Professor Carlton begins with a discussion of the proper interpretation of the Coase Theorem: not that property rights always lead to the efficient outcome, but rather that sometimes property rights are not well-defined and that the presence of transaction costs often matters. Topic today: how patent enforcement is not being conducted today in the most efficient way.
Carlton first discusses the traditional justifications for and criticisms of the patent system. Justifications or goals are to create financial incentives for invention and disclosure. Criticisms are that granting patents when ideas are obvious is inefficient; growth of weak patents; surge in litigation and damage awards. Chart shows how number of U.S. patents granted begins to surge in the 1980s, coincident with the establishment of the Federal Circuit. More important is the composition of patents: software related patents have gone up from about 1/4 of all patents granted to about 1/2 since 1991. (Source: GAO Analysis of USPTO data.) Similarly, the number of patent cases filed from 1992-2012 has surged from about 1,500 to about 4,500; in the last three years, it has particularly surged. The number of defendants in software patent cases has also gone way up, to over 6,000 in 2011 (compared with a little more than 2,000 non-software-related defendants).
Carlton then moves on to discuss nonpracticing entities (a.k.a trolls or patent assertion entities), whom he characterizes as collectives (presumably because they collect patents into portfolios?). From an antitrust point of view, if you're creating a portfolio of patents that do much the same thing, you eliminate potential competition that otherwise would arise if the similar-functioning patents were owned by competitors. In addition, some patents are complements. Independent pricing of complements leads to the Cournot complements problem: the aggregate royalty will be higher than is efficient, where independent ownership of the complementary patents exists. Furthermore, absent NPEs, firm 1 and firm 2 might cross-license where each believes the other is infringing its patents. Hence the need for a defensive patent portfolio. NPEs, however, which don't practice, can't be countersued. So in 2010, roughly 30% of patent cases involve NPEs; 65% two years later.
The other collective relevant to the story is standard setting organizations (SSOs). Problem is that manufacturers practice before a patent is declared standard essential and before a royalty is set. Standard may have thousands of patents. Leads to holdup. Thus, standard setting "can create market power where none would otherwise exist." So SSOs require members to reveal their SEPs and to agree to license on FRAND terms. Problem is that terms are ambiguous and can lead to litigation. Diagram shows the smartphone patent litigation thicket; most of the companies involved are practicing entities, but some are NPEs.
"Abuse of market power in an SSO: patent owner harms firms in the SSO or vice versa." The vice versa is that the industry members could gang up on the patent owner and refuse to use the patent absent favorable terms, but Carlton believes this concern [i.e., so-called reverse holdup] gets more attention than it should. "Subset of firms can harm other rivals when unanimity is not required." Alternatively, "some or all existing firms can harm entrants." Charging differential royalties would be a mechanism. Thus, SSOs require FRAND rates to prevent abuse.
Why not decide the rate upfront? One reason is fear on the part of the SSO members that this would violate section 1 of the Sherman Act, although the DOJ has indicated it would apply the rule of reason, not per se liability. Nevertheless, SSOs aren't taking the bait. Slide also indicates that this would be time-consuming. But what does FRAND mean?
Reasonable: royalty should not include holdup value, should be that which would have negotiated ex ante.
Nondiscriminatory: this part hasn't received as much attention. Purpose is to prevent the ganging up by one subset of SSO members against another that Carlton referred to above. But we still need a definition of "similarly situated." Two possibilities: firms with same ex ante incremental value pay the same rate, or all firms using a common component incorporating the patent pay the same rate. The first makes economic sense: if a patent lowers my cost by $1 and your cost by $1, we are similarly situated. But does not require that every firm pay the same royalty. An alternative would be the second option, if the first is too complicated.
Subtleties and real world complications: one of them is dynamics. The value of the patent may change over time. How do you distinguish the value of the patent going up over time, over the practice of holdup? Interpret FRAND to mean you commit to not increasing the rate over time, or allow most favored nations clauses.
Another complication: injunctions. Allowing a court to determine the FRAND rate seems consistent with the FRAND commitment. Or a court could say "I think the rate the patentee is offering is FRAND, and I will issue an injunction until you agree to it." Other than this, injunctions have no place in FRAND proceedings. The ITC's ability to issue injunctions is therefore wholly improper.
"Limitations of FRAND: FRAND stems from a voluntary agreement among members of the SSO. It does not apply to non-members."
Other reforms may help, e.g., attorneys' fees, tightening requirements for software patents, increase filing fees. Or this: "Antitrust liability for SSOs that fail to constrain royalties for SEPs," because they set up an institution that allows for the creation of market power that wouldn't otherwise exist and not putting into place a mechanism for constraining it. If that happens, we'll see more upfront commitments about royalty rates, and firms acquiring portfolios of complementary SEPs to avoid the Cournot complements problem.
Question from audience: any noninterested parties in SSOs? Carlton: I believe that both customers and industry members can participate.
Question from audience: Is there a danger in considering the SEP issue in isolation from the rest of the patent system? Lowering value of FRAND-encumbers technology in comparison to others? Carlton: FRAND doesn't commit you to charge FRAND for purposes other than that which is relevant to the standard. But there is a problem is a nonmember owns an SEP and wants to charge a supra-FRAND rate. Some SSOs require that absent a FRAND commitment they'll use a different technology, if possible.
Question from Danny Sokol: you mentioned two possibilities, competition law or reforming the patent system. But which is optimal? Carlton: I understand that a lot of cases are being brought in Germany now, forum shopping. As to which solution is better, no question in my mind that reforming the patent system is better. But given how hard it is to do this, we're stuck with antitrust issues.
Carlton also notes that patents tend to be concentrated in a few industries; others prefer trade secrets. Refers to Petra Moser's scholarship. Is trade secrets enough of a protection for software? Some say no, it can be reverse-engineered. But what is the value added of this stuff?