Academic Testimony

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



2 and










19 9:30 A.M. TO 4:30 P.M.

24 Reported and transcribed by:

25 Kathleen Carr Meheen, CSR 8748



2 Morning Session:


4 Deputy General Counsel for Policy Studies

5 Federal Trade Commission

6 and


8 Attorney, Legal Policy Section

9 Antitrust Division, U.S. Department of Justice


12 Morning Session:

13 Aaron Edlin

14 Joseph Farrell

15 Howard Shelanski




2 Afternoon Session:


4 Assistant General Counsel for Policy Studies

5 Federal Trade Commission

6 and


8 Economist

9 Antitrust Division, U.S. Department of Justice


12 Afternoon Session:

13 Timothy Bresnahan

14 Richard Gilbert

15 Daniel Rubinfeld

16 Carl Shapiro



1 C O N T E N T S


3 Introduction

4 Presentations:

5    Joseph Farrell

6    Aaron Edlin

7    Howard Shelanski

8 Moderated Discussion

9 Lunch Recess


12 Introduction

13 Presentations:

14    Timothy Bresnahan

15    Richard Gilbert

16    Daniel Rubinfeld

17    Carl Shapiro

18 Moderated Discussion



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

2 * * * * *

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

4 General Counsel for Policy Studies at the Federal Trade

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

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

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

8 Department of Justice.

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

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

11 please turn off your cell phones, Blackberries, anything

12 that might ring or clang or make noise.

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

14 structure, we request that the audience not make any

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

16 to limit it to the moderators and the panelists.

17 Before introducing our speakers and starting our

18 panel discussion, I would again like to thank the

19 University of California at Berkeley for hosting the

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

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

22 again, Richard Gilbert and Carl Shapiro for offering us

23 the facilities and making the necessary arrangements.

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

25 Law & Technology and the Haas Business School for


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

2 And those who have provided us with logistical support,

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

4 thank you again.

5 We're honored to have assembled this morning a

6 distinguished group of the finest lawyers from the

7 University of California Berkeley to offer their testimony

8 in connection with these hearings. They will provide

9 their perspectives on various themes and issues related to

10 the complex area of Section 2 jurisprudence, including

11 some research and economic analysis.

12 We've gathered seven panelists for today's

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

14 our morning panelists.

15 This morning's panelists are Aaron Edlin, the

16 Richard Jennings Professor of Law, University of

17 California Berkeley; Joseph Farrell, Professor of

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

19 and Howard Shelanski, here, Associate Dean and Professor

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

21 Technology.

22 Our format this morning will be pretty simple.

23 Each speaker will make an opening presentation from twenty

24 to thirty minutes. After the presentations are finished,

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


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

2 moderated discussion with our panelists.

3 We're scheduled to conclude this morning's

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

5 hearing from our panelists.

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

7 to thank are the panelists themselves. We appreciate the

8 time and effort and your willingness to share your

9 insights with us to make this a successful hearing.

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

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

12 add.

13 MR. MATELIS: Thank you, Bill.

14 The Department of Justice's Antitrust Division

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

16 conduct hearings. We are delighted that such esteemed

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

18 And the Antitrust Division takes particular

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

20 in the Antitrust Division as Deputy Assistant Attorneys

21 General for Economics.

22 We expect that today's panelists will discuss a

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

24 conduct and antitrust laws and we look forward to the

25 presentations and the panel discussions that follow.


1 On behalf of the Antitrust Division, I would

2 like to take this opportunity to thank the Berkeley Center

3 for Law and Technology and the Competition Policy Center

4 at the University of California Berkeley for hosting us

5 today.

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

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

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

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

10 appreciative.

11 Finally I'd like to thank Bill and his

12 colleagues at the FTC for all their hard work in

13 organizing today's panel and assembling the great speakers

14 that we have lined up today. Thank you.

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

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

17 now holds the Richard Jennings Chair and professorships in

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

19 served on the economic side as Senior Economist at the

20 Council of Economic Advisers during the years of the

21 Clinton Whitehouse. He is co-author with Professors

22 Areeda and Kaplow of one of the leading casebooks on

23 antitrust and he has published many articles dealing with

24 competition policy and antitrust law.

25 Aaron?


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

2 the slides.

3 MR. COHEN: And yesterday we had the

4 representative from Microsoft [laughter].

5 MR. EDLIN: Maybe we could switch speakers?

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

7 then.

8 Joe is Professor of Economics here at Berkeley.

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

10 The Journal of Industrial Economics, and former President

11 of the Industrial Organization Society

12 Professor Farrell was Chief Economist at the

13 Federal Communications Commission in 1996 to 1997 and was

14 Deputy Assistant Attorney General for Economics at the

15 Antitrust Division of the Department of Justice from 2000

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

17 Science and Telecommunications Board of the National

18 Academies of Science.

19 Joe

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

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

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

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

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

25 unilateral conduct field seems to have drifted a long way


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

2 worry that it leads to bad policy.

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

4 back to some first principles. Because the field has

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

6 clearly I think understood exactly what those first

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

8 about what they might be.

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

10 that distinguishes quite importantly between the final

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

12 should be economic efficiency, and the protections and the

13 process involved in antitrust enforcement. And it does

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

15 economic efficiency, each case should be analyzed or each

16 transaction should be analyzed along the lines of economic

17 efficiency.

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

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

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

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

22 thoroughly convincing evidence that my economic value for

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

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

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


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

2 economic market system that we have operates by enforcing

3 the property rights of the iPod. And that enforcement

4 does not look directly at whether the enforcement is in

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

6 antitrust often does something rather similar, okay?

7 So, before I get to the first substantive slide

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

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

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

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

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

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

14 other people's trade.

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

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

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

18 techniques by which a firm might be able to restrain

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

20 the core problems.

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

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

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

24 Northeast Airlines offers a five hundred dollar fare. And

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


1 buy it. No better deal is available.

2 An entrant that I've called Sprite would happily

3 sell at three hundred dollars a similar product.

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

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

6 everybody recognizes that if Sprite enters and offers the

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

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

9 profit competing against the two hundred dollar fare.

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

11 consumers continue to pay five hundred dollars.

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

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

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

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

16 what it is that's wrong there.

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

18 on discussions that Aaron Edlin and I have been having

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

20 that Sprite's willingness to sell at three hundred

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

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

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

24 be able to extract five hundred dollars from consumers,

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


1 hundred dollars. Okay.

2 And you might think that normally in a

3 competitive process, whatever that means, not only ought

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

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

6 Well, the mechanics we just went through.

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

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

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

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

11 have to compete against five hundred but against two

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

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

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

15 wrong in the competitive process.

16 So, stepping back, and here are some first

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

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

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

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

21 in the microeconomics classes and the way that

22 professional economists typically think about real world

23 problems.

24 What we spend the most time with undergraduates

25 on is that you can get to an economically efficient


1 outcome via price-taking perfectly competitive

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

3 the price-taking equilibrium, whether it would be

4 efficient or not, is unrealistic and unobtainable in many

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

6 nothing else, large economies of scale make that a

7 nonstarter.

8 And it's also interesting to note that antitrust

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

10 eschews many opportunities to move toward price-taking

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

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

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

14 behavior. And not only is that potentially difficult and

15 problematic to do, but antitrust seems to take the

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

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

18 controversial, but that's my impression.

19 The second approach to economic efficiency, which is

20 less juicy material for teaching undergraduates because it

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

22 to those who teach undergraduate micro classes, but is

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

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

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


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

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

3 unmanageably large but possibly not, that would have an

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

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

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

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

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

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

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

11 inefficiencies that go along with that.

12 So, Sprite and consumers jointly would like to

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

14 doesn't that happen?

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

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

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

18 of possible outcomes such that no group of consumers and

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

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

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

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

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

24 could all do better by doing something else.

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


1 imagine everybody getting together. But, conditional on

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

3 have a reasonable shot at finding a smaller and more

4 manageable blocking coalition.

5 What's a blocking coalition? A blocking

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

7 do better than the status quo given their endowments and

8 abilities to trade and so on.

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

10 competitive equilibrium analysis, we have core analysis.

11 And it suggests a rather different process. Instead of

12 suggesting a process where we kind of hammer on the

13 economy until most firms are somewhere close to

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

15 actually feasible in many important sectors of the

16 economy, it suggests a process where we protect the

17 ability of these blocking coalitions to work around any

18 inefficiencies.

19 So, a perspective on antitrust is this: That

20 antitrust protects the process of forming blocking

21 coalitions that block bad outcomes. And how does it

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

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

24 level, things that thwart the formation of blocking

25 coalitions that would otherwise prevent bad outcomes.


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

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

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

4 contracts of course are protected by antitrust. Some of

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

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

7 So, back to the Northeast and Sprite example,

8 Northeast is getting five hundred. Sprite and consumers

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

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

11 hundred dollar fare is not something that would survive in

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

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

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

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

16 blocking coalition by making clear that if the blocking

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

18 with the two hundred dollar fare.

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

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

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

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

23 is charging three hundred, Northeast then does cut its

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

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


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

2 had this three hundred dollar outcome. Northeast is

3 forming a blocking coalition with consumers to block it

4 with a two hundred fare.

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

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

7 blocking coalition against Northeast's five hundred

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

9 genuinely is at some level somewhat part of the

10 competitive process and somewhat a fundamental undermining

11 thwarting blocking of the competitive process. Okay.

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

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

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

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

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

17 relevant. Knowing whether Northeast made in any sense a

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

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

20 relevant. Okay.

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

22 specific rules and policies that have come to dominate the

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

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

25 actually work our way through and figure out what the


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

2 help.

3 So, what does this suggest about predatory

4 pricing. It suggests most fundamentally that predatory is

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

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

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

8 the entrant expects to have to compete against is very

9 different from the price that consumers actually end up

10 paying.

11 So, is Northeast's price cut primarily a

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

13 essence of the competitive process, or an

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

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

16 essence of an antitrust offense. Okay.

17 So, one way to answer this that is sensible

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

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

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

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

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

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

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

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


1 Certainly good for consumers.

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

3 that consumers really end up paying five hundred and they

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

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

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

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

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

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

10 it out.

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

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

13 it. The incumbent is restraining trade when given its

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

15 coalition, a potential blocking coalition, that would make

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

17 better off, but the incumbent strategically thwarts the

18 formation of that blocking coalition.

19 So, we saw one possible way in which the

20 incumbent might thwart the formation of a blocking

21 coalition, threatening that if that coalition starts to

22 form, then the price it charges will change.

23 Another way you might do that is through some

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

25 particularly favorable deals to some pivotal members of


1 this blocking coalition while expropriating others. I

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

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

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

5 work, it seems like that is really disrupting the

6 competitive process.

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

8 suggestion of what the competitive process is, none of

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

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

11 would be an important aspect of a direct inquiry into

12 economic efficiency. Right? Because if Northeast

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

14 inelastic, then having Northeast charging five hundred

15 dollars might be more efficient than having Sprite come in

16 and serving customers.

17 And I claim that Northeast thwarting this entry

18 would be a thwarting of the competitive process without

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

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

21 blocking coalitions as the process whereby we move towards

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

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

24 insist in the interim that they actually have to increase

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


1 formation of blocking coalitions without that inquiry,

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

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

4 economically efficient

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

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

7 protecting the process of forming blocking coalitions. We

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

9 efficiency and it is not necessary and may actually be

10 counterproductive to ask about economic efficiency at each

11 step.

12 That does not mean that I'm advocating a

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

14 criterion is actually economic efficiency. At each step,

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

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

17 consumers a better deal, then the entrant likes the

18 formation of this blocking coalition. So, the question

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

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

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

22 final goal is consumer surplus.

23 So, that freedom to trade principle is, we

24 think, an intriguing and promising way to understand

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


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

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

3 difficulties in some cases, like the hypothetical I was

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

5 of unilateral conduct that gets challenged in antitrust

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

7 And let me sketch this out

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

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

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

11 freedom to trade is potentially at risk where the entrant

12 has to compete against the low price, but consumers

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

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

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

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

17 conquer, exclusive dealing and so on.

18 But, in the case of many challenged practices,

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

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

21 incumbent being asked to stick around but just do

22 something different.

