Academic Testimony

This document is available in three formats: this web page (for browsing content), PDF (comparable to original document formatting), and WordPerfect.




















199:30 A.M. TO 4:30 P.M.





24Reported and transcribed by:

25Kathleen Carr Meheen, CSR 8748



2Morning Session:


4Deputy General Counsel for Policy Studies

5Federal Trade Commission



8Attorney, Legal Policy Section

9Antitrust Division, U.S. Department of Justice



12Morning Session:

13Aaron Edlin

14Joseph Farrell

15Howard Shelanski













2Afternoon Session:


4Assistant General Counsel for Policy Studies

5Federal Trade Commission




9Antitrust Division, U.S. Department of Justice



12Afternoon Session:

13Timothy Bresnahan

14Richard Gilbert

15Daniel Rubinfeld

16Carl Shapiro











1C O N T E N T S




5   Joseph Farrell

6   Aaron Edlin

7   Howard Shelanski

8Moderated Discussion

9Lunch Recess





14   Timothy Bresnahan

15   Richard Gilbert

16   Daniel Rubinfeld

17   Carl Shapiro

18Moderated Discussion









1P R O C E E D I N G S

2* * * * *

3MR. COHEN: Good morning. I'm Joe Cohen, Deputy

4General Counsel for Policy Studies at the Federal Trade

5Commission and I'm going to be one of the moderators at

6this morning's session. My co-moderator is Joe Matelis,

7an attorney in the Antitrust Division at the U.S.

8Department of Justice.

9Before I start I'd like to cover a couple of

10housekeeping rules. First, as a courtesy to our speakers,

11please turn off your cell phones, Blackberries, anything

12that might ring or clang or make noise.

13Second, because these are set up as in a hearing

14structure, we request that the audience not make any

15comments or ask any questions during the session. We have

16to limit it to the moderators and the panelists.

17Before introducing our speakers and starting our

18panel discussion, I would again like to thank the

19University of California at Berkeley for hosting the

20FTC/DOJ Section 2 hearing sessions yesterday and today.

21In particular I'd like to thank Howard Shelanski, once

22again, Richard Gilbert and Carl Shapiro for offering us

23the facilities and making the necessary arrangements.

24I'd also like to thank the Berkeley Center for

25Law & Technology and the Haas Business School for


1providing the facilities, videotaping, web casting, etc.

2And those who have provided us with logistical support,

3Bob Pardue and others, I thanked you all once already, but

4thank you again.

5We're honored to have assembled this morning a

6distinguished group of the finest lawyers from the

7University of California Berkeley to offer their testimony

8in connection with these hearings. They will provide

9their perspectives on various themes and issues related to

10the complex area of Section 2 jurisprudence, including

11some research and economic analysis.

12We've gathered seven panelists for today's

13sessions. Four will talk this afternoon and three will be

14our morning panelists.

15This morning's panelists are Aaron Edlin, the

16Richard Jennings Professor of Law, University of

17California Berkeley; Joseph Farrell, Professor of

18Economics at -- right here at the University of Berkeley,

19and Howard Shelanski, here, Associate Dean and Professor

20of Law and Director of the Berkeley Center for Law and


22Our format this morning will be pretty simple.

23Each speaker will make an opening presentation from twenty

24to thirty minutes. After the presentations are finished,

25we're going to take a break, probably for about fifteen


1minutes, and then we'll come back, reconvene, and have a

2moderated discussion with our panelists.

3We're scheduled to conclude this morning's

4session at approximately noon. So, we look forward to

5hearing from our panelists.

6And before we begin, the last group that I want

7to thank are the panelists themselves. We appreciate the

8time and effort and your willingness to share your

9insights with us to make this a successful hearing.

10I'd now like to turn to my DOJ colleague, Joe

11Matelis, our co-moderator, for any remarks he'd like to


13MR. MATELIS: Thank you, Bill.

14The Department of Justice's Antitrust Division

15is very pleased to participate in today's single-firm

16conduct hearings. We are delighted that such esteemed

17panelists have agreed to share their views with us today.

18And the Antitrust Division takes particular

19pride in noting that five of today's panelists have served

20in the Antitrust Division as Deputy Assistant Attorneys

21General for Economics.

22We expect that today's panelists will discuss a

23wide range of topics that arise in evaluating single-firm

24conduct and antitrust laws and we look forward to the

25presentations and the panel discussions that follow.


1On behalf of the Antitrust Division, I would

2like to take this opportunity to thank the Berkeley Center

3for Law and Technology and the Competition Policy Center

4at the University of California Berkeley for hosting us


6Also on behalf of the Antitrust Division, I'd

7like to thank Joe, Aaron and Howard for agreeing to

8volunteer your time and share your insights with us. It's

9a great public service that you're doing and we're very


11Finally I'd like to thank Bill and his

12colleagues at the FTC for all their hard work in

13organizing today's panel and assembling the great speakers

14that we have lined up today. Thank you.

15MR. COHEN: Our first speaker is going to be

16Aaron Edlin, who has taught at Berkeley since 1993. He

17now holds the Richard Jennings Chair and professorships in

18both the economic department and the law school. He's

19served on the economic side as Senior Economist at the

20Council of Economic Advisers during the years of the

21Clinton Whitehouse. He is co-author with Professors

22Areeda and Kaplow of one of the leading casebooks on

23antitrust and he has published many articles dealing with

24competition policy and antitrust law.



1MR. EDLIN: Thank you. Let's see how we get to

2the slides.

3MR. COHEN: And yesterday we had the

4representative from Microsoft [laughter].

5MR. EDLIN: Maybe we could switch speakers?

6MR. COHEN: I am going to introduce Joe Farrell,


8Joe is Professor of Economics here at Berkeley.

9He's a Fellow of the Econometric Society, former Editor of

10The Journal of Industrial Economics, and former President

11of the Industrial Organization Society

12Professor Farrell was Chief Economist at the

13Federal Communications Commission in 1996 to 1997 and was

14Deputy Assistant Attorney General for Economics at the

15Antitrust Division of the Department of Justice from 2000

16to 2001. From 2001 to 2004, he served on the Computer

17Science and Telecommunications Board of the National

18Academies of Science.

19 Joe

20MR. FARRELL: Thank you. So, who am I and why

21am I here? We've just heard who I am. Why am I here?

22Because I've drifted into antitrust from economics. I

23think that's true of a lot of the people here. And one of

24the things that's most striking is that the whole

25unilateral conduct field seems to have drifted a long way


1from first principles. And it's unsatisfying to me and I

2worry that it leads to bad policy.

3So, what I'd like to do is to try to bring us

4back to some first principles. Because the field has

5drifted so far from first principles, it's not even

6clearly I think understood exactly what those first

7principles are. And I'm going to put forward a suggestion

8about what they might be.

9The suggestion I'm going to put forward is one

10that distinguishes quite importantly between the final

11goal of antitrust, which I think most of us agree is and

12should be economic efficiency, and the protections and the

13process involved in antitrust enforcement. And it does

14not logically follow that, just because the final goal is

15economic efficiency, each case should be analyzed or each

16transaction should be analyzed along the lines of economic


18Just to give you a simple example, if I go into

19a store and take an iPod off the shelf and put it in my

20pocket and walk out, that's typically illegal if I didn't

21do more than that. And it's illegal even if I can show by

22thoroughly convincing evidence that my economic value for

23the iPod exceeds the store's replacement cost. In other

24words, it was an efficient transaction for me to steal the

25iPod. Well, that doesn't cut any ice in law enforcement


1as I understand it and probably shouldn't. And the

2economic market system that we have operates by enforcing

3the property rights of the iPod. And that enforcement

4does not look directly at whether the enforcement is in

5the instant efficient or not. And I'm going to claim that

6antitrust often does something rather similar, okay?

7So, before I get to the first substantive slide

8with the provocative title "Analyze This," let me say

9that, as I understand it, the fundamental of antitrust is

10that you are not supposed to restrain trade. That doesn't

11mean you are not supposed to restrain your own trade.

12People often comment that it's all right to restrain your

13own trade. What you're not meant to do is to restrain

14other people's trade.

15And you might ask, well, how can you possibly

16restrain other people's trade unless you actually tie them

17up or something. Well, it turns out that there are

18techniques by which a firm might be able to restrain

19others' trade. And those techniques it seems to me are

20the core problems.

21So, that's all setup. Let's come to my purely

22hypothetical example, "Analyze This." So, let's think

23about the airline market. An airline that I've called

24Northeast Airlines offers a five hundred dollar fare. And

25it's the only airline that's in that market, so consumers


1buy it. No better deal is available.

2An entrant that I've called Sprite would happily

3sell at three hundred dollars a similar product.

4Consumers would prefer that deal. So, why doesn't it

5happen? Well, it doesn't happen in this instance because

6everybody recognizes that if Sprite enters and offers the

7three hundred dollar deal, Northeast will cut its price to

8two hundred dollars. And Sprite is unable to make a

9profit competing against the two hundred dollar fare.

10So, Sprite anticipates that, doesn't enter, and

11consumers continue to pay five hundred dollars.

12So, before we get into, well, what law might it

13violate and what policies are there and so on, I'd like to

14observe that something is clearly wrong there. And let's

15delve a little bit in a first principle kind of way into

16what it is that's wrong there.

17What's wrong I would argue -- and this is based

18on discussions that Aaron Edlin and I have been having

19over a pretty protracted period of time. What's wrong is

20that Sprite's willingness to sell at three hundred

21dollars, which consumers would prefer to the status quo, ought

22to block Northeast's ability to charge those consumers five

23hundred dollars. In other words, Northeast ought not to

24be able to extract five hundred dollars from consumers,

25given Sprite is willing to sell them the product for three


1hundred dollars. Okay.

2And you might think that normally in a

3competitive process, whatever that means, not only ought

4it to block it but it would. And here it doesn't. And

5what are the mechanics of how it doesn't.

6Well, the mechanics we just went through.

7Northeast, intentionally or not, thwarts Sprite's and

8consumers' joint wish, given Northeast's five hundred

9dollar price, to trade at three hundred dollars. And the

10way that that works is that if Sprite came in it would not

11have to compete against five hundred but against two

12hundred, and it can't compete against two hundred.

13I am saying nothing yet about what's illegal.

14I'm just saying this is an instance of something going

15wrong in the competitive process.

16So, stepping back, and here are some first

17principles, okay. Economists study by and large two approaches

18to economic efficiency. And there's a little bit of a

19disconnect, I think, between the formal material that you

20spend a lot of time banging into the undergraduates' heads

21in the microeconomics classes and the way that

22professional economists typically think about real world


24What we spend the most time with undergraduates

25on is that you can get to an economically efficient


1outcome via price-taking perfectly competitive

2equilibrium. Okay. However, it's sort of obvious that

3the price-taking equilibrium, whether it would be

4efficient or not, is unrealistic and unobtainable in many

5sectors of the economy that are of antitrust concern. If

6nothing else, large economies of scale make that a


8And it's also interesting to note that antitrust

9doesn't just move cautiously, but I would say proudly

10eschews many opportunities to move toward price-taking

11equilibrium. So, in particular, if you have a legitimate

12monopoly, quote, unquote, there is no attempt to try to

13force you to do anything that's closer to price-taking

14behavior. And not only is that potentially difficult and

15problematic to do, but antitrust seems to take the

16attitude, it's difficult, but we wouldn't try even if we

17thought we could do it. Now maybe that's a little

18controversial, but that's my impression.

19The second approach to economic efficiency, which is

20less juicy material for teaching undergraduates because it

21has less of the mid-level mathematics that seems to appeal

22to those who teach undergraduate micro classes, but is

23actually probably more important, is based a little bit on

24the Coase theorem, that's kind of the extreme expression of

25it, or in formal economic terms is often called the core


1of the economy. And that's the idea that if there is some

2inefficiency, then there's some group of people, possibly

3unmanageably large but possibly not, that would have an

4incentive to contract around it. Okay. And therefore we

5think about just how difficult would that be, and if it

6wouldn't be all that difficult, then we predict that the

7inefficiency will either go away or won't be all that big.

8So, for example, it's not exactly an

9inefficiency but it's a problem for the consumers that

10Northeast is charging such a high fare, and there are

11inefficiencies that go along with that.

12So, Sprite and consumers jointly would like to

13contract around that high fare. And the question is: Why

14doesn't that happen?

15So, just to give you a little bit of jargon so

16as to make you feel that there's real substance to this

17talk, what economists call the core of an economy is a set

18of possible outcomes such that no group of consumers and

19firms could find an alternative that's better for all of

20them. Okay. And the core contains only outcomes that are

21economically efficient, of course, because if you have an

22outcome that's inefficient, then the grand coalition, as

23we call it, that is, the set of all consumers and firms,

24could all do better by doing something else.

25Of course that's not a very realistic process to


1imagine everybody getting together. But, conditional on

2knowing that something inefficient is not in the core, we

3have a reasonable shot at finding a smaller and more

4manageable blocking coalition.

5What's a blocking coalition? A blocking

6coalition is a group of consumers and firms that can all

7do better than the status quo given their endowments and

8abilities to trade and so on.

9So, in parallel, if you like, with the

10competitive equilibrium analysis, we have core analysis.

11And it suggests a rather different process. Instead of

12suggesting a process where we kind of hammer on the

13economy until most firms are somewhere close to

14price-taking, okay, and which, as I mentioned, is not

15actually feasible in many important sectors of the

16economy, it suggests a process where we protect the

17ability of these blocking coalitions to work around any


19 So, a perspective on antitrust is this: That

20antitrust protects the process of forming blocking

21coalitions that block bad outcomes. And how does it

22protect that? Antitrust is -- it says certain things are

23illegal. What sorts of things are illegal? Well, at some

24level, things that thwart the formation of blocking

25coalitions that would otherwise prevent bad outcomes.


1VThat's three negatives, which is a very large number of

2negatives, okay, but that's the way it is, okay.

3So, the last bullet, just to remind you, not all

4contracts of course are protected by antitrust. Some of

5them are illegal, so there's a little bit of a thorny

6issue there, but I'll just note that in passing.

7So, back to the Northeast and Sprite example,

8Northeast is getting five hundred. Sprite and consumers

9would all be better off trading at three hundred. So,

10that's a blocking coalition that tells us that the five

11hundred dollar fare is not something that would survive in

12the core. And, in particular, there's this particular

13blocking coalition. And Northeast, and, again, I am not

14saying whether they do it on purpose or it's a natural

15outcome of the way the market works, but thwarts the

16blocking coalition by making clear that if the blocking

17coalition tries to form, Northeast will block that in turn

18with the two hundred dollar fare.

19So, how do we assess Northeast's price cut from

20five hundred to two hundred dollars? It seems to me

21there's a very difficult and fundamental tension here. In

22the instant, that is, if Sprite has actually entered and

23is charging three hundred, Northeast then does cut its

24price to two hundred, and the two hundred kind of is then

25the outcome that we're looking at, well, that seems like


1part of the competitive process as I've described it. We

2had this three hundred dollar outcome. Northeast is

3forming a blocking coalition with consumers to block it

4with a two hundred fare.

5However, in its ex ante impact, the prospect of

6this two hundred dollars thwarts the formation of Sprite's

7blocking coalition against Northeast's five hundred

8dollars. So, depending on which way you look at this, it

9genuinely is at some level somewhat part of the

10competitive process and somewhat a fundamental undermining

11thwarting blocking of the competitive process. Okay.

12Well, that's a pretty fundamental tension. How

13are we going to deal with it? I don't know exactly. I

14don't even know approximately. But one thing that's

15pretty clear I think out of this discussion, knowing what

16Northeast's costs are doesn't tell you anything very

17relevant. Knowing whether Northeast made in any sense a

18sacrifice with this price cut in some actual or but-for

19sense isn't really relevant or doesn't seem to be

20relevant. Okay.

21So, there's a difficult question here. And the

22specific rules and policies that have come to dominate the

23law on this kind of behavior don't look as if they're

24going to be of any help because, of course, until we

25actually work our way through and figure out what the


1right answer is, you don't quite know what will be of


3So, what does this suggest about predatory

4pricing. It suggests most fundamentally that predatory is

5an adjective that doesn't apply to the level of price. It

6applies to a pattern of pricing. And, in particular, it

7applies to a pattern of pricing such that the price that

8the entrant expects to have to compete against is very

9different from the price that consumers actually end up


11So, is Northeast's price cut primarily a

12blocking coalition to Sprite's three hundred that's the

13essence of the competitive process, or an

14out-of-equilibrium threat to thwart consumers and Sprite

15from blocking the five hundred. That I think might be the

16essence of an antitrust offense. Okay.

17So, one way to answer this that is sensible

18seeming but a little bit ad hoc, departing a bit perhaps

19from first principles, but perhaps not, is to say, well,

20you sort of want to look at how stable that two hundred

21dollars is. If that's really what you've arrived at and

22now you are there and you're going to sort of stay there,

23then that's sort of how the process is meant to work. We

24had originally five hundred, then three hundred, now we've

25got two hundred, and we've got there and that's good.


1Certainly good for consumers.

2If, on the other hand, what happens is mostly

3that consumers really end up paying five hundred and they

4only pay three hundred or two hundred in the rare and

5short-lived cases where Sprite makes a mistake and enters,

6then that seems like a failure of the process. And,

7again, it doesn't seem to me that there's much prospect

8that sacrifice tests or cost tests are going to be very

9helpful here. So, we don't know until we sort of figure

10it out.

11So, this suggests to Aaron and me a principle we

12call freedom to trade. It's a nice phrase, but we mean

13it. The incumbent is restraining trade when given its

14pricing, etc., etc., etc., and there's a blocking

15coalition, a potential blocking coalition, that would make

16all its, that is, the blocking coalition's, participants

17better off, but the incumbent strategically thwarts the

18formation of that blocking coalition.

19So, we saw one possible way in which the

20incumbent might thwart the formation of a blocking

21coalition, threatening that if that coalition starts to

22form, then the price it charges will change.

23Another way you might do that is through some

24kind of divide-and-conquer strategy that says, offer

25particularly favorable deals to some pivotal members of


1this blocking coalition while expropriating others. I

2don't want to get into the game theory of how it can work

3and how it can fail. The fact is it can sometimes work,

4but the point I really want to stress here is, when it can

5work, it seems like that is really disrupting the

6competitive process.

7Now, notice that none of this, according to my

8suggestion of what the competitive process is, none of

9this asks, well, just how unpleasant is it for Northeast

10if Sprite comes in and takes away its customers. And that

11would be an important aspect of a direct inquiry into

12economic efficiency. Right? Because if Northeast

13actually has very low costs, and if demand is fairly

14inelastic, then having Northeast charging five hundred

15dollars might be more efficient than having Sprite come in

16and serving customers.

17And I claim that Northeast thwarting this entry

18would be a thwarting of the competitive process without

19asking about that. Okay? So, as I said in the beginning,

20it seems to me that if we're looking at the formation of

21blocking coalitions as the process whereby we move towards

22the core and that's what's economic efficiency, when we

23talk about the formation of blocking coalitions, we don't

24insist in the interim that they actually have to increase

25efficiency. Instead, we know that if you allow the


1formation of blocking coalitions without that inquiry,

2that process, when it settles down, will get you to

3something that's in the core and therefore really is

4economically efficient

5So, it seems to me that that captures a lot of

6the spirit of the competitive process, that we're

7protecting the process of forming blocking coalitions. We

8believe that in the long run that will lead to economic

9efficiency and it is not necessary and may actually be

10counterproductive to ask about economic efficiency at each


12That does not mean that I'm advocating a

13consumer surplus criterion. Instead, I'm assuming that the final

14criterion is actually economic efficiency. At each step,

15we do actually look at what consumers want because it's

16presumed, I guess, that if an entrant is willing to offer

17consumers a better deal, then the entrant likes the

18formation of this blocking coalition. So, the question

19becomes: Do consumers also like it. But the fact that

20there's a sense in which we're looking at consumer

21preferences at each step, does not at all imply that the

22final goal is consumer surplus.

23So, that freedom to trade principle is, we

24think, an intriguing and promising way to understand

25antitrust starting -- or a lot of antitrust, anyway,


1starting from first principles. How far does it get you?

2It gets you to understand, or at least understand the

3difficulties in some cases, like the hypothetical I was

4talking about and some others. But there's a huge range

5of unilateral conduct that gets challenged in antitrust

6that it really doesn't directly help you to understand.

7And let me sketch this out

8And in order to help this, what we're going to

9do is to introduce a different phrase, also a good phrase,

10"level playing field." So, the observations is that

11freedom to trade is potentially at risk where the entrant

12has to compete against the low price, but consumers

13actually pay a high price. That is the case in my

14Northeast/Sprite hypothetical. And I am going to say that

15the playing field is level if those prices are equal.

16That helps us understand, perhaps, predation, divide and

17conquer, exclusive dealing and so on.

18But, in the case of many challenged practices,

19if the incumbent were simply to go away, consumers would

20not be better off. So, a frequent allegation involves the

21incumbent being asked to stick around but just do

22something different.

23So, you can put a lot of unilateral conduct

24complaints into the following framework. The incumbent is

25offering two trades to consumers, not as alternatives


1typically. I'm going say a price of one hundred dollars

2for Product A and a price of five dollars for Product B.

3And the discrepancy there is meant to reduce confusion

4about which is which. Okay?

5And as a potential blocking coalition, sort of,

6when entrant and consumers enter B at a price of three

7dollars. In other words, there's somebody out there who

8would love to supply B for three dollars, but the entrant

9simply can't do A, so the incumbent is a monopolist in A.

10And the incumbent says, using one technique or another, if

11you want to buy my A, you have to buy my B, or more

12generally links A to B. Okay.

13So, the incumbent might refuse to trade in A if

14the customer deals with the entrant in B, or it might

15raise the price of A from a hundred to, let's say, a

16hundred and ten, which would swamp, of course, any gains

17from buying B at three instead of five. And given that

18we're assuming that there's a monopoly in A, by the way,

19that may well not involve a big profit penalty for the


21Now, if you look in B, it should look like

22freedom to trade is violated and certainly the playing

23field is not level. But in A and B together, there isn't

24a potential blocking coalition. Nobody but the incumbent,

25I assume, can do A, and consumers don't want to just get


1the cheaper B and not get A. So, if you take the freedom

2to trade criterion strictly, there is no potential

3blocking coalition, so there can't be a risk that the

4incumbent is thwarting a potential blocking coalition.

5 And really what this comes down to is: What's

6the right unit of analysis. Should we be looking at A and

7B together? Should we look at B separately? What should

8we do?

9By the way, I tried to avoid using the term

10"market" in talking about A and B because there's no

11particular reason to think that A and B will be defined in

12the usual way of antitrust markets.

13So, just to illustrate this, in case it's

14getting a little too abstract, a few of the traditional

15boxes, so if A is the tying good, B is the tied good, and

16the incumbent is somehow linking trade of the tied good to

17trade of the tying good.

18Exclusive dealing, A is a bunch of widgets that

19the consumer wants to buy, and B is other widgets, maybe

20it's a different date or maybe just more of them today or

21maybe a different place or something.

22If you look at aftermarkets, A might be the

23original equipment and B might be service to the


25So, in all of these cases, it's not uncommon for


1there to be someone who wants to make a better offer in B

2and is stymied by some sort of linkage with A.

3So, what have I learned from all this? The

4setup and the going back to first principles has, at least

5for me, clarified the goal and the technique of antitrust.

6I've come to think that, although price-taking equilibrium

7does conduce to economic efficiency and is typically a

8good thing, and is certainly not inconsistent with

9analysis of the kind that gets us towards the core,

10nevertheless the latter is more fundamental to the ideas

11of antitrust than is price-taking equilibrium.

