other factors. This growth is quite similar to the remarkable growth in patent propensity that Hall and Ziedonis (2001) found in the semiconductor industry. Productivity-based explanations are unlikely to account for even half of the rise in software patent propensity. The pattern of the residual increase is consistent with the explanation that changes in patent law made software
View Full Document
Unlocking...