23 So, you can put a lot of unilateral conduct

24 complaints into the following framework. The incumbent is

25 offering two trades to consumers, not as alternatives


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

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

3 And the discrepancy there is meant to reduce confusion

4 about which is which. Okay?

5 And as a potential blocking coalition, sort of,

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

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

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

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

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

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

12 generally links A to B. Okay.

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

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

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

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

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

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

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

20 incumbent.

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

22 freedom to trade is violated and certainly the playing

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

24 a potential blocking coalition. Nobody but the incumbent,

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


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

2 to trade criterion strictly, there is no potential

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

4 incumbent is thwarting a potential blocking coalition.

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

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

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

8 we do?

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

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

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

12 the usual way of antitrust markets.

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

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

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

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

17 trade of the tying good.

18 Exclusive dealing, A is a bunch of widgets that

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

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

21 maybe a different place or something.

22 If you look at aftermarkets, A might be the

23 original equipment and B might be service to the

24 equipment.

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


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

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

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

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

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

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

7 does conduce to economic efficiency and is typically a

8 good thing, and is certainly not inconsistent with

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

10 nevertheless the latter is more fundamental to the ideas

11 of antitrust than is price-taking equilibrium.

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

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

14 understanding, that protecting competition as a process is

15 potentially, and I think actually very different from

16 imposing on each step of the process a requirement that

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

18 think that that's the final goal.

19 Trying to go much beyond that, based closely on

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

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

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

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

24 rules of law and policies that govern unilateral conduct

25 in antitrust has emerged from the kind of slightly vague


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

2 first principles.

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

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

5 brings the thorns into sharper focus. And the particular

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

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

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

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

10 field or freedom to trade in B.

11 Is that always illegitimate? That would be a

12 strict interpretation of freedom to trade as the only

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

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

15 Ideally, based firmly on these same first

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

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

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

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

20 principles.

21 Thank you.

22 (Applause.)

23 MR. COHEN: Where are we, Aaron?

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

25 I think, be able to project the slides.


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

2 Howard?

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

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

5 Aaron Edlin.

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

7 Let's do the show.

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

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

10 bandied about a lot recently and increasingly over the

11 past two decades.

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

13 against these necessary and sufficient tests for

14 monopolization."

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

16 question in Section 2 is: When is exclusion

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

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

19 monopoly excludes competitors by consistently charging low

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

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

22 What that example goes to prove, however, is

23 that we need something other than exclusion to be

24 anticompetitive. So, the question is: What plus

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


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

2 the "what" is.

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

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

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

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

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

8 as a result of lessened competition.

9 There is of course another variant, which is

10 extreme sacrifice, which comes more directly out of

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

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

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

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

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

16 of lessened competition.

17 More recently, Greg Werden and Doug Melamed put

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

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

20 economic sense but for a lessening of competition.

21 These sacrifice tests are on the move, or have

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

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

24 first advocated in the predatory pricing context. Thanks

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


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

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

3 They've also been on the move from sufficiency

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

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

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

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

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

9 says that extreme sacrifice is necessary in pricing cases.

10 The DOJ "Trinko" brief advocates the no economic

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

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

13 prove Bell Atlantic's refusal to deal caused Bell

14 Atlantic's short-term economic losses."

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

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

17 sacrifice is necessary to violation. And Werden and

18 Melamed have quite explicitly argued that no economic

19 sense is the unifying principle of Section 2 violations.

20 My fundamental contention which I've been

21 arguing for years is that sacrifice is not needed for

22 anticompetitive effect and frequently not needed.

23 My "Yale Law Journal" article argues that this

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

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


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

2 that is exclusionary and anticompetitive. There I explain

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

4 as Joe Farrell explained, even though prices will remain

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

6 maximizing.

7 I ask rhetorically: If sacrifice is wrong

8 headed in the predatory pricing context, why are we

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

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

11 refusal to sell at retail prices to Highlands increased

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

13 that it certainly is conceivable, perhaps even likely,

14 that when Ski Company refused to sell at retail to

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

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

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

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

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

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

21 refused to sell at retail.

22 But would that mean that the refusal was any

23 less exclusionary or anticompetitive? I think not. The

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

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


1 emphasized what they took to be consumer harm, the

2 revisionist claims of Trinko about "Aspen"

3 notwithstanding.

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

5 fairly clear that sacrifice is not necessary for

6 anticompetitive effect are submarine patents. If you seek

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

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

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

10 Many people have been talking about an extreme

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

12 Werden and Melamed, and fellow travelers with them,

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

14 cost of the dynamite triggers liability. There is a

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

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

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

18 lessening of competition, which justifies the cost of

19 paying for dynamite.

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

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

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

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

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

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


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

2 dynamite.

3 Well, now I gather that the dynamite has a

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

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

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

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

8 test is not looking at the right thing.

9 If the sacrifice test is not looking at the

10 right thing, neither is extreme sacrifice. Extreme

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

12 anticompetitive effect. Consider the American Airlines

13 case brought by the DOJ unsuccessfully. The judge thought

14 there that the extra plane was profitable if you ignore

15 effects on other planes. I suggest that everyone reread

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

17 you think that the extreme sacrifice test might make

18 sense, as the judge did.

19 Marginal revenue, as every economist and econ 1

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

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

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

23 What that means is that monopolies with lots of market

24 power can sacrifice enormously without triggering the

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


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

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

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

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

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

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

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

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

9 large profits have a substantial and much larger license

10 to blow up their competitors than other firms.

11 Rhetorically I'll ask why.

12 Consider the no economic sense test. Does that

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

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

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

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

17 or prevent entry.

18 Suppose there is no economic sense in charging

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

20 prevents others from entering. Well, the no economic

21 sense test condemns that limit pricing. But note that

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

23 easy case on slide two.

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

25 grants a safe harbor for charging the low price.


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

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

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

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

5 This smells bad to me.

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

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

8 the no economic sense test emphasize again that the cost

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

10 might be negative in the dump fee hypothetical.

11 My claim would be that blowing up your

12 competitor's factory is anticompetitive regardless of the

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

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

15 the operating profits, regardless of whether you pass the

16 no economic sense test.

17 The fundamental problem in my view with all

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

19 any kind of first principles that are attractive. They

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

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

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

23 can offer consumers - by which rivals who can offer

24 consumers higher utility actually get to provide that

25 higher utility.


1 The tests don't flow from any other principles

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

3 Now, when someone like me points out that there

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

5 action is anticompetitive, what you quickly bump into,

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

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

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

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

10 A common argument is that you need a hurdle to

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

12 for anticompetitive effect, but the plaintiffs should be

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

14 of cases from chilling legitimate competition.

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

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

17 want to eliminate these false positives, you could

18 eliminate Section 2 entirely, or you could eliminate

19 Section 2 for any plaintiffs whose name begins with A

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

21 Erecting arbitrary hurdles because the right

22 test is difficult to administer properly is, I would

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

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

25 test.


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

2 administrative difficulties truly make false positives a

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

4 that much discussion by the refrainers about false

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

6 arbitrary hurdles or abandoning the right test. You could

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

8 improve jury instructions. You could create procedural

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

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

11 that modern examples of clear false positives are pretty

12 rare.

13 What are my conclusions? That patience is

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

15 at least better ones, and that administrative difficulties

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

17 been used to do so.

18 Thank you.

19 (Applause.)

20 MR. COHEN: Okay. Our last presenter this

21 morning is Howard Shelanski, Professor of Law at Berkeley

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

23 of the Berkeley Center for Law and Technology. His

24 research focuses on antitrust policy and regulation.

25 On the economic side, from 1999 to 2000,


1 Professor Shelanski served as Chief Economist of the

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

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

4 Economic Advisers at the White House.

5 On the law side, Professor Shelanski served as a

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

7 We welcome your presentation.

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

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

10 presentation that at least in some aspects will connect to

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

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

13 choose among different kinds of tests for enforcement

14 under Section 2.

15 And I want to speak specifically about

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

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

18 the wake of the "Trinko" case.

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

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

21 the Federal Trade Commission are reviewing enforcement

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

23 parallel efforts ongoing, indeed some undertaken in recent

24 years by the Federal Trade Commission, to rethink and

25 reform intellectual property rights, and particularly to


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

2 intellectual property to foreclose competition with weak

3 or questionable IP rights.

4 And I think that the potential outcomes of IP

5 reform could matter for aspects of antitrust reforms, and

6 notably for policy toward unilateral refusals to deal.

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

8 Section 2 enforcement, and in particular thinking about

9 unilateral refusals to deal, antitrust reform efforts

10 should not ignore intellectual property reform processes

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

12 antitrust authorities should keep an eye on IP reform and

13 take into account how it might affect enforcement policies

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

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

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

17 interactive system. But I also want to talk about

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

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

20 IP reform is likely to reduce the strength or availability

21 of intellectual property protections, antitrust

22 authorities might consider enforcing less strictly against

23 refusals to deal.

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

25 there is a presumption against requiring a firm to deal


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

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

3 duties 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

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

7 imposed a duty to deal.

8 I agree with Aaron Edlin that Justice Scalia

9 engaged in some revisionism by finding profit sacrifice in

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

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

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

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

14 of 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.

17 The question that's left for the antitrust

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

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

20 about some policy issues that might arise, some background

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

22 answer to that question of what standard to use in

23 imposing a duty to deal.

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

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


1 anticompetitive harm. There are many cases where refusals

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

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

4 to innovate or produce something itself.

5 Mandatory dealing in cases where there isn't

6 anticompetitive harm could impede investment and

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

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

9 the previous rounds of these hearings, Former Assistant

10 Attorney General for Antitrust Eupate has some testimony

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

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

13 argument by Professor Steven Fallon that suggests the

14 evidence for such innovation deterrence is thin. But we

15 have to at least keep in mind the possibility that

16 mandatory dealing could impede investment.

17 I think that one of the bigger concerns is that

18 enforcement of a duty to deal might reduce competitive

19 innovation and production not by the firms being forced to

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

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

22 alternative.

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

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

25 many refusals to deal that are not anticompetitive and


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

2 justify its presumption against the duty to deal. But

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

4 evil per se.

5 So, refusals to deal can have anticompetitive

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

7 refusals to deal from enforcement.

8 Now, I want to suggest that one necessary

9 condition for such harm is that competitors and third

10 parties face economic barriers to providing the goods at

11 issue or that competitors and third parties face legal

12 barriers to providing the goods at issue.

13 And I would suggest we should not impose duties

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

15 competitive supply do not exist. There you get very

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

17 innovation either on the supply or the demand side.

18 But what about refusals that could be

19 anticompetitive, for which there are economic barriers or

20 legal barriers. There are several standards that we could

21 use to identify those situations and to decide whether or

22 not to enforce a duty to deal.

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

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

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


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

2 of deterring innovation on either the supply or the demand

3 side.

4 Alternatively, we have a range of rule of reason

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

6 phrase them as potential consumer welfare tests, the kind

7 of tests that Professor Salop proposed in an earlier round

8 of these hearings; a business justification test, which

9 Kolasky suggested in that same round; and a profit

10 sacrifice test of various stringency, ranging right up to

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

12 Melamed has articulated.

13 Then we have the old line essential facilities

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

15 Court has never adopted. One could quibble about what

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

17 the Appellate Court for the essential facilities approach,

18 notably in the Seventh Circuit.

19 So, how should the Justice Department and the

20 Federal Trade Commission choose among these various

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

22 approach because per se legality fails to block cases

23 where the only effect is anticompetitive. And while often

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

25 innovation and investment incentives, there isn't clear


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

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

3 I don't much like the essential facilities

4 approach either because it does ignore some legitimate

5 business justifications. And I think that it may too

6 easily allow access and deter innovation and investment by

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

8 concern is it requires a quasi-regulatory solution.

9 While I fully agree with my colleague Aaron that

10 we should not let administrative difficulties justify a

11 bad test, we shouldn't ignore administrative difficulties

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

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

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

15 been articulated in the appellate courts.

16 Well, this leads to the rule of reason

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

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

19 heard a lot of very interesting and provocative arguments

20 for the specific nature of the test.

21 I want to oversimplify by assuming that if you

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

23 that you can differentiate them along a spectrum from

24 relatively strong enforcement to relatively weak

25 enforcement. In other words, they can be differentiated


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

2 anticompetitive by how strictly they would enforce against

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

4 duty to deal.

