I have argued elsewhere that the principal economic justification for awarding permanent injunctions in patent infringement cases comes down to information costs: in other words, that the parties are in a better position than is a court to determine what the patent is worth in terms of how much it is likely to contribute to the infringer’s ability to increase profits or reduce costs. (Note that when the infringer is a less efficient user of the invention than is the patentee, you wouldn’t expect the parties to reach any agreement. The patentee would simply exclude the infringer, which is the correct outcome assuming that we want to preserve the patent incentive. In such a case, an adequate alternative to an injunction would be a court-ordered royalty that would exceed the amount the infringer would be willing to pay. When the infringer is more efficient, you would expect the parties to bargain toward a license following entry of the injunction, if they haven't done so already.) Thus when courts award ongoing royalties they may generate two costs that otherwise would be avoided: the adjudication costs incurred as a result of having to litigate the amount of the royalty, above what would have been incurred if the court had awarded an injunction instead; and error costs resulting from the court setting the amount of the royalty too high or too low. Those error costs would take the form of either unnecessary deadweight loss if the royalty is set higher than what the parties would have agreed to, or the social costs of whatever consequences follow from setting the royalty too low and thus potentially undermining the patent incentive. We can denote adjudication costs with the letter A and error costs with the letter E.
On the other hand, there is also a social cost to entering an injunction. One possible cost is that an injunction will limit access to some product that is necessary to public health or safety, and it is this cost that might justify compulsory licensing in some cases, but I will put that matter to one side for now and focus instead on the cost that is more likely to arise in a broader range of cases, namely patent holdup. The touchstone of patent holdup is that the patentee can use an injunction to extract a royalty that reflects not only the ex ante value of the patent but also the ex post costs that the infringer would face in trying to design around (or, to put it another way, some of the value of complementary technologies with which the patented invention interacts). I view this as a social cost because it results in greater deadweight loss than is necessary to preserve the patent incentive; the patentee winds up getting a royalty that is higher than the value of the patent’s contribution to the state of the art. Let’s call the costs arising from patent holdup H.
If my analysis is correct, then courts should award the prevailing patentee a permanent injunction when A+ E > H, and they should deny an injunction when A + E < H.
To be sure, no one can quantify these variables; but I think we can draw some inferences as to whether A + E is likely to be greater or less than H in a large number of cases. First, there are likely to be cases where the cost of holdup is small. A principal example would again be when the patentee would not agree to any royalty because the patentee is the more efficient user and therefore prefers exclusion to licensing. There’s no holdup because the patentee will not be extracting any future rents from the infringer. Second, while traditionally some analysts have believed that what I am calling A and E are high, that may not always be the case. In particular, where the court is awarding past damages that provide a basis for calculating a future royalty, A may not be all that significant. While some might express concern that courts are more likely to undervalue than overvalue the patent (so that error costs will be potentially large and will undermine the patent incentive), I’m not sure that’s necessarily right. In cases in which courts grant compulsory licenses (say, for drugs), that may be a risk, because the whole point of that type of compulsory licensing is to make the invention available at a below-market price. If we exclude those cases from our consideration and focus on more common examples, I’m not sure there’s any theoretical reason to expect awards systematically to be biased one way or the other. I suppose that if infringers can somehow calculate that they will negotiate when a court is likely to award an overcompensatory remedy and to infringe when it is likely to award an undercompensatory remedy, self-selection would lead to systematic undercompensation. But that seems to require a good deal of foresight on the part of infringers, as well a willingness to ignore the high cost of attorney fees and the risk of enhanced damages if the defendant knowingly infringes. If anything, the risk of overcompensation might be higher if courts award ongoing royalties based on a higher royalty rate than they use to calculated damages for past infringement, which (unfortunately) is the current U.S. practice. Third, courts can look for indicia of when holdup risk is substantial, based on whether the infringing product incorporates numerous patents, whether the patented invention would be easy to design around, and so on.
Thus my own view is that in cases in which holdup risk is large, courts should generally deny injunctions based on the above analysis. When holdup risk is not so high, which I think would be the majority of cases, I’d be willing to assume that A and E are large enough to militate in favor of injunctive relief, though ultimately I suppose the magnitude of those variables is an empirical question.