First speaker is Thomas Knight, economics Ph.D. candidate at the University of Florida. Coauthored paper with Roger Blair. In deciding what constitutes a reasonable rate, can economic analysis determine the ex ante bargaining solution? Economics can get us part of the way, but in terms of selecting a particular one, it only gets us so far. What would have happened had the standard not been adopted? Can consider alternatives, etc., but can't necessarily identify a unique outcome as the correct one. Bargaining outcomes are determined by the negotiating process. Nash bargaining is axiomatic but doesn't tell us how the outcome is arrived at. Often comes to 50/50 split. Or there could be an ultimatum game; accept or reject, not counteroffers, winner-take-all. Idea of a single reasonable rate not supported by economic theory.
Second speaker is Keith Hylton, Boston University, "A Unified Framework for Competition Policy and Innovation Policy". Common to view them as policies generated separately. But they can be viewed as two sides of the same coin. Simple model of optimal penalty and its implications. Optimal fine is unambiguously less than the amount that internalizes consumer harm. A more lenient policy than the Chicago School result. Optimal "fine" actually may be a reward or subsidy. Innovation gets greater weight than it typically gets in competition policy. Slide shows the familiar downward sloping demand curve, deadweight loss (D), consumer surplus (W), transfer from consumers to producer (T). But there can also be an efficiency gain if marginal cost decreases, designated "E" on the diagram. Internalizing consumer harm: static penalty = (T+D)/P + C, where P is probability enforcer will go after monopolizing firm and C is the cost of enforcement. But now assume that innovation can create monopoly power. This changes the story. Optimal innovation subsidy = (- W/P) + C. Give the monopolist a subsidy for investing in lowering marginal cost. The dynamic penalty is (1 - theta)(static penalty) + theta(innovation subsidy). What is theta? It lies between 1 and 0 (but can't be 0). Bias toward innovation. Theta "is a function of the ratio of the sensitivity (elasticity) of innovation with respect to penalty to the sensitivity of monopolization with respect to penalty." Hylton believes the model implies more antitrust deference with respect to, e.g., pharmaceuticals: need to protect innovation incentives. [My thought: yes, we need to protect innovation incentives, but it's always a matter of degree. Not all drugs are life saving drugs, access is also important, etc.] Maybe don't inject antitrust law into FRAND disputes; rely on patent and contract law. [I'm largely in agreement with that, for reasons I will discuss in connection with my paper.]
Third speaker: Allan Shampine (executive VP at Compass Lexicon). Holdup is one problem; strategic action is another. During standard setting process, members could try to rig the system, e.g., agree to charge each other a low rate and a higher rate to others ex post. SSOs realize these are potential problems; FRAND intended to address them. "Reasonable" according to most economists means the ex ante approach. Figuring out exactly how to do it can be tricky, though. And setting rates in advance is costly and often would be wasted effort, where standard wasn't successful. But to say it's hard doesn't just mean it's difficult but also that there is huge uncertainty. Ex ante approach is correct theoretically, but also we need something that works. Sometimes ex ante announcements are best. Arbitration can reduce costs of determining. Shampine suggests thinking about the nondiscrimination aspect (citing work with Dennis Carlton). But what is "similarly situated"? A more precise definition: firms are similarly situated if they have the same incremental value relative to whatever we are going to use, but suffers from same problem as ex ante approach. Suggests using a broader version of nondiscrimination. Smallest compliant element: everyone pays same rate. Not a perfect solution, but ensures that people won't be disadvantaged relative to one another. Can help to reduce holdup. Does potentially leave money on the table for the patent holder. But can be very hard to tell is there is holdup or not. Advantage of simple nondiscrimination rule is simplicity. Probably rates should not rise over time, because difficult to tell why. Another issue: FRAND is not a cure-all. Plenty of things that come up in patent negotiations that FRAND is not designed to address, such as royalty stacking. How do you apportion when many of the players are not at the table? Though there is a lot of good thinking about this going on; still, FRAND may not be the best way to do it. FRAND also doesn't help with weak patents. Perhaps you could increase filing fees to deter frivolous assertions, or prohibit software patents, or whatever. Or maybe licensees could sue SSO for antitrust violation--creating monopoly power. To follow up with something that Knight said, the ex ante approach can tell you what the maximum surplus is, but not how to split it up. Still can be helpful to set upper bounds.