12I also think that it's important to understand,

13and I have made some steps in my own mind at least to

14understanding, that protecting competition as a process is

15potentially, and I think actually very different from

16imposing on each step of the process a requirement that

17has to increase, let's say, economic efficiency, if you

18think that that's the final goal.

19Trying to go much beyond that, based closely on

20first principles as I've been trying to do, turns out to

21be quite thorny. Okay. And I think there's a lesson in

22there, which is it reinforces what you might already have

23known or believed, which is a lot of the rules of thumb,

24rules of law and policies that govern unilateral conduct

25in antitrust has emerged from the kind of slightly vague


1process that hasn't really linked them very tightly to

2first principles.

3So, to me, it reinforces that these are thorny

4issues. The positive message is, at least for me, it

5brings the thorns into sharper focus. And the particular

6thorn that I think is pervasive here and is brought into

7sharper focus is when, how, in what circumstances, in what

8ways can one in some sense require the incumbent to hold

9fixed its offer in A, and then we analyze level playing

10field or freedom to trade in B.

11Is that always illegitimate? That would be a

12strict interpretation of freedom to trade as the only

13criterion. Is it always legitimate? That would be the

14opposite, I guess. Or is there something in between?

15Ideally, based firmly on these same first

16principles. So, it's not a question of saying, well,

17let's consider a hypothetical and figure out what we

18intuitively think. But I'd like to work towards getting

19there in a way that's closely linked to these first


21Thank you.


23MR. COHEN: Where are we, Aaron?

24MR. EDLIN: I will after the break, or any time

25I think, be able to project the slides.


1MR. COHEN: Okay, should we then go on to


3MR. EDLIN: No. I am ready to present,

4MR. COHEN: Fine. We're now going to turn to

5Aaron Edlin.

6MR. EDLIN: Look at that, okay. Great progress.

7Let's do the show.

8So, the title is, "Sacrifice, Extreme Sacrifice,

9and No Economic Sense," three criteria that have been

10bandied about a lot recently and increasingly over the

11past two decades.

12After the colon, the title is: "The case

13against these necessary and sufficient tests for


15So, of course the big question, the $64,000

16question in Section 2 is: When is exclusion

17anticompetitive and when is it not? The easy case that we

18all understand, presumably, as to how to answer is, if a

19monopoly excludes competitors by consistently charging low

20prices, well that is anticompetitive. It's the essence of

21the competitive process. It's good for consumers.

22What that example goes to prove, however, is

23that we need something other than exclusion to be

24anticompetitive. So, the question is: What plus

25exclusion is anticompetitive. The "what" is clearly not


1consistently low prices. The question, though, is what

2the "what" is.

3And three possible whats have been, as I said,

4bandied about a lot of late. They all are basic

5sacrifice tests. The basic sacrifice suggested in "Aspen"

6and "Trinko" is foregoing profits now or in one line of

7business to make more later or in another line of business

8as a result of lessened competition.

9There is of course another variant, which is

10extreme sacrifice, which comes more directly out of

11predatory pricing, and you see it applied in "Barry

12Wright" and "American Airlines," which is that the test is

13really about actually losing money, not just not making as

14much as you could, pricing below cost and losing money to

15make more later or in another line of business as a result

16of lessened competition.

17More recently, Greg Werden and Doug Melamed put

18forward, and a DOJ "Trinko" brief puts forward a no

19economic sense test, which is that the action makes no

20economic sense but for a lessening of competition.

21These sacrifice tests are on the move, or have

22been on the move. In one sense from pricing cases to

23non-pricing cases. My reading is that they began and were

24first advocated in the predatory pricing context. Thanks

25to "Areeda and Turner" and "Willig." And they later


1spread to non-pricing contexts. Thanks, for example, to

2"Aspen," "Trinko" and "Covad."

3They've also been on the move from sufficiency

4once other elements are shown, which is to say, from

5something that's helpful in making a case to something

6that's necessary for the plaintiff to make a case. So, in

7"Barry Wright," we see that there's been no violation,

8where above cost, where the pricing is above cost, which

9says that extreme sacrifice is necessary in pricing cases.

10 The DOJ "Trinko" brief advocates the no economic

11sense test as necessary. "Covad" assumes that sacrifice

12is necessary. Doug Ginsberg writes, "'Covad' will have to

13prove Bell Atlantic's refusal to deal caused Bell

14Atlantic's short-term economic losses."

15Scalia's "Trinko" interpretation of "Aspen,"

16which I think is a bit revisionist, is that Ski Company

17sacrifice is necessary to violation. And Werden and

18Melamed have quite explicitly argued that no economic

19sense is the unifying principle of Section 2 violations.

20My fundamental contention which I've been

21arguing for years is that sacrifice is not needed for

22anticompetitive effect and frequently not needed.

23My "Yale Law Journal" article argues that this

24is true for what I call above cost predatory pricing. And

25if you think that below cost is part of the definition of


1predatory pricing, then what I mean is above cost pricing

2that is exclusionary and anticompetitive. There I explain

3how consumers can be hurt by threats to lower prices, much

4as Joe Farrell explained, even though prices will remain

5above cost, and perhaps even though prices may be profit


7I ask rhetorically: If sacrifice is wrong

8headed in the predatory pricing context, why are we

9extending it to non-pricing cases? Consider "Aspen."

10Now, suppose, as I think is likely, that Ski Company's

11refusal to sell at retail prices to Highlands increased

12Ski's retail sales to skiers. What I'm thinking there is

13that it certainly is conceivable, perhaps even likely,

14that when Ski Company refused to sell at retail to

15Highlands, what that meant was that, sure, they sold a

16couple less tickets as part of Highlands' adventure packs.

17However, on the other hand, what likely happened was that

18the consumer decided, or many of them did, that they would

19buy a whole week of skiing at Ski Company. So, there may

20have been no sacrifice there of profits, even though they

21refused to sell at retail.

22But would that mean that the refusal was any

23less exclusionary or anticompetitive? I think not. The

24"Aspen" court didn't just rest on what I think is a shaky

25notion of Ski Company's sacrifice, but they also


1emphasized what they took to be consumer harm, the

2revisionist claims of Trinko about "Aspen"


4Another case or set of cases where I think it's

5fairly clear that sacrifice is not necessary for

6anticompetitive effect are submarine patents. If you seek

7a patented process into an industry standard, that may not

8involve sacrifice of any kind that I can see. But that

9fact doesn't make it a good thing to do.

10Many people have been talking about an extreme

11case where Firm A blows up a competitor's plant. Now,

12Werden and Melamed, and fellow travelers with them,

13emphasize that this isn't a problem for them because the

14cost of the dynamite triggers liability. There is a

15sacrifice; you had to pay for the dynamite. And that is

16what triggers liability and means that there's no economic

17sense to blowing up your competitor's plant but for the

18lessening of competition, which justifies the cost of

19paying for dynamite.

20Like Joe Farrell, I don't -- this reasoning

21doesn't grab me and I feel a great suspicion that the cost

22of the dynamite could really be important here. But one

23way of saying that is to change the hypo. What if Firm A

24is avoiding a dump fee by deposing of surplus dynamite in

25this way. If they didn't blow up the competitor's plant,


1they would have had to pay a dump fee to dispose of the


3Well, now I gather that the dynamite has a

4negative cost. So, according to the no economic sense

5test or the sacrifice test, there should be no liability.

6Well, this just can't be. It can't be that it should

7hinge on that. This suggests to me that the sacrifice

8test is not looking at the right thing.

9If the sacrifice test is not looking at the

10right thing, neither is extreme sacrifice. Extreme

11sacrifice, that is losses, are certainly not needed for

12anticompetitive effect. Consider the American Airlines

13case brought by the DOJ unsuccessfully. The judge thought

14there that the extra plane was profitable if you ignore

15effects on other planes. I suggest that everyone reread

16footnote 13 of that case over and over and over again if

17you think that the extreme sacrifice test might make

18sense, as the judge did.

19Marginal revenue, as every economist and econ 1

20student knows, is less than price. For firms with lots of

21market power, which you might think are one of the focuses

22of Section 2, marginal revenue is much lower than price.

23What that means is that monopolies with lots of market

24power can sacrifice enormously without triggering the

25extreme sacrifice test. I think, as I pointed out


1previously, it is very ironic to give such firms a

2license, such a license, such a grand license to exclude.

3Let's go back and consider the case of blowing

4up your competitor's factory. Could it be a violation

5only if the dynamite is so expensive that its cost exceeds

6Firm A's operating profits? It seems outlandish to me on

7its face, but the extreme sacrifice test says yes.

8And I'll point out that in that case, firms with

9large profits have a substantial and much larger license

10to blow up their competitors than other firms.

11Rhetorically I'll ask why.

12Consider the no economic sense test. Does that

13make sense? Well, apply it to limit pricing. Consider a

14firm that could charge a high price and make lots of

15money, for a while anyway, but this firm chooses a low

16price, less profitable for now. Why? In order to delay

17or prevent entry.

18Suppose there is no economic sense in charging

19this low price before there is entry, except that it

20prevents others from entering. Well, the no economic

21sense test condemns that limit pricing. But note that

22that's the essence of competition. It's what I had as the

23easy case on slide two.

24Werden doesn't apply the test here. Instead he

25grants a safe harbor for charging the low price.


1Now, if your test would condemn this case and so

2you have to make an exception and grant a safe harbor

3because it's so obvious that this is procompetitive, I'd

4suggest that the test is not getting at the fundamentals.

5This smells bad to me.

6Back to blowing up the competitor's factory, a

7la "Conwood" discussion, Werden, page 425. Proponents of

8the no economic sense test emphasize again that the cost

9of the dynamite makes it illegal. As I pointed out, costs

10might be negative in the dump fee hypothetical.

11My claim would be that blowing up your

12competitor's factory is anticompetitive regardless of the

13cost of the dynamite, regardless of whether it has a

14negative cost, a small positive cost, or costs more than

15the operating profits, regardless of whether you pass the

16no economic sense test.

17The fundamental problem in my view with all

18these sacrifice tests is that these tests don't flow from

19any kind of first principles that are attractive. They

20don't flow from consumer welfare or from efficiency. They

21also don't flow from a notion of how the competitive

22process would work, for example, a process by which rivals

23can offer consumers - by which rivals who can offer

24consumers higher utility actually get to provide that

25higher utility.


1The tests don't flow from any other principles

2I've been able to discern from reading about them.

3Now, when someone like me points out that there

4are many cases where the tests are not satisfied but the

5action is anticompetitive, what you quickly bump into,

6both in the commentary and in the cases, is a refrain

7about false positives. It's a chorus. Fears and claims

8about these false positives abound. However, I'd suggest

9a modern example that I can put forward are pretty scarce.

10A common argument is that you need a hurdle to

11avoid these false positives. So, sacrifice is not needed

12for anticompetitive effect, but the plaintiffs should be

13required to show it anyway, in order to prevent an avalanche

14of cases from chilling legitimate competition.

15To me, when I hear that, I wonder, why not just

16tax plaintiffs, if that's the goal. Or, if you really

17want to eliminate these false positives, you could

18eliminate Section 2 entirely, or you could eliminate

19Section 2 for any plaintiffs whose name begins with A

20through M, then you get rid of half the false positives.

21Erecting arbitrary hurdles because the right

22test is difficult to administer properly is, I would

23argue, wrong-headed. What commentators should do, and

24ultimately courts, is seek, as best they can, the right



1Now, once you've sought the right test, if

2administrative difficulties truly make false positives a

3bigger problem than false negatives, and there is not all

4that much discussion by the refrainers about false

5negatives, there is an answer which doesn't involve

6arbitrary hurdles or abandoning the right test. You could

7raise the standard of proof in that case. You could

8improve jury instructions. You could create procedural

9hurdles like "Dauber" to require rigorous evidence. We

10have a number of those. And, again, I think you'll find

11that modern examples of clear false positives are pretty


13What are my conclusions? That patience is

14needed. We should be searching for the right standard, or

15at least better ones, and that administrative difficulties

16don't justify arbitrary tests. And too often they have

17been used to do so.

18Thank you.


20MR. COHEN: Okay. Our last presenter this

21morning is Howard Shelanski, Professor of Law at Berkeley

22here, where he is also Associate Dean and the co-director

23of the Berkeley Center for Law and Technology. His

24research focuses on antitrust policy and regulation.

25On the economic side, from 1999 to 2000,


1Professor Shelanski served as Chief Economist of the

2Federal Communications Commission, and in 1998 to 1999, he

3was a Senior Economist to the President's Council of

4Economic Advisers at the White House.

5On the law side, Professor Shelanski served as a

6clerk to U.S. Supreme Court Justice Antonin Scalia.

7We welcome your presentation.

8MR. SHELANSKI: Thanks very much, Bill. I'm

9really happy to be here. And I want to make a

10presentation that at least in some aspects will connect to

11what my colleague Aaron Edlin was just talking about, in

12the sense that it may give some insights into how to

13choose among different kinds of tests for enforcement

14under Section 2.

15And I want to speak specifically about

16enforcement in the area of unilateral refusals to deal, an

17area that has, I think, become particularly challenging in

18the wake of the "Trinko" case.

19And the broad point that I want to make is this:

20That at the same time that the Department of Justice and

21the Federal Trade Commission are reviewing enforcement

22policy for Section 2 of the Sherman Act, there are

23parallel efforts ongoing, indeed some undertaken in recent

24years by the Federal Trade Commission, to rethink and

25reform intellectual property rights, and particularly to


1reform it in a way that makes it harder for firms to use

2intellectual property to foreclose competition with weak

3or questionable IP rights.

4And I think that the potential outcomes of IP

5reform could matter for aspects of antitrust reforms, and

6notably for policy toward unilateral refusals to deal.

7So, my main point is that, in thinking about

8Section 2 enforcement, and in particular thinking about

9unilateral refusals to deal, antitrust reform efforts

10should not ignore intellectual property reform processes

11So, I have a general suggestion, which is that

12antitrust authorities should keep an eye on IP reform and

13take into account how it might affect enforcement policies

14under Section 2. Not a terribly original idea in broad.

15Louis Kaplow in 1984 wrote a very nice paper talking about

16how antitrust and IP should be thought of as part of an

17interactive system. But I also want to talk about

18specific conjecture and, as we get further along, you'll

19see why I refer to it as merely conjecture, which is, if

20IP reform is likely to reduce the strength or availability

21of intellectual property protections, antitrust

22authorities might consider enforcing less strictly against

23refusals to deal.

24Now, let me try to explain why. Under "Trinko,"

25there is a presumption against requiring a firm to deal


1with competitors. Now, there are many things one can read

2into "Trinko". "Trinko" adopts a very strong line against

3duties to deal for firms in the unilateral context. But

4"Trinko" did preserve "Aspen". Very interestingly,

5"Aspen", which is a hard case to teach to students and in

6many ways a hard case to explain. "Aspen" is a case that

7imposed a duty to deal.

8I agree with Aaron Edlin that Justice Scalia

9engaged in some revisionism by finding profit sacrifice in

10that case, but inherently what "Aspen" says is, if there

11is nothing that you gain by refusing to deal, then we are

12going to assume that what you gained is a reduction in

13competition that inures to your benefit. That's one way

14of looking at it. But "Aspen" still exists after

15"Trinko". We have a strong presumption articulated in the

16"Trinko" decision against imposing duties to deal.

17The question that's left for the antitrust

18agencies is the following: Okay, where do we impose the

19duty to deal or not. So, I want to talk a little bit

20about some policy issues that might arise, some background

21issues, and then talk about how IP reform might affect the

22answer to that question of what standard to use in

23imposing a duty to deal.

24Well, the first thing that we need to keep in

25mind of course is that only some refusals to deal cause


1anticompetitive harm. There are many cases where refusals

2to deal will cause competitive supply to enter the market,

3would cause a firm to invent around the refusal to deal or

4to innovate or produce something itself.

5Mandatory dealing in cases where there isn't

6anticompetitive harm could impede investment and

7innovation by the firms being forced to deal. So, that's

8an argument one often hears. If you go back to some of

9the previous rounds of these hearings, Former Assistant

10Attorney General for Antitrust Eupate has some testimony

11saying exactly this, if you force firms to deal, they're

12not going to innovate. There's some interesting counter

13argument by Professor Steven Fallon that suggests the

14evidence for such innovation deterrence is thin. But we

15have to at least keep in mind the possibility that

16mandatory dealing could impede investment.

17I think that one of the bigger concerns is that

18enforcement of a duty to deal might reduce competitive

19innovation and production not by the firms being forced to

20deal, but by other firms in the marketplace or by the

21would-be buyer, by creating a quasi-regulated purchase


23So, "Trinko" takes into account all of these

24possibilities, that there isn't a lot of -- that there are

25many refusals to deal that are not anticompetitive and


1imposing a duty to deal in fact may have consequences to

2justify its presumption against the duty to deal. But

3"Trinko" does not necessarily mean refusals to deal are

4evil per se.

5So, refusals to deal can have anticompetitive

6harms. And we would not necessarily want to exempt those

7refusals to deal from enforcement.

8Now, I want to suggest that one necessary

9condition for such harm is that competitors and third

10parties face economic barriers to providing the goods at

11issue or that competitors and third parties face legal

12barriers to providing the goods at issue.

13And I would suggest we should not impose duties

14to deal in goods for which economic or legal barriers to

15competitive supply do not exist. There you get very

16little pay off and you may creat some deterrent effects to

17innovation either on the supply or the demand side.

18But what about refusals that could be

19anticompetitive, for which there are economic barriers or

20legal barriers. There are several standards that we could

21use to identify those situations and to decide whether or

22not to enforce a duty to deal.

23So, one thing we could do is to say, listen, we

24should have per se legality for refusals to deal. This is

25in the spirit of "Trinko", it's a strong reading of


1"Trinko", but it's a very clean line and we avoid any risk

2of deterring innovation on either the supply or the demand


4Alternatively, we have a range of rule of reason

5approaches. And I'm just going to very simplistically

6phrase them as potential consumer welfare tests, the kind

7of tests that Professor Salop proposed in an earlier round

8of these hearings; a business justification test, which

9Kolasky suggested in that same round; and a profit

10sacrifice test of various stringency, ranging right up to

11a no business sense kind of test of the kind that Doug

12Melamed has articulated.

13Then we have the old line essential facilities

14approach, which as Justice Scalia tells us, the Supreme

15Court has never adopted. One could quibble about what

16"Onertel" means, but there is some precedent certainly in

17the Appellate Court for the essential facilities approach,

18notably in the Seventh Circuit.

19So, how should the Justice Department and the

20Federal Trade Commission choose among these various

21approaches? Well, I don't much like the per se legality

22approach because per se legality fails to block cases

23where the only effect is anticompetitive. And while often

24justified on the grounds of preserving the refusing firm's

25innovation and investment incentives, there isn't clear


1evidence that that is [unintelligible]. And I think

2you're likely to have poor welfare effects here.

3I don't much like the essential facilities

4approach either because it does ignore some legitimate

5business justifications. And I think that it may too

6easily allow access and deter innovation and investment by

7the buyer or the third parties. And more -- of great

8concern is it requires a quasi-regulatory solution.

9While I fully agree with my colleague Aaron that

10we should not let administrative difficulties justify a

11bad test, we shouldn't ignore administrative difficulties

12in the test that we actually choose to administer. And

13there's some hard pricing questions that emerge any time

14that we follow the full essential facilities test as it's

15been articulated in the appellate courts.

16Well, this leads to the rule of reason

17alternatives. And I'm not going to exactly say which rule

18of reason alternative I think is best. I think we've

19heard a lot of very interesting and provocative arguments

20for the specific nature of the test.

21I want to oversimplify by assuming that if you

22took all of the rules of reason tests that are proposed

23that you can differentiate them along a spectrum from

24relatively strong enforcement to relatively weak

25enforcement. In other words, they can be differentiated


1according to the likelihood that we'll find conduct to be

2anticompetitive by how strictly they would enforce against

3refusals to deal and how likely they would be to impose a

4duty to deal.

5So, the policy for the courts and the antitrust

6agencies I think may be how stringent or generous the rule

7of reason test to choose for judging refusals to deal. I

8think that IP rights, intellectual property rights, might

9affect the answer. And here's why.

10Intellectual property rights are a primary

11source of legal barriers to competitive provision of goods

12that an incumbent refuses to sell to rivals. We heard in

13the testimony yesterday from some of the company

14witnesses, notably QUALCOMM and a couple of others, that

15they're very concerned about any rule that might require

16them to deal in particular ways with their intellectual

17property. Intellectual property rights grant them a legal

18ability to give them the ability to impose a legal barrier

19to invent around to innovations that would replicate their

20invention, and therefore gives power, creates an effect

21out of their refusal to deal or refusal to deal on

22particular terms.

23But, logically, any reduction in the strength

24and availability of IP protections could reduce the pool

25of goods for which there are legal barriers to competitive


1supply. There is an empirical question buried in here

2that I will return to at the end. But I think that IP

3reform could therefore affect the frequency with which

4refusals to deal weaken the conditions for being

5anticompetitive, in turn affecting the likelihood that

6enforcement of the duty to deal was warranted.

7So, what's the benefit of a more discerning

8intellectual property policy if IP reform reduces a firm's

9ability to use IP protections to block competitive supply

10and innovation, then IP reform can limit the need for rule

11of reason exceptions to Trinko's presumption against

12mandatory dealing with rivals.

13Now, one might say, okay, fine, why not have

14intellectual property reform and a fairly liberal duty to

15deal. Won't that unblock lots anticompetitive refusals to


17Well, both intellectual property reform and

18duties to deal aim to reduce barriers to competitive

19supply and innovation, but I think that their individual

20welfare effects may not be additive if they're undertaken


22Suppose that we do not have IP reform and that

23there is some good that is being used anticompetitively to

24block competitive supply. The duty to deal can increase

25welfare with no risk of deterring investment or innovation


1by the would-be buyer or third parties. The would-be

2buyer or third parties could be blocked by an intellectual

3property barrier to competitive supply or innovation, and

4so requiring that the refusing to sell or deal doesn't

5block any innovation on the demand side by the would-be

6buyer or by third parties. It might deter innovation and

7investment by the incumbent. That is something that we

8need to think about.

9With reduction of legal barriers through IP

10reform, however, the duty to deal now can undermine new

11competition and innovation, reducing welfare. So, the

12firm that is refusing to deal and the good that is

13protected by intellectual property, if they now have a

14weaker intellectual property right, we might want to say,

15well, let's not make them deal because now there's an

16invent around or a replication that didn't exist before.

17So, IP reform raises the likelihood, whether to

18any significant level is another question, but it raises

19the likelihood of false positives in antitrust enforcement

20through imposition of a duty to deal where the conditions

21for anticompetitive harm as a legal barrier do not hold.

22So, let's take a little bit of a closer look at

23the implications of IP reform for Section 2 reform. There

24are several kinds of proposals for intellectual property

25reform that could bear on the effects of refusals to deal.


1There's just some broad examples

2There are proposals to raise the bar for

3patentability: better pre and post grant opposition

4procedures; more transparent review, both in initial grant

5and post grant of patent grants or annuities.

6There are also proposals to reduce consequences

7of patentability: a narrowed patentable subject matter,

8for example, cutting software out of patentable subject

9matters; expanded research exceptions and reduced

10presumptions of harm in injunction proceedings which might

11push parties to the bargaining table; and limit refusals

12to deal. And these are proposals that can be found in the

13National Academy of Sciences' proposal, in the Federal

14Trade Commission's report of a couple of years ago; in

15draft statute that floated around in 2004; and in a

16variety of ongoing documents one can find these proposals.