5 So, the policy for the courts and the antitrust

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

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

8 think that IP rights, intellectual property rights, might

9 affect the answer. And here's why.

10 Intellectual property rights are a primary

11 source of legal barriers to competitive provision of goods

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

13 the testimony yesterday from some of the company

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

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

16 them to deal in particular ways with their intellectual

17 property. Intellectual property rights grant them a legal

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

19 to invent around to innovations that would replicate their

20 invention, and therefore gives power, creates an effect

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

22 particular terms.

23 But, logically, any reduction in the strength

24 and availability of IP protections could reduce the pool

25 of goods for which there are legal barriers to competitive


1 supply. There is an empirical question buried in here

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

3 reform could therefore affect the frequency with which

4 refusals to deal weaken the conditions for being

5 anticompetitive, in turn affecting the likelihood that

6 enforcement of the duty to deal was warranted.

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

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

9 ability to use IP protections to block competitive supply

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

11 of reason exceptions to Trinko's presumption against

12 mandatory dealing with rivals.

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

14 intellectual property reform and a fairly liberal duty to

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

16 deal.

17 Well, both intellectual property reform and

18 duties to deal aim to reduce barriers to competitive

19 supply and innovation, but I think that their individual

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

21 together.

22 Suppose that we do not have IP reform and that

23 there is some good that is being used anticompetitively to

24 block competitive supply. The duty to deal can increase

25 welfare with no risk of deterring investment or innovation


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

2 buyer or third parties could be blocked by an intellectual

3 property barrier to competitive supply or innovation, and

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

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

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

7 investment by the incumbent. That is something that we

8 need to think about.

9 With reduction of legal barriers through IP

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

11 competition and innovation, reducing welfare. So, the

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

13 protected by intellectual property, if they now have a

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

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

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

17 So, IP reform raises the likelihood, whether to

18 any significant level is another question, but it raises

19 the likelihood of false positives in antitrust enforcement

20 through imposition of a duty to deal where the conditions

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

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

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

24 are several kinds of proposals for intellectual property

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


1 There's just some broad examples

2 There are proposals to raise the bar for

3 patentability: better pre and post grant opposition

4 procedures; more transparent review, both in initial grant

5 and post grant of patent grants or annuities.

6 There are also proposals to reduce consequences

7 of patentability: a narrowed patentable subject matter,

8 for example, cutting software out of patentable subject

9 matters; expanded research exceptions and reduced

10 presumptions of harm in injunction proceedings which might

11 push parties to the bargaining table; and limit refusals

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

13 National Academy of Sciences' proposal, in the Federal

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

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

16 variety of ongoing documents one can find these proposals.

17 So, the effects of these proposals would likely

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

19 make those protections less expansive. Some of the most

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

21 important point, would reduce the ability of incumbents to

22 foreclose competitive provision of goods through the

23 exercise of intellectual property rights.

24 Depending on circumstances, these refined IP

25 protections could have varying effects on incentives to


1 deal. The reduced ability to foreclose competitive

2 innovation through the enforcement of an intellectual

3 property right might make an incumbent more eager to sell

4 to rivals because it would expect greater competitive

5 entry in the relevant property market than existed

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

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

8 Alternatively, an incumbent may be less eager to

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

10 likelihood of competitive entry compared to what would

11 occur if it keeps the good to itself.

12 And which of these incentive effects occurs

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

14 degree to which the selling firm is vertically integrated.

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

16 But I think on the whole refined intellectual

17 property could reduce the incidence and the impact of

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

19 could reduce the willingness to deal with rivals by

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

21 innovative alternatives to its technology. But I think

22 this effect is most likely where the goods involved are

23 easy to reverse engineer and replicate. And these in

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

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


1 eventually be able to market.

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

3 intellectual property protections should either reduce

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

5 effects of refusing to deal by opening the door to

6 competitive supply and innovation.

7 So, what are the implications for Section 2

8 reform? The latter effect, competitive reinvention or

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

10 be preserved. Antitrust reform should not impede a

11 competitive reinvention because they should not provide an

12 alternative or option to competitive entry or invention or

13 innovation where it is feasible to occur.

14 So, I think that if intellectual property reform

15 reduced legal barriers to competitive production of the

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

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

18 presumption against mandated dealing could create a

19 quasi-regulatory alternative to buyers that is unnecessary

20 and unhelpful to economic welfare.

21 So, that's some questions to investigate before

22 we know whether intellectual property reform is actually

23 going to matter.

24 Several key questions. First of all, how likely

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


1 consequences of IP protections for competition. I think

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

3 political and very contentious. What will emerge from

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

5 hard to know exactly what.

6 The next question is really an empirical one and

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

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

9 protected goods for which the barrier to competitive

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

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

12 considerations I'm suggesting can be put aside as

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

14 particular industries or markets, then refusal to deal

15 policy should recognize the welfare and complexities that

16 intellectual property reform might introduce.

17 And the final question is: What effects will

18 applied intellectual property protections have on the

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

20 that's an interesting question to investigate.

21 So, I have some tentative conclusions.

22 The rule of reason approach for refusals to deal

23 has potential advantages over either per se legality or

24 the essential facilities test.

25 The policy problem is to decide how strict a


1 test the courts and agencies should apply in assessing the

2 reasonability of refusals to deal with rivals. And the

3 potential results of intellectual property reform may be a

4 relevant consideration in that choice, with more refined

5 intellectual property rights weighing in favor of less

6 strict enforcement against refusals to deal.

7 Thank you.

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

9 going to take a break for roughly fifteen minutes.

10 (A brief recess was taken.)

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

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

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

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

15 that were raised by the other panelists.

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

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

18 get this transcript.

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

20 collusion over here on the right.

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

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

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

24 or comments.

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


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

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

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

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

5 keeps its price low.

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

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

8 proposals, as compared with other tests that are typically

9 used in this area, would increase the ability of

10 competitive firms already in the market to raise rivals'

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

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

13 then basically telling the five hundred dollar firm, you

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

15 some kind of antitrust scrutiny that you will find

16 unpleasant.

17 Is the raising of rivals' cost prospect greater

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

19 just something that I think ought to be thought about

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

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

22 consideration of whether or not the five hundred dollar

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

24 out, the monopolist has the greater ability to sacrifice

25 profits because it has obviously much higher net profits.


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

2 competitive strategy angle here, if the five hundred

3 dollar price is not three hundred dollars above the

4 competitive equilibrium, but a hundred dollars over the

5 competitive equilibrium, we might worry a little bit less

6 about the five hundred dollar price being the one that

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

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

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

10 the current test? And would that matter to your

11 recommendation of what do in in a particular case?

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

13 one.

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

15 probably right. I would take pretty strongly the

16 perspective that the competitive process is about having

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

18 competitive equilibrium price is likely to be, and that

19 therefore enforcement of competition policy and antitrust

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

21 the competitive price would be X.

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

23 process, as I understand it, operates through the

24 formation of a blocking coalition that make the

25 participants better off, without an inquiry into how much


1 the incumbent loses from this entry.

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

3 literature, the usual citation is the Mankiw and Whinston

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

5 way that regulation has traditionally treated

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

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

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

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

10 competition policy is I think that the competitive process

11 works precisely by ignoring the effects on the incumbent.

12 And obviously if you want to increase welfare in the

13 small, ignoring something like that that could be quite

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

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

16 rather well.

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

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

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

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

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

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

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

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

25 precedence.


1 I actually think that you're on pretty good

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

3 understood your comments about "Barry Wright" correctly,

4 that that case made the mistake of thinking that downward

5 pricing was more important than the competitive process.

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

7 know if that's unfair or not.

8 And certainly in Arizona against the Maricopa

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

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

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

12 the competitive process and making sure it works well.

13 So, there might be some legal standing for you

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

15 processor oriented thinking instead of the price oriented

16 thinking that polluted the predatory pricing process.

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

18 if you want to respond to that.

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

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

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

22 goes.

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

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

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


1 very apt to know, like costs.

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

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

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

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

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

7 may want to only worry about firms thwarting rivals from

8 providing very substantial value increases to consumers,

9 and not worry about situations where they are only

10 providing minimal value increases. And if the prices are

11 already pretty close to the competitive level, then you

12 won't find rivals offering to provide very substantial

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

14 antitrust interferes very much in those circumstances.

15 But now you wanted to respond to what he just

16 said.

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

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

19 There's no doubt that sacrifice tests and cost

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

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

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

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

24 use intent evidence in an antitrust case.

25 It seems to me that intent is what you can


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

2 careful using intent evidence. Obviously there is the

3 pervasive problem of testosterone poisoned sales managers.

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

5 available evidence as to contemporaneous estimates of

6 likely effects.

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

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

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

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

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

12 cost information might be useful.

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

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

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

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

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

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

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

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

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

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

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

24 the permanent price.

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


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

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

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

4 determinative role there, but it might be relevant to

5 thinking about that question.

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

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

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

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

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

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

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

13 all the various situations to determine whether something

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

15 hope for? Is this something beyond our ability?

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

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

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

19 here.

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

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

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

23 circumstance. Whatever test you think is right is going

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

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


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

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

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

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

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

6 wrong answer.

7 MR. COHEN: Anyone else?

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

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

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

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

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

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

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

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

16 than settings which involve some of type of tortious

17 conduct? Are there generalities that might guide us?

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

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

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

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

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

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

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

25 for substantial diminution in competition, well, that


1 suggests substantial diminution in competition is likely.

2 So, the test can be very relevant from that

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

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

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

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

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

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

9 instances, you included a temporal dimension. You

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

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

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

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

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

15 to comment on those temporal formulations?

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

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

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

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

20 the first place.

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

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

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

24 what time scale you're looking over.

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


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

2 that might tell you about something about what time scale

3 you're looking over.

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

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

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

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

8 MR. MATELIS: This is a question about false

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

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

11 I suppose a slightly more spirited defense of

12 the concept of false positives, which the Supreme Court

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

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

15 likely to fix false positives, whereas false negatives

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

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

18 pricing jurisprudence, where plaintiffs were winning cases

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

20 Is this really a concern? Is the Supreme Court

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

22 concern overstated in general? How should this play a

23 role in devising antitrust policy?

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

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


1 the question as I think you intended.

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

3 Type 3 error [laughter].

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

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

6 results oriented test like efficiency or consumer welfare,

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

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

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

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

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

12 application you may get it wrong.

13 And the question is: Will people so fear that

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

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

16 whether that's process or results interpreted.

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

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

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

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

21 apply something that substantively doesn't make much

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

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

24 coming from bad jury instructions, make better jury

25 instructions. If they are coming from courts having an


1 insignificant standard of proof where it seems sufficient

2 to allege that something bad happened rather than to

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

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

5 that something really very bad happened, rather than just

6 a little bad.

7 So, I see the problem of false positives as

8 being less in the precedents than in the applications of

9 the facts.

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

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

12 positive standpoint. They look worse from the false

13 positive standpoint at the beginning when the rule is

14 articulated, then after there has been experience gained

15 in its application.

16 I think that, as an agency gains familiarity

17 with the application of a rule, understanding of what

18 certain fact patterns really mean, as courts get more

19 experiences with reviewing cases and get a body of

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

21 more frightening aspects of the rule may be damped down

22 and you may get beneficial application.

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

24 false positives between public enforcement and private

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


1 agencies' abilities to gain a body of knowledge and

2 understanding that they then bring to bear in their

3 enforcement discretion under any given rules.

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

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

6 any particular district judge's reading perhaps outside of

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

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

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

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

11 development of a body of precedence and knowledge in the

12 judiciaries than you get in the agencies.

13 So, I think false positive may be worse for

14 private enforcement than for public enforcement. But, on

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

16 application is the key issue. The deterrence effect is

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

18 think it can be offset in light of experience.

19 MR. MATELIS: Anything to add, Joe?

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

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

22 question based off of something Aaron has mentioned twice

23 now.

24 What are better jury instructions that we should

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


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

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

3 instructing them?

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

5 the two best candidates that I think we should be

6 instructing them clearly on, whatever we think the right

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

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

9 process oriented test, which is that someone is being

10 blocked from providing higher value to consumers, which is

11 a process oriented test.

12 And the instruction should of course distinguish

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

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

15 that's not the issue.

