An Empirical Look at Software Patents

JPMH writes "At last, some quantitative analysis about software patents. This paper by Bessen and Hunt looks at the econometrics of who is being granted sw patents. It finds that far from encouraging innovation, software patenting is associated with firms which have *lower* then expected R&D spending as a proportion of sales (and even lower as a proportion of costs). Also, the companies which are most orientated towards software patenting (ie which have the largest proportion of their new patents in software) are the same companies who for their size are the most aggressive in the area of non-software patenting generally. This suggests that software patents are an unusually cheap and easy option for firms trying to build strategic "patent thickets" around their positions. Software patents have become a substitute, not an incentive, for innovation.
(Note: the key EU parliament committee vote in the debate on new legislation on software patenting in Europe is now expected on 10 June, not May 21. Still time to write those letters)." Read on for some interesting findings from the study.

"The paper itself is quite technical, with a detailed presentation of the mathematical economics involved, so here are some of Bessen and Hunt's more interesting results (see especially the tables at the end of the paper for more details):

• 69% of software patents in 1995-99 were assigned to manufacturing companies, despite these only employing 10% of programmers and analysts.
• Pure software publishers obtain unusually few patents per $10m R&D, only about a quarter of the rate for the whole economy. (But software services firms, and especially IBM, patent significantly more).
• In the 1980s, when software patents were hard to obtain, firms whose patents were mostly in software had unusally low numbers of patents in total (allowing for their size, R&D spending etc).
• Now this has reversed, and the firms with the largest proportion of their patents in software are those which are also granted the largest number of non-software patents (again, allowing for their size, R&D spending etc).
• A higher proportion of software patents is associated with a reduction in R&D as a proportion of sales, and an even larger reduction in R&D as a proportion of costs.
• Without the substitution of patent activity for research, "thumbnail calculations imply that at the end of the 90s R&D would have been about 10-15% higher" for the economy as a whole."