17So, the effects of these proposals would likely

18be to make fewer goods subject to IP protections and to

19make those protections less expansive. Some of the most

20prominently discussed IP reforms, and I think this is the

21important point, would reduce the ability of incumbents to

22foreclose competitive provision of goods through the

23exercise of intellectual property rights.

24Depending on circumstances, these refined IP

25protections could have varying effects on incentives to


1deal. The reduced ability to foreclose competitive

2innovation through the enforcement of an intellectual

3property right might make an incumbent more eager to sell

4to rivals because it would expect greater competitive

5entry in the relevant property market than existed

6pre-reform, and the incumbent may therefore want to take

7the sales for itself for as long as it can.

8Alternatively, an incumbent may be less eager to

9deal if the sale to others would raise the speed or

10likelihood of competitive entry compared to what would

11occur if it keeps the good to itself.

12And which of these incentive effects occurs

13would depend very much on the nature of the good, the

14degree to which the selling firm is vertically integrated.

15There are a number of questions that are factored in.

16But I think on the whole refined intellectual

17property could reduce the incidence and the impact of

18refusals to deal. It is true that refined IP protections

19could reduce the willingness to deal with rivals by

20reducing an incumbent's ability to block replication of or

21innovative alternatives to its technology. But I think

22this effect is most likely where the goods involved are

23easy to reverse engineer and replicate. And these in

24turn, I think, are the goods where refusals to deal would

25be less harmful because the would be-buyer or others will


1eventually be able to market.

2So, on the whole, I think we'll find

3intellectual property protections should either reduce

4incentives to refusals to deal, or reduce the long-term

5effects of refusing to deal by opening the door to

6competitive supply and innovation.

7So, what are the implications for Section 2

8reform? The latter effect, competitive reinvention or

9replication of the goods at issue in a refusal case should

10be preserved. Antitrust reform should not impede a

11competitive reinvention because they should not provide an

12alternative or option to competitive entry or invention or

13innovation where it is feasible to occur.

14So, I think that if intellectual property reform

15reduced legal barriers to competitive production of the

16relevant good, Section 2 should be less willing to require

17the incumbent to deal. Broad exemptions to the "Trinko"

18presumption against mandated dealing could create a

19quasi-regulatory alternative to buyers that is unnecessary

20and unhelpful to economic welfare.

21So, that's some questions to investigate before

22we know whether intellectual property reform is actually

23going to matter.

24Several key questions. First of all, how likely

25is IP reform and to what extent will it refine the


1consequences of IP protections for competition. I think

2to question these efforts are under way. They're very

3political and very contentious. What will emerge from

4them is unclear. I think something will, but I think it's

5hard to know exactly what.

6The next question is really an empirical one and

7I think lies at the core of what I'm suggesting today:

8How much of a problem with refusal to deal stems from IP

9protected goods for which the barrier to competitive

10supply is a legal one rather than an economic one that

11stems from scale or something else. If not much, then the

12considerations I'm suggesting can be put aside as

13Section 2 reform proceeds. But if a lot, even if only in

14particular industries or markets, then refusal to deal

15policy should recognize the welfare and complexities that

16intellectual property reform might introduce.

17And the final question is: What effects will

18applied intellectual property protections have on the

19incentive of incumbent firms to deal with rivals. I think

20that's an interesting question to investigate.

21So, I have some tentative conclusions.

22The rule of reason approach for refusals to deal

23has potential advantages over either per se legality or

24the essential facilities test.

25The policy problem is to decide how strict a


1test the courts and agencies should apply in assessing the

2reasonability of refusals to deal with rivals. And the

3potential results of intellectual property reform may be a

4relevant consideration in that choice, with more refined

5intellectual property rights weighing in favor of less

6strict enforcement against refusals to deal.

7Thank you.

8MR. COHEN: Thank you very much Howard we're now

9going to take a break for roughly fifteen minutes.

10(A brief recess was taken.)

11MR. COHEN: Fine. Before we begin our questions

12and round-table discussions, I think a way to start this

13second session would be to give each of our speakers a few

14minutes to respond to or comment upon some of the issues

15that were raised by the other panelists.

16You can go in whichever order you prefer. We do

17ask as a reminder to speak into the microphone so we can

18get this transcript.

19MR. SHELANSKI: I'll start because I expect

20collusion over here on the right.

21So, I really enjoyed Aaron's and Joe's related

22presentations and I think that they are both in the core

23respects correct. I do have just a couple of observations

24or comments.

25So, one suggestion I would make is if you take


1Aaron's presentation and Joe's presentation and put them

2together, you could take them as saying that, if a firm

3cuts price in response to entry, one test is that it is

4not acting anticompetitively, it's in a safe harbor if it

5keeps its price low.

6And I just wonder -- the question I would have

7or the thing I would ask them to consider is whether their

8proposals, as compared with other tests that are typically

9used in this area, would increase the ability of

10competitive firms already in the market to raise rivals'

11costs by entering, for example, on the airline route that

12was at five hundred, bringing it down to two hundred, and

13then basically telling the five hundred dollar firm, you

14either need to cut your price and keep it there or face

15some kind of antitrust scrutiny that you will find


17Is the raising of rivals' cost prospect greater

18under proposal than under others? I don't know. It's

19just something that I think ought to be thought about

20The other comment that I have is that I am not

21fully persuaded that costs don't matter at all in the

22consideration of whether or not the five hundred dollar

23price is a problem or not. Obviously, as Aaron points

24out, the monopolist has the greater ability to sacrifice

25profits because it has obviously much higher net profits.


1But I wonder, again, and this may relate to the

2competitive strategy angle here, if the five hundred

3dollar price is not three hundred dollars above the

4competitive equilibrium, but a hundred dollars over the

5competitive equilibrium, we might worry a little bit less

6about the five hundred dollar price being the one that

7we're running into in the market because someone decides

8not to enter at four hundred dollars. Don't we have to

9look at costs to know how great a welfare loss there is to

10the current test? And would that matter to your

11recommendation of what do in in a particular case?

12MR. FARRELL: Well, let me start with that last


14I think if we knew everything, then you're

15probably right. I would take pretty strongly the

16perspective that the competitive process is about having

17policies that don't require us to know what the

18competitive equilibrium price is likely to be, and that

19therefore enforcement of competition policy and antitrust

20should not depend upon on our being able to say we think

21the competitive price would be X.

22And that's part of why I think the competitive

23process, as I understand it, operates through the

24formation of a blocking coalition that make the

25participants better off, without an inquiry into how much


1the incumbent loses from this entry.

2So, if you look at the entry in the oligopoly

3literature, the usual citation is the Mankiw and Whinston

4article, 1986 or thereabouts. And if you think about the

5way that regulation has traditionally treated

6cream-skimming and loss of income and profits due to entry

7and, think in terms of access pricing to control and deal

8with that, all of that it seems to me is extremely foreign

9to competition policy. And the reason it's foreign to

10competition policy is I think that the competitive process

11works precisely by ignoring the effects on the incumbent.

12And obviously if you want to increase welfare in the

13small, ignoring something like that that could be quite

14important is a stupid thing to do. But I think as part of

15an overall process, it's brilliant and seems to work

16rather well.

17And I think there are times, perhaps many times,

18when many, perhaps all of us, get confused about that.

19Because there's no doubt, I think there's a consensus that

20the eventual goal of all of this is economic efficiency.

21So, it's always very tempting to look at economic

22efficiency in each instance, and perhaps often is right to

23do so, but I think it's often wrong to do so.

24MR. SHELANSKI: And just a comment here on legal



1I actually think that you're on pretty good

2ground with some recent legal precedent. I mean, if I

3understood your comments about "Barry Wright" correctly,

4that that case made the mistake of thinking that downward

5pricing was more important than the competitive process.

6Maybe that's a way of summarizing your critique. I don't

7know if that's unfair or not.

8And certainly in Arizona against the Maricopa

9Medical Association case, even though that was a Section 1

10case, the Supreme Court said fairly strongly that we don't

11care about direction price level. What we care about is

12the competitive process and making sure it works well.

13So, there might be some legal standing for you

14to argue that your proposal is more in keeping with modern

15processor oriented thinking instead of the price oriented

16thinking that polluted the predatory pricing process.

17MR. FARRELL: I have something else to say, but

18if you want to respond to that.

19MR. EDLIN: Well, I wasn't going to respond to

20that. I was going to respond to what he said previously,

21which I suppose is not the rule as to how a conversation


23But I think Joe is right that, to the extent we

24can, we're certainly better off having an antitrust

25jurisprudence that doesn't focus on things that we are not


1very apt to know, like costs.

2And as to Howard's point, which is certainly

3correct, that if price is close already to the competitive

4equilibrium, then you shouldn't worry very much about what

5happens no matter what. I agree with that. And one thing

6that -- this gets to the last slide I had, which is, you

7may want to only worry about firms thwarting rivals from

8providing very substantial value increases to consumers,

9and not worry about situations where they are only

10providing minimal value increases. And if the prices are

11already pretty close to the competitive level, then you

12won't find rivals offering to provide very substantial

13value increases to consumers, and so we won't find that

14antitrust interferes very much in those circumstances.

15But now you wanted to respond to what he just


17MR. FARRELL: Well, I wanted to say something

18else about the role of costs in all of this.

19There's no doubt that sacrifice tests and cost

20tests can be illuminating concerning intent. And it's a

21bit of a paradox, I think, or piquant at least, that

22many of the same people who are very keen on sacrifice

23tests are also the first ones to lay into any attempt to

24use intent evidence in an antitrust case.

25It seems to me that intent is what you can


1sometimes infer from sacrifice tests, and one needs to be

2careful using intent evidence. Obviously there is the

3pervasive problem of testosterone poisoned sales managers.

4But thoughtful, high level intent may often be the best

5available evidence as to contemporaneous estimates of

6likely effects.

7And so I don't think we should be either too

8credulous or too rude about intent evidence. It's a kind

9of evidence, and it seems to me it's the kind of evidence

10that's most directly brought out by looking at sacrifice.

11Let me say one other thing, though, about how

12cost information might be useful.

13If it's right, as I suggested at one point, that

14you'd want to look at, in my hypothetical Northeast two

15hundred dollar price, and in some sense try to gauge

16whether that is where we've now got to, or whether it's

17just a quick and short-lived fighting price that will

18disappear as soon as the entrant has gone away and will be

19back to five hundred, if that's an important question,

20which it may well be, then it's perhaps somewhat

21informative to look at Northeast's costs, because if two

22hundred is below Northeast's cost, you might say, well,

23that more or less rules out the possibility that it's now

24the permanent price.

25Of course, there's a lot of other evidence about


1what the permanent price must be, such as what actually

2happened post exit versus what was happening pre-entry.

3And so I certainly don't see that costs would play a

4determinative role there, but it might be relevant to

5thinking about that question.

6MR. COHEN: Okay. I think we'll start things

7off by building on some of Aaron's testimony.

8I'll try the first question. Given the critique

9that you supplied of some of the existing tests as to

10whether conduct is exclusionary, what's your thinking as

11to whether it's sensible to be looking for any single test

12that captures all the elements of what we would want in

13all the various situations to determine whether something

14is exclusionary or not? Is this something that we could

15hope for? Is this something beyond our ability?

16MR. EDLIN: Well, I'd say it's always reasonable

17to hope, and physicists will hope for the grand unified

18theory and they may find it, and we should similarly hope


20Now, however, I think that what you should not

21hope for is that you'll find the right unified test and it

22will be easy to apply to the facts in any given

23circumstance. Whatever test you think is right is going

24to necessarily lead to huge factual disputes as to how the

25test comes out under the circumstance. I think a lot of


1people are driven by a desire to get away from that

2problem. And I think ultimately there are only two ways

3to get away from that problem, and one is per se legality

4and the other one is per se illegality, and both of them

5are very convenient, but I think that both of them are the

6wrong answer.

7MR. COHEN: Anyone else?

8Another way of trying to get at sort of the same

9set of issues, I guess, do you have any principles in mind

10that might help us determine areas in which any given test

11is more likely to work in a given setting than another

12setting? For example, are we more likely to have success

13with one of these tests in any price or non-price context?

14Are we more likely to have success with one of these tests

15in a setting where the issue is tying up inputs rather

16than settings which involve some of type of tortious

17conduct? Are there generalities that might guide us?

18MR. EDLIN: I think the main generality I would

19have is that one is more likely to have success with the

20test when it's seen from a sufficiency point of view than

21from a necessity point of view. And it -- or viewed

22differently, that these things are very -- can be very

23helpful evidence, either, as Joe said of intent, or of

24likely effect, which is to say, if you would not do it but

25for substantial diminution in competition, well, that


1suggests substantial diminution in competition is likely.

2So, the test can be very relevant from that

3point of view. It's when you start to push the

4implication sign the other way, which is what's been

5happening, that I think there's real danger. And the

6danger is across all of the categories that you listed.

7MR. COHEN: I noticed when you went through some

8of the variance of these tests, in a couple of the

9instances, you included a temporal dimension. You

10included short-term sacrifice for long-term profits.

11Does anybody regard the short-term/long-term

12distinction as something that's really needed here? Is it

13just a sacrifice in general? And if short-term/long-term

14matters, what are we talking about for time? Anybody want

15to comment on those temporal formulations?

16MR. FARRELL: Well, I'll make a perhaps slightly

17rude comment. Usually when you don't know quite what

18version of the test you mean, it's because you're not

19really clear on the logic of why the test makes sense in

20the first place.

21So, I think, for example, if you're trying to

22infer intent, then you'd want to ask yourself, all right,

23what is it exactly that the argument here is saying and

24what time scale you're looking over.

25If you're wanting to say there's no possibility


1that this is a price you would charge in the long run,

2that might tell you about something about what time scale

3you're looking over.

4So, I would go back to the underlying logic.

5And if you don't know how to go back to the underlying

6logic, that's a sign that there are deeper problems than

7just not knowing for what time scale to evaluate things.

8MR. MATELIS: This is a question about false

9positives and false negatives, which you mentioned, Aaron,

10and I'd be interested in all the panels' views.

11I suppose a slightly more spirited defense of

12the concept of false positives, which the Supreme Court

13has mentioned in just about every Section 2 case in the

14last twenty-five years, is that the competitive process is

15likely to fix false positives, whereas false negatives

16become ingrained in precedent and we're stuck with them

17for many, many years, as we were for decades in predatory

18pricing jurisprudence, where plaintiffs were winning cases

19where today I think everyone would agree they might not.

20Is this really a concern? Is the Supreme Court

21wrong stressing the idea of false positives, or is the

22concern overstated in general? How should this play a

23role in devising antitrust policy?

24MR. EDLIN: Well, I think you flipped the false

25positives and false negatives there, so I'll try to answer


1the question as I think you intended.

2MR. FARRELL: It's what statisticians know as

3Type 3 error [laughter].

4MR. EDLIN: So, as I see it, if you find what

5you consider to be the right test, whether that is a final

6results oriented test like efficiency or consumer welfare,

7or whether it's a process type test such as the freedom to

8trade that Joe and I are suggesting, I think the problem

9of false positives is not so much one of legal precedents

10but one of application, which is to say, if you've got the

11right test, then the real fundamental problem is, in its

12application you may get it wrong.

13And the question is: Will people so fear that

14when the test is applied to them that it will be gotten

15wrong that they don't do many procompetitive things,

16whether that's process or results interpreted.

17And I think we are so far from such a situation

18today that it just doesn't concern me very much. But if

19we were in that situation, I again don't think the right

20thing to do would be to say, well, let's find -- let's

21apply something that substantively doesn't make much

22sense. Rather, I think you should look at the source of

23where the false positives are coming from. If they're

24coming from bad jury instructions, make better jury

25instructions. If they are coming from courts having an


1insignificant standard of proof where it seems sufficient

2to allege that something bad happened rather than to

3really prove it, then we should crank up the standard of

4proof. And if -- and/or you say that you have to show

5that something really very bad happened, rather than just

6a little bad.

7So, I see the problem of false positives as

8being less in the precedents than in the applications of

9the facts.

10MR. SHELANSKI: I agree with Aaron. I would

11just add that I think a lot of rules look bad from a false

12positive standpoint. They look worse from the false

13positive standpoint at the beginning when the rule is

14articulated, then after there has been experience gained

15in its application.

16I think that, as an agency gains familiarity

17with the application of a rule, understanding of what

18certain fact patterns really mean, as courts get more

19experiences with reviewing cases and get a body of

20precedence and a body of jury instructions, some of the

21more frightening aspects of the rule may be damped down

22and you may get beneficial application.

23I do think there's a difference with respect to

24false positives between public enforcement and private

25enforcement under Section 2. I have a lot of faith in the


1agencies' abilities to gain a body of knowledge and

2understanding that they then bring to bear in their

3enforcement discretion under any given rules.

4I think with the courts, where there's a perhaps

5much less coherent body of learning, you have to rely on

6any particular district judge's reading perhaps outside of

7its own circuit and perhaps outside of its own circuit,

8and rely on a cohesive body of understanding. And this is

9not -- I am not trying to bash the capability of judges.

10I'm trying to just suggest you may get a less coherent

11development of a body of precedence and knowledge in the

12judiciaries than you get in the agencies.

13So, I think false positive may be worse for

14private enforcement than for public enforcement. But, on

15the whole, I would agree with Aaron, I think the

16application is the key issue. The deterrence effect is

17probably overemphasized in a lot of what one reads, and I

18think it can be offset in light of experience.

19MR. MATELIS: Anything to add, Joe?

20MR. FARRELL: No. I'll reserve my time.

21MR. MATELIS: Okay. Again this is a general

22question based off of something Aaron has mentioned twice


24What are better jury instructions that we should

25be giving juries in Section 2 cases? This might be


1another way of saying, if we don't want to instruct them

2on the no economic sense test, on what should we be

3instructing them?

4MR. EDLIN: Well, I think that the two best --

5the two best candidates that I think we should be

6instructing them clearly on, whatever we think the right

7test is, and the two best candidates that I have are a

8results oriented test, which is consumer welfare, or a

9process oriented test, which is that someone is being

10blocked from providing higher value to consumers, which is

11a process oriented test.

12And the instruction should of course distinguish

13all of the standard worries that people have, such as that

14it's not sufficient that rivals are losing money, and

15that's not the issue.

16What I'm really getting at there is, if you

17really -- I think the first thing before suggesting

18approving jury instructions is to come to a clear

19understanding of what antitrust is trying to accomplish.

20The second thing is to see if there really are a

21lot of false positives, and I don't see them. Right now I

22would say the improvement to jury instructions would be to

23not focus on tests that I think are nonsensical, which is

24the primary problem with them now.

25MR. MATELIS: Howard, Joe?


1MR. FARRELL: Well, in the unlikely event that I

2ever end up on an antitrust trust jury, I guess what I

3would want to hear is: The following specific questions

4have been given some prominence, but you the jury should

5please interpret them to the extent possible in light of

6the kind of fundamental things that Aaron was mentioning.

7MR. COHEN: Okay, let's turn a few questions to

8Joe's presentation.

9I really started with three questions, but as I

10think about it more, they come together into one. I'll

11throw it out in various forms.

12You talked some time early on about whether the

13results of not being able to successfully form a blocking

14coalition results from actions of the five hundred dollar

15airline, whether it happened intentionally or not, I think

16you said at one point, or another time you phrased it,

17whether it's a natural outcome of the way the market


19But then your rule you were trying to focus on

20where there's really a problem, you talked about whether

21the incumbent, the five hundred dollar incumbent,

22strategically thwarts the coalition.

23I'm going to ask you to try to give us some

24content about what you mean about "strategically thwarts."

25And maybe you can think about it in terms of a question of


1whether this approach would make it unlawful for a low

2cost producer merely to develop the reputation as an

3aggressive price competitor.

4Sort of a third way of asking the same question:

5What's happened to the bad conduct element of Section 2 in

6this core analysis?

7MR. FARRELL: Well, so first off, as I

8understand it, where we're surrounded by lawyers here, I

9don't think there is a bad conduct. There's an

10anticompetitive component, anticompetitive conduct.

11And if you accept the ideas that are being put

12forward about what anticompetitiveness means, then there

13can be conduct that is anticompetitive that is harmful for

14competition that isn't necessarily bad in any sense other

15than being harmful to competition.

16Now, there certainly has been a body of thought

17and especially shorthand that says you want it to be bad

18as well in some other way. That I think -- I try to

19interpret that in the following way. Let's suppose that

20in the course of trial, imagine it takes place in this

21order although it wouldn't have to, it's been shown that

22the defendant did some things that harmed consumers by

23excluding competition and were not, let's say, highly

24efficient. And I'm pulling together ideas of various

25sources here, I think.


1And now we ask, well, was it bad conduct? Well,

2from an economist's point of view, it seems as if in the

3instance it has just been shown to be bad conduct. So,

4the question is what further requirement is being asked

5for here.

6I think the further requirement that's being

7asked for here is the following: That this conduct -- if

8this conduct is condemned, it will have some sort of

9deterrent effect on conduct that sounds like this when

10described. And that deterrent effect will extend of

11course to other places where the competitive implications

12of the conduct might be a little bit different.

13And so what you want in addition to finding this

14conduct was inefficiently anticompetitive and

15anti-consumer here, you want some degree of confidence

16that similar-sounding conduct is going to tend to be not

17such a good thing or a bad thing, in other circumstances

18where maybe it won't be inefficiently anti-consumer,


20Well, that puts a lot of weight on the

21psychological or even philosophical concept of conduct

22that sounds like this. There's a philosopher named I

23believe Grice, who really tested foundations of that kind

24of thing by inventing a word, grue, g-r-u-e, which means

25green up until this morning or blue after this morning.


1And so all of your past observations that trees are green

2are also observations that trees are grue. What do you

3predict the tree color will be this afternoon.

4Obviously that's playing with words in the way

5that philosophers love to do, but it does suffice to make

6the point that, if what you are looking for in a, quote,

7"bad conduct" problem is something along the lines of

8similar conduct that is going to be bad in other

9circumstances, you need a concept of what's similar. And

10that's not really an economic concept, as far as I can

11tell. It's some sort of intuitive or possibly legal


13MR. COHEN: Anyone else?

14I'll shift ahead because your comments invite


16What kind of difficulties would you expect

17courts have in operationalizing something like this? I

18would hate to go in and try to tell them that trees are

19green in the morning but blue later.

20MR. FARRELL: Well, just to be clear, at least

21in my own mind, I would be delighted if judges were to

22listen, and when we get around to writing, read this kind

23of stuff. But I am not convinced that it's ready for

24courts yet.

25What I think I would like courts to do is put up


1a lot of resistance to the incorrect tests that are being

2bandied about on the pretext of administrability, bright

3line, sort of vaguely right, perhaps, maybe, although we

4can't exactly tell you why.

5And I would like to see courts, led by the

6Supreme Court, say, look, we really have not sorted out

7yet what administrable concrete tests we need to apply for

8Section 2 liability. For the time being let's do

9so-and-so, but that's not meant to be the final answer.

10Because I think it's pretty clear that nobody is

11in a position to say yet what the final answer should be.

12And I think there's a huge danger, given the way courts

13and lawyers tend to think and talk, that things are going

14to congeal prematurely.

15MR. COHEN: I'm wondering if you're at a point

16yet where you could predict if there are particular types

17of conduct where the analysis you're thinking of is really

18likely to lead to different results than you've been

19getting through viewing perfect competition as the goal?

20You may go through a different process. Do you have any

21idea where the results are likely to come up?

22MR. FARRELL: No. I think the salient

23differences are going to be based on the question of how

24closely you try to examine direct efficiency consequences

25versus trusting the competitive process to do that and not


1requiring it in the narrow instance.