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

17 really -- I think the first thing before suggesting

18 approving jury instructions is to come to a clear

19 understanding of what antitrust is trying to accomplish.

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

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

22 would say the improvement to jury instructions would be to

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

24 the primary problem with them now.

25 MR. MATELIS: Howard, Joe?


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

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

3 would want to hear is: The following specific questions

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

5 please interpret them to the extent possible in light of

6 the kind of fundamental things that Aaron was mentioning.

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

8 Joe's presentation.

9 I really started with three questions, but as I

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

11 throw it out in various forms.

12 You talked some time early on about whether the

13 results of not being able to successfully form a blocking

14 coalition results from actions of the five hundred dollar

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

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

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

18 worked.

19 But then your rule you were trying to focus on

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

21 the incumbent, the five hundred dollar incumbent,

22 strategically thwarts the coalition.

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

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

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


1 whether this approach would make it unlawful for a low

2 cost producer merely to develop the reputation as an

3 aggressive price competitor.

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

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

6 this core analysis?

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

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

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

10 anticompetitive component, anticompetitive conduct.

11 And if you accept the ideas that are being put

12 forward about what anticompetitiveness means, then there

13 can be conduct that is anticompetitive that is harmful for

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

15 than being harmful to competition.

16 Now, there certainly has been a body of thought

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

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

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

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

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

22 the defendant did some things that harmed consumers by

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

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

25 sources here, I think.


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

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

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

4 the question is what further requirement is being asked

5 for here.

6 I think the further requirement that's being

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

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

9 deterrent effect on conduct that sounds like this when

10 described. And that deterrent effect will extend of

11 course to other places where the competitive implications

12 of the conduct might be a little bit different.

13 And so what you want in addition to finding this

14 conduct was inefficiently anticompetitive and

15 anti-consumer here, you want some degree of confidence

16 that similar-sounding conduct is going to tend to be not

17 such a good thing or a bad thing, in other circumstances

18 where maybe it won't be inefficiently anti-consumer,

19 anticompetitive.

20 Well, that puts a lot of weight on the

21 psychological or even philosophical concept of conduct

22 that sounds like this. There's a philosopher named I

23 believe Grice, who really tested foundations of that kind

24 of thing by inventing a word, grue, g-r-u-e, which means

25 green up until this morning or blue after this morning.


1 And so all of your past observations that trees are green

2 are also observations that trees are grue. What do you

3 predict the tree color will be this afternoon.

4 Obviously that's playing with words in the way

5 that philosophers love to do, but it does suffice to make

6 the point that, if what you are looking for in a, quote,

7 "bad conduct" problem is something along the lines of

8 similar conduct that is going to be bad in other

9 circumstances, you need a concept of what's similar. And

10 that's not really an economic concept, as far as I can

11 tell. It's some sort of intuitive or possibly legal

12 concept.

13 MR. COHEN: Anyone else?

14 I'll shift ahead because your comments invite

15 this.

16 What kind of difficulties would you expect

17 courts have in operationalizing something like this? I

18 would hate to go in and try to tell them that trees are

19 green in the morning but blue later.

20 MR. FARRELL: Well, just to be clear, at least

21 in my own mind, I would be delighted if judges were to

22 listen, and when we get around to writing, read this kind

23 of stuff. But I am not convinced that it's ready for

24 courts yet.

25 What I think I would like courts to do is put up


1 a lot of resistance to the incorrect tests that are being

2 bandied about on the pretext of administrability, bright

3 line, sort of vaguely right, perhaps, maybe, although we

4 can't exactly tell you why.

5 And I would like to see courts, led by the

6 Supreme Court, say, look, we really have not sorted out

7 yet what administrable concrete tests we need to apply for

8 Section 2 liability. For the time being let's do

9 so-and-so, but that's not meant to be the final answer.

10 Because I think it's pretty clear that nobody is

11 in a position to say yet what the final answer should be.

12 And I think there's a huge danger, given the way courts

13 and lawyers tend to think and talk, that things are going

14 to congeal prematurely.

15 MR. COHEN: I'm wondering if you're at a point

16 yet where you could predict if there are particular types

17 of conduct where the analysis you're thinking of is really

18 likely to lead to different results than you've been

19 getting through viewing perfect competition as the goal?

20 You may go through a different process. Do you have any

21 idea where the results are likely to come up?

22 MR. FARRELL: No. I think the salient

23 differences are going to be based on the question of how

24 closely you try to examine direct efficiency consequences

25 versus trusting the competitive process to do that and not


1 requiring it in the narrow instance.

2 You know, technically if there is a perfectly

3 competitive equilibrium in an economy, it is then in the

4 core. And so I don't think there is a substantive

5 tension between the two. I think it's more a question of

6 what process each one suggests to you.

7 It seems to me the core -- and let me stress,

8 I'm not suggesting ever examining an outcome to see

9 whether it is in the core. I'm suggesting the process

10 that is suggested by that, which is, make it relatively

11 easy, or don't allow it to be made artificially difficult

12 to form blocking coalitions.

13 Whether there is a similar process that is

14 suggested by thinking about perfect competition, I am not

15 quite so sure. You know, economists have talked for a

16 long time about the fact that perfect competition is

17 describable as an outcome, and we don't have a very good

18 story about how you get there. There's the infamous

19 Walrasian auctioneer. That's obviously not a process that

20 takes place in reality, let alone is protectable by

21 antitrust.

22 It seems to me that thinking about the coalition

23 formation model gives you a stronger suggestion about what

24 process to protect than thinking about perfect

25 competition.


1 MR. EDLIN: I'll hazard a guess, which is, if

2 you thought about things a little more the way that Joe

3 and I think about things, then you would find that the

4 Department of Justice would probably have won the American

5 Airlines case; that entry would be easier in many

6 industries because monopoly or dominant firms would have

7 more limited ability to thwart entry; more attempts by

8 monopolies to prevent entry by tying goods together would

9 be illegal, but not all; and those would be the kinds of

10 things that you would see in terms of substantive outcome

11 differences.

12 MR. SHELANSKI: I will just add that I think the

13 process emphasis, while extremely important theoretically

14 and at some level is absolutely correct economically does

15 have some pragmatic difficulties.

16 I actually really worry about instructing juries

17 on the process as opposed to outcomes. And you can

18 combine the two to halve their inquiry, but I think the

19 confusion between competition and competitor is one very

20 easily sown in juries.

21 And connected to your question earlier about

22 false positives, I think that as a firm, faced

23 particularly with a private suit, knowing the instruction

24 is going to the jury about process, you're worried about

25 looking aggressive, worried about looking the bad guy, and


1 you get a lot of hidden false positives through

2 settlement, particularly in the private cases.

3 So, I do think it's worth thinking a lot more

4 about the pragmatic implications of the process

5 instruction of going forward.

6 MR. COHEN: Finally, for Joe.

7 The theory that you've explained depends on the

8 formation of these blocking coalitions. There are

9 obviously impediments to this. You recognize them and

10 they may not always be formed, but at least there's an

11 incentive to do them.

12 Have you thought about how we should take into

13 account the fact that not all of these coalitions will

14 ever form in the first place, that there maybe information

15 problems or the cost that prevents them from happening?

16 How do we bridge from incentive to actual assumption that

17 they're there and therefore that their losses are

18 significant?

19 MR. FARRELL: I don't. I mean, I think, as I

20 think I mentioned, the way you prove that a competitive --

21 that everything in the core is Pareto efficient, is by

22 pointing to the so-called grand coalition of everybody, if

23 it was prey to inefficient, then in theory this grand

24 coalition could block. That's obviously not going to

25 happen.


1 So, I think any policy, including antitrust, is

2 not going to be able to get us all the way to Pareto

3 efficiency, whether it thinks of it in terms of central

4 planning, price-taking equilibrium or the core.

5 Now, as related more directly on a practical

6 point, which is, well, what happens if -- this is I think

7 maybe what you were getting at with the bad act question.

8 What happens if we have a not very good outcome in the

9 status quo and the blocking coalition that, quote, ought

10 unquote, to form doesn't form, not because of anything

11 that the incumbent does, but just because it's really hard

12 to form.

13 Well, I think at some level that could be a

14 competition policy question. There might be changes that

15 could be made in the way the market works to make it more

16 likely that such coalitions would form.

17 If it were a competition policy question, it

18 wouldn't necessarily be an antitrust question. I think

19 they're potentially distinct areas. And it might be

20 neither. It might just be, well, that's too bad, that's

21 one of the imperfections of the world.

22 MR. MATELIS: At the beginning of these

23 hearings, both the Assistant Attorney General and the

24 Chairman of the FTC stressed the importance of safe

25 harbors for guiding businesses that are seeking to comply


1 with the antitrust laws.

2 And, Joe, I have a question for you. The

3 examples in your presentation were responses of a firm to

4 new entry. Northeast's response to Sprite's entry and the

5 A and B product potential responses at the new entry.

6 Are there responses to new entry that, you know,

7 looking at things through the core, should be within a

8 safe harbor and something that firms should always feel

9 comfortable doing?

10 MR. FARRELL: Well, I'm sure there are, but just

11 as I don't know exactly what the right rules for

12 liabilities should be in a practical sense here, I also

13 don't know what the right rules for safe harbor should be.

14 I mean, one can give the following answer, which

15 is sort of in the spirit of something Tim Bresnahan has

16 said, and you will be hearing from him this afternoon,

17 that the safe harbor is to make your money by being nice

18 to consumers, not to make your money by being the other

19 stuff you can be. That's not quite the way Tim put it,

20 but he had a somewhat similar line which maybe you can get

21 out of him if you ask him.

22 MR. COHEN: Directing some questions to Howard

23 Shelanski's presentation.

24 You focused very much on intellectual property,

25 the effects of possible changes in that area, bleeding


1 over into how we might look at Section 2 issues.

2 If we're looking at Section 2 issues, we're not

3 likely to have differential treatment of instances in

4 which there are lateral refusals for intellectual

5 properties versus others.

6 Would your rule somehow -- are you envisioning

7 somehow distinguishing between the two, or just a one size

8 fits all modification?

9 MR. SHELANSKI: One size fits all is what I'm

10 looking at. I'm actually not so much proposing a

11 particular rule, because I agree with you there should not

12 be two rules. Obviously the precedent is a little choppy

13 between the various circuit courts on the extent to which

14 you get special Section 2 protections for intellectual

15 property.

16 But my view is you should not have a separate

17 rule. And I was really looking at the macro level. If

18 you take the total pool of goods that firms refuse to deal

19 with, some of them are going to impose barriers because

20 they're legally protected, legally blocked by IP.

21 The smaller the pool of goods where there's an

22 anticompetitive refusal to deal, the less enforcement

23 minded you want to be against refusals to deal.

24 So, for me it's really an adjustment mechanism

25 about how permissive or strict a unitary rule you apply.


1 I mean, if you were to look and see, boy, a lot of these

2 refusals to deal cases have at their core intellectual

3 property. Then I think intellectual property would not,

4 say, have a different rule for those cases versus others,

5 but it would say we can have a more permissive rule

6 towards refusals if we had intellectual property

7 enforcement.

8 MR. COHEN: One thing that you mentioned a

9 number of times in your talk was issues about the degree

10 to which imposing liability or not imposing liability for

11 refusals to deal might affect innovation, might affect

12 efforts invent around whatever problem there is.

13 It's a little unfair, I know you gave a

14 theoretical presentation, but of course we're very

15 interested in anything empirical.

16 Do you have any -- can you give any summary or

17 are there any indications of what there is out there in

18 the way of empirical evidence on this?

19 MR. SHELANSKI: If I can cheat a little bit, I

20 think I can. So, I did raise that issue of demand side

21 innovation and competitive supply because I feel that in

22 the discussion about duties to deals there's been

23 overemphasis on deterring the initial innovation by the

24 supplier. I think that's extremely important. And I

25 wouldn't want to see a situation where we punished


1 innovation per se. So, I want to be very careful. But I

2 wanted to build into the demand side there's innovation on

3 both sides of the enforcement question.

4 So, here's a possible place to look for some

5 empirical support, and this is contentious. I would go to

6 the regulatory arena and I would look at the unbundling

7 obligations of the Telecommunications Act of 1996.

