As I discuss in my book (pages 281-84), under many circumstances European competition law considers no-challenge clauses in patent license agreements to be anticompetitive. At the same time, however, it generally permits the parties to include a clause that terminates a patent license agreement if the licensee challenges patent validity. When such a clause is present, challenges are not without risk, because if the license terminates and the patent is found to be valid, the former licensee is now an infringer.
In the U.S., by contrast, since Lear v. Adkins licensees are not estopped from challenging patent validity (e.g., by terminating the license and raising invalidity as a defense to a claim for patent infringement), and since the Supreme Court's 2007 decision in MedImmune licensees in good standing may challenge patent validity by way of a declaratory judgment action (i.e., without terminating the license). It's not altogether clear, however, whether a no-challenge clause or a clause that results in termination of the license in the event of a challenge would be enforceable (although a 2012 Second Circuit decision held that no-challenge clauses in prelitigation settlement agreements are not enforceable). See my book, pages 154-57 & n.323. See also this recently-published comment by Melissa Brenner.
As I also discuss in the book, it's also not entirely clear (to me) what the optimal rule is. I can imagine at least four possible rules. Rule 1 would be that licensees cannot challenge validity unless the license specifically permits them to do so. Rule 2 would be that licensees may challenge unless the license specifically forbids them from doing so. Rule 3 would permit challenges even if the licensor and licensee would be willing to agree not to permit them, but it would allow the parties to agree to some sort of consequence (such as termination) in the event of a challenge. (This is, basically, the European rule as I understand it.) Rule 4 would be the same as Rule 3, except that it wouldn't allow for any adverse consequences to the licensee in the event of a challenge. From a law-and-economics perspective, the choice between Rules 1 and 2 would seem to rest on Coasean considerations (i.e., are transaction costs likely to be higher or lower under Rule 1 or 2?). Rules 3 and 4 would seem suboptimal, since they prevent the parties from bargaining toward a mutually agreeable outcome, unless (for example) such an outcome threatens substantial negative externalities. And if those externalities are high enough, 4 might be preferable to 3.
In the present context, there clearly is an externality if the licensor and licensee make it difficult or impossible for the licensee to challenge validity, because some invalid patents may wind up going unchallenged (for good, or at least for a period of time). How substantial that externality is, is hard to gauge. Some invalid patents may cause little social harm in terms of lost consumer surplus and unnecessary transaction costs, others a lot. In addition, some systems may provide ample opportunities for other entities to challenge patent validity (e.g., by allowing anyone to commence an invalidation proceeding), others not. On the other hand, the easier we make it for licensees to challenge validity, the more costly licensing potentially becomes because patentees are likely to demand some sort of premium to make up for what now becomes an unshiftable risk. Empirically, though, we don't know how much more costly it becomes or whether it deters any socially beneficial licensing. Perhaps the effects are small.
As we shall see in a post I plan to publish later this coming week, as a policy matter all of this might have some relevance to the current debate in Europe over whether, or under what conditions, owners of standard essential patents should be able to obtain injunctive relief for infringement of their patents.