2You know, technically if there is a perfectly

3competitive equilibrium in an economy, it is then in the

4core. And so I don't think there is a substantive

5tension between the two. I think it's more a question of

6what process each one suggests to you.

7 It seems to me the core -- and let me stress,

8I'm not suggesting ever examining an outcome to see

9whether it is in the core. I'm suggesting the process

10that is suggested by that, which is, make it relatively

11easy, or don't allow it to be made artificially difficult

12to form blocking coalitions.

13Whether there is a similar process that is

14suggested by thinking about perfect competition, I am not

15quite so sure. You know, economists have talked for a

16long time about the fact that perfect competition is

17describable as an outcome, and we don't have a very good

18story about how you get there. There's the infamous

19Walrasian auctioneer. That's obviously not a process that

20takes place in reality, let alone is protectable by


22It seems to me that thinking about the coalition

23formation model gives you a stronger suggestion about what

24process to protect than thinking about perfect



1MR. EDLIN: I'll hazard a guess, which is, if

2you thought about things a little more the way that Joe

3and I think about things, then you would find that the

4Department of Justice would probably have won the American

5Airlines case; that entry would be easier in many

6industries because monopoly or dominant firms would have

7more limited ability to thwart entry; more attempts by

8monopolies to prevent entry by tying goods together would

9be illegal, but not all; and those would be the kinds of

10things that you would see in terms of substantive outcome


12 MR. SHELANSKI: I will just add that I think the

13process emphasis, while extremely important theoretically

14and at some level is absolutely correct economically does

15have some pragmatic difficulties.

16I actually really worry about instructing juries

17on the process as opposed to outcomes. And you can

18combine the two to halve their inquiry, but I think the

19confusion between competition and competitor is one very

20easily sown in juries.

21And connected to your question earlier about

22false positives, I think that as a firm, faced

23particularly with a private suit, knowing the instruction

24is going to the jury about process, you're worried about

25looking aggressive, worried about looking the bad guy, and


1you get a lot of hidden false positives through

2settlement, particularly in the private cases.

3So, I do think it's worth thinking a lot more

4about the pragmatic implications of the process

5instruction of going forward.

6MR. COHEN: Finally, for Joe.

7The theory that you've explained depends on the

8formation of these blocking coalitions. There are

9obviously impediments to this. You recognize them and

10they may not always be formed, but at least there's an

11incentive to do them.

12Have you thought about how we should take into

13account the fact that not all of these coalitions will

14ever form in the first place, that there maybe information

15problems or the cost that prevents them from happening?

16How do we bridge from incentive to actual assumption that

17they're there and therefore that their losses are


19MR. FARRELL: I don't. I mean, I think, as I

20think I mentioned, the way you prove that a competitive --

21that everything in the core is Pareto efficient, is by

22pointing to the so-called grand coalition of everybody, if

23it was prey to inefficient, then in theory this grand

24coalition could block. That's obviously not going to



1So, I think any policy, including antitrust, is

2not going to be able to get us all the way to Pareto

3efficiency, whether it thinks of it in terms of central

4planning, price-taking equilibrium or the core.

5Now, as related more directly on a practical

6point, which is, well, what happens if -- this is I think

7maybe what you were getting at with the bad act question.

8What happens if we have a not very good outcome in the

9status quo and the blocking coalition that, quote, ought

10unquote, to form doesn't form, not because of anything

11that the incumbent does, but just because it's really hard

12to form.

13Well, I think at some level that could be a

14competition policy question. There might be changes that

15could be made in the way the market works to make it more

16likely that such coalitions would form.

17If it were a competition policy question, it

18wouldn't necessarily be an antitrust question. I think

19they're potentially distinct areas. And it might be

20neither. It might just be, well, that's too bad, that's

21one of the imperfections of the world.

22MR. MATELIS: At the beginning of these

23hearings, both the Assistant Attorney General and the

24Chairman of the FTC stressed the importance of safe

25harbors for guiding businesses that are seeking to comply


1with the antitrust laws.

2And, Joe, I have a question for you. The

3examples in your presentation were responses of a firm to

4new entry. Northeast's response to Sprite's entry and the

5A and B product potential responses at the new entry.

6Are there responses to new entry that, you know,

7looking at things through the core, should be within a

8safe harbor and something that firms should always feel

9comfortable doing?

10MR. FARRELL: Well, I'm sure there are, but just

11as I don't know exactly what the right rules for

12liabilities should be in a practical sense here, I also

13don't know what the right rules for safe harbor should be.

14I mean, one can give the following answer, which

15is sort of in the spirit of something Tim Bresnahan has

16said, and you will be hearing from him this afternoon,

17that the safe harbor is to make your money by being nice

18to consumers, not to make your money by being the other

19stuff you can be. That's not quite the way Tim put it,

20but he had a somewhat similar line which maybe you can get

21out of him if you ask him.

22MR. COHEN: Directing some questions to Howard

23Shelanski's presentation.

24You focused very much on intellectual property,

25the effects of possible changes in that area, bleeding


1over into how we might look at Section 2 issues.

2If we're looking at Section 2 issues, we're not

3likely to have differential treatment of instances in

4which there are lateral refusals for intellectual

5properties versus others.

6Would your rule somehow -- are you envisioning

7somehow distinguishing between the two, or just a one size

8fits all modification?

9MR. SHELANSKI: One size fits all is what I'm

10looking at. I'm actually not so much proposing a

11particular rule, because I agree with you there should not

12be two rules. Obviously the precedent is a little choppy

13between the various circuit courts on the extent to which

14you get special Section 2 protections for intellectual


16But my view is you should not have a separate

17rule. And I was really looking at the macro level. If

18you take the total pool of goods that firms refuse to deal

19with, some of them are going to impose barriers because

20they're legally protected, legally blocked by IP.

21The smaller the pool of goods where there's an

22anticompetitive refusal to deal, the less enforcement

23minded you want to be against refusals to deal.

24So, for me it's really an adjustment mechanism

25about how permissive or strict a unitary rule you apply.


1I mean, if you were to look and see, boy, a lot of these

2refusals to deal cases have at their core intellectual

3property. Then I think intellectual property would not,

4say, have a different rule for those cases versus others,

5but it would say we can have a more permissive rule

6towards refusals if we had intellectual property


8MR. COHEN: One thing that you mentioned a

9number of times in your talk was issues about the degree

10to which imposing liability or not imposing liability for

11refusals to deal might affect innovation, might affect

12efforts invent around whatever problem there is.

13It's a little unfair, I know you gave a

14theoretical presentation, but of course we're very

15interested in anything empirical.

16Do you have any -- can you give any summary or

17are there any indications of what there is out there in

18the way of empirical evidence on this?

19MR. SHELANSKI: If I can cheat a little bit, I

20think I can. So, I did raise that issue of demand side

21innovation and competitive supply because I feel that in

22the discussion about duties to deals there's been

23overemphasis on deterring the initial innovation by the

24supplier. I think that's extremely important. And I

25wouldn't want to see a situation where we punished


1innovation per se. So, I want to be very careful. But I

2wanted to build into the demand side there's innovation on

3both sides of the enforcement question.

4So, here's a possible place to look for some

5empirical support, and this is contentious. I would go to

6the regulatory arena and I would look at the unbundling

7obligations of the Telecommunications Act of 1996.

8There are allegations that overly permissive

9access for competitors to incumbent networks reduced the

10degree to which these new entrants built their own

11facilities and their own networks, therefore leading to

12less vigorous competitive entry.

13I think there's a lot of debate over the extent

14to which this is true, but there is some empirical

15evidence that after the FCC repealed a very permissive

16access to the incumbent platform under what some would

17argue were subsidized rates -- there is a legitimate

18dispute over that -- that after they repealed that access,

19there was a lot more facilities-based entry, a lot more

20actual building and installment of competitive facilities.

21This does suggest that a duty to deal, which

22would then include some kinds of terms of dealing, runs

23the risk of stopping entry of competitive assets into

24other markets. And the telecommunications market might be

25one place to look for such evidence. And there is some


1literature out there with competing arguments about

2whether the essential facilities treatment or the duty to

3deal imposed by the Telecommunications Act of 1996 on

4incumbent networks deterred and chased out new competitive


6MR. FARRELL: I think part of the reason why

7people have focused on incentives of the original

8invention or the original investment is that, of course,

9that innovation or investment directly leads to social


11Duplicative investment is -- I want to avoid

12taking too narrow a view here, but nevertheless, at some

13level duplicative investment is wasteful. And while

14having some of it may well be part of the process and

15negotiating for voluntary access in the shadow of the

16threat when you look at the investment is probably a

17bigger part of the process, I think it's actually wrong to

18treat reducing the incentive for duplicative investment as

19a policy downside in itself.

20Now, it might actually be a kind of shorthand or

21a proxy for some other harms that you think come out of

22more mandated sharing than other policies would give you.

23But I think one wants to be wary of that shorthand.

24MR. SHELANSKI: I'll disagree slightly. I think

25you're right that that's something to be taken into



2I think the market conditions under which that

3duplicative entry would be welfare decreasing are fairly

4specialized. I don't know how common they are. I think

5it needs to be taken into account. But while it's a

6consideration, I am not sure that it's a big enough

7problem that I would discount -- I certainly wouldn't

8discount the value of at least some competitive investment

9or duplicative investment, especially where it's not

10economically blocked. There's not some kind of natural

11monopoly or scale kind of argument that would make that

12investment a not be beneficial end, but where there's

13simply a legal barrier to producing something that could

14be produced fairly cheaply. Software would be an example.

15MR. COHEN: Just one more. I'm going to return

16to something that Joe just mentioned a couple answers ago.

17You drew the distinction in a sense between a

18competition issue and an antitrust issue. Another way of

19phrasing some of the same points we've already been going


21To the panel just generally: Do you see a

22difference in your analysis between a competition issue in

23the sense of maximizing efficiency, and an antitrust issue

24in the sense of what should be a legal violation?

25MR. FARRELL: I'm certainly very open to that, I


1think. First of all, I would not phrase a competition

2issue quite as maximizing efficiency, for all the reasons

3we spent all morning talking about.

4But I think it's perfectly possible for a

5competition agency, let's say, to discover that

6such-and-such a market would work a lot more competitively

7with these ground rules than with those ground rules. And

8to try to use its influence, perhaps even its legal

9authority, to have the better rules rather than the less

10good rules apply.

11And that doesn't necessarily involve anybody

12having, quote, done anything wrong. And so I think

13there's potentially a difference between competition would

14work better in such-and-such a way than with the status

15quo, and saying so-and-so has committed an antitrust


17So, yes, I think there's probably a big area

18there, actually.

19MR. COHEN: Okay. Do any of the panelist have

20any final points they want to make?

21MR. EDLIN: I'm in favor of lunch.

22MR. COHEN: Okay, we vote for lunch here.

23I again want to thank all of our panelists for

24their thoughtful and insightful remarks. I ask the

25audience to please join me in a round of applause for our




3MR. COHEN: And our afternoon session will begin

4promptly at 1:30.

5(Whereupon, at 11:59 a.m., a lunch recess was























2(1:30 P.M.)

3MS. GRIMM: Good afternoon. I would like to

4welcome everyone to our afternoon session. And I'm glad

5that you all could be with us today.

6I am Karen Grimm. I am Assistant General

7Counsel for Policy Studies at the Federal Trade

8Commission. I am going to be moderating the session this

9afternoon, along with June Lee, who is an economist at the

10Antitrust Division of the U.S. Department of Justice.

11Before we start, I would like to just go through

12two housekeeping details. First of all, as a courtesy to

13our speakers, please turn off all your cell phones,

14Blackberries, and other devices

15And, secondly, because these are hearings, we

16request that the audience not make any comments or ask any

17questions during the presentation.

18This afternoon we are honored to have another

19group of distinguished economists from the University of

20California at Berkeley and Stanford University to offer

21their testimony in these series of Section 2 hearings.

22Our afternoon panelists, like those this

23morning, will provide their perspectives on various issues

24related to the complex area of Section 2 jurisprudence and



1Our panelists this afternoon are Timothy

2Bresnahan, who is the Landau Professor of Technology and

3the Economy in the economics department at Stanford

4University; Richard Gilbert, who is a professor of

5economics at the University of California Berkeley and the

6chair of the Berkeley Competition Policy Center; Daniel

7Rubinfeld, who is the Robert L. Bridges Professor of Law

8and Professor of Economics at the University of California

9Berkeley; and Carl Shapiro, who is the TransAmerica

10Professor of Business Strategy and Professor of Economics

11and the Director of the Institute of Business and Economic

12Research at the University of California Berkeley.

13Our first three panelists will make

14presentations, and Professor Shapiro will be participating

15in the discussion with his fellow panelists.

16Our format this afternoon is as follows: Each

17speaker will make a 20 to 30 minute presentation. After

18all the presentations have been completed, we will take

19about a 15 minute break. And after that break we will

20reconvene for a round-table discussion. We are scheduled

21to conclude this session about 4:30.

22I would like to thank all of you for being with

23us here today. I want to thank all of our panelists for

24coming and for their participation. We very much

25appreciate the time and effort all of them have put into


1preparing their presentations and their willingness to

2share their insights with us.

3I would now like to turn the podium over to my

4DOJ colleague and co-moderator, June Lee, for any remarks

5she would like to make

6Ms. Lee: The Antitrust Division of the

7Department of Justice is pleased to co-sponsor today's

8single-firm conduct hearing. As noted by Joe Matelis this

9morning, five of today's panelists were Deputy Assistant

10Attorneys General in the Antitrust Division. Four of the

11five are in the panel. I thank them for participating

12and, like Karen, for sharing their insights. I look

13forward to their presentations in what I'm sure will be a

14lively discussion.

15 I join Joe in thanking the Competition Policy

16Center and the Berkeley Center For Law And Technology at

17the University of California Berkeley for hosting these

18hearings. And I thank Karen and her colleagues at the FTC

19for their work in organizing today's hearing and

20assembling the august panel we have today.


22MS. GRIMM: Our first speaker this afternoon is

23Timothy Bresnahan, who is Landau Professor of Technology

24and the Economy at Stanford University and Chair of the

25department of economics.


1He is Director of the Center for Research in

2Employment and Economic Growth in the Stanford Institute

3for Economic Policy Research. He also has served as Chief

4Economist of the Antitrust Division of the U.S. Department

5of Justice.

6 His research interests lie in the economic of

7industry, especially of high technology industry.

8Professor Bresnahan received his B.A. from

9Haverford College and his master's degree and Ph.D. in

10economics from Princeton University.


12MR. BRESNAHAN: Thanks for that very nice

13introduction. Let me see if I can find my slides.

14While I'm finding my slides, let me confess that

15in my role as department chair, I worked with the agencies

16in a failed effort to bring these hearings to Stanford

17rather than Berkeley. If you think of the reputations of

18those two great universities, you might infer that signals

19a leftward shift in the antitrust enforcement effort.

20But I don't think that's what it signals. If

21you either look over here to my left or at the brochure

22from the Competition Policy Center, you can see why

23Berkeley is an enormous center of academic influence in

24this area. This was the right place to put it.

25I want to talk about monopolization (Section 2)


1cases. And my real agenda is to normalize them, to

2regularize them within antitrust analysis.

3We have a tendency in talking about Section 2

4matters to immediately leap to the most difficult part,

5which is the part that's about alternative efficiency

6theories of whatever business practice it is that's

7challenged in the Section 2 matter.

8I think that that makes Section 2 matters more

9difficult than they need to be, and I'm going to propose a

10different approach, not inconsistent with what we've

11done in the past, and which we'll see in a minute, not

12inconsistent with recent court decisions.

13I'm going to suggest a different approach where

14we look at competitive effects first. It's not very

15surprising that I want to look at competitive effects

16first since I'm an economist.

17And then I think I'm going to argue it's going

18to make thinking about whether a Section 2 case is

19procompetitive much easier than starting from that very

20difficult question of whether the challenged practices are

21an act of competing rather than anticompetitive act. So,

22I'm going to start with competitive effects.

23There's been a good bit of action in the courts

24in Section 2 lately. You know, I see three big topics

25here: boundaries with other parts of the law, notably with


1patent and copyright law; predatory pricing is another area.

2What I want to talk about are bundling and

3related practices. So, vertical Section 2 cases where a

4monopolist commits monopolization or is alleged to

5commit monopolization through bundling its monopoly

6product with something else, or through contractual

7restrictions that amount to de facto bundling.

8I'm also going to talk about Microsoft and

9Dentsply in some detail, but Dentsply first. This partly

10reflects the idea that I think that the folks who do

11judicial decisions have the same economics in mind that

12I'm going to talk about this afternoon. And it's partly

13that there are three cases, two recent cases in this area.

14And I found those two, again another confession, much

15easier to read than I found the LePage's case, which was a

16struggle for me, although I am not sure it's inconsistent

17with what I'm going to say.

18So, any Section 2 inquiry I think has at its

19heart an economic structure if it's a rule of reason

20inquiry. Any rule of reasoning inquiry has economics in

21it. I think the economics enters at two distinct places.

22It has to enter in market power. You need economists to

23figure out market power. And I want to say, as I've been

24saying for a quarter of a century, the thing about market

25power is sometimes it's a useful shortcut in antitrust


1enforcement. We should be thinking about competitive

2effects when we're thinking about market power, particularly

3I would encourage the agencies, when picking cases in the

4merger area or in the Section 2 area, to pick cases where

5there's potentially a substantial change in the conditions

6of competition in the market and significant impact on the

7economy. That's not the same as market power. That's a

8change in market power.

9The other place where economics matters is in

10thinking about the causal flow from the acts which are

11alleged to be anticompetitive in a Section 2 case to the

12changes in market power. And I'm going to argue, this is

13my theme for the afternoon, you can gain a lot of clarity

14about a Section 2 case by bringing the competitive effects

15and causation arguments to the forefront. And I think that's

16consistent with the three bundling cases I cited, bundling

17or tying cases, I cited on the previous page.

18Section 2 cases are never going to be easy.

19Let's be real. There's a reason for that. This is I

20think the hardest part. Almost all conduct which would be

21exclusionary in some context would be an ordinary and

22competitive business practice in some other industry. So,

23it's necessarily context specific. That makes it

24difficult I think for attorneys to get their heads around

25Section 2 matters all the time because it seems like


1there's a fairly unstructured rule of reason analysis in a

2Section 2 case.

3I'm going to argue again that monopolization can

4lead you to a fairly structured economic competitive

5effects decision. Let me do that right away. I'll do it

6in Dentsply first.

7This is a Department of Justice case. I know a

8part of the history of it. I believe it was brought when

9Dan Rubinfeld was Chief Economist. It was litigated when

10I was Chief Economist. And I just learned from Professor

11Shapiro that it was under investigation on his watch.

12MR. GILBERT: It was under investigation at the

13FTC before I was at DOJ.

14MR. BRESNAHAN: Exactly. We are lucky that

15prefabricated artificial teeth is not a market which

16changed quite so quickly as computer software. But I note

17that the other case I am going to talk about, Microsoft,

18has a similarly long, long series of investigations before

19there was a serious enforcement action.

20So, what's the story of Dentsply? Why did the

21Department of Justice bring a Section 2 action?

22So, part of it, there is a market definition,

23there is monopoly power, and there is, in the current

24market, a monopoly in prefabricated artificial teeth.

25There are some small sellers, but there is one great big


1seller named Dentsply.

2Now, here's the competitive effects part. And

3this is something I think that's a reason that's going to

4make cases fairly rare in monopolization. While there is

5a monopoly in prefabricated artificial teeth, there could

6be substantially more competition in the market from a

7number of non-Dentsply like artificial teeth -- prefab

8artificial teeth providers who are smaller, very small at

9the time the case was brought, and typically lower priced.

10And it's the difference between the competitive

11regime there is, monopoly, and the competitive regime

12there could be, much less monopoly, which is the

13competitive effect that I think we should bring to the


15If it's inevitable, if Dentsply has a

16monopoly that cannot be changed, if there is some barrier to

17entry which cannot be lowered by any earthly force, there

18can still be a monopoly but how can there be

19monopolization? Monopolization I think needs to be

20cause of a change in the competitive regime or prevention of

21the change to a competitive regime that otherwise might arise.

22Now, in this case, the mechanism, you need a bad

23act as well as a competitive effect to have a Section 2

24case. The mechanism by which Dentsply prevented the

25emergence of competition from these other firms was


1exclusive contracts with dealers. They were dealers who

2supply dental laboratories with all kinds of things, but

3in particular with prefabricated artificial teeth. And

4those contracts block the laboratory from sourcing another

5firm's teeth, preventing the American consumer from

6having an effective prefabricated tooth choice.

7You know, there's a market in everything. Some

8of it might be competitive. As you get older, you get

9more serious about the importance of health care markets

10for having a competitive organization. And, Lord knows,

11there is not enough competition in most health care markets.

12So, I want to bring to the forefront, the

13horizontal competitive effects. Impact, if there's a

14Section 2 case, the impact of the bad acts, the contracts

15in this case, is to reduce competition in the market for

16prefab artificial teeth. So, it's possible that there are

17two competitive regimes, one with monopoly and the other

18with competition.

19And I want to push to the second, the vertical

20restraints logic, that the economic effects of these

21contracts, these exclusive contacts, is to change that

22competitive regime.

23You know, it seems to me that you can, in the

24course of investigating an alleged Section 2 violation,

25discard an enormous number of cases just by thinking about


1-- not about the efficiency theory of the supposed bad

2act, but rather just thinking about the anticompetitive

3theory. The inquiry would ask: is it possible that there

4could be less competition and also there could be more

5competition in this industry? Is it possible that if the

6dealer contracts weren't exclusive that then there

7could be competition? Without a "yes" to both, further

8inquiry is not going to lead a Section 2 case. The second

9question, the exclusivity of the dealer contracts having

10sufficient impact to change the compeitive regime, that is

11not a small inquiry. There is a lot of assumptions under


13 There are at least two base assumptions. The

14monopolist, Dentsply, is in a position to compel the

15dealers to accept these exclusive contracts. That's not

16going to be true in all industries. There can't, for example,

17be the possibility of some other parallel distribution segment

18which can grow up and distribute the competitive prefab

19teeth. Furthermore, while the distribution channel firms must

20not be in a position to resist Dentsply, Dentsply's competitors

21must need the distribution channel. Thus, the distribution

22channel must be dependant upon Dentsply but depended upon by

23the competitors. Not all distribution channels will satisfy

24both conditions. So, there's a reason that these exclusive

25dealership contractors have bite. Bite, it was entirely


1accidental pun. I think if we could go down the path of the

2Dentsply puns, they would be very unhappy for us.

3But I mean to emphasize that there are two

4dualities just in the competitive effects part of a

5Section 2 case, which means, before you get to the hard thing

6about efficiencies, you could throw a lot of cases out. It

7has to be possible that there's two competitive regimes,

8monopoly and more competitive, and it has to be possible

9that the bad act works to move the market between them,

10and that itself has two steps. The little guys, the

11potential competitive providers of these competitive teeth,

12have to need the distributors. The distributors need to be a

13powerful hard-to-replace force. And the existing monopolist,

14Dentsply, has to be able to kick around the distributors.

15So, you've got two dualities, it's monopoly, but

16it might be more competitive. And the distributors are

17important, but the monopolist is in a position to either

18bribe them or compel them to prevent the outbreak of

19competition, competition which would be plausibly in their


21Those two dual tests I think will weed out a lot

22of cases before you begin this open-ended discussion of

23whether these particular contracts are efficient. So,

24here is how I graph it. You've got -- your centerpiece

25should be the anticompetitive effects. So, in


1monopolization case, the effects are anticompetitive.

