Gaétan de Rassenfosse and Adam Jaffe have posted a new empirical working paper, Are Patent Fees Effective at Weeding Out Low-Quality Patents? Short answer: yes. Longer answer:
The authors used the difference-in-differences technique to study the effect of the Patent Law Amendment Act of 1982, which increased the fees for maintaining a patent to full term from $239 to $3,200. They looked at four indicators of patent quality: citations received, number of claims at grant (as an indicator of breadth), number of jurisdictions in which patent protection is sought, number of times renewed. These factors have been shown to be correlated with economic importance, although the correlation is sometimes murky. And a very important caveat is that economic value is only way to think about patent quality; others might define low-quality patents as those that are legally invalid, or as those where the patent grant was unnecessary to induce the invention. It is unclear how these three very different definitions are related, and each is important in different contexts.
So to the extent we care about rough proxies for economic value, how were those affected by the large increase in fees in 1982? The authors estimate that 24-30% of patents in the lowest quality quintile were filtered out by the change, although the increase in average quality was not uniform across patentees: the effect for the largest patentees (with portfolios of over 100 patents) was 1/3 to 1/2 as big as the effect for the smallest patentees.
This new paper provides a useful piece of historical evidence that contributes to a growing scholarly interest in the use of fees as a policy lever in patent law. For example, Ayres & Parchomovsky (2007) discuss the use of renewal fees to regulate patents, and Bessen & Love (2013) argue for the use of fees to limit abusive patent practices. And a terrific new piece by Neel Sukhatme explores how the PTO could use its regulatory monopoly power to set differential patent prices in the public interest (and uses some of his own empirical research to suggest actual differential prices).