8 There are allegations that overly permissive

9 access for competitors to incumbent networks reduced the

10 degree to which these new entrants built their own

11 facilities and their own networks, therefore leading to

12 less vigorous competitive entry.

13 I think there's a lot of debate over the extent

14 to which this is true, but there is some empirical

15 evidence that after the FCC repealed a very permissive

16 access to the incumbent platform under what some would

17 argue were subsidized rates -- there is a legitimate

18 dispute over that -- that after they repealed that access,

19 there was a lot more facilities-based entry, a lot more

20 actual building and installment of competitive facilities.

21 This does suggest that a duty to deal, which

22 would then include some kinds of terms of dealing, runs

23 the risk of stopping entry of competitive assets into

24 other markets. And the telecommunications market might be

25 one place to look for such evidence. And there is some


1 literature out there with competing arguments about

2 whether the essential facilities treatment or the duty to

3 deal imposed by the Telecommunications Act of 1996 on

4 incumbent networks deterred and chased out new competitive

5 essence.

6 MR. FARRELL: I think part of the reason why

7 people have focused on incentives of the original

8 invention or the original investment is that, of course,

9 that innovation or investment directly leads to social

10 benefits.

11 Duplicative investment is -- I want to avoid

12 taking too narrow a view here, but nevertheless, at some

13 level duplicative investment is wasteful. And while

14 having some of it may well be part of the process and

15 negotiating for voluntary access in the shadow of the

16 threat when you look at the investment is probably a

17 bigger part of the process, I think it's actually wrong to

18 treat reducing the incentive for duplicative investment as

19 a policy downside in itself.

20 Now, it might actually be a kind of shorthand or

21 a proxy for some other harms that you think come out of

22 more mandated sharing than other policies would give you.

23 But I think one wants to be wary of that shorthand.

24 MR. SHELANSKI: I'll disagree slightly. I think

25 you're right that that's something to be taken into


1 account.

2 I think the market conditions under which that

3 duplicative entry would be welfare decreasing are fairly

4 specialized. I don't know how common they are. I think

5 it needs to be taken into account. But while it's a

6 consideration, I am not sure that it's a big enough

7 problem that I would discount -- I certainly wouldn't

8 discount the value of at least some competitive investment

9 or duplicative investment, especially where it's not

10 economically blocked. There's not some kind of natural

11 monopoly or scale kind of argument that would make that

12 investment a not be beneficial end, but where there's

13 simply a legal barrier to producing something that could

14 be produced fairly cheaply. Software would be an example.

15 MR. COHEN: Just one more. I'm going to return

16 to something that Joe just mentioned a couple answers ago.

17 You drew the distinction in a sense between a

18 competition issue and an antitrust issue. Another way of

19 phrasing some of the same points we've already been going

20 over.

21 To the panel just generally: Do you see a

22 difference in your analysis between a competition issue in

23 the sense of maximizing efficiency, and an antitrust issue

24 in the sense of what should be a legal violation?

25 MR. FARRELL: I'm certainly very open to that, I


1 think. First of all, I would not phrase a competition

2 issue quite as maximizing efficiency, for all the reasons

3 we spent all morning talking about.

4 But I think it's perfectly possible for a

5 competition agency, let's say, to discover that

6 such-and-such a market would work a lot more competitively

7 with these ground rules than with those ground rules. And

8 to try to use its influence, perhaps even its legal

9 authority, to have the better rules rather than the less

10 good rules apply.

11 And that doesn't necessarily involve anybody

12 having, quote, done anything wrong. And so I think

13 there's potentially a difference between competition would

14 work better in such-and-such a way than with the status

15 quo, and saying so-and-so has committed an antitrust

16 offense.

17 So, yes, I think there's probably a big area

18 there, actually.

19 MR. COHEN: Okay. Do any of the panelist have

20 any final points they want to make?

21 MR. EDLIN: I'm in favor of lunch.

22 MR. COHEN: Okay, we vote for lunch here.

23 I again want to thank all of our panelists for

24 their thoughtful and insightful remarks. I ask the

25 audience to please join me in a round of applause for our


1 speakers.

2 (Applause.)

3 MR. COHEN: And our afternoon session will begin

4 promptly at 1:30.

5 (Whereupon, at 11:59 a.m., a lunch recess was

6 taken.)




2 (1:30 P.M.)

3 MS. GRIMM: Good afternoon. I would like to

4 welcome everyone to our afternoon session. And I'm glad

5 that you all could be with us today.

6 I am Karen Grimm. I am Assistant General

7 Counsel for Policy Studies at the Federal Trade

8 Commission. I am going to be moderating the session this

9 afternoon, along with June Lee, who is an economist at the

10 Antitrust Division of the U.S. Department of Justice.

11 Before we start, I would like to just go through

12 two housekeeping details. First of all, as a courtesy to

13 our speakers, please turn off all your cell phones,

14 Blackberries, and other devices

15 And, secondly, because these are hearings, we

16 request that the audience not make any comments or ask any

17 questions during the presentation.

18 This afternoon we are honored to have another

19 group of distinguished economists from the University of

20 California at Berkeley and Stanford University to offer

21 their testimony in these series of Section 2 hearings.

22 Our afternoon panelists, like those this

23 morning, will provide their perspectives on various issues

24 related to the complex area of Section 2 jurisprudence and

25 enforcement.


1 Our panelists this afternoon are Timothy

2 Bresnahan, who is the Landau Professor of Technology and

3 the Economy in the economics department at Stanford

4 University; Richard Gilbert, who is a professor of

5 economics at the University of California Berkeley and the

6 chair of the Berkeley Competition Policy Center; Daniel

7 Rubinfeld, who is the Robert L. Bridges Professor of Law

8 and Professor of Economics at the University of California

9 Berkeley; and Carl Shapiro, who is the TransAmerica

10 Professor of Business Strategy and Professor of Economics

11 and the Director of the Institute of Business and Economic

12 Research at the University of California Berkeley.

13 Our first three panelists will make

14 presentations, and Professor Shapiro will be participating

15 in the discussion with his fellow panelists.

16 Our format this afternoon is as follows: Each

17 speaker will make a 20 to 30 minute presentation. After

18 all the presentations have been completed, we will take

19 about a 15 minute break. And after that break we will

20 reconvene for a round-table discussion. We are scheduled

21 to conclude this session about 4:30.

22 I would like to thank all of you for being with

23 us here today. I want to thank all of our panelists for

24 coming and for their participation. We very much

25 appreciate the time and effort all of them have put into


1 preparing their presentations and their willingness to

2 share their insights with us.

3 I would now like to turn the podium over to my

4 DOJ colleague and co-moderator, June Lee, for any remarks

5 she would like to make

6 Ms. Lee: The Antitrust Division of the

7 Department of Justice is pleased to co-sponsor today's

8 single-firm conduct hearing. As noted by Joe Matelis this

9 morning, five of today's panelists were Deputy Assistant

10 Attorneys General in the Antitrust Division. Four of the

11 five are in the panel. I thank them for participating

12 and, like Karen, for sharing their insights. I look

13 forward to their presentations in what I'm sure will be a

14 lively discussion.

15 I join Joe in thanking the Competition Policy

16 Center and the Berkeley Center For Law And Technology at

17 the University of California Berkeley for hosting these

18 hearings. And I thank Karen and her colleagues at the FTC

19 for their work in organizing today's hearing and

20 assembling the august panel we have today.

21 Karen.

22 MS. GRIMM: Our first speaker this afternoon is

23 Timothy Bresnahan, who is Landau Professor of Technology

24 and the Economy at Stanford University and Chair of the

25 department of economics.


1 He is Director of the Center for Research in

2 Employment and Economic Growth in the Stanford Institute

3 for Economic Policy Research. He also has served as Chief

4 Economist of the Antitrust Division of the U.S. Department

5 of Justice.

6 His research interests lie in the economic of

7 industry, especially of high technology industry.

8 Professor Bresnahan received his B.A. from

9 Haverford College and his master's degree and Ph.D. in

10 economics from Princeton University.

11 Tim.

12 MR. BRESNAHAN: Thanks for that very nice

13 introduction. Let me see if I can find my slides.

14 While I'm finding my slides, let me confess that

15 in my role as department chair, I worked with the agencies

16 in a failed effort to bring these hearings to Stanford

17 rather than Berkeley. If you think of the reputations of

18 those two great universities, you might infer that signals

19 a leftward shift in the antitrust enforcement effort.

20 But I don't think that's what it signals. If

21 you either look over here to my left or at the brochure

22 from the Competition Policy Center, you can see why

23 Berkeley is an enormous center of academic influence in

24 this area. This was the right place to put it.

25 I want to talk about monopolization (Section 2)


1 cases. And my real agenda is to normalize them, to

2 regularize them within antitrust analysis.

3 We have a tendency in talking about Section 2

4 matters to immediately leap to the most difficult part,

5 which is the part that's about alternative efficiency

6 theories of whatever business practice it is that's

7 challenged in the Section 2 matter.

8 I think that that makes Section 2 matters more

9 difficult than they need to be, and I'm going to propose a

10 different approach, not inconsistent with what we've

11 done in the past, and which we'll see in a minute, not

12 inconsistent with recent court decisions.

13 I'm going to suggest a different approach where

14 we look at competitive effects first. It's not very

15 surprising that I want to look at competitive effects

16 first since I'm an economist.

17 And then I think I'm going to argue it's going

18 to make thinking about whether a Section 2 case is

19 procompetitive much easier than starting from that very

20 difficult question of whether the challenged practices are

21 an act of competing rather than anticompetitive act. So,

22 I'm going to start with competitive effects.

23 There's been a good bit of action in the courts

24 in Section 2 lately. You know, I see three big topics

25 here: boundaries with other parts of the law, notably with


1 patent and copyright law; predatory pricing is another area.

2 What I want to talk about are bundling and

3 related practices. So, vertical Section 2 cases where a

4 monopolist commits monopolization or is alleged to

5 commit monopolization through bundling its monopoly

6 product with something else, or through contractual

7 restrictions that amount to de facto bundling.

8 I'm also going to talk about Microsoft and

9 Dentsply in some detail, but Dentsply first. This partly

10 reflects the idea that I think that the folks who do

11 judicial decisions have the same economics in mind that

12 I'm going to talk about this afternoon. And it's partly

13 that there are three cases, two recent cases in this area.

14 And I found those two, again another confession, much

15 easier to read than I found the LePage's case, which was a

16 struggle for me, although I am not sure it's inconsistent

17 with what I'm going to say.

18 So, any Section 2 inquiry I think has at its

19 heart an economic structure if it's a rule of reason

20 inquiry. Any rule of reasoning inquiry has economics in

21 it. I think the economics enters at two distinct places.

22 It has to enter in market power. You need economists to

23 figure out market power. And I want to say, as I've been

24 saying for a quarter of a century, the thing about market

25 power is sometimes it's a useful shortcut in antitrust


1 enforcement. We should be thinking about competitive

2 effects when we're thinking about market power, particularly

3 I would encourage the agencies, when picking cases in the

4 merger area or in the Section 2 area, to pick cases where

5 there's potentially a substantial change in the conditions

6 of competition in the market and significant impact on the

7 economy. That's not the same as market power. That's a

8 change in market power.

9 The other place where economics matters is in

10 thinking about the causal flow from the acts which are

11 alleged to be anticompetitive in a Section 2 case to the

12 changes in market power. And I'm going to argue, this is

13 my theme for the afternoon, you can gain a lot of clarity

14 about a Section 2 case by bringing the competitive effects

15 and causation arguments to the forefront. And I think that's

16 consistent with the three bundling cases I cited, bundling

17 or tying cases, I cited on the previous page.

18 Section 2 cases are never going to be easy.

19 Let's be real. There's a reason for that. This is I

20 think the hardest part. Almost all conduct which would be

21 exclusionary in some context would be an ordinary and

22 competitive business practice in some other industry. So,

23 it's necessarily context specific. That makes it

24 difficult I think for attorneys to get their heads around

25 Section 2 matters all the time because it seems like


1 there's a fairly unstructured rule of reason analysis in a

2 Section 2 case.

3 I'm going to argue again that monopolization can

4 lead you to a fairly structured economic competitive

5 effects decision. Let me do that right away. I'll do it

6 in Dentsply first.