2There is an exclusionary act, in this case the contracts,

3which is keeping us in a higher market power monopoly, in

4this case industry regime rather than a lesser market


6And, as I said, that's a lot for the plaintiff

7to show. In the case of the agencies, that's a lot for

8them to show. And I want to urge a review of whether we

9can show these things early in a case. When I said to

10kind of regularize Section 2 review, you know, it's just

11like merger review, is there a competitive effect this merger

12is going to do? Is there a competitive effect these are bad

13practices are going to have, too? Is it really true that

14there is more market power in the current regime but there

15could be less market power? And that is the centerpiece,

16that there is this causation, there's these

17exclusive contracts, which exist because the existing

18monopolist wants to maintain a monopoly, or what's keeping

19us in the less competitive regime rather than the more

20competitive regime.

21And I think if you do both that causation

22carefully and that competitive effects carefully that

23would make Section 2 cases look a lot more like ordinary

24antitrust analysis.

25So, I said a number of times that that's a lot


1to show. It has to be possible that the competitive

2regime could change; it has to be possible that the bad

3acts are what's preventing the competitive regime from

4changing; there has to not be another explanation of why

5the competitive regime is not changing.

6We spend so much time in Section 2. Here's my

7one slide. I think I only have one slide and it's sort of

8ordinary analysis. We spend so much time thinking about

9whether there's an efficiency theory of the

10anticompetitive acts. And that is important. But, you

11know, I guess I would say, solve the problem with whether

12there's a harm to competition first and then worry about

13if there's an efficiency theory.

14A lot of this efficiency discussion -- and here

15I'm echoing Professor Farrell's earlier remarks in these

16hearings -- we're driving in the direction of that world

17of pure economic theory where we can figure out in a

18quantitatively precise and reliable way whether

19the consumer of the industry is better off with the

20existing industry structure, including its contracts,

21versus some counterfactual regime where the contracts

22would be gone and there would be less efficiency

23presumably from the contracts, but also more competition.

24In economic theory, the author of the model knows

25everything and could calculate how well off consumers are in


1another world. In the real world, the ability of

2empirical economics, even with the very high level of

3inquisitory abilities of the enforcement agencies to figure

4out what would happen in that but-for world in enough detail

5to calculate social welfare seems to me to be a waste of


7So, I would say, plaintiff has to show that

8there is an anticompetitive effect and that it's causal.

9And defendant gets to rebut that. Defendant has to show

10that their practices are efficient. Plaintiff gets to

11rebut that.

12If the world is not tired of hearing from me

13about the Microsoft case, let me talk about that one too.

14I mostly want to emphasize its parallels to Dentsply.

15Again, my competitive effects story is in the

16graph here, I think I'm very close to the D.C. Circuit's

17logic here. The competitive story is slightly different

18because the industries are slightly different. And this is

19one of the inevitable costs of Section 2. Section 2 cases

20are rare. They arise in those industries where there is the

21possibility of a big change in competitive circumstances.

22That's not most industries and that's probably

23idiosyncratic industries. Certainly these two, the teeth

24and the software, are both idiosyncratic.

25So, what's the state of the market? There is a


1Windows monopoly in operating systems on PCs. That was

2true when the case was brought. I got the year wrong. It

3was tendered to the Department of Justice in 1997 or so.

4There could have been dynamic competition for the operating

5system market if the mass use of the Internet led to new

6standards in new markets.

7So, here -- well, in the case of Dentsply, there

8was a monopoly and could have been competition in the

9market for prefab artificial teeth. In the case of

10operating system software, there is a monopoly and the

11industry in the past had had dynamic competition where

12entrants in many important software products had replaced

13incumbents. And in other important software markets, they

14had given incumbents a terrible scare and created

15incentives to get some real innovation out of them. In

16these software markets, there is persistent static

17monopoly, but there could be the prospect of

18Schumpeterian competition. So, that's the two competitive

19regimes that you get the competitive effects on.

20The other part of Microsoft is really quite

21similar to Dentsply. What kept the world in the monopoly

22regime rather than in the potentially more competitive

23regime, a regime where say a Linux might have taken a run

24at the position of Microsoft Windows on the desktop?

25It was a distribution case just like Dentsply,


1how could actual distributors, and a wide number of

2different kinds of complementors, with other third parties

3that would have worked with something like Linux on the

4desktop prevented a market test for the Internet entrepreneurs,

5and thereby ultimately prevented Schumpeterian competition in

6the operating systems market.

7So, again, this is the stuff that the Antitrust

8Division had to prove in Microsoft. That's why it's such a

9long case. Two potential competitive regimes. One, the

10present one in operating systems and other infrastructure

11software on the PC, which is about ten years you've had

12very little competition in those industries, but in the

13same industry in the previous twenty years before that you had

14it all the time. Maybe there could have been, certainly the

15Microsoft guys thought there could have been, dynamic

16competition against some of those valuable position if the

17Internet entrepreneurs had succeeded.

18Some of this was more complicated and it's vertical

19in more senses. The Internet entrepreneurs wVere not

20horizontal competitors for Windows. The browser

21was a complement. So, this was vertical restrictions to

22prevent vertical disintegration. The vertical

23disintegration would have permitted horizontal (dynamic)

24competition in the operating systems market. So, it's a

25good thing that the history of the industry had so much


1vertical disintegration causing horizontal competition for

2the market, and that the Microsoft guys in their internal

3documents were so clear about that that such a complex

4case could be argued.

5So, there's Andy Grove. Everybody has seen

6Andy's slide a hundred times. The way you get competition

7is you get a vertical disintegration. Andy was, when he

8wrote this, the CEO of Intel. Mr. Gates of Microsoft has

9said this many times as well.

10This was the essence of the antitrust case,

11that the internal documents, used that model of

12vertical disintegration leading to horizontal competition,

13provided evidence for the potential change in the competitive


15Now, again, I want to say, these cases are going

16to be rare. There's not a lot of industries where

17vertical disintegration is the key trigger for horizontal

18competition. It happens to be in infrastructural or mass

19market software on your personal computer that that's

20true, and it's been true since the industry was founded.

21But, the cases where there can be causation from a

22vertical restriction to horizontal competition are going

23to be reasonably rare. This was one.

24I would emphasize again, look for evidence of

25that causal change before you go worrying about



2This part is pretty much the same as

3Dentsply in many ways. Microsoft is more complicated

4because it's vertical in two senses: vertical

5restrictions to prevent vertical disintegration, and

6vertical disintegration in turn preventing horizontal


8But what was really important in the competitive

9effects in the case was that chain of causation did lead

10to blocking of a threat which could have led to the kind

11of dynamic and very valuable competition we had seen over

12the previous twenty years in this industry.

13Microsoft -- this other pragmatic, question about

14when to bring a Section 2 case, it's helpful to have a

15defendant that tries to prove entirely implausible things

16like, there's no market power in Windows. It was a bad moment

17for their economics expert witness, I think.

18The other very unwise thing

19that Microsoft chose to prove was that their reaction to

20the widespread mass market use of the Internet wasn't

21strategic, even though there were hundreds and hundreds

22and hundreds of internal documents saying that it was

23strategic. The CEO, whose memo I just quoted saying, this

24is a terrible threat to us, chose to testify that he had

25no idea what the threatening firm was doing at the time.


1So, defendent's trying to prove that it wasn't

2strategic, trying to prove that there was market power,

3made it somewhat easier for the government to prevail.

4These are complicated cases. The agencies are not always

5going to prove both dualities, that there could be a change

6in market conditions and that the distribution system is

7essential causally to keeping an out.

8So, here's another one with a slide. The ultimate

9remedy chosen in Microsoft was to require divestiture of

10all applications, including the browser and Microsoft office.

11This was not on Richard's, Carl's or Dan's watch.

12This one is on my watch. And I have to say, I had to put

13up this slide. There slide -- actually there is a long

14history of this particular slide. When Dennis Yao, who

15was my roommate in high school, was a Commissioner in the

16FTC in 1989 or 1990, called me and said, you know, we

17figured out we don't want to go after IBM and Microsoft

18together, should we go after Microsoft. And the metaphor

19immediately leapt to my mind, you're going to be like a

20dog that's chasing a fire truck, you know, they're rolling

21down a little street, noisy, illegal as hell,

22anticompetitive as hell, but what are you going to do with

23it when you catch it?

24As it worked out, they didn't catch Microsoft.

25I did. And the dog in this picture turned out in actual


1history to be me. What did we get? Not any remedy which

2changed the conditions of competition. Ultimately, there

3was an entirely ineffectual settlement in the United States

4and a mildly effectual settlement in the EU. Certainly not

5enough remotely to have the kind of competitive conditions

6change that was possible from the widespread use of the


8So, there's another problem with the agencies,

9bringing large, complicated antitrust cases. The

10counter example here would be, of course, U.S. v.

11AT&T. The United States was incredibly well served by

12that case. During the long interval between the AT&T

13breakup and the soon-to-happen reestablishment of the

14Bell System, we were incredibly well served to have

15vertical disintegration in telephony. The fact that we

16had vertical disintegration in telephony at the moment in

17history when, for example, technologists finally figured

18out how to have mass market use of online services. That

19was incredibly fortunate and that resulted from the antitrust

20case. But they can also fizzle. And even if you win a

21case, there can be severe problems in finding a remedy

22that the antitrust system will undertake.

23So, let me go to my bottom line. I really want us

24to turn around. These cases are going to be hard to prove and

25I want us to turn around and think about both the potential


1for a competitive effect, meaning there could be change in

2the conditions of competition. The form of that change was

3different with the two cases I talked about. Second, think

4about a causal link between the alleged act and monopoly. I

5would bring those to the fore. Those would be my framework

6for thinking about a Section 2 case.

7But of course that discussion is only about the

8question of whether there is an antitrust case. This doesn't

9remove from the agencies or any other plaintiff, but particularly

10not for the agencies, the problem of thinking about whether

11there's enough of a harm to competition at stake to justify

12any intervention. I guess I would say that in an cases like

13AT&T or Microsoft, where you've got a substantial impediment

14to technical progress in an infrastructure industry, that matters

15to the whole economy, arising from the lack of competition.

16That one might get you over the hump. But there are other

17metrics that can be used, such as the size of the differnce

18between the two competitive regimes and the importance to


20And also to think through whether there might be

21an efficiency defense, whether there might be more harm

22than good done by the antitrust intervention. I don't

23want to take that away, but I do want to say that I would

24emphasize -- I would emphasize thinking through whether

25there is an antitrust case in a perfectly ordinary


1antitrust analytical way, competitive effects and


3Thank you very much.

4MS. GRIMM: Thank you very much.

5Our next speaker is Professor Rich Gilbert, who

6is Professor of Economics of the University of California

7at Berkeley.

8From 1993 to 1995, he was Deputy Assistant

9Attorney General in the Antitrust Division of the U.S.

10Department of Justice, where he led the efforts that

11developed joint Department of Justice and Federal Trade

12Commission "Antitrust Guidelines for the Licensing of

13Intellectual Property."

14Professor Gilbert has served as an Associate

15Editor of the "The Journal of Industrial Economics," "The

16Journal of Economic Theory," and "The Review of Industrial


18Professor Gilbert research specialties include

19antitrust economics, intellectual property, and research

20and development.

21He earned his Ph.D. from Stanford University in

221976. He received a Bachelor of Science degree in

23Electrical Engineering in 1966 and a Master of Science

24degree in 1967 both, from Cornell university.

25Professor Gilbert.


1MR. GILBERT: Thank you very much, Karen.

2While I figure out how to find my talk here, I

3will thank you for bringing these hearings to Berkeley.

4We're very glad we could be able to host these hearings.

5And here we go.

6I'm going to talk about a very narrow slice of

7conduct that could invoke Section 2 liability, namely

8innovation or product design, and ask the question of

9whether innovation, certain types of innovations can be a

10source of Section 2 or contribute to Section 2 liability.

11Now, I don't think many people would argue that

12innovation is great for the economy. Nevertheless, there

13are quite a number of cases that have alleged that

14innovation or product design has contributed to

15monopolization. Of course, Microsoft, as we just heard,

16is one. A slew of cases involving IBM and standardization

17for complimentary products, the use of complimentary

18products. There are some interesting cases on the horizon

19in the prescription drug industry that raise innovation

20issues in a Section 2 sort of context.

21So, I'm going to be reviewing some of these

22cases and asking whether we could have a standard, we've

23heard a lot about standards this morning to evaluate

24Section 2 type conduct, whether any of these standards is

25useful for evaluating innovation. Maybe I will give you


1my punch line right away. I think the answer is no, and

2try to tell you why.

3I'll begin -- let's see. I'm going to begin

4with a very simple model. I hope not to raise the fear

5factor too much and talk about letters here. If you are

6worried about this, you can replace any letters with

7numbers. So, I want to talk about a very simple model of


9I have here an old technology. It has a social

10value, v, zero, for each use. You could use, say, fifty

11dollars for v, zero. A new technology could come along

12with a higher social value, maybe a hundred dollars, for

13each use. I'm going to strip away marginal cost to keep

14things as simple as possible. There are a bunch of users,

15say there's a thousand users, if you want. And there's

16some R&D costs.

17Now, in this simple model, the innovation is

18socially desirable, I mean, it's still the small one can

19be as simple as possible if the total incremental social

20value exceeds the cost of the innovation.

21So, we have our thousand consumers and they each

22use this technology in one application, the extra value of

23the innovation is fifty dollars, so that would be fifty

24thousand dollars. The question is: Does that cover the



1Now, in terms of whether the innovation is

2privately profitable, there's a price that the innovator

3can collect for the new technology and it's profitable if

4the price it can collect times the number of people who

5buy it, assuming they all buy it, in fact covers the cost.

6So, the first that I want to make, and there is

7a paper that should be coming out in "Competition Policy

8International" on this topic, the first point is to say,

9innovations can be socially desirable but not privately

10profitable, or you can have innovations that are privately

11profitable but not socially desirable.

12So, the first point is a very simple point:

13That innovation can go any way -- there can be any order

14in evaluating social and private profitableness. It's not

15like a price -- innovation is like a price change in some

16respects. If you come out with an innovation for a

17product, it's like reducing its quality-adjusted price,

18and you can make an analogy between innovation and, say,

19predatory pricing. If you reduce the quality-adjusted

20price, that leads to the exit of competitor, and then you

21raise your price again, that has a sort of predatory

22flavor to it.

23But unlike pricing, where lower price certainly

24lowers the price above marginal cost is a good thing, we

25really don't know if more or less innovation is a good


1thing unless you do the whole analysis.

2So, the standards I want to talk about, these

3came up this morning, I want to talk about different rules

4of reason which I interpret as either a total rule of

5reason, which looks at all of the economic value

6associated with some conduct, whether it's value to

7consumers or value to producers.

8And then there's probably the more popular

9consumer rule of reason analysis which focuses on

10consumers, and some people would say is at the heart of

11antitrust analysis, at least according to, say, Steve

12Sala, although others such as Joe Farrell and Mike Katz,

13and Ken Hirers from the antitrust division, have advocated

14a total rule of reason standard.

15Then there's the profit sacrifice test in one of

16its many forms. There's the no economic sense test.

17We've heard a little bit about that this morning. And

18then I'll talk a little bit about sham innovation.

19So, a total real of reason analysis, in a sense

20it's the right thing to do if you are, sort of by

21definition, an economist, it's the right thing to do

22because it ask whether total surplus is increased from

23some activity. And even if that makes producers

24relatively better off than consumers, at least there's the

25possibility that those producer profits will flow


1eventually to consumer benefit, or that somehow producers

2can bribe consumers to get it all right.

3But the problem of course is that you can have

4the price being either larger or smaller than the

5incremental social benefit. And all of the analysis would

6have to be done when the innovation decisions from the

7perspective of the decisions that are actually made, which

8means what we call an ex ante analysis. And this really I

9think sets up innovation as being distinctly different

10from other conduct. Because when you talk about

11innovation, it's absolutely necessary to keep going

12backwards and backwards to what are the incentive effects

13of whatever rules or policies you have in place, what are

14their incentive effects for innovation in the first place.

15And now it's easy to say, well, of course that's

16right, of course we're going to take that into account.

17But I want to ask you, if you have been in these hearings,

18how many times have people really gone backwards and said,

19what are the implications of what we're doing for the

20kinds of decisions that people are make that could have

21developed and could develop new products or new processes

22or whatever ten years from now. And I would say you

23haven't heard it very many times.

24So, it very easy to lose sight of these

25incentive effects. And on top of that, if you did a total


1rule of reason analysis, the analysis that you would have

2to do is hugely complex. You have to really take into

3account all spillovers, how innovation affects consumers

4and firms in other industries, and those we know can be

5very, very large. And with the complexity, you can lead

6easily to false positive and false negatives. I'm not

7going to say Type 1 and Type 2 because I always forget

8which one is which, so I will just say false positives and

9false negatives, and you can figure out which one is a

10positive and which one is a negative on your own.

11Too much enforcement or too little enforcement.

12Portion. It can go either way.

13A consumer rule of reason analysis. Again, it's

14very complex. The problems are similar to those that

15arise in a total rule of reason analysis. Again, the

16ex ante problems, the uncertainties, the spillover

17effects, etc. And as well can lead to conclusions that

18just simply don't make sense. This is particularly a

19problem in innovation. You could have an innovation that

20just saves millions of dollars in production cost, but

21maybe it leads to a nickel increase in price, which

22certainly could happen. And would you want to say that

23this is an anticompetitive innovation because consumers

24are slightly worse off, despite the fact that it's

25generated enormous savings and efficiencies on the


1producer's side.

2Well, I know that people can differ on that, but

3my view is that it just doesn't make any sense to discount

4all of those efficiencies. Now, you can say that you're

5looking at a merger case or you're looking at other

6conduct that doesn't involve product design, that those

7kinds of efficiencies are not likely to be huge or have

8not been demonstrated to be huge, but when you're talking

9directly about innovations these efficiencies exist as

10part of the innovation. So, you can't discount them.

11A profit sacrifice test. There are, of course,

12different versions of a profit sacrifice test. And I'm

13going to quote Janusz Ordover's and Bobby Willig's

14definition: "Predatory intentions are present if a

15practice would be unprofitable without the exit that it

16causes but profitable with the exit." Now, Ordover and

17Willig also say this is just talking about predatory

18intent not facts, they add a lot of other conditions in

19their analysis that make this analysis considerably more

20elaborate, and in many ways closer to a total rule of

21reason analysis. So, this is just the basic idea of a

22profit sacrifice test.

23Now, the profit sacrifice test, I am not the

24first to say this, it doesn't seem to me to make any sense

25to innovation, even though it was in fact developed


1originally to talk about innovation as well as price --

2predatory pricing. The problem of course first of all is

3that innovation almost always involves a profit sacrifice.

4It's called investing in research and development. That's

5what you do.

6It's also the case that innovation, if it really

7works, probably excludes competitors. So, exclusion is

8sometimes a direct result of producing a really good

9mousetrap. The other mousetraps can't compete.

10Now -- and furthermore, and this is absolutely

11crucial, is that we need to know how much market power

12after the innovation occurs is necessary to justify the

13investment in innovation in the first place. And you can

14make statements about whether innovation creates too much

15or too little market power relative to its social value.

16But the social value is very hard to calculate. And the

17amount of power or pricing power that is necessary to

18evoke the right amount of investment in research and

19development is simply a very hard question. So, I

20conclude, based on this, that a profit sacrifice test

21really doesn't do very much to inform this analysis.

22What about a no economic sense test. I am going

23to use Greg Werden's version of this. He says: "Conduct

24is not exclusionary or predatory unless it would make no

25economic sense for the defendant but for the tendency to


1eliminate or lessen competition."

2Now, you can see that, with all the negatives

3again, the no economic sense test is really a test of the

4absence of predation. So, if it makes sense to do this

5activity, then it's not predatory.

6Now, although it's not really clear in the no

7economic sense test what no economic sense means, there

8are two interpretations of this, certainly as applied to

9innovation. One is that it's not profitable. No

10reasonable firm would have dumped all of this money into a

11new product design unless it had a purpose of excluding

12competition. A second interpretation is that innovation

13really always makes economic sense because it's just a

14good thing that firms do.

15Depending upon which one of these

16interpretations you have, if it's the first one, then the

17no economic sense test is very similar to the profit

18sacrifice test. Now, if it's the second one, the no

19economic sense test is similar to really whether

20innovation is a sham, meaning whether it's a fraud or not.

21I think it's the case, and I know that Werden has said

22that his view of the no economic sense test as applied to

23innovation is the second version, not the first version.

24And I also know that he has views of conduct that do not

25in fact involve a profit sacrifice, even though there was


1some discussion this morning that they're the same. His

2example was a world with no arson laws and flush with

3matches. So, you can go out there and burn down anybody

4you want, including your competitors.

5So, let me review a little bit of some cases

6involving predatory innovation, particularly with respect

7to complimentary products, products that interact with

8other products. Those are almost totally -- well, they're

9not entirely, but to a great extent they have to do with

10changes to the interface stands. That was certainly the

11case with the IBM peripherals litigation, a bunch of these

12in the late 1970s, whether it's other people's disk drives

13would hook into and work with IBM mainframe computers.

14And the Microsoft case. And there's been a few others.

15As a general conclusion in looking through these

16cases, well, you can find a lot of lower court decisions,

17a general conclusion is that in nearly all of these cases,

18weak evidence of efficiencies was sufficient to avoid

19liability for predatory innovation.

20So, after there was lots of talk about whether

21there was monopoly power or not, or whether or not there

22was a monopoly of market effect, competitive effect. The

23final analysis, they said -- these courts generally said,

24well, we can think of it as an efficiency reason for this

25conduct, therefore it's okay.


1To my knowledge, only the Microsoft case,

2Microsoft 4 as it is sometimes affectionately called,

3purported to apply a rule of reason analysis to

4innovation. So, let's talk about Microsoft a little bit.

5The Microsoft case actually came up with a road

6map to kind of evaluate innovation. There actually five

7steps to the road map. I'm going to condense them to


9The plaintiff first must demonstrate that the

10conduct that harmed consumers had an economic

11anticompetitive effect. Second, if a plaintiff

12successfully demonstrates anticompetitive effect, then the

13monopolist may prefer a procompetitive justification for

14its conduct. So, the second step is the monopolist,

15alleged monopolist can talk facts and say, we have a

16reason for doing this. And then the third step says,

17well, the plaintiff can now come back and rebut the

18monopolist's justification. Or, if it can't actually

19justifiably rebut it, it can demonstrate that the

20anticompetitive effect was bigger than the procompetitive

21benefit and outweighs it. So, it can do a rule of reason

22analysis is what it says.

23Well, let me just review what happened in the

24Microsoft case. There were many allegations having to do

25with Java standards and with various contracting policies


1with lots of different players in the industry.

2There were really three design elements that

3were challenged in the Microsoft case. One was not having

4Internet Explorer in the Add/Remove programs utility. The

5other was designing Windows so as in certain circumstances

6to override the user's choice of a default browser other

7than Internet Explorer. And the third one was commingling

8browser and operating system code.

9Now, interestingly, the court concluded that

10Microsoft offered no procompetitive justifications for the

11first and the third, and these were held by the court to

12contribute to the Section 2 violation. But then the court

13also concluded that plaintiffs -- that Microsoft did offer

14a justification for the second element of its conduct,

15that is, the overriding of user's choice of the browser,

16which the plaintiff did not rebut, and therefore in fact

17the Microsoft court never got to the third step. So, the

18court never got to the rule of reason balancing in the

19third step because either it was anticompetitive with no

20efficiencies or there were efficiencies and the plaintiff

21didn't come back. So, maybe Tim will explain or Dan will

22elaborate on this, but this is my reading of what happened

23with the court.

