TIIP Newsletter

What is a software patent? Who gets them, and why do these firms obtain so many? Does the availability of software patents contribute to innovation? These are some of the questions that this working paper addresses.

During the 1970s, software patents were difficult to obtain. Over time it became gradually easier, and by the mid 1990s, federal courts treated the patentability of computer programs in much the same was as they did any other technology. The result has been a dramatic increase in the number of software patents granted—over 20,000 a year in recent years.

One difficulty in evaluating the effects of this shift is that no official definition or classification of software patents exists. After examining and classifying 400 patents, the authors developed a keyword search to identify software patents[1] and identified over 130,000 granted between 1976 and 1999 by the U.S. Patent and Trademark Office (USPTO).

The authors examined the distribution of software patents across industries. Between 1994 and 1997, manufacturing firms received three of every four software patents granted—the machinery and electronics industries alone obtained a majority. Software publishers, meanwhile, obtained only 5 percent of software patents. Controlling for investments in R&D and other factors, the authors found that firms in the machinery, electronics, and instrument industries obtained software patents at a rate four to ten times higher than firms in the software industry. These industries also hold a majority of patents of all kinds issued to firms in the U.S.

In fact, an industry’s share of software patents was much more closely related to its share of all patents obtained than to its share of inputs associated with software creation. For example, manufacturing firms employed roughly one in ten programmers, and the machinery and electronics industries employed only one in twenty. Software firms employed a third of all programmers.

Using regression analysis, the authors found that only one third of the total growth in software patents—about 16 percent a year—could be explained by changes in firm characteristics like R&D, capital spending, or employment of computer programmers at the industry level. The remainder can be explained by changes in the legal treatment of software inventions, increases in productivity among programmers, or both. They argue that according to the available evidence, it seems that less than half of this amount can be explained by productivity growth.

All things being equal, firms successfully applied for 50 percent more software patents in 1991 than in 1987; by 1996, that rate had exceeded 150 percent. There was no difference in the growth in software patent propensity among younger and older firms, except among those in the software industry. Here, new firms tended to patent less intensively than incumbents.

One way to answer this question is to investigate the effects of patents on firms’ investments in R&D. The authors ask whether software patents and R&D are complementary inputs or substitutes in the production of profits. Ordinarily, we would expect them to be complements—more R&D should lead to more inventions, making additional patents easier to obtain. Similarly, stronger patent protection should increase the expected return from successful R&D projects, stimulating more R&D.

This is not what the authors found. They examined the relationship between changes in firms’ R&D intensity (the ratio of R&D to sales), changes in the cost of other inputs, and changes in their focus on software patents—the percentage of software patents among all of their new patents—over five-year intervals.[2] The authors found that during the 1990s, all else equal, firms who increased their focus on software patents tended to reduce their R&D intensity relative to their peers. This suggests that in the 1990s, software patents substituted for R&D. This negative relationship was found only in certain industries, specifically those industries noted strategic patenting.

Taking the analysis literally, if the number of software patents grew only as rapidly as that of all other patents after 1991, the average R&D intensity of U.S. firms would be about 7 percent higher than was actually recorded in 1997. This amounts to about $9 billion in additional private R&D.

These patterns seem inconsistent with the expectation that granting more and stronger intellectual property rights on software inventions stimulates R&D. But suppose instead that firms assemble large patent portfolios in order to extract royalties from competitors and to defend themselves from similar behavior by their rivals. In theory at least, extensive competition in patents, rather than in inventions, may occur when firms rely on similar technologies and the cost of assembling large portfolios is not very high. In such an environment, firms may compete to tax one another’s inventions by demanding royalties. In the process, they can reduce one another’s incentive to engage in R&D.[3] Under these circumstances, firms may find themselves competing not in the marketplace but in court.

Bessen, James. 2003. “Patent Thickets: Strategic Patenting of Complex Technologies,” Research on Innovation Working Paper.

Bessen, James and Robert M. Hunt. 2004. “The Software Patent Experiment,” Federal Reserve Bank of Philadelphia Business Review, 3^rd Quarter.

* The views expressed here are those of the author and do not necessarily represent the views of the Federal Reserve Bank of Philadelphia or the Federal Reserve System.

[1] The authors also compare their definition and method to others found in the literature.

[2] This approach is equivalent to estimating changes in a firm’s demand for inputs (in this case for R&D) induced by changes in their price, including the relative cost of obtaining software patents.

[3] For a theoretical model of such behavior, see Jim Bessen’s 2003 working paper.

An Empirical Look at Software Patents

This work is licensed under a Creative Commons License.