7 This is a Department of Justice case. I know a

8 part of the history of it. I believe it was brought when

9 Dan Rubinfeld was Chief Economist. It was litigated when

10 I was Chief Economist. And I just learned from Professor

11 Shapiro that it was under investigation on his watch.

12 MR. GILBERT: It was under investigation at the

13 FTC before I was at DOJ.

14 MR. BRESNAHAN: Exactly. We are lucky that

15 prefabricated artificial teeth is not a market which

16 changed quite so quickly as computer software. But I note

17 that the other case I am going to talk about, Microsoft,

18 has a similarly long, long series of investigations before

19 there was a serious enforcement action.

20 So, what's the story of Dentsply? Why did the

21 Department of Justice bring a Section 2 action?

22 So, part of it, there is a market definition,

23 there is monopoly power, and there is, in the current

24 market, a monopoly in prefabricated artificial teeth.

25 There are some small sellers, but there is one great big


1 seller named Dentsply.

2 Now, here's the competitive effects part. And

3 this is something I think that's a reason that's going to

4 make cases fairly rare in monopolization. While there is

5 a monopoly in prefabricated artificial teeth, there could

6 be substantially more competition in the market from a

7 number of non-Dentsply like artificial teeth -- prefab

8 artificial teeth providers who are smaller, very small at

9 the time the case was brought, and typically lower priced.

10 And it's the difference between the competitive

11 regime there is, monopoly, and the competitive regime

12 there could be, much less monopoly, which is the

13 competitive effect that I think we should bring to the

14 forefront.

15 If it's inevitable, if Dentsply has a

16 monopoly that cannot be changed, if there is some barrier to

17 entry which cannot be lowered by any earthly force, there

18 can still be a monopoly but how can there be

19 monopolization? Monopolization I think needs to be

20 cause of a change in the competitive regime or prevention of

21 the change to a competitive regime that otherwise might arise.

22 Now, in this case, the mechanism, you need a bad

23 act as well as a competitive effect to have a Section 2

24 case. The mechanism by which Dentsply prevented the

25 emergence of competition from these other firms was


1 exclusive contracts with dealers. They were dealers who

2 supply dental laboratories with all kinds of things, but

3 in particular with prefabricated artificial teeth. And

4 those contracts block the laboratory from sourcing another

5 firm's teeth, preventing the American consumer from

6 having an effective prefabricated tooth choice.

7 You know, there's a market in everything. Some

8 of it might be competitive. As you get older, you get

9 more serious about the importance of health care markets

10 for having a competitive organization. And, Lord knows,

11 there is not enough competition in most health care markets.

12 So, I want to bring to the forefront, the

13 horizontal competitive effects. Impact, if there's a

14 Section 2 case, the impact of the bad acts, the contracts

15 in this case, is to reduce competition in the market for

16 prefab artificial teeth. So, it's possible that there are

17 two competitive regimes, one with monopoly and the other

18 with competition.

19 And I want to push to the second, the vertical

20 restraints logic, that the economic effects of these

21 contracts, these exclusive contacts, is to change that

22 competitive regime.

23 You know, it seems to me that you can, in the

24 course of investigating an alleged Section 2 violation,

25 discard an enormous number of cases just by thinking about


1 -- not about the efficiency theory of the supposed bad

2 act, but rather just thinking about the anticompetitive

3 theory. The inquiry would ask: is it possible that there

4 could be less competition and also there could be more

5 competition in this industry? Is it possible that if the

6 dealer contracts weren't exclusive that then there

7 could be competition? Without a "yes" to both, further

8 inquiry is not going to lead a Section 2 case. The second

9 question, the exclusivity of the dealer contracts having

10 sufficient impact to change the compeitive regime, that is

11 not a small inquiry. There is a lot of assumptions under

12 there.

13 There are at least two base assumptions. The

14 monopolist, Dentsply, is in a position to compel the

15 dealers to accept these exclusive contracts. That's not

16 going to be true in all industries. There can't, for example,

17 be the possibility of some other parallel distribution segment

18 which can grow up and distribute the competitive prefab

19 teeth. Furthermore, while the distribution channel firms must

20 not be in a position to resist Dentsply, Dentsply's competitors

21 must need the distribution channel. Thus, the distribution

22 channel must be dependant upon Dentsply but depended upon by

23 the competitors. Not all distribution channels will satisfy

24 both conditions. So, there's a reason that these exclusive

25 dealership contractors have bite. Bite, it was entirely


1 accidental pun. I think if we could go down the path of the

2 Dentsply puns, they would be very unhappy for us.

3 But I mean to emphasize that there are two

4 dualities just in the competitive effects part of a

5 Section 2 case, which means, before you get to the hard thing

6 about efficiencies, you could throw a lot of cases out. It

7 has to be possible that there's two competitive regimes,

8 monopoly and more competitive, and it has to be possible

9 that the bad act works to move the market between them,

10 and that itself has two steps. The little guys, the

11 potential competitive providers of these competitive teeth,

12 have to need the distributors. The distributors need to be a

13 powerful hard-to-replace force. And the existing monopolist,

14 Dentsply, has to be able to kick around the distributors.

15 So, you've got two dualities, it's monopoly, but

16 it might be more competitive. And the distributors are

17 important, but the monopolist is in a position to either

18 bribe them or compel them to prevent the outbreak of

19 competition, competition which would be plausibly in their

20 interests.

21 Those two dual tests I think will weed out a lot

22 of cases before you begin this open-ended discussion of

23 whether these particular contracts are efficient. So,

24 here is how I graph it. You've got -- your centerpiece

25 should be the anticompetitive effects. So, in


1 monopolization case, the effects are anticompetitive.

2 There is an exclusionary act, in this case the contracts,

3 which is keeping us in a higher market power monopoly, in

4 this case industry regime rather than a lesser market

5 power.

6 And, as I said, that's a lot for the plaintiff

7 to show. In the case of the agencies, that's a lot for

8 them to show. And I want to urge a review of whether we

9 can show these things early in a case. When I said to

10 kind of regularize Section 2 review, you know, it's just

11 like merger review, is there a competitive effect this merger

12 is going to do? Is there a competitive effect these are bad

13 practices are going to have, too? Is it really true that

14 there is more market power in the current regime but there

15 could be less market power? And that is the centerpiece,

16 that there is this causation, there's these

17 exclusive contracts, which exist because the existing

18 monopolist wants to maintain a monopoly, or what's keeping

19 us in the less competitive regime rather than the more

20 competitive regime.

21 And I think if you do both that causation

22 carefully and that competitive effects carefully that

23 would make Section 2 cases look a lot more like ordinary

24 antitrust analysis.

25 So, I said a number of times that that's a lot


1 to show. It has to be possible that the competitive

2 regime could change; it has to be possible that the bad

3 acts are what's preventing the competitive regime from

4 changing; there has to not be another explanation of why

5 the competitive regime is not changing.

6 We spend so much time in Section 2. Here's my

7 one slide. I think I only have one slide and it's sort of

8 ordinary analysis. We spend so much time thinking about

9 whether there's an efficiency theory of the

10 anticompetitive acts. And that is important. But, you

11 know, I guess I would say, solve the problem with whether

12 there's a harm to competition first and then worry about

13 if there's an efficiency theory.

14 A lot of this efficiency discussion -- and here

15 I'm echoing Professor Farrell's earlier remarks in these

16 hearings -- we're driving in the direction of that world

17 of pure economic theory where we can figure out in a

18 quantitatively precise and reliable way whether

19 the consumer of the industry is better off with the

20 existing industry structure, including its contracts,

21 versus some counterfactual regime where the contracts

22 would be gone and there would be less efficiency

23 presumably from the contracts, but also more competition.

24 In economic theory, the author of the model knows

25 everything and could calculate how well off consumers are in


1 another world. In the real world, the ability of

2 empirical economics, even with the very high level of

3 inquisitory abilities of the enforcement agencies to figure

4 out what would happen in that but-for world in enough detail

5 to calculate social welfare seems to me to be a waste of

6 time.

7 So, I would say, plaintiff has to show that

8 there is an anticompetitive effect and that it's causal.

9 And defendant gets to rebut that. Defendant has to show

10 that their practices are efficient. Plaintiff gets to

11 rebut that.

12 If the world is not tired of hearing from me

13 about the Microsoft case, let me talk about that one too.

14 I mostly want to emphasize its parallels to Dentsply.

15 Again, my competitive effects story is in the

16 graph here, I think I'm very close to the D.C. Circuit's

17 logic here. The competitive story is slightly different

18 because the industries are slightly different. And this is

19 one of the inevitable costs of Section 2. Section 2 cases

20 are rare. They arise in those industries where there is the

21 possibility of a big change in competitive circumstances.

22 That's not most industries and that's probably

23 idiosyncratic industries. Certainly these two, the teeth

24 and the software, are both idiosyncratic.

25 So, what's the state of the market? There is a


1 Windows monopoly in operating systems on PCs. That was

2 true when the case was brought. I got the year wrong. It

3 was tendered to the Department of Justice in 1997 or so.

4 There could have been dynamic competition for the operating

5 system market if the mass use of the Internet led to new

6 standards in new markets.

7 So, here -- well, in the case of Dentsply, there

8 was a monopoly and could have been competition in the

9 market for prefab artificial teeth. In the case of

10 operating system software, there is a monopoly and the

11 industry in the past had had dynamic competition where

12 entrants in many important software products had replaced

13 incumbents. And in other important software markets, they

14 had given incumbents a terrible scare and created

15 incentives to get some real innovation out of them. In

16 these software markets, there is persistent static

17 monopoly, but there could be the prospect of

18 Schumpeterian competition. So, that's the two competitive

19 regimes that you get the competitive effects on.

20 The other part of Microsoft is really quite

21 similar to Dentsply. What kept the world in the monopoly

22 regime rather than in the potentially more competitive

23 regime, a regime where say a Linux might have taken a run

24 at the position of Microsoft Windows on the desktop?

25 It was a distribution case just like Dentsply,


1 how could actual distributors, and a wide number of

2 different kinds of complementors, with other third parties

3 that would have worked with something like Linux on the

4 desktop prevented a market test for the Internet entrepreneurs,

5 and thereby ultimately prevented Schumpeterian competition in

6 the operating systems market.

7 So, again, this is the stuff that the Antitrust

8 Division had to prove in Microsoft. That's why it's such a

9 long case. Two potential competitive regimes. One, the

10 present one in operating systems and other infrastructure

11 software on the PC, which is about ten years you've had

12 very little competition in those industries, but in the

13 same industry in the previous twenty years before that you had

14 it all the time. Maybe there could have been, certainly the

15 Microsoft guys thought there could have been, dynamic

16 competition against some of those valuable position if the

17 Internet entrepreneurs had succeeded.

18 Some of this was more complicated and it's vertical

19 in more senses. The Internet entrepreneurs wVere not

20 horizontal competitors for Windows. The browser

21 was a complement. So, this was vertical restrictions to

22 prevent vertical disintegration. The vertical

23 disintegration would have permitted horizontal (dynamic)

24 competition in the operating systems market. So, it's a

25 good thing that the history of the industry had so much


1 vertical disintegration causing horizontal competition for

2 the market, and that the Microsoft guys in their internal

3 documents were so clear about that that such a complex

4 case could be argued.

5 So, there's Andy Grove. Everybody has seen

6 Andy's slide a hundred times. The way you get competition

7 is you get a vertical disintegration. Andy was, when he

8 wrote this, the CEO of Intel. Mr. Gates of Microsoft has

9 said this many times as well.

10 This was the essence of the antitrust case,

11 that the internal documents, used that model of

12 vertical disintegration leading to horizontal competition,

13 provided evidence for the potential change in the competitive

14 regime.

15 Now, again, I want to say, these cases are going

16 to be rare. There's not a lot of industries where

17 vertical disintegration is the key trigger for horizontal

18 competition. It happens to be in infrastructural or mass

19 market software on your personal computer that that's

20 true, and it's been true since the industry was founded.

21 But, the cases where there can be causation from a

22 vertical restriction to horizontal competition are going

23 to be reasonably rare. This was one.

24 I would emphasize again, look for evidence of

25 that causal change before you go worrying about


1 efficiencies.