24So, the practical effect of what happened in the

25Microsoft court's analysis was really, I think, similar to


1the no economic sense test of the first variety. That is,

2was there some reason for this conduct. If there was,

3it's okay.

4Now I want to turn to another area that I find

5quite interesting. As they say, this is emerging

6antitrust. This is that drug patents may delay generic

7competition. So, the innovation that contributes to these

8drug patents can have competitive effect. It can have

9competitive effect both through the nature of generic

10substitution and also because of the specific elements of

11the Hatch-Waxman Act, which impose a 30-month stay on

12generic competition if you have a patent.

13So, one of these cases is called Tricor, which

14is actually a drug called phenofibrate. It's used to

15control triglyceride and cholesterol levels. And I should

16acknowledge I have been a consultant in this case. A

17second case is Prilosec and Nexium, which Prilosec is a

18common drug prescribed for heartburn, gastric reflux, and

19then your more serious conditions like esophageal and

20duodenal ulcers. It turns out that Nexium is what is

21called an isomer of the chemical that's in Prilosec. It's

22basically the same molecules. It's been rearranged a

23little bit. And it's supposed to have some advantages for

24the esophageal and duodenal ulcers, but not for heartburn.

25The allegations that came up in both of these


1cases is that the innovations are costly but minor

2improvements, that they're contrary to the intent of the

3Hatch-Waxman legislation to promote generic competition,

4and they have large adverse competitive effects by

5delaying generic competition.

6 Now, I think certainly if you just take a

7snapshot of competition, once these drugs exist, anything

8that delays generic competition has at lease the

9possibility of a competitive effect. But it's important

10to recognize that the Hatch-Waxman legislation was a trade

11off between more generic competition and more protection

12for patented drugs. In fact, the first three letters of

13the Hatch-Waxman Act are patent term restorations. I

14think it was designed to protect pioneer drugs, as well as

15promote generic competition.

16Product line extensions certainly increase

17incentives for drug innovation. If you actually look at

18the respective patent terms for prescription drugs,

19patented prescription drugs, it's actually quite short.

20It's one of the shortest of all industries because of all

21the FDA delays and regulations required to actually

22produce the drugs. And it's very hard to assess these

23benefits from these innovations.

24So, I think instead of looking at any of these

25standards to inform a Section 2 analysis for innovation, I


1find all of them seriously lacking. I think instead you

2can turn to consistency with other rules.

3And let's talk about something we've heard

4before, talk about the process rather than the outcome.

5That was discussion was featured in this morning's session

6to great extent by members of the panel talking about the

7process rather than the outcome.

8So, the quote that I'm quoting here is by a

9distinguished economist, but not anyone from our group.

10It's from an economist who works for the Oakland Athletics

11who was quoted by Michael Lewis in "Moneyball," and he was

12actually talking about how to hire baseball players, but I

13think his insight here is equally applicable to antitrust

14policy, "We have to look at process, not outcomes."

15So, if we think about making an analogy between

16innovation effects, and the effects and rules that are

17applied to other conduct, I want to argue that, in many

18innovation cases, the effects of the innovation are very

19similar to the effects of a unilateral refusal to deal.

20When you're talking about, say, if IBM refuses to make

21mainframes compatible able with third parties' components,

22it's a lot like saying, well, one day Microsoft gets up

23and says, I don't want to work with these third party

24people anymore, I want to build computers just for myself.

25Microsoft refuses to make Windows compatible with other


1browsers. Or a generic drug manufacturer refuses to

2supply a drug that generics can copy.

3In effect, this conduct looks a lot like a

4unilateral refusal to deal. Now, these days, after

5"Verizon v. Trinko", seems like unilateral refusals to

6deal have a long way to go before they can generate

7antitrust liability.

8Now, I don't want to state that as a categorical

9fact, or that "Verizon v. Trinko," that all the words in

10"Verizon v. Trinko" were necessarily the greatest words

11that have ever been uttered in all of antitrust policy. I

12am not sure it's the greatest policy.

13But my only point is that if you are going to

14have a policy that gives considerable deference to a

15decision by a single firm about who that firm will deal

16with or supply, it just seems odd that one wouldn't have a

17more strict policy, more intervention policy with respect

18to innovations that have very similar effects.

19So, I'm not saying -- again I want to emphasize

20that I'm not saying that we should have policies that say

21that unilateral refusals to deal with per se legal, I

22don't think that's necessarily the right thing. But if we

23are going to have such a policy, then consistency seems to

24say that if you unilateral innovations that have similar

25effects should not be treated more severely.


1So, one of my conclusions here is that all of

2the rule of reason and profit sacrifice tests have limited

3value to evaluate what is sometimes called predatory

4innovation. It's hard to do; likely to get the wrong

5answer; very hard to look al at the incentive effects that

6are necessary to really thinking about innovation.

7The no economic sense test is better, but only

8if it's interpreted as a test of sham innovation because

9otherwise it comes out just like or very similar to a

10profit sacrifice test.

11And my other conclusion is that this is what

12courts in fact almost always have done with very few

13exceptions in the way they've treated these cases and it's

14probably as reasonable an approach as any.

15MS. GRIMM: Our third presenter this afternoon

16is Daniel Rubinfeld, who is the Robert L. Bridges

17Professor of Law and Professor of Economics at the

18University of California at Berkeley, where he has taught

19since 1983

20He has also served as Deputy Assistant Attorney

21General for Antitrust in the U.S. Department of Justice,

22as well as in various capacities with the President's

23Council of Economic Advisors, the National Academy of

24Sciences, the Urban Institute, and the National Bureau of

25Economic Research.


1Professor Rubinfeld's major books include

2"Econometric Models and Economic Forecasts" and

3"Microeconomics." Recent publications include, "Antitrust

4Enforcement in Dynamic Network Industries" in "The

5Antitrust Bulletin," 1998; and "Empirical Methods in

6Antitrust: Review and Evidence" in "American Law and

7Economics Review."

8He is President of the American Law and

9Economics Association.

10Professor Rubinfeld received his B.A. from

11Princeton University in 1967; his M.S. and Ph.D. from the

12Massachusetts Institute of Technology.


14MR. RUBINFELD: Thanks very much. I really,

15like everyone else, appreciate the opportunity to appear

16before you today. It's been about eight or nine years

17since I left the Antitrust Division and I guess,

18understandably I've aged about eight or nine years during

19that time, and I find as one gets older one tends to

20reflect back on the past, perhaps more than one should.

21But what I'm going to do in my comments today is to really

22do some reflection on what happened, and I might hit on

23some of the previous commentators' issues, but see I can do it in

24a way that will be constructive for the agencies as you

25think about forming your policies.


1So, the first point I want to make is why I

2think it's really important to have an active Section 2

3jurisprudence. And I want to look back and talk about the

4legacy of "U.S. vs. Microsoft" for antitrust enforcement.

5And, finally, I want to look at bundling and talk about

6the legacy of "LePage's vs. 3M".

7I should say, to make it clear, that I had an

8interest in both of those cases. I helped to prosecute

9the Microsoft case. And I have consulted for 3M with

10respect to some of the issues that arose in its appellate

11case. I was not involved in the LePage's case itself, but

12I was involved in thinking about some of the appellate

13issues. So, I have taken a pretty close look at the Third

14Circuit opinion in that case.

15If you're interested in some of the deeper

16comments I am going to give today, they will appear in

17two articles. One is an article that Doug Melamed and

18myself are completing for our forthcoming volume in

19which we are looking at the lessons of the Microsoft case.

20And the second is an article I published a year or so ago,

21looking at the bundling in the "LePaige's vs. 3M" case.

22Before I go on to the cases, as far as the active

23Section 2 jurisprudence is concerned, I guess history

24affects how one views things, and I can be very quick, I

25can just say, having been involved in actually bringing


1both Microsoft and Dentsply, both of which I thought was

2the right thing to do, and the D.C. Circuit and the Third

3Circuit in both cases have written opinions that were

4supportive of that decision, I'm proud to have been

5involved in both of those cases, and I think that shows,

6consistent with what Tim Bresnahan said, it shows the kind

7of active Section 2 jurisprudence that I think makes


9Both cases had a particular set of facts

10associated with them that told a story that made them the

11right cases to bring, viable cases. And I think the

12agencies need to be careful because there is not going to

13be a lot of good Section 2 cases. So, you need to be

14careful and active and watchful for the appropriate

15opportunities in the future.

16So, having said that, let me go on and take a

17look at "U.S. vs. Microsoft". And I am going focus now

18really on sort of what we've learned from the case in a

19very broad perspective. I'm not going to try to go into

20some of the technical details unless we have discussion


22It's sometimes easy to forget, since this is

23almost ten years ago when at least my version of Microsoft

24was brought, that people were barely talking about network

25effects. Now it's taken for granted that in high tech


1it's common to face industries in which network effects

2matters and that enters into the economics and to the law,

3legal thinking about the cases.

4I see one of the legacies of Microsoft is sort

5of helping to bring us from the pre-network effect world

6to a world where network effects are often the core of the


8Next important is people are thinking somewhat

9differently now than they were before about barriers to

10entry. When we originally think about investigating the

11Microsoft case, obviously barriers to entry was something

12that I paid a lot of attention to. We became convinced

13that there was a significant barrier to entry, but it's

14not the usual one you might imagine. It had to do with

15the fact that in order to have a successful operating

16system, you really needed to have successful applications.

17There was what we called a two-level entry problem. And

18we spent a lot of time developing the underlying economics

19that describe this applications barrier to entry.

20One of the things that people forget, actually I

21almost forget myself, is that the term "application

22barrier to entry" did not exist, at least to my

23knowledge, prior to our work. We coined and reiterated it

24every time we could at trial until the judge finally got it

25into his mind.


1And it was fun to watch the trial, by the way,

2because at the beginning of the trial, Microsoft disavowed

3the application "barrier to entry." By the end of the

4trial it was being discussed by them as if it were a

5common coin of the realm.

6So, let's remember that that was one, for better

7or worse, I think for better, one of the legacies of the

8Microsoft case.

9The other thing is, as you all know, the case

10involved tying, but it was different than the classic kind

11of tying case, which is usually thought of leveraging

12market power from a market where a firm has substantial

13market power to use some related power where it does not

14necessarily have significant market power.

15But this case did involve tying as well as

16bundling. And it was a non-leveraged form of tying. And

17now it's not, I think unusual to think about bundling in

18that context in certain cases where it was probably quite

19radical at the time.

20The other thing is that the case brought to our

21mind a different way, a different perspective of thinking

22about market definitions. As Tim suggested earlier today,

23there's always been a lot of talk about Schumpeterian

24competition and certainly the agencies have been aware of

25it for a long time.


1In this case, to one degree or another,

2Schumpeterian competition really came to the forefront

3because, in the debate about market definition and market

4power, Microsoft took the position that it was the threat

5of entry by competitors that really not only restrained

6this market definition, this market power, but also in

7fact meant that the market should be defined very broadly.

8Microsoft argued for an extremely broad market definition

9that included almost all operating systems, from

10hand-helds pretty much up through mainframe computers, and

11argued that it had no market power over that relevant


13I still remember one particular trial exhibit

14which Microsoft presented which sort of brought this issue

15to the front. And the exhibit said that Microsoft faces

16substantial competition from known and unknown

17competition. And my view, which was borne out, by the

18way, by the Circuit Court opinion, is that when you have

19to defend your market power or lack of it by describing

20competition that no one knows about yet, you really have a

21fairly weak position.

22And if you read the D.C. Circuit opinion, I

23think the D.C. Circuit got it right, as they did in most

24areas, they said, the nascent competition really could be

25important but it really has to be competition which is


1expected with reasonable certainty to actually be there in

2the marketplace at some period in the future, thinking

3about two years would be the relevant time period.

4But the fact that someone might come along and

5take away your market power isn't sufficient. I think the

6court was pretty clear about that. And it's basically the

7right place to be.

8As far as legal issues I see coming out of the

9case, there are about five. I'd like to highlight, again,

10without getting into the technical/legal side of the case,

11the first thing which I think we now take for granted, or

12at least I hope we do, which is that the same antitrust

13principles apply in dynamic high tech industries as apply

14in the other industries. The application of course might

15be somewhat different, but the principals are the same.

16And I quote Judge Posner, who really says what I

17have in mind, which is that antitrust doctrine really is

18pretty well situated to allow us to handle high tech

19industries. We don't need to rewrite Section 2, in my


21Up until Rich started speaking earlier, I would

22have said hardly anyone remembers that there are IP issues

23raised in Microsoft. Rich laid them out pretty well.

24And so, what I wanted to say is that the court

25makes it pretty clear that the same general antitrust


1principles that apply to conduct involving intellectual

2property that apply to any other form of property underc

3the antitrust laws.

4Originally, at one point in the case, Microsoft

5actually claimed that their IP rights covered the entire

6desktop, at least with respect to the first boot up of

7their operating system. The court made it very clear that

8(a) that was too expansive an interpretation, and (b) that

9it was appropriate for the Sherman Act and the courts to

10really look at the IP issues. You did not get a free ride

11just because you did in fact have some legitimate

12intellectual property.

13And Rich described in detail and correctly where

14the court finally came out about these specific IP issues.

15With respect to product design, as I interpret

16the court opinion, it makes clear that the court is going

17to give pretty wide deference to firms that are designing

18new products, along the lines Rich described. But the

19court also said this is an area that's open for viable

20investigation. And where particular aspects of

21Microsoft's product design excluded rivals, the court did

22shift the burden to Microsoft to establish a

23procompetitive justification for the design. There is no

24safe harbor just because you're involved in innovation or

25product design. And the removal of the Add/Remove utility


1which Rich described was one good example of that. The

2court was very clear that was problematic and there was no

3procompetitive justification given that I can see in the


5There's also an issue in this kind of Section 2

6case as to whether you ought to kind of just describe the

7case with kind of a broad brush or kind of go into the

8practices with fine detail. My sense, my personal sense

9during the trial was that there were times when the

10defense seemed to say, we want to just talk very broadly

11about the rights of a dominant firm to engage in certain

12kinds of potentially procompetitive activities. And the

13government, as I saw it, focused really in with apparent

14detail about the details surrounding each of these kinds

15of conduct.

16And I read the D.C. Circuit as basically saying

17that any aspect, the explicit, discrete aspect of

18monopolist conduct that tends to exclude rivals may be

19illegal, unless there's a legitimate procompetitive

20justification for that particular conduct.

21So, there is at least a burden-shifting aspect

22to some of the illegal rules that flow from the Microsoft

23case, which I think is appropriate.

24There is an issue about whether you ought to

25focus on rules or cases, specific facts. Here, as you


1know, the court, the appellate court, on the time claims,

2suggested that the per se rule didn't apply because of the

3particular attributes of platform software. So, we're now

4left in a somewhat unclear world that may apply mostly to

5Section 1, but also has Section 2 implications as to how

6to treat tying.

7And I have to say here, as an economist, you may

8not be surprised to hear that I'm pretty sympathetic with

9the comments of the court. I think it's really hard to,

10as an economist, come up with per se rules that would

11apply in this kind of high tech context.

12Of course we don't know quite where that would

13have ended up because the Department of Justice chose not

14to appeal that part of the D.C. Circuit's ruling.

15With respect to causation, I see the case telling

16us conduct that violates the antitrust laws only if it

17injures competition. Causation can be inferred when

18exclusionary conduct is aimed at producers of nascent

19competitive technologies, as well as when it's aimed at

20producers of established substitutes.

21So, basically the court spelled out causation along

22the lines Tim suggested, and I think the court makes it

23pretty clear that that's necessary and that the government

24succeeded in that effort.

25What about profit sacrifice? Here we could


1debate exactly how to characterize the case. I would say

2that the case we put forward did really involve a profit

3sacrifice test. My definition would be that conduct is

4anticompetitive when it would not make business sense for

5the defendant but for its tendency to exclude rivals and

6create or maintain market power for the defendant.

7This is kind of a crude paraphrase. If you go

8back and read the details of the case, you'll see a more

9formal definition. It is a variant on a profit sacrifice

10test. I wouldn't say it's quite a no nonsense test, but

11it's pretty close.

12Now, that's not what the D.C. Circuit said.

13What the D.C. Circuit said was quite close to what Rich

14Gilbert said earlier. The court said that the conduct is

15anticompetitive if it harms the competitive process and

16either it's not shown to further efficiency or to have

17some other procompetitive justification or the

18anticompetitive harm outweighs its procompetitive benefit.

19So, the D.C. Circuit was suggesting more of a balancing

20test than a profit sacrifice test.

21And this leaves us with the question of what we

22should do if we find Section 2 type conduct that harms

23competition and furthers a legitimate purpose should we

24have a balancing test.

25Now, I should say here, I am not entirely sure


1of where I would end up, but I lean strongly towards the

2profit sacrifice test, at least in most cases, because I

3think it's easier to operationalize. We could debate

4about how to exactly operationalize it, but I think Tim

5suggested that, in most of these cases, it's just not

6possible to sit down and do a fully complete balancing

7rule of reason analysis. We don't have the time or the

8information available. And the cost, by the way,

9including the cost to the parties, would be tremendous.

10And I think in most situations, a profit sacrifice test

11would get us to the right place. I think you can try to

12find some counter-examples, but I think you have to work

13hard to do it. So, I am on the side of the folks who

14think we ought to just refine the profit sacrifice test.

15Okay, let me switch to my other case of

16interest, "LePage's vs. 3M". You have heard about it a

17little bit already. This was the case involving bundled

18rebates offered by 3M in the market for transparent tape.

193M was facing substantial competition from LePage's, not a

20new entrant, but an entrant that had become very

21successful in the production and sale of private label


23 And the question was: Were 3M's programs,

24specific bundling programs, anticompetitive and a

25violation of Section 2.


1Now, here I'm very critical of the Third Circuit

2opinion generally for two reasons. One is that the

3opinion itself does not, in my mind, in any way provide

4any clear guidance as to how firms ought to behalf when

5they do have a dominant position and they are deciding

6what kind of business practice to engage in. And I think

7any clear legal rule ought to do so.

8And, secondly, I actually think that I have been

9unable to come up with what I think is any coherent theory

10of predation or any Section 2 theory which fits the facts

11of the 3M case. In my view, the Third Circuit was a

12little bit loose in how they actually borrowed and used

13facts of the case. I actually went back and read most of

14the record in the LePage's case and I cannot find a theory

15that I find coherent that actually fits the facts of the


17And the thing to remember is that bundling

18itself of course is quite ubiquitous and often is

19procompetitive. So, if we generate a legal rule, we want

20someone else to define those relatively few cases where

21bundling is a problem and distinguish it from the majority

22of cases where bundling is procompetitive. So, we're

23looking for those particular situations. Lack of clarity

24is a problem.

25Let me briefly take a few minutes and just very


1quickly tell you about 3M's programs. There were a whole

2bunch of programs being attacked, but the two that

3involved bundled rebates were, first, the executive growth

4fund program. And the thing that's key about this program

5was it was actually I think a one-year program and it was

6a pilot program for a small number of customers.

7Now, what it did do was it set up growth targets

8for six different errant divisions of 3M, which would

9cover a lot of office supply products. And firms actually

10had to meet target goals in each of these divisions.

11Now, my view is that the executive growth fund

12program -- let me be clear that this is my view and not

133M's view. My view is that, had this program been

14expansive and had it covered all customers rather than

15just a few, and had it continued for a number of years, it

16could well have been an anticompetitive program. I don't

17think it was because it was too narrow. It had no ability

18really to substantially exclude competitors because many

19of the key competitors, Walmart being the most important,

20were not covered by this program. But it had the

21potential if it continued to actually be restrictive

22because of the specific design of the program.

23But for various reasons, which I think relate

24partly to the demands of some customers, including

25Walmart, 3M changed its program to a partnership growth


1program, and this program did involve discounts in six

2different areas, but there were no specific targets to

3reach in each of the areas. Basically you got a rebate

4based on the aggregate of all your purchases in all six

5categories. So, this amounted to a somewhat complex

6discount program, volume discount program.

7And my view is that the PGF program, as it's

8called, was not anticompetitive, even though the court

9felt otherwise.

10So, if you go back and look at the LePage's

11trial and ask -- take a look at the trial and ask if the

12trial helps to support some of those theories of

13competition, I would say no. I didn't see any testimony

14in the record about economies of scale or scope, which

15would be important, particularly to get at the issue of

16whether LePage's or any other competitor would remain

17viable in the face of these practices.

18There was no predatory pricing claim.

19Plaintiffs agreed that LePage's was pricing above cost.

20In fact, by my calculations, even if you took all of the

21discount programs at 3M, no matter what the products were,

22attribute all the discounts to tape, it would still be

23pricing above cost.

24 I didn't see anything about profit sacrifice

25that I could infer from the opinion. So, there was


1nothing that fit my particular interest in pursuing these

2kinds of Section 2 cases.

3There was no time claim at all. It was a

4bundling case, not a tying case. There was also no showing

5of market power with respect to any product other than

6transparent tape. So, the kind of leveraging theory you

7might expect to see in a time case was not present either.

8Now, the jury did find, interestingly, no

9exclusion under Section 1, but they did find a violation

10under Section 2. So, this leaves me with a puzzle of what

11the legacy is of "LePage's vs. 3M". I think for a while

12the Commission may have though this case was unusual, but

13it's pretty clear now that the Third Circuit opinion has,

14let's say, encouraged a lot of litigation surrounding

15these kinds of practices.

16So, I went back and asked myself, what should

17the principles be here. And I would say, speaking very

18broadly, if the rebates associated with bundling reduce

19consumer welfare by impairing rivals' ability to make

20competitive offers to potential customers, that's going to

21be something generally that's going to give me concern. I

22am not going to say it's necessarily anticompetitive, but

23that would give me great pause.

24And that general rule takes into account

25efficiencies and allows price increases by firms, as long


1as they don't impair rivals' ability to compete. But that

2general rule is really not very helpful from a process

3point of view. It's really too broad to make applicable.

4So, I would say the following. I'd say, there

5are conditions under which one may be anticompetitive, but

6none of them fit LePage's.

7And, just quickly, because I think we're running

8out of time, here's some examples of situations in which I

9think bundling might be anticompetitive, none of which

10fits the LePage's case.

11The first would be traditional contractual tying

12of the kind that we saw in Jefferson Parish. The second

13would be predation through profit sacrifice of the kind

14where bundling was used in the form it was in the

15Microsoft case, and perhaps I'd include Dentsply there as

16well. The third might be monopoly maintenance through the

17creation of barriers to entry, which is, at least my

18interpretation of "SmithKline versus Eli Lilly," a case I

19was not involved in, where at least the court stated that

20the sale of monopoly products were used to harm

21competition in a non-monopoly market.

22Now, where does this leave us? We need a

23workable test. I wish I could come here and tell you I

24figured out what that test is. I have read many papers

25written by folks in the agencies and elsewhere suggesting


1various tests.

2I still not have seen one that I am entirely

3happy with, but a couple things strike me as important

4when and if we get such a test. One is that, weakening a

5rival should not be sufficient to condemn a monopolist,

6otherwise we will be discouraging firms from innovating

7and growing and being successful, which I think would be

8harmful to our competitive process.

9Secondly, while it would be very nice to have an

10incremental cost benefit test for certain kinds of

11bundling, there are a lot of difficulties in putting that

12test into play that I won't bore you with here. So, we

13have more work to do there.

