Monday, 27 October 2008

Patent stacking -- the truth at last ...

The Journal of Competition Law and Economics (JCLE), published three times a year by Oxford University Press, carries occasional articles that are of great interest to aficionados of intellectual property. This issue features an article by Harvard Professor Einer Elhauge, "Do patent holdup and royalty stacking lead to systematically excessive royalties?". According to the abstract:
"Some recent literature has concluded that patent remedies result in systematically excessive royalties because of holdup and stacking problems. This article shows that this literature is mistaken. The royalty rates predicted by the holdup models are often (plausibly most of the time) below the true optimal rate. Further, those predicted royalty rates are overstated because of incorrect assumptions about constant demand, one-shot bargaining, and informational symmetry. Although this literature concludes that overcompensation problems are exacerbated by doctrines measuring damages using past negotiated royalties, in fact such doctrines exacerbate undercompensation problems. Undercompensation problems are further increased to the extent that juries cannot measure damages with perfect accuracy, a problem that persists even if damages are just as likely to be overestimated as underestimated. Nor do the royalty rates predicted by the holdup model apply if there is competition in the downstream product market or upstream market for inventions. Royalty stacking does not lead to royalties that exceed the optimal rate, contrary to this literature, but in fact tends to produce royalties that are at or below the optimal rate".
The article is not an easy read for lawyers on account of the algebra, but the conclusions are clear enough -- and in many respects extremely encouraging.

1 comment:

Jordan said...

Makes me wonder if patent damages have gotten so specialised if it would be good policy to move damages away from the jury and to a special master in charge of assessing the amount.