2 This part is pretty much the same as

3 Dentsply in many ways. Microsoft is more complicated

4 because it's vertical in two senses: vertical

5 restrictions to prevent vertical disintegration, and

6 vertical disintegration in turn preventing horizontal

7 competition.

8 But what was really important in the competitive

9 effects in the case was that chain of causation did lead

10 to blocking of a threat which could have led to the kind

11 of dynamic and very valuable competition we had seen over

12 the previous twenty years in this industry.

13 Microsoft -- this other pragmatic, question about

14 when to bring a Section 2 case, it's helpful to have a

15 defendant that tries to prove entirely implausible things

16 like, there's no market power in Windows. It was a bad moment

17 for their economics expert witness, I think.

18 The other very unwise thing

19 that Microsoft chose to prove was that their reaction to

20 the widespread mass market use of the Internet wasn't

21 strategic, even though there were hundreds and hundreds

22 and hundreds of internal documents saying that it was

23 strategic. The CEO, whose memo I just quoted saying, this

24 is a terrible threat to us, chose to testify that he had

25 no idea what the threatening firm was doing at the time.


1 So, defendent's trying to prove that it wasn't

2 strategic, trying to prove that there was market power,

3 made it somewhat easier for the government to prevail.

4 These are complicated cases. The agencies are not always

5 going to prove both dualities, that there could be a change

6 in market conditions and that the distribution system is

7 essential causally to keeping an out.

8 So, here's another one with a slide. The ultimate

9 remedy chosen in Microsoft was to require divestiture of

10 all applications, including the browser and Microsoft office.

11 This was not on Richard's, Carl's or Dan's watch.

12 This one is on my watch. And I have to say, I had to put

13 up this slide. There slide -- actually there is a long

14 history of this particular slide. When Dennis Yao, who

15 was my roommate in high school, was a Commissioner in the

16 FTC in 1989 or 1990, called me and said, you know, we

17 figured out we don't want to go after IBM and Microsoft

18 together, should we go after Microsoft. And the metaphor

19 immediately leapt to my mind, you're going to be like a

20 dog that's chasing a fire truck, you know, they're rolling

21 down a little street, noisy, illegal as hell,

22 anticompetitive as hell, but what are you going to do with

23 it when you catch it?

24 As it worked out, they didn't catch Microsoft.

25 I did. And the dog in this picture turned out in actual


1 history to be me. What did we get? Not any remedy which

2 changed the conditions of competition. Ultimately, there

3 was an entirely ineffectual settlement in the United States

4 and a mildly effectual settlement in the EU. Certainly not

5 enough remotely to have the kind of competitive conditions

6 change that was possible from the widespread use of the

7 Internet.

8 So, there's another problem with the agencies,

9 bringing large, complicated antitrust cases. The

10 counter example here would be, of course, U.S. v.

11 AT&T. The United States was incredibly well served by

12 that case. During the long interval between the AT&T

13 breakup and the soon-to-happen reestablishment of the

14 Bell System, we were incredibly well served to have

15 vertical disintegration in telephony. The fact that we

16 had vertical disintegration in telephony at the moment in

17 history when, for example, technologists finally figured

18 out how to have mass market use of online services. That

19 was incredibly fortunate and that resulted from the antitrust

20 case. But they can also fizzle. And even if you win a

21 case, there can be severe problems in finding a remedy

22 that the antitrust system will undertake.

23 So, let me go to my bottom line. I really want us

24 to turn around. These cases are going to be hard to prove and

25 I want us to turn around and think about both the potential


1 for a competitive effect, meaning there could be change in

2 the conditions of competition. The form of that change was

3 different with the two cases I talked about. Second, think

4 about a causal link between the alleged act and monopoly. I

5 would bring those to the fore. Those would be my framework

6 for thinking about a Section 2 case.

7 But of course that discussion is only about the

8 question of whether there is an antitrust case. This doesn't

9 remove from the agencies or any other plaintiff, but particularly

10 not for the agencies, the problem of thinking about whether

11 there's enough of a harm to competition at stake to justify

12 any intervention. I guess I would say that in an cases like

13 AT&T or Microsoft, where you've got a substantial impediment

14 to technical progress in an infrastructure industry, that matters

15 to the whole economy, arising from the lack of competition.

16 That one might get you over the hump. But there are other

17 metrics that can be used, such as the size of the differnce

18 between the two competitive regimes and the importance to

19 consumers.

20 And also to think through whether there might be

21 an efficiency defense, whether there might be more harm

22 than good done by the antitrust intervention. I don't

23 want to take that away, but I do want to say that I would

24 emphasize -- I would emphasize thinking through whether

25 there is an antitrust case in a perfectly ordinary


1 antitrust analytical way, competitive effects and

2 causation.

3 Thank you very much.

4 MS. GRIMM: Thank you very much.

5 Our next speaker is Professor Rich Gilbert, who

6 is Professor of Economics of the University of California

7 at Berkeley.

8 From 1993 to 1995, he was Deputy Assistant

9 Attorney General in the Antitrust Division of the U.S.

10 Department of Justice, where he led the efforts that

11 developed joint Department of Justice and Federal Trade

12 Commission "Antitrust Guidelines for the Licensing of

13 Intellectual Property."

14 Professor Gilbert has served as an Associate

15 Editor of the "The Journal of Industrial Economics," "The

16 Journal of Economic Theory," and "The Review of Industrial

17 Organization."

18 Professor Gilbert research specialties include

19 antitrust economics, intellectual property, and research

20 and development.

21 He earned his Ph.D. from Stanford University in

22 1976. He received a Bachelor of Science degree in

23 Electrical Engineering in 1966 and a Master of Science

24 degree in 1967 both, from Cornell university.

25 Professor Gilbert.


1 MR. GILBERT: Thank you very much, Karen.

2 While I figure out how to find my talk here, I

3 will thank you for bringing these hearings to Berkeley.

4 We're very glad we could be able to host these hearings.

5 And here we go.

6 I'm going to talk about a very narrow slice of

7 conduct that could invoke Section 2 liability, namely

8 innovation or product design, and ask the question of

9 whether innovation, certain types of innovations can be a

10 source of Section 2 or contribute to Section 2 liability.

11 Now, I don't think many people would argue that

12 innovation is great for the economy. Nevertheless, there

13 are quite a number of cases that have alleged that

14 innovation or product design has contributed to

15 monopolization. Of course, Microsoft, as we just heard,

16 is one. A slew of cases involving IBM and standardization

17 for complimentary products, the use of complimentary

18 products. There are some interesting cases on the horizon

19 in the prescription drug industry that raise innovation

20 issues in a Section 2 sort of context.

21 So, I'm going to be reviewing some of these

22 cases and asking whether we could have a standard, we've

23 heard a lot about standards this morning to evaluate

24 Section 2 type conduct, whether any of these standards is

25 useful for evaluating innovation. Maybe I will give you


1 my punch line right away. I think the answer is no, and

2 try to tell you why.

3 I'll begin -- let's see. I'm going to begin

4 with a very simple model. I hope not to raise the fear

5 factor too much and talk about letters here. If you are

6 worried about this, you can replace any letters with

7 numbers. So, I want to talk about a very simple model of

8 innovation.

9 I have here an old technology. It has a social

10 value, v, zero, for each use. You could use, say, fifty

11 dollars for v, zero. A new technology could come along

12 with a higher social value, maybe a hundred dollars, for

13 each use. I'm going to strip away marginal cost to keep

14 things as simple as possible. There are a bunch of users,

15 say there's a thousand users, if you want. And there's

16 some R&D costs.

17 Now, in this simple model, the innovation is

18 socially desirable, I mean, it's still the small one can

19 be as simple as possible if the total incremental social

20 value exceeds the cost of the innovation.

21 So, we have our thousand consumers and they each

22 use this technology in one application, the extra value of

23 the innovation is fifty dollars, so that would be fifty

24 thousand dollars. The question is: Does that cover the

25 cost?


1 Now, in terms of whether the innovation is

2 privately profitable, there's a price that the innovator

3 can collect for the new technology and it's profitable if

4 the price it can collect times the number of people who

5 buy it, assuming they all buy it, in fact covers the cost.

6 So, the first that I want to make, and there is

7 a paper that should be coming out in "Competition Policy

8 International" on this topic, the first point is to say,

9 innovations can be socially desirable but not privately

10 profitable, or you can have innovations that are privately

11 profitable but not socially desirable.

12 So, the first point is a very simple point:

13 That innovation can go any way -- there can be any order

14 in evaluating social and private profitableness. It's not

15 like a price -- innovation is like a price change in some

16 respects. If you come out with an innovation for a

17 product, it's like reducing its quality-adjusted price,

18 and you can make an analogy between innovation and, say,

19 predatory pricing. If you reduce the quality-adjusted

20 price, that leads to the exit of competitor, and then you

21 raise your price again, that has a sort of predatory

22 flavor to it.

23 But unlike pricing, where lower price certainly

24 lowers the price above marginal cost is a good thing, we

25 really don't know if more or less innovation is a good


1 thing unless you do the whole analysis.

2 So, the standards I want to talk about, these

3 came up this morning, I want to talk about different rules

4 of reason which I interpret as either a total rule of

5 reason, which looks at all of the economic value

6 associated with some conduct, whether it's value to

7 consumers or value to producers.

8 And then there's probably the more popular

9 consumer rule of reason analysis which focuses on

10 consumers, and some people would say is at the heart of

11 antitrust analysis, at least according to, say, Steve

12 Sala, although others such as Joe Farrell and Mike Katz,

13 and Ken Hirers from the antitrust division, have advocated

14 a total rule of reason standard.

15 Then there's the profit sacrifice test in one of

16 its many forms. There's the no economic sense test.

17 We've heard a little bit about that this morning. And

18 then I'll talk a little bit about sham innovation.

19 So, a total real of reason analysis, in a sense

20 it's the right thing to do if you are, sort of by

21 definition, an economist, it's the right thing to do

22 because it ask whether total surplus is increased from

23 some activity. And even if that makes producers

24 relatively better off than consumers, at least there's the

25 possibility that those producer profits will flow


1 eventually to consumer benefit, or that somehow producers

2 can bribe consumers to get it all right.

3 But the problem of course is that you can have

4 the price being either larger or smaller than the

5 incremental social benefit. And all of the analysis would

6 have to be done when the innovation decisions from the

7 perspective of the decisions that are actually made, which

8 means what we call an ex ante analysis. And this really I

9 think sets up innovation as being distinctly different

10 from other conduct. Because when you talk about

11 innovation, it's absolutely necessary to keep going

12 backwards and backwards to what are the incentive effects

13 of whatever rules or policies you have in place, what are

14 their incentive effects for innovation in the first place.

15 And now it's easy to say, well, of course that's

16 right, of course we're going to take that into account.

17 But I want to ask you, if you have been in these hearings,

18 how many times have people really gone backwards and said,

19 what are the implications of what we're doing for the

20 kinds of decisions that people are make that could have

21 developed and could develop new products or new processes

22 or whatever ten years from now. And I would say you

23 haven't heard it very many times.

24 So, it very easy to lose sight of these

25 incentive effects. And on top of that, if you did a total


1 rule of reason analysis, the analysis that you would have

2 to do is hugely complex. You have to really take into

3 account all spillovers, how innovation affects consumers

4 and firms in other industries, and those we know can be

5 very, very large. And with the complexity, you can lead

6 easily to false positive and false negatives. I'm not

7 going to say Type 1 and Type 2 because I always forget

8 which one is which, so I will just say false positives and

9 false negatives, and you can figure out which one is a

10 positive and which one is a negative on your own.

11 Too much enforcement or too little enforcement.

12 Portion. It can go either way.

13 A consumer rule of reason analysis. Again, it's

14 very complex. The problems are similar to those that

15 arise in a total rule of reason analysis. Again, the

16 ex ante problems, the uncertainties, the spillover

17 effects, etc. And as well can lead to conclusions that

18 just simply don't make sense. This is particularly a

19 problem in innovation. You could have an innovation that

20 just saves millions of dollars in production cost, but

21 maybe it leads to a nickel increase in price, which

22 certainly could happen. And would you want to say that

23 this is an anticompetitive innovation because consumers

24 are slightly worse off, despite the fact that it's

25 generated enormous savings and efficiencies on the


1 producer's side.