14Third, we might say that for a bundled rebate

15program to be anticompetitive, it at least necessarily

16ought to be the case that the incremental costs associated

17with the available discounts exceed the incremental

18profits associated with the incremental sales that

19generate. If you take that language, I think you can

20create a viable safe harbor at least that would at least

21give firms some comfort that certain practices would be

22presumed to be legitimate.

23And I actually believe, having done my work in

24LePage's, that the behavior of 3M would actually satisfy

25this safe harbor test. But you don't want to condemn


1nondiscriminatory price cuts in single markets and you

2want to be careful not to penalize policies that exclude

3less efficient competitors.

4That's a different issue because if you want a

5test that's workable for a firm that's engaged in a

6policy, it's very hard to say you shouldn't exclude a less

7efficient competitor because the firm is not going to know

8typically whether its competitors are more or less


10So, this test really is not going to be a

11perfect test and probably never will be.

12 A workable rule should be one that's clear and

13manageable. We don't want businesses to say what I hear a

14lot in recent years, which is we have no idea which

15practices we can engage in or not because anything that

16seems to have any bundling aspect to it could lead to a

17Third Circuit lawsuit.

18Now, as far as the thoughts I have given you, I

19just happened to go back and look on the web recently at

20the AMC's tentative recommendations. I assume they're

21still tentative. And I found myself in agreement with

22their recommendations in the areas I am talking about.

23There are some other areas I would disagree.

24But I noticed that the AMC tentatively is

25recommending no need to revise the antitrust laws to apply


1to high tech industries. And I agree very strongly with


3The AMC is proposing no need for Congress to

4amend Section 2. And I agree strongly with that as well.

5And, finally, it looks like the AMC is thinking

6of recommending additional clarity and improvement in

7Section 2, particularly with respect to areas such as

8bundling. And I agree strongly with that as well.

9Thank you very much.


11MS. GRIMM: I'd like to thank all of our


13We are going to take a 15-minute break now.

14We'll reconvene in 15 minute for our round-table


16(A brief recess was taken.)

17 MS. GRIMM: Before we get to our questions and

18round-table discussion, I would like to introduce our

19fourth panelist, who will discuss some of the ideas that

20have been advanced by our other panelists this afternoon,

21as well as some of his own ideas about Section 2.

22Carl Shapiro, our fourth panelist, is the

23Transamerica Professor of Business Strategy at the Haas

24School of Business at the University of California at

25Berkeley. He also is Director of the Institute of


1Business and Economic Research and Professor of Economics

2in the Economics Department at U.C. Berkeley.

3He earned his Ph.D. in economics at MIT in 1981;

4taught at Princeton University during the 1980s; and has

5been at Berkeley since 1990.

6He has been editor of the "Journal of Economic

7Perspectives," and a Fellow for the Center for Advanced

8Study in the Behavioral Sciences.

9Professor Shapiro has published extensively and

10his current research interests include antitrust

11economics, intellectual property and licensing, product

12standards and compatibility, and the economics of networks

13and interconnection.

14Professor Shapiro served as Deputy Assistant

15Attorney General for Economics in the Antitrust Division

16of the U.S. Department of Justice in 1995 and 1996.


18MR. SHAPIRO: Thank you very much. I don't have

19any slides. I am going to cover some ideas I have and

20then comment on and kind of get the discussion going about

21each of the previous panelists.

22You probably already picked up the theme here

23is that we get up here and we reminisce about the cases

24that were brought or investigated while we were at the

25Antitrust Division. Okay? We really appreciate you


1coming out here because we all have a love for the

2Antitrust Division. FTC, too. And we sort of appreciate

3your coming out here so we don't have to go again to D.C.

4One of the themes that we've picked up here and

5throughout many of these hearings is that Section 2 cases

6are inherently really hard because it's a single-firm

7conduct and it's not like a cartel case. They're really

8hard and there's always elements and you have to be very


10And I don't disagree with any of that, but I

11want to focus on they seem to be harder than they need to

12be in some cases. And it's one of my themes, intersecting

13with the role of patents and plus innovation and

14Section 2.

15And I'm going to depart from the DOJ reminiscing

16and actually talk about the Unocal case, which was brought

17by the FTC, and which I served as an expert witness for

18complaint counsel. And that was litigated at the -- by an

19administrative law judge, before the administrative law


21So, let me just quickly remind you of that case

22or tell you about the case. So, Unocal had some patents

23-- had patents -- came to have patents during the '90s on

24reformulated gasoline. The State of California through

25the California Air Resources Board, CARB, established


1regulations for gasoline in order to make the

2cleaner-burning reformulated gasoline.

3And it came to pass that the regulations that

4were adopted, that Unocal's patents, apparently or very

5likely, many of the refineries would have to infringe

6those patents for a large fraction of the gasoline they

7would make if it would comply with the state regulations.

8So, and the allegation was that Unocal had acted

9deceptively by leading the industry members to believe

10that its patents would be -- that either it did not have

11patents or would make them available on a royalty-free

12basis. That was the representation when the regulations

13were being formulated and that Unocal then later sought to

14get royalties. That was the allegation of deceptive


16So, you would think -- well, let's say I would

17think, at least, maybe you would think, that this should

18be the sort of Section 2 case, and I guess it was FTC

19Section 5, and I'm not distinguishing those for my purpose

20here, that it would be relatively straightforward.

21 Big factual question about whether Unocal acted

22deceptively. They vigorously denied that they did so.

23The FTC or certainly complaint counsel was arguing they

24had. I simply assumed that they had for the purposes of

25evaluating market power and competitive effect. That was


1the fact question. If they had not engaged in any

2deception, I believed there was nothing to the case. That

3was my understanding, as I recall it.

4So, if they acted deceptively, and let's take

5the really cleanest version, they led people to believe

6patents would be available on a royalty-free basis.

7Regulations are selected. Literally billions of dollars

8are invested by refiners to comply with these regulations,

9made CARB gasoline, as it is called, and then they

10asserted patents.

11So, the reason I would say this should be, to my

12way of viewing, a relatively simple case because the

13conduct alleged and assumed by me, as an expert at least,

14deception is not something that we have to wring our hands

15over, oh, is that something that's procompetitive, is it

16important that companies engage in that sometimes. It's

17not like discounting. It's not like product innovation.


19So, now then the question is, okay, we don't

20really have to worry about stifling deception, okay. So,

21does it have a significant effect on prices, on market

22power? And if they represented that the patents would be

23available royalty-free and are later seeking something

24like five cents a gallon, to throw out a number, for

25pretty much the whole industry a very large fraction of


1the gasoline that would be produced, well, that's a price

2increase. There's very strong evidence that would be

3passed to the final consumers, motorists. Not that that

4matters so much because, even if not, it would be borne by

5the direct customers of the technology, refiners, who

6would be using the technology. And so you get right away

7the competitive effects without any real business

8justification for the conduct that's alleged or


10And yet, Unocal raised many, many arguments.

11We do not know how the administrative law judge or the

12commission or subsequent appeals court might have reacted

13to these. We do know from other cases, the case of --

14the Rambus case. There are a variety of Rambus

15cases that also involve similar allegations regarding

16standards and patents. And we know from other cases I

17won't get into that the courts have tended to say, well,

18wait a minute, you have a patent and so you get some

19market power associated with the patent, and so we should

20be very careful not to jump on -- not to conclude that,

21just because there's market power, somehow it has to do

22with anticompetitive conduct, because patents may very

23well confer market power in a perfectly desirable way.

24So, I guess I'm raising a concern that what

25should be a simple case, there seems to be, in some


1quarters at least, sort of a worship of patents that

2therefore mixes up market power attributable to the

3innovation versus market power -- additional market power

4that comes about from conduct, just the sort of thing that

5Tim was mentioning, actually, look at additional effects

6of the conduct.

7And the economic opportunities of hold up I

8think are very clear, going back at least to Oliver

9Williamson, my distinguished colleague here at Berkeley,

10and yet these were denied essentially by Unocal and its

11economic expert. That is to say, the notion that once

12refiners had invested enormous sums in order to comply

13with the regulations, that would necessarily put Unocal in

14a stronger bargaining position to get royalties that they

15could not have gotten earlier.

16So, I would say it's relatively fundamental

17economic principles, fairly clear fact pattern, and yet we

18have -- and, for example, the whole Antitrust debate about

19defining the relevant market. Defendants can often, in

20this case at least, try to make that very complicated,

21exactly which technologies are in the market and which

22ones are substitutes, and what was the best alternative,

23and how good was it, and how much -- they even argued, our

24technology is so good that people would have picked it

25anyhow and, therefore, even if we engaged in deception, it


1wouldn't matter.

2Well, I just don't think that's right because

3there's additional market power that results from lock in.

4So, sometimes the elements that we always think

5of for Section 2 cases: defining the market, measuring

6the market power; being cognizant of preexisting market

7power, in this case because of patents, I think we need to

8be careful not to lose sight of what may be a simple or

9more direct argument that can get us to analysis without

10doing -- without necessarily following some of these steps

11and without getting tied up particularly in market

12definition. And, again, Tim, I know, emphasized that he

13really, as most economists, if we can, we want to get to

14competitive effects. And market definition may or may not

15be helpful in getting us there in market shares.

16And if you think about the cases I've described

17today, measuring exactly which share of how much of the

18gasoline infringes or might infringe and what other

19technologies are being used is a distraction,

20fundamentally a distraction to what's being looked at


22And that came in in terms of remedy as well. My

23testimony was, we should restore competition, which means

24they should license these patents on a royalty-free basis,

25as they had represented under my working assumption. And


1yet Unocal argued that, well, our technology is so good

2that we should be able to charge more than that, even if

3we engaged in deception, because under competition somehow

4they would have been able to charge a lot.

5Then you ask, well, then why did you act

6deceptively. And they say, well, we didn't. Well, what

7if you have. So, you go back and forth. All right.

8So, while I'm not expecting the DOJ or FTC to

9suggest that we throw out market definition, for example,

10in Section 2 cases. I do think looking for shortcuts that

11are reliable is a good thing to do.

12 Let me go on to say something about the previous

13speakers now that I've made some points about some of my

14own thoughts about Unocal.

15So, Tim first, Professor Bresnahan. Very

16gracious of him to come up here to Berkeley and appreciate

17his kind words about Berkeley. I will try to reciprocate

18and I will make two trips to Stanford in the next week for

19conferences there, and with pleasure.

20I took some of what you said, Tim, to be

21suggesting that we could think of screening cases based on

22whether there's a theory of harm that the conduct would

23lead to a significant increase in market power, or let's

24put that differently, relax the constraints on pricing

25that are facing the firm that's accused, or the defendant



2And I think that's a really good way to go. So,

3I support that.

4One way I like to think about it is we could ask

5if the conduct is directed at certain competitors or maybe

6at certain distributors who then would be important for

7certain other competitors in your Dentsply case, we could

8ask, if the conduct was really effective and eliminated

9those competitors, a certain class or group of

10competitors, would the firm be able to significantly raise

11price. Or, alternatively, if those competitors were fully

12enabled, would that lead prices to fall significantly.

13If that's true, then we need to proceed further

14in the inquiry. If not, because the price is really

15governed by some other set of dynamics, you know, in the

16case of patented drugs, if you get rid of the generic

17competition, that would usually lead to a higher price,

18but it could be in some cases that competition from other

19patented drugs is what's driving price or, in principle,

20that sort of competition, and then we could stop that

21inquiry if the targets were not really providing sufficient

22competitive discipline. So, I am very supportive of that


24 You said at some point, Tim, that it was very

25hard to do some sort of balancing, you know, particularly


1quantifying the balancing of net effects, harm to

2consumers, benefits to consumers. And so I guess the

3economic theorists, I guess that's going to include me

4now, may like to measure all these things and do this in

5our models, but in practice that balancing would be hard

6to do. It is hard to do.

7One thing we might do is then focus more on the

8competitive process, rather than necessarily a particular


10But you also said the defendant could show that

11the practices were efficient and that would be a defense.

12So, if there was anticompetitive danger, the defense could

13come back and say the practices were efficient. I don't

14know what that means in practice. I guess I'd like to

15hear more from you about that. Because there is typically

16going to be some story about, oh, this has lower prices

17for some customers so it's efficient, or this is going to

18prevent free riding, so I need to have exclusive dealing

19here. There's going to be some efficiency story and I

20don't understand how you can avoid doing some balancing

21after the efficiency flag is raised and now are we done.

22I don't think you mean they're done just because the

23defense raises the efficiency argument. So, what happens


25My last comment was on -- I don't want to get


1into Microsoft. Believe me, I really don't want to get

2into Microsoft. But you did mention -- I like your term,

3the "remedy fizzle." I don't know if you coined that

4term, but I like it. You took some responsibility, I

5think --

6MR. BRESNAHAN: I lived that term.

7MR. SHAPIRO: For years, right? I just wanted

8to share the responsibility because, having testified for

9the states at the remedy phase, I want to share that

10responsibility with you.

11MR. SHAPIRO: Rich -- next, Rich Gilbert. I

12really liked to hear what you had to say about interfaces,

13Rich, because this seems to me -- I kept coming -- this

14came up when I heard you talk about IBM and Microsoft and

15other examples, it seems to me, going back to at least

16IBM, and probably selling machines in the 19th century or

17something, you've often got this pattern where, I have a

18product and I innovate, I improve it and, as part of

19improving it, I change the interface or I start producing

20a complementary product that needs to be compatible and

21it's innovative and very often intellectual property

22rights are used to control or secure an interface. And

23yet we know from the telecommunications, we know from

24other network industries, that controlling interfaces can

25lead to a certain octopus-like nature from what might be a


1secure monopoly in one product initially.

2And speaking for myself, I get really torn

3because I feel like, well, fine, the monopolist, if you

4want to call them, improved their product. Integration,

5where different components are integrated together, is a

6very important element of improved performance, and so how

7are we going to draw these boundaries. You know, do we

8want to treat interfaces differently, for example, either

9under a copyright or patents or how does it intersect with

10antitrust. I think these things are hard and I wonder if

11you want to say more about that.

12I was -- it was shocking to me, I have to say,

13to have an economist tell lawyers to focus on the process

14rather than the outcome. I just --

15 MR. GILBERT: Not the first today.

16MR. SHAPIRO: I know, it's true. This is all

17the more shocking because lawyers are very good at process

18in my experience and economists are always thinking about

19these outcomes and are often blind to the process. So, I

20just -- I don't know, we might have to revoke your card.

21I don't know.

22And then -- well, I guess I was maybe not

23shocked, but a little surprised that you said, well, the

24courts have done fine because all of this is hard. If

25it's sham innovation that's your standard at the end, that


1seems very hard for plaintiffs. And maybe that's what you

2want. I mean, what would it take -- what would count

3as a sham? Could you give us an example? For example, to

4say where, well, the product is a little better but they

5didn't have to do it this way, for example. What would be

6a sham? You know, I think it's sort of ironic when I

7think about Microsoft -- I said I wouldn't talk about it

8much -- but one of the things Microsoft really pushed

9throughout the trial was freedom to design their product

10the way they wanted to and the great benefit of

11integrating different features, as opposed to more

12components or modular.

13Well, what is it now, eight, ten years later? I

14think they're really having trouble because what the

15computer science community always does know is, no, that's

16not good design. Good design is modular and basically

17people on the other side are telling Microsoft, you

18wouldn't do this except for strategic reasons. And now in

19a way that's sort of spaghetti code or the increasingly

20complexity of Windows has made it very, very hard for them

21to meet deadlines in terms of coming out with new versions

22and a lot of other problems they've had.

23So, what would you do in that case to say, well,

24you don't have to design it this way, or maybe you don't

25want to go there if it's not a sham. Any company can


1choose how to design their product, even if it's not

2something they would choose to do except for strategic or

3exclusionary reasons. Or is that too intensive. I don't


5But maybe, and you can confirm this, Rich,

6you're saying it's so hard to do these cases, that it's

7true a sham innovation standard is very hard for a

8plaintiff, but that's okay and we're just not going to get

9many cases. And maybe that's where we're at. Is that

10what you support?

11Dan. I will finish soon here. Dan, there's a

12lot to say, but I noticed you were emphasizing the

13somewhat novel nature of network effects and the coining

14of the application "barrier to entry" in the mid to late

15'90s by you and Joel Klein, I guess.

16I have to tell a little story. So, Mike Katz

17and I did work on network effects going back to the '80s.

18And so we're working -- (laughter). No, that's neither

19here nor there. Academics can do anything, but until it

20comes into practice... So -- but I just want to tell a

21little story around that.

22So, we're working in the early '80s and we're

23working on the network effect. And actually personal

24computers and computer software is a good example of

25applications -- that was our example, actually,


1applications that run on an operating system.

2And Mike said to me -- and we're getting kind of

3excited about this and I guess we got published in a top

4journal, and Mike says, this is great, but I have to tell

5you, I have a friend who is doing a lot more with this.

6Not a friend. I should say, a former classmate. So, he

7says, back when he was at Harvard, there was this guy and

8he was making a lot of money on this. The guy's name was

9Bill Gates.

10So, we often think, oh, we work out these

11theories, but often after somebody else puts them into

12practice and understands them pretty well, then the law

13can kind of catch up with that and maybe academics as


15Okay, I'll leave it at that.

16MS. GRIMM: Tim, would you like to start off

17here and respond?

18MR. BRESNAHAN: Yes, I want to start off. I'm

19not sure I want to respond. I really like Carl's

20restatement of my screening idea. That was exactly what I

21was trying to say.

22Let me take on hard-to-balance because I don't

23think I'm against balancing. And I want to use the

24example of sham innovation because I think that's pretty



1The art of balancing, I'm against two things

2that sounds like balancing. One is a burden-shifting

3argument that suggests either an efficiency defense,

4defendant has to show that one rule really is better than

5the other quantitatively, or in a plaintiff's case where

6some sort of efficiency defense has been raised, an

7argument that plaintiff has to show that the world is

8going to be better off without the market power.

9I think that those procedures in which one party

10or the other has to sort of calculate the counterattack

11from the rule with precision are not going to go very far.

12And I guess I wouldn't go all the way to saying

13we should only like the competitive process. But, you

14know, a courtroom is a hostile environment for numbers.

15That's just a fact. There are things that courts are

16better at than numbers. So, a quantitative balancing I

17think is going to be very difficult.

18If we were going to have something, for

19example, bigger than sham innovation, what if a

20court were going to say, you know, cutting off future

21races to replace Office and Windows, cutting off the

22widespread distribution of new innovations in the PC

23business sounds like a lot of harm to competition to me.

24There's maybe a lot of zeros at the end of the numbers.

25Mixing the code between the early stage browser and


1the operating system, you know, you really got to hold

2your nose to call that innovation. Maybe there was

3something innovative to it. Maybe there were some

4benefits to integration, but it doesn't sound very

5innovative to me. So in this case the balance is

6pretty obvious.

7At that level of a balancing test, I'd be very

8comfortable, and I think I'd be comfortable with a broader

9definition than just the innovation has to be literally a

10sham. I guess I'd be comfortable with the view that the

11court can feel that the efficiencies are either clearly

12smaller or clearly -- not smaller in a quantitative sense,

13but in a salient sense or in a quality of evidence sense

14than the market power or vice versa. So, I'd be in favor

15of balancing. I just don't want to do it first.

16And I think the question that Rich raised

17earlier about, all the traditional tests are going

18to look pretty bad for innovation, I guess I would want

19a balancing test in that area. There's a lot of things

20that can get labeled as innovation. There's a lot of

21things which may seem like "innovation" to the defendant

22but which are dramatically less innovative than what

23other firms in the industry can do. I think this is

24one of the enduring lessons of the Microsoft case.

25On one of my trips to Silicon Valley to discuss


1the Microsoft case, I talked to a roomful of people and somebody

2said, weren't they accused of "innovating too fast." And

3somebody else said, they can't possibly be guilty of

4innovating too fast; those guys (Microsoft) have never

5innovated too fast in their lives; they never innovate fast

6enough. And stuff like that will come out in a courtroom.

7For this reason, I think that a standard that

8innovation has to be a sham is too narrow.

9MS. GRIMM: Professor Gilbert?

10MR. GILBERT: Well, when I started this project

11of looking at standards for innovation, I did a lot of

12reading. And one of the papers I came across was the

13paper by a Mark Popofsky. And Mark, in that paper,

14advocated basically different standards for different

15types of conduct, very much a process-oriented approach.

16 And my initial reaction when I read that paper

17was I sort of reeled back and said, oh, this doesn't make

18any sense at all where we're going to put everything that

19goes on in the economy in a separate category and have a

20different set of antitrust rules for it. I guess at that

21point I still had my economist card.

22But the more I looked at this area, the more I

23started to think, how do we actually do this analysis and

24what do you have to take into account to do the analysis

25right, the more I was led to the conclusion that maybe


1Mark got it right, that there were certain things that you

2do and a lot of things you can't do, and that different

3standards apply to different types of conduct.

4I mean, certainly the failure to innovate is not

5an antitrust violation, even though it's really what we're

6concerned about or should be concerned about.

7Other problems in this -- along this line, I

8have a paper with Mike Reardon where we look at

9technological tying. And the point of that paper is that

10there are lots of different outcomes. And even if you had

11really good information, you could do an analysis and you

12really could examine the problem, you don't know which

13equilibrium outcome is going to occur in the market. And

14there could be good outcomes from technological tying and

15there could be bad outcomes from technological tying. But

16putting a court into the position of trying to figure out

17which equilibrium the market is at and which one is

18better, that's a tough place to be.

19But I do understand that a lot of this conduct

20can have very undesirable consequences. If there are less

21restrictive alternatives, and you can identify them and

22really carve them out from the conduct, well, that's

23great. But unfortunately, lots of times the restriction

24that goes along with an innovation is inherent in the

25innovation. That's where it's difficult. I think, of


1course, if you can separate it out, that's fine, it's a

2lot easier.

3You mentioned IP protection. Yeah, it would be

4nice if we could -- it's hard to find an academic these

5days who wouldn't like to see lesser IP protections, and

6particularly for things that have network externalities,

7the other barriers to entry like interface standards. But

8that's a little bit out of our area.

9Let me talk a little bit about sham innovation.

10Again, I'm very sympathetic to the concept that just

11calling it innovation should not be able to protect all

12kinds of undesirable conduct and consequences. That just

13seems pretty obvious.

14But how you actually measure how discrete an

15innovation has to be before it is not a sham brings you

16right into the kind of numbers that Tim was saying are

17very hard for a court or anybody else to do. What number

18is big enough? And it's not just the innovation need,

19it's when the innovation occurs and how it occurs. Is it

20rolled out in every market, does that make it a sham or


22And I come back to this unilateral refusal to

23deal analogy. Without defending -- I don't want to defend

24a "Trinko" approach, but I just find it very odd that

25innovation that has similar consequences should be held to


1a higher standard.

2So, I still think there are things that are

3unlawful. I don't think that innovation should be able to

4protect all kinds of activity. But when you are looking

5hard to do, it's almost impossibly hard to take into

6account all of the incentive effects and the chilling

7effects if you get it wrong.

8And the bottom line, it seems to me, is that

9most of the time we're not going to have a problem and you

10should just be careful about chilling innovation by

11intervening where there might be a problem unless you're

12absolutely, absolutely sure that that's the case.

13MS. GRIMM: Professor Rubinfeld?