Moderator opens it up to questions from the audience.
Bill Page asks Keith Hylton: does you model assume there will be both efficiency gains and losses? For naked monopolization, the static model works fine. Hylton: yes, but for a lot of conduct it isn't naked monopolization.
Haw asks Hylton about the diagram and its relation to the Williamson diagram. Does the product exist already? Hylton: no; if penalty is too large, product won't exist. Haw: in period 0, no market. When does marginal cost go down? Hylton: When monopolization creates efficiency and increases monopolization.
Wickelgren: so there is a period 1 where the firm could enter, produce, and face competition, but it won't if it can't enjoy a monopoly? Hylton : yes. Wickelgren: could we view this as two separate problems? You enter the market and create a consumer surplus of W + T + D. You could get a reward of W + T + D but then there will be competition. After optimal reward, penalty for anticompetitive action after the reward?
I asked whether the reduction in marginal cost was the innovation. Answer: not necessarily.The innovation precedes this, then there is (say) an exclusive dealing arrangement that has both efficiency gains and increases monopolization.
Page: why not just say that efficiency conduct is legal? [Sorry, I didn't quite catch the response and discussion.]
I asked Shampine: Economists often think that price discrimination isn't necessarily a bad thing. In your model, is the principal concern that nondiscrimination provides an additional safeguard against holdup, or is it a safeguard against coalitions of SSO members engaging in favoritism towards themselves, or what? Shampine: both. But strategic action is the principal rationale.
Wickelgren: asks Knight about models with asymmetric information. Knight: Nash bargaining says little about strategic interaction. Wickelgren: what if we take a normative view that we want complete information, etc., and apply that, instead of a positive view? Knight: courts could. And do. Shampine: I don't know why courts are so scared of Nash and game theory. But if we are dealing with hypothetical information, once you figure out what you're bargaining over, you have to turn over to game theory; why should that be a problem? Could suggest normative approach, e.g., this would be efficient if they did it this way.
John Golden, University of Texas: To me, the real problem is how you assign value to these things. Circularity to the whole process; these are legally generated rights, bargaining in shadow of law. I'm not sure that worrying about the bargaining process is the key thing. And juries can't process. Problem is that parties are often orders of magnitude apart. Hylton responds: I'll try to bridge these two. One issue is determination of royalties; the other is whether FRAND = injunction waiver. On royalties, agreement to ex post determination of rates by somebody. I don't see much of an antitrust/innovation conflict there. The conflict is when the FTC or EC thinks a FRAND commitment means no injunction; introducing something from antitrust law that the parties didn't agree to. [I don't agree with that.] Shampine: true enough that if parties are orders of magnitude off, just picking the upper bound is helpful. Figuring out how you split it is a less important issue.
My question: if you always pick, say, the mean of the range of possible outcomes, wouldn't you expect to get things right in the aggregate over time? In other words, how important is it to get the "right" outcome in every case if we are on average right? Knight: basic intuition seems correct. investment decisions made ex post so if on average right, should be good enough. Though can be complications in figuring the "average."
David Taylor: maybe the expectations approach I suggested isn't good for smaller firms. Bigger impact on them if you get it wrong. Knight: yes, could be heterogeneity.
David Killough: Innovation just means something that is different, not necessarily better. Want to reward people who make better things. Patents designed to produce innovations. Market tells you what is better. is that implied here?
Shampine: very good point. There is a spectrum. Targeting rewards so as to maximize innovation a tricky problem. Hylton: I agree. In my model, the company that made an innovation that had no value would get a penalty, not a subsidy. W would be 0.