Suppose that A owns a patent and that B is contemplating the launch of a product that arguably would incorporate that patent. Suppose further that both parties believe that (1) there is a 70% probability the patent is valid, (2) there is an independent 80% probability that B’s product would incorporate the patent; and (3) the present value of B’s expected profit from the arguable use of the patent, over and above what B would earn from using the next-best noninfringing alternative, is $2 million. If the parties are equally good bargainers (meaning that they will share the gains from trade on a 50-50 basis), one would expect them to agree to a royalty under which B would pay A $560,000, calculated as follows: $2 million x 0.7 x 0.8 x 0.5 = $560,000.
As I have noted previously, citing work by Kalos & Putnam, in the event B infringes rather than licenses, A should recover $1 million, not $560,000. The reason is that if A knew she could recover only $560,000 if she went to trial, her expected payoff from going to trial would be $560,000 x 0.7 x 0.8 = $313,000. In other words, there would be a double discounting problem unless the royalty awarded at the patent infringement trial attempts to replicate what A and B would have agreed to ex ante, on the assumption that they both knew the patent to be valid and infringed.
However, my discussion the other day of remedies for the wrongful litigation or enforcement of patent rights has led me to realize that the above analysis is incomplete. In particular, two other variables must be considered: first, whether a patent infringement trial would decide both infringement and validity (as is the case in many countries, including the United States), or only infringement (as in several other countries, including Germany); and second, whether a subsequent decision finding the patent invalid would entitle B to restitution of any damages B paid to A (as in Germany) or not (as in the U.S., U.K., and France). The answer, it seems to me, is that if infringement and validity are not decided in the same forum, B should be entitled to restitution (unless the damages award is discounted by the ex ante probability of invalidity). Otherwise A may be getting a windfall, because her expected payoff from going to trial is $800,000 (that is, $1 million x 0.8), whereas she would have agreed to only $560,000 ex ante. Similarly, if infringement and validity are decided in the same forum, B should not be entitled to restitution, because otherwise A’s (ultimate) expected payoff would be less than $560,000.
Seen in this light, the rules followed in Germany and in the U.S. seem to make sense. As I understand it, German courts, like U.S. courts, do not take into account the possibility that A and B would have agreed to a discount reflecting the possibility of noninfringement or invalidity. See Benkard, Patentgesetz, 10th ed., § 139, para. 64 (“Ein Abschlag, den vernünftige Vertragspartner mit Rücksicht auf eine vielfach gegebene Unsicherheit über materielle Schutzfähigkeit und Bestand des Patents machen, kann hier unberücksichtigt bleibe, da der Verletzer sich nicht dem Risiko aussetzt, für eine nicht schutzwürdige Erfindung zahlen zu müssen . . . .”) One possible wrinkle, though, is this: in Germany, licensing partners sometimes do agree that the licensee will pay an upfront premium which the licensor gets to keep in the event the patent is invalidated. (If the patent is invalidated, the licensor is not required to give back royalties the licensee has already paid, even though the patentee may be required to give back its damages to the infringer.) This rule would risk making the licensee worse off than the infringer, so German courts increase the royalty the infringer must pay above the normal rate in order to compensate for the infringer’s having avoided this risk. In a sense, then, perhaps one could say that to some extent German courts may take the probability of invalidity into account in awarding royalties. See Thomas Kühnen, Patent Litigation Proceedings in Germany: A Handbook for Practitioners 541 (Frank Peterreins tr., 6th ed. 2013) (stating that a factor taken into account in terms of increasing the royalty is that “unlike the licensee, the infringer bears no risk of payment for invalid property rights, the level of the additional amount depending on the extent to which the patent in suit is actually in danger of being revoked”). I'm not sure how German courts would go about estimating "the extent to which the patent in suit is actually in danger of being revoked," though sometimes estimates of this nature are made in connection with motions for preliminary injunctions.