14MR. RUBINFELD: I don't have anything to offer

15specifically on that debate. I just have a couple quick


17First of all, most of my good ideas actually

18come from Carl Shapiro one way or another. So, my only

19intimation was trying to get the courts to see that as


21The other thing -- that actually was a serious

22comment. But the other slightly more serious comment is

23that there is an interesting theme I've noticed just at

24least from this group, and that is, when we -- before we



1went off to Washington in one extent or another, we were,

2let's say each of us in our own way, somewhat more

3theoretically inclined in thinking about some of these

4issues. And the effect of the Washington experience I

5think on all of us to one degree or another is really for

6us to worry about finding something that's really

7operational that will actually help the agencies and

8others really resolve practical problems.

9And so the emphasis on process, and I would put

10it as sort of finding workable kind of second best

11solutions, is the natural thing to think about. And I

12think that's something I do a lot of.

13In another context, for example, I was struck in

14a lot of mergers I worked on that we had, I think, at the

15division, and also probably at the FTC as well, some very

16sophisticated simulation software, which only as far as I

17could tell one or two people understood, and not all of

18them were in the agency. If you know the folks I'm

19talking about, you know what I mean.

20And it would have taken in many cases something

21like six to eight weeks to make it actually functional,

22which is hard to do under a Hart-Scott-Rodino. So, after I

23left, I actually, with my co-author, Roy Epstein, wrote

24some new software and came up with a much simplified

25procedure which, while greatly simplified, actually is


1something you can do within the thirty-day period.

2So, a lot of our work has been driven by that

3common theme. And I think with respect to sham

4litigation, that's sort of the same issue I think we're

5all heading towards, which is, we see a problem and now we

6have to sort of help to think about what would be a

7workable solution for the courts.

8MS. GRIMM: I'd like to give our -- all of you

9panelists an opportunity to kind of question each other,

10if you'd like to, as Carl did for all of you, or to

11respond to any of the points made by each other. And then

12we'll ask a couple questions on our own.

13MR. BRESNAHAN: I'd like to take the bait that

14Carl offered us in discussing the Unocal matter, because I

15bet that most economists would agree with him that,

16if there's some amount of market power or power to

17exclude associated with a patent, and if some act,

18deception is an extreme, but there might be others,

19some act or deception to embed it into an interface

20standard, or maybe even just embedding it in an interface

21standard in a way that doesn't have any technical

22benefits, there's some act that extends the coverage of

23that patent and gives the firm that holds the patent a lot

24more market power than it would otherwise have, that

25that's very troubling.


1 And this is one of the disciplinary divides I

2think you see between economists and attorneys.

3Economists are more eager to take that position.

4I suspect that one of the problems with that is

5that, patent law hasn't been particularly successful --

6forgetting antitrust law for a minute. Patent law hasn't

7been particularly successful at delineating the power to

8exclude in any particular patent conveys on its owner.

9So, when you get into these cases in the

10pharmaceutical industry where the patent on the original

11molecule is running out but there's a new patent on, the

12same molecule but packaged into a lozenge form or something

13like that, that it's actually not completely transparent,

14what's the right answer to the question, "how much

15market power does the patent provide?" And when the

16pharmaceutical firm starts playing Carom shots off the

17enormous complexities of the regulatory process under

18Hatch-Waxman, what is the answer to the question, "how

19much market power was conveyed by the original patent?"

20So that even if we're fairly comfortable with

21the idea that creation of additional market power beyond

22what the patent originally would have given that can be

23a thing that can be very hard to determine in a legal


25There probably is a near consensus among academic


1economists that patent policy in the United States over

2protects the patent holder. I think I agree with Carl on

3that. There's this other problem that patent policy is

4too vague, that patents simply don't look like property rights

5here. You have to go to courts or to the regulatory

6system to find out who owns what. And that -- the antitrust

7doctrine, Carl quoted the traditional antitrust doctrine that,

8intellectual property law if what it is and we ask

9whether there's additional market power on top of that.

10That may be more attractive in its economics than its law

11because it's hard to determine how much market power there

12would have been absent the anticompetitive acts.

13MR. GILBERT: I'd kind of like to reinforce what

14Tim said earlier, Carl, and I think also Dan as well.

15While a lot of our discussions today might be

16interpreted as suggesting that Section 2 analysis is very

17hard to do and therefore we shouldn't do it, and there's a

18lot of ways in which I think that's absolutely wrong, and

19that is Section 2 analysis isn't that hard and should be

20done, I do think that the law creates a road map to make

21Section 2 analysis unnecessarily difficult. You've got to

22have -- you know, you've got to identify the market, the

23product market, the geographic market, you have standing,

24you have all of these things. In all of these cases, I

25know cases I have been involved in, I'm sure everybody


1else, it seems like you never get to the question.

2You know, the relevant question is: Does the

3conduct really raise prices. And most of the time that's

4pretty obvious whether it does or doesn't and you don't

5have to do all this other stuff. And I think the law

6often puts us in a position of having to go through this

7kind of rogue set of steps that's in many ways very, very


9MR. SHAPIRO: Well, two things. The first one

10is to emphasize my concerns about the fetish over patents

11in intellectual property rights, therefore in some cases

12being a little blind to the fact that they can be

13leveraged, if you want to use that word, and you can get

14more power than was granted with the patent, particularly

15with patents that are very iffy. And there's a whole set

16of these questions about that.

17I mean, I guess it's outside of Section 2, but

18these pharmaceutical settlements cases, like the Shering

19case the FTC brought, and where the Second Circuit has

20gone with those cases was the tamoxifen case and seeing

21the patent as, oh, well, even if you paid off a competitor

22to leave because you have a patent, somehow it's okay, it

23doesn't mean you've stated an antitrust claim, that's

24something the -- you know, even if that's outside

25Section 2, that thinking is something that both agencies


1should really head off.

2And I guess there's an IP report still coming.

3There's -- that seems to be a very important role to

4delineate the importance of patents, yes, and the reward,

5yes, but there's a limited power that is granted, and

6beyond that, we can have abuses.

7I would shift topics a little bit and actually

8ask a question of Dan that I skipped when I was standing


10I'm curious, Dan, in your discussion of

11LePage's, whether you -- I guess you favor a bright line

12test of comparing price to marginal cost for additional

13units sold in a bundle. Or maybe, what about comparing

14marginal revenue to marginal cost to see whether the extra

15sale and bundling was profitable or not, a kind of profit

16sacrifice test.

17So, would you favor either of those? I mean,

18you're objecting to LePage's as being vague. So, here are

19two potential standards that are a lot more specific. I

20guess I'm talking about a safe harbor, either if the price

21is above marginal cost or if the marginal revenue is above

22marginal cost, then the bundling is okay. Of course, even

23if it's not, we assume you want to look first back to

24scope and so forth. So, there's two questions related to



1If the program is limited, there's only a few

2customers or a short period of time, if that's the case,

3would you just wave it through? It just doesn't matter

4what the structure of the program is to you because it

5couldn't have anticompetitive effects or not?

6And then related to that, I don't know if you're

7familiar with the EU's approach to this, but they're

8required to share methodology and calculating volume

9discounts, multi-product or single product, and whether

10you think that's something that the U.S. should pick up


12MR. RUBINFELD: Good questions, Carl. I

13actually am not familiar with the EU side, so I am not

14going to try to answer that.

15With respect to the workable test, you're right,

16I was suggesting just a safe harbor and I think I would

17accept your clarification. I was looking for a profit

18sacrifice kind of test, so I would compare marginal

19revenue and marginal cost, that's if marginal revenue is

20different from price, but only to get a safe harbor.

21The problem in extending that test is that,

22while I think there's some bundling cases which I think

23are appropriately seen as really being an extension of a

24predatory pricing case and probably ought to come under

25Brooke Group, I think there are other kinds of bundling


1practices which probably are not seen that way. So, the

2safe harbor I don't think ought to be seen as

3characterizing all, all types of bundling. Other types of

4bundling might seem more smart with respect to other kinds

5of exclusionary conduct of the kind we talked about

6earlier today.

7The other thing that you asked me about my point

8about the effect of this initial program being very

9limited. To me that is quite important because -- I may

10hear something to the contrary in a second -- but it seems

11to me that if there's a practice that cannot be shown to

12either have the effect and be sufficiently exclusionary

13that it makes a competitor not viable or perhaps even has

14no effect on its ability to operate at an efficient scale.

15I don't see how that practice ought to be considered


17So, I think you do -- in my opinion, you do have

18to show that if there's exclusion, it's substantial enough

19to really matter from the point of view of the potential

20competitiveness of the firm that's being affected.

21We can debate whether we should focus on volume

22scale or efficient scale, but certainly there ought to be

23some measurable effect.

24MS. LEE: Dan, you had said in your presentation

25that a variant on the profit sacrifice test would be


1appropriate to use as a general standard for all Section 2


3 I was hoping that you could refine that a little

4bit, in particular, you know, how is this different from

5the traditional profit sacrifice test, whatever that may

6be, and how does it differ from the no economic sense


8MR. RUBINFELD: That's a great question. I

9think I really can't -- without going back to my drawing

10board for maybe a few years, I don't think I can answer

11that very well.

12The reason why I was saying a variant in my

13comments is that I have been trying to follow some of the

14debate in the literature among the folks who prefer more

15of a balancing test to a profit sacrifice test. And it's

16not that hard to come up with hypotheticals that would

17defeat almost any version of a profit sacrifice test under

18certain circumstances.

19And so what I was imagining was that one would

20be able to come up with either a more robust rule that was

21not subject to too many of these hypotheticals, or maybe a

22complex rule that said under certain circumstances we do

23the test one way and under other circumstances another.

24But, unfortunately, I don't really have an

25answer to that question. I am hoping, Jim, that you and


1others at the Division will work hard to give me an


3MS. LEE: Okay. That was a good way to deflect

4the question.

5MR. RUBINFELD: Others here may have an answer.

6MS. LEE: Let me also get you to react to Tim's

7proposal in terms of how we should evaluate Section 2

8cases, I would call it a step-wise rule of reason. Tim,

9please feel free to disagree with me if you don't think I

10am charactering that appropriately.

11MR. RUBINFELD: You are asking me that question?

12MS. LEE: Yes. How would it be different from a

13variant of the profit sacrifice test that you think would

14be appropriate.

15MR. RUBINFELD: Well, I guess without being too

16specific, I have some of the same reactions I guess others

17on the panel have expressed based upon my own experience

18both in the Division and working on private cases, and

19that is the cases often get bogged down in complex debates

20about issues like market definition, without really

21talking about competitive effects.

22So, I'm actually -- at the level Tim is talking

23about, I'm very symptomatic with his suggestion. I think

24the pharmaceutical cases for me are really an excellent

25example of that. I have been involved in a number of


1these where there's a huge battle about market definition,

2which can be a very tricky issue in pharma cases for a lot

3of reasons, and yet I thought that -- the answer to the

4question, how you define the relevant market, at least if

5you are using the guidelines, really has almost no impact

6on whether there's a competitive effect.

7If you think that a generic would have entered

8earlier, and the generic most of the time is going to

9enter at a substantial discount off the price of the brand

10product, there is likely to be an effect. It's going to

11be the rare case where competition is driven just by other

12branded products.

13Now, if you think that's the case, then the real

14battle is going to be on issues such as causation, whether

15the practice itself had procompetitive benefits, and so

16on. So, there will still be a lot to debate, but the

17debate will be about whether this competitive effect A and

18B, whether there are justifications that say that that

19procompetitive effect was worth it.

20Rather than debate, which can get pretty far off

21the subject, or market power -- certainly most, if not

22all, successful brand products generate a lot of market

23power. That's the point of Hatch-Waxman to some extent,

24or the point of patent laws generally. And -- but the

25point of Hatch-Waxman in part is to encourage entry to


1benefit consumers. And the effect of that entry is going

2to be to reduce some of that market power.

3And I don't think any of that should be very

4controversial and yet I have seen a number of cases where

5the battles over whether firms have market power seem to

6take prominence. And so a process that in my view would

7move us more quickly to the heart of the cases would be a

8constructive process.

9Ms. Lee: Tim, let me ask you to clarify

10something that I didn't quite understand about your

11proposed way of analysis.

12In particular you had suggested that, well, if

13you look at -- if you first establish a causal effect

14between the act and then the effect, this gets you around

15the whole complex processes of trying to figure out what

16the appropriate but-for world is when you do the

17traditional sort of economic efficiency analysis.

18I don't quite see that in terms of, to establish

19causality, don't you have to establish in some sense what

20the world would have been absent the exclusionary act?

21MR. BRESNAHAN: That's a good question.

22I agree that to establish causality you need to

23say what the world would have been like in a competitive

24sense absent the anticompetitive act.

25I think the force of my argument is to -- is


1really procedural. It's to move the things which are

2going to be most difficult for courts to do back in this

3sequence. So, I mean, you heard us all economists say,

4it's often easier to see whether there's a competitive

5effect than to get market power right. I think that's

6probably going to be true.

7Certainly if there's a Section 2 case there,

8it's going to be easy to see what the competitive effect

9is. And then if you can't see it, there's no Section 2

10case there.

11Similarly, that the challenged conduct causes

12the market to be less competitive, that's an inquiry that

13can be undertaken within the four walls of what causes

14competition, without any balancing against the efficiency

15of the challenged conduct. Does it change the conditions

16of the competition? And I bet a lot of cases will follow

17thereto, and that's within the four walls of ordinary

18antitrust analysis. Is the reason that the market is less

19competitive because the challenged conduct raises entry

20barriers, raises them in a way that, you know, the

21entrants and third parties can't get around to the

22relevant time frame. Those are all difficult tests to


24So, most Section 2 inquiries should fall by the

25wayside. I just want them to fall by the wayside cheaply.


1And then you come to the last thing, which as

2we've all said is really, really hard, you know, you've

3got causation, there's some challenged conduct which is

4changing the conditions of competition, but there's also

5something good about it. You know, it's innovative or

6it's a price cut so it's especially good for customers,

7and now we've got to do this balancing, which I think is a

8very, very difficult thing to do.

9So, I just want to reduce the incidence of the

10balancing. Rather than leaping to that right away, go

11through other things first and discard cases. And I

12think that the causation -- the causation inquiry which

13says, is the challenged conduct holding entry barriers

14high is an easier counter-factual inquiry than, is the

15extent to which it's holding entry barriers high worse

16than its countervailing efficiency. It's got one less


18So that would be how I would proceed. And the

19basic idea is to save wear and tear on the system, which

20is potentially the result.

21MS. LEE: Thank you for the clarification.

22Rich, I wanted to ask you, you had said you have

23became more sympathetic to the idea that in different

24Section 2 matters different standards should apply.

25How would one go about determining the best


1standards to apply in each situation?

2MR. GILBERT: Again, a very good question.

3Certainly what sets innovation apart is the

4temporal linkage and very complicated linkage between the

5conduct at issue and the investment research and

6development that create the innovation and the prospects

7that any antitrust venture that would show that kind of

8very beneficial investment. And suppose you had a case

9where you didn't think that linkage was all that

10important, so you intervene in that case. But then if you

11do that, that also creates a precedence for there being

12other cases the linkage could be very important, and you

13definitely don't want to chill innovation in those other


15If you think about how some of those early cases

16-- if some of those early cases came out differently,

17because almost all the cases that I can see ultimately

18basically are pretty close to a sham innovation test. If

19they had done something very different from that, what the

20implications would be for people actually involved in

21product design could be kind of interesting.

22Now, there is a lot of conduct where I don't

23think those issue are at all significant. You know, they

24may be present to some extent, but they're just not

25significant. And so if you're talking about ordinary


1exclusive dealing or bundling or whatever, I think in many

2of those cases you can if not forget about, certainly

3discount, the more complicated intertemporal effects. And

4the analysis I think becomes much easier. And the sort of

5rule of reason analysis becomes much more possible.

6Weighing of benefits and costs becomes more reasonable.

7MS. LEE: Carl, do you have anything you want to

8say in addition to what you said already about general

9standards? You had said in your comments that you were

10very sympathetic to a standards approach.

11Is there anything else you would like to add?

12MR. SHAPIRO: Well, you called it a structured

13-- what did you call it?

14MS. LEE: No, I called it a step-wise.

15MR. SHAPIRO: Good, that's the ticket.

16MS. LEE: I think that's what it was.

17MR. SHAPIRO: So, I think of it in terms of

18screens. Traditionally, the monopoly power screen. You

19have a lot of power, and if you don't, then Section 2

20doesn't apply.

21I think I would push for: Does the conduct hold

22up the prospect to leading to significant increase in

23market power, okay, as actually a better question to use

24as a screen.

25Now, the reason I think the traditional screen


1has been applied, it's been assumed if you don't have any

2power to start you, you can't manufacture something from

3nothing. And that may be true in a lot of cases, although

4not always. Maybe deception turns up.

5Furthermore, even if you have power to begin

6with, if the conduct couldn't add much to it, maybe you

7have a patent, then we can dismiss that case, we don't

8have to go anywhere. So, you would get something knocked

9out on this increment screen that you wouldn't get knocked

10out based on a preexisting traditional power screen.

11So, I think it's a lot more closely tied to what

12Tim was saying at the top of the program here about

13looking at effects and increment. And there are ways to

14do that, implement that, and I have written about that and

15other people have, too. So, that's a general concept I

16think that cuts across a lot of cases.

17At the same time, I agree with Rich that -- and

18I think Dan -- well, profit sacrifice may apply in some

19cases but not others, so then you have to be more nuanced.

20You know, profit sacrifice would not apply in the Unocal


22MS. LEE: So, let me ask you the same question I

23asked Rich.

24Do you have a suggestion about the methodology

25of figuring out, well, which is the best approach in each


1type of matter?

2MR. SHAPIRO: It would be very unwise for me to

3get into that at this late hour.

4MS. GRIMM: I just have one question on

5remedies, and this is for Tim. Again, on the Microsoft

6remedy which you labeled a fizzle and you said the remedy

7in AT&T from your point of view was successful.

8I was wondering if could share any views with us

9on appropriate remedies in Section 2 cases, perhaps

10structural versus the conduct remedies.

11MR. BRESNAHAN: I'm almost certain there's no

12general law of remedies in Section 2 cases because

13Section 2 cases are so context specific and so fact dense.

14You know, in the structural remedy that was

15negotiated rather than imposed by a court in AT&T, I think

16the logic of that was caused by an attempt to minimize the

17harm to competition and innovation by walling off the rest

18of the industry (by vertical disintegration) from the

19necessarily regulated sector of telephony, local phones.

20And that's just a very specific argument.

21So, some principle that has remedies that are

22reasonably proportional to the harm to competition that's

23been proved, I think it's going to -- I think it's going

24to be very hard to go farther than that to a broader abstract



1MR. SHAPIRO: If I could just make a quick

2comment. I thought about Microsoft remedies in context

3here. At one end you have, sin no more, don't do what you

4did before, narrowly defined, maybe defined to reflect the

5market changing. And, you know, that doesn't seem to me

6that does much to restore competition if there's been real

7damage with some lasting effect, okay, if the case was

8significant to begin with.

9One of the things that was interesting in that

10case was that -- and I think it's true in a lot of cases

11-- it's very hard to know exactly what the effects are.

12So, you can't say, ah, we're trying to engineer the market

13to return to a certain state and that's what we mean by

14restoring competition.

15So, again, in that context, really the case was

16about raising entry barriers, as Tim put. My view was,

17you should have a remedy that lowered entry barriers and

18then come what may. Maybe entry will occur, maybe it


20But sin no more seems to me it's probably going

21to be too weak in most cases where the case was worth

22bringing to begin with.

23MS. LEE: Let me ask a follow-up to that.

24If the only suitable remedy is a sin no more

25remedy, do you think the agency should bring a Section 2


1case in that instance?

2MR. SHAPIRO: Well, there still could be some

3deterrent effects. And there are private cases that

4follow on, for example, that could have a major role. And

5there were private cases in the Microsoft case that

6involved a lot of money.

7So, it could well be. I guess I would hope if

8it's a major case that either agency could come up with

9something a little more effective and maybe even creative.

10But, at the same time, partly from the Microsoft

11experience, it's very hard for a court to impose a remedy

12when the company says this is crazy, it won't work, you'll

13destroy all sorts of good things, and the government

14agency, you know, yeah, there's information but it's hard

15to know. So, I think it's very hard. And so if you are

16stuck with sin no more, it could still be worth bringing,


18MS. LEE: Let me just solicit the other

19panelists about that. Anything different or anything to


21MR. BRESNAHAN: Yes, I guess I'd be more

22conservative on this ground than Carl. It's hard to get a

23lot in deterrence in this area of antitrust law because

24it's so hard to -- you know, we're never going to have a

25doctrine that says these specific practices are


1anticompetitive. I mean, guys will just know not to do

2those particular practices. It's much more complex than


4So, other than generally wanting to keep the

5idea that there might be this prosecution of particularly

6egregious anticompetitive acts, this is not a great area

7where you can get an awful lot of deterrence out of --

8you know, out of a case where there's a remedy that

9doesn't do anything.

10So, I'd be less -- I'd put less emphasis on

11deterrence and, more emphasis on the view that it

12should really be looking for cases where you can make

13a big difference for the American consumer.

14I mean, before I was in government, in

15connection with Microsoft, I took the position, don't

16bring it unless you're going to do something really

17big, which I went on to say, probably meant don't bring

18it, although that turned out to be wrong. The government

19did ask for a remedy that would have changed the

20conditions of competition.

21I think these experiences are rare, important

22and efficacious in the first instance, and seeking

23deterrence only, you know, only perhaps in flagrant


25MR. GILBERT: Sometimes, not always of course,


1the case that dominance leads to conduct that is

2persistent and durable, that companies in dominant

3positions tend to do the same sort of anticompetitive

4things. And it's also the case that that dominance is

5persistent, that even if you try to break it up, forces

6are going to tend to recreate it. And I wouldn't say

7that's always true, but that's sometimes true.

8But I also say that, even in those cases where

9you cannot have a real structural remedy, that structural

10remedies wouldn't be very effective, a big case like this

11brought by DOJ or FTC has a lot of consequences for these

12companies. And I think you have a significant deterrence


14MR. RUBINFELD: The only thing that I was going

15to add is, these remedies come out of course not in just

16in court decisions we're talking about, but also in

17consent decrees that are reached. And I think it makes a

18big difference how you craft a consent decree. You know,

19I can think of some cases which I was involved in where we

20literally got a promise never to do A again and nothing

21more. There were other cases where the consent decree

22really laid out fairly carefully what we meant by not

23doing it again, not only for this company, but also the

24consent decree sent a clear message since the consent

25decree can be part of the public record.


1So, you can get some deterrence even in a

2situation where the structural remedy doesn't work if you

3craft the right consent decree. And, obviously, it

4depends on every case, but I think obviously the agencies

5should and I am sure do think hard about exactly how to

6did that. And that's an important exercise.

7MR. SHAPIRO: Let me just clarify. There was

8kind of a sin no more at one extreme and then I heard a

9couple of people talking about structural remedies.

10There's a lot of running room in between.

11MS. LEE: Agreed.

12MS. GRIMM: Well, my watch says it is 4:30. I

13would like to thank all of our panelists for being here

14this afternoon and sharing with us their very insightful


16I would also like to thank again the University

17of California at Berkeley for their hospitality.

18Would everyone please join me in giving our

19panelists a round of applause.


21(Whereupon, at 4:30 p.m., the hearing was






1C E R T I F I C A T I O N O F R E P O R T E R




5DATE: JANUARY 31, 2007


7I HEREBY CERTIFY that the transcript contained

8herein is a full and accurate transcript of the notes

9taken by me at the hearing on the above cause before the

10FEDERAL TRADE COMMISSION to the best of my knowledge and


 DATED: February 22, 2007













Updated June 25, 2015

Was this page helpful?

Was this page helpful?
Yes No