2 Well, I know that people can differ on that, but

3 my view is that it just doesn't make any sense to discount

4 all of those efficiencies. Now, you can say that you're

5 looking at a merger case or you're looking at other

6 conduct that doesn't involve product design, that those

7 kinds of efficiencies are not likely to be huge or have

8 not been demonstrated to be huge, but when you're talking

9 directly about innovations these efficiencies exist as

10 part of the innovation. So, you can't discount them.

11 A profit sacrifice test. There are, of course,

12 different versions of a profit sacrifice test. And I'm

13 going to quote Janusz Ordover's and Bobby Willig's

14 definition: "Predatory intentions are present if a

15 practice would be unprofitable without the exit that it

16 causes but profitable with the exit." Now, Ordover and

17 Willig also say this is just talking about predatory

18 intent not facts, they add a lot of other conditions in

19 their analysis that make this analysis considerably more

20 elaborate, and in many ways closer to a total rule of

21 reason analysis. So, this is just the basic idea of a

22 profit sacrifice test.

23 Now, the profit sacrifice test, I am not the

24 first to say this, it doesn't seem to me to make any sense

25 to innovation, even though it was in fact developed


1 originally to talk about innovation as well as price --

2 predatory pricing. The problem of course first of all is

3 that innovation almost always involves a profit sacrifice.

4 It's called investing in research and development. That's

5 what you do.

6 It's also the case that innovation, if it really

7 works, probably excludes competitors. So, exclusion is

8 sometimes a direct result of producing a really good

9 mousetrap. The other mousetraps can't compete.

10 Now -- and furthermore, and this is absolutely

11 crucial, is that we need to know how much market power

12 after the innovation occurs is necessary to justify the

13 investment in innovation in the first place. And you can

14 make statements about whether innovation creates too much

15 or too little market power relative to its social value.

16 But the social value is very hard to calculate. And the

17 amount of power or pricing power that is necessary to

18 evoke the right amount of investment in research and

19 development is simply a very hard question. So, I

20 conclude, based on this, that a profit sacrifice test

21 really doesn't do very much to inform this analysis.

22 What about a no economic sense test. I am going

23 to use Greg Werden's version of this. He says: "Conduct

24 is not exclusionary or predatory unless it would make no

25 economic sense for the defendant but for the tendency to


1 eliminate or lessen competition."

2 Now, you can see that, with all the negatives

3 again, the no economic sense test is really a test of the

4 absence of predation. So, if it makes sense to do this

5 activity, then it's not predatory.

6 Now, although it's not really clear in the no

7 economic sense test what no economic sense means, there

8 are two interpretations of this, certainly as applied to

9 innovation. One is that it's not profitable. No

10 reasonable firm would have dumped all of this money into a

11 new product design unless it had a purpose of excluding

12 competition. A second interpretation is that innovation

13 really always makes economic sense because it's just a

14 good thing that firms do.

15 Depending upon which one of these

16 interpretations you have, if it's the first one, then the

17 no economic sense test is very similar to the profit

18 sacrifice test. Now, if it's the second one, the no

19 economic sense test is similar to really whether

20 innovation is a sham, meaning whether it's a fraud or not.

21 I think it's the case, and I know that Werden has said

22 that his view of the no economic sense test as applied to

23 innovation is the second version, not the first version.

24 And I also know that he has views of conduct that do not

25 in fact involve a profit sacrifice, even though there was


1 some discussion this morning that they're the same. His

2 example was a world with no arson laws and flush with

3 matches. So, you can go out there and burn down anybody

4 you want, including your competitors.

5 So, let me review a little bit of some cases

6 involving predatory innovation, particularly with respect

7 to complimentary products, products that interact with

8 other products. Those are almost totally -- well, they're

9 not entirely, but to a great extent they have to do with

10 changes to the interface stands. That was certainly the

11 case with the IBM peripherals litigation, a bunch of these

12 in the late 1970s, whether it's other people's disk drives

13 would hook into and work with IBM mainframe computers.

14 And the Microsoft case. And there's been a few others.

15 As a general conclusion in looking through these

16 cases, well, you can find a lot of lower court decisions,

17 a general conclusion is that in nearly all of these cases,

18 weak evidence of efficiencies was sufficient to avoid

19 liability for predatory innovation.

20 So, after there was lots of talk about whether

21 there was monopoly power or not, or whether or not there

22 was a monopoly of market effect, competitive effect. The

23 final analysis, they said -- these courts generally said,

24 well, we can think of it as an efficiency reason for this

25 conduct, therefore it's okay.


1 To my knowledge, only the Microsoft case,

2 Microsoft 4 as it is sometimes affectionately called,

3 purported to apply a rule of reason analysis to

4 innovation. So, let's talk about Microsoft a little bit.

5 The Microsoft case actually came up with a road

6 map to kind of evaluate innovation. There actually five

7 steps to the road map. I'm going to condense them to

8 three.

9 The plaintiff first must demonstrate that the

10 conduct that harmed consumers had an economic

11 anticompetitive effect. Second, if a plaintiff

12 successfully demonstrates anticompetitive effect, then the

13 monopolist may prefer a procompetitive justification for

14 its conduct. So, the second step is the monopolist,

15 alleged monopolist can talk facts and say, we have a

16 reason for doing this. And then the third step says,

17 well, the plaintiff can now come back and rebut the

18 monopolist's justification. Or, if it can't actually

19 justifiably rebut it, it can demonstrate that the

20 anticompetitive effect was bigger than the procompetitive

21 benefit and outweighs it. So, it can do a rule of reason

22 analysis is what it says.

23 Well, let me just review what happened in the

24 Microsoft case. There were many allegations having to do

25 with Java standards and with various contracting policies


1 with lots of different players in the industry.

2 There were really three design elements that

3 were challenged in the Microsoft case. One was not having

4 Internet Explorer in the Add/Remove programs utility. The

5 other was designing Windows so as in certain circumstances

6 to override the user's choice of a default browser other

7 than Internet Explorer. And the third one was commingling

8 browser and operating system code.

9 Now, interestingly, the court concluded that

10 Microsoft offered no procompetitive justifications for the

11 first and the third, and these were held by the court to

12 contribute to the Section 2 violation. But then the court

13 also concluded that plaintiffs -- that Microsoft did offer

14 a justification for the second element of its conduct,

15 that is, the overriding of user's choice of the browser,

16 which the plaintiff did not rebut, and therefore in fact

17 the Microsoft court never got to the third step. So, the

18 court never got to the rule of reason balancing in the

19 third step because either it was anticompetitive with no

20 efficiencies or there were efficiencies and the plaintiff

21 didn't come back. So, maybe Tim will explain or Dan will

22 elaborate on this, but this is my reading of what happened

23 with the court.

24 So, the practical effect of what happened in the

25 Microsoft court's analysis was really, I think, similar to


1 the no economic sense test of the first variety. That is,

2 was there some reason for this conduct. If there was,

3 it's okay.

4 Now I want to turn to another area that I find

5 quite interesting. As they say, this is emerging

6 antitrust. This is that drug patents may delay generic

7 competition. So, the innovation that contributes to these

8 drug patents can have competitive effect. It can have

9 competitive effect both through the nature of generic

10 substitution and also because of the specific elements of

11 the Hatch-Waxman Act, which impose a 30-month stay on

12 generic competition if you have a patent.

13 So, one of these cases is called Tricor, which

14 is actually a drug called phenofibrate. It's used to

15 control triglyceride and cholesterol levels. And I should

16 acknowledge I have been a consultant in this case. A

17 second case is Prilosec and Nexium, which Prilosec is a

18 common drug prescribed for heartburn, gastric reflux, and

19 then your more serious conditions like esophageal and

20 duodenal ulcers. It turns out that Nexium is what is

21 called an isomer of the chemical that's in Prilosec. It's

22 basically the same molecules. It's been rearranged a

23 little bit. And it's supposed to have some advantages for

24 the esophageal and duodenal ulcers, but not for heartburn.

25 The allegations that came up in both of these


1 cases is that the innovations are costly but minor

2 improvements, that they're contrary to the intent of the

3 Hatch-Waxman legislation to promote generic competition,

4 and they have large adverse competitive effects by

5 delaying generic competition.

6 Now, I think certainly if you just take a

7 snapshot of competition, once these drugs exist, anything

8 that delays generic competition has at lease the

9 possibility of a competitive effect. But it's important

10 to recognize that the Hatch-Waxman legislation was a trade

11 off between more generic competition and more protection

12 for patented drugs. In fact, the first three letters of

13 the Hatch-Waxman Act are patent term restorations. I

14 think it was designed to protect pioneer drugs, as well as

15 promote generic competition.

16 Product line extensions certainly increase

17 incentives for drug innovation. If you actually look at

18 the respective patent terms for prescription drugs,

19 patented prescription drugs, it's actually quite short.

20 It's one of the shortest of all industries because of all

21 the FDA delays and regulations required to actually

22 produce the drugs. And it's very hard to assess these

23 benefits from these innovations.

24 So, I think instead of looking at any of these

25 standards to inform a Section 2 analysis for innovation, I


1 find all of them seriously lacking. I think instead you

2 can turn to consistency with other rules.

3 And let's talk about something we've heard

4 before, talk about the process rather than the outcome.

5 That was discussion was featured in this morning's session

6 to great extent by members of the panel talking about the

7 process rather than the outcome.

8 So, the quote that I'm quoting here is by a

9 distinguished economist, but not anyone from our group.

10 It's from an economist who works for the Oakland Athletics

11 who was quoted by Michael Lewis in "Moneyball," and he was

12 actually talking about how to hire baseball players, but I

13 think his insight here is equally applicable to antitrust

14 policy, "We have to look at process, not outcomes."

15 So, if we think about making an analogy between

16 innovation effects, and the effects and rules that are

17 applied to other conduct, I want to argue that, in many

18 innovation cases, the effects of the innovation are very

19 similar to the effects of a unilateral refusal to deal.

20 When you're talking about, say, if IBM refuses to make

21 mainframes compatible able with third parties' components,

22 it's a lot like saying, well, one day Microsoft gets up

23 and says, I don't want to work with these third party

24 people anymore, I want to build computers just for myself.

25 Microsoft refuses to make Windows compatible with other


1 browsers. Or a generic drug manufacturer refuses to

2 supply a drug that generics can copy.

3 In effect, this conduct looks a lot like a

4 unilateral refusal to deal. Now, these days, after

5 "Verizon v. Trinko", seems like unilateral refusals to

6 deal have a long way to go before they can generate

7 antitrust liability.

8 Now, I don't want to state that as a categorical

9 fact, or that "Verizon v. Trinko," that all the words in

10 "Verizon v. Trinko" were necessarily the greatest words

11 that have ever been uttered in all of antitrust policy. I

12 am not sure it's the greatest policy.

13 But my only point is that if you are going to

14 have a policy that gives considerable deference to a

15 decision by a single firm about who that firm will deal

16 with or supply, it just seems odd that one wouldn't have a

17 more strict policy, more intervention policy with respect

18 to innovations that have very similar effects.

19 So, I'm not saying -- again I want to emphasize

20 that I'm not saying that we should have policies that say

21 that unilateral refusals to deal with per se legal, I

22 don't think that's necessarily the right thing. But if we

23 are going to have such a policy, then consistency seems to

24 say that if you unilateral innovations that have similar

25 effects should not be treated more severely.


1 So, one of my conclusions here is that all of

2 the rule of reason and profit sacrifice tests have limited

3 value to evaluate what is sometimes called predatory

4 innovation. It's hard to do; likely to get the wrong

5 answer; very hard to look al at the incentive effects that

6 are necessary to really thinking about innovation.

7 The no economic sense test is better, but only

8 if it's interpreted as a test of sham innovation because

9 otherwise it comes out just like or very similar to a

10 profit sacrifice test.

11 And my other conclusion is that this is what

12 courts in fact almost always have done with very few

13 exceptions in the way they've treated these cases and it's

14 probably as reasonable an approach as any.

15 MS. GRIMM: Our third presenter this afternoon

16 is Daniel Rubinfeld, who is the Robert L. Bridges

17 Professor of Law and Professor of Economics at the

18 University of California at Berkeley, where he has taught

19 since 1983

20 He has also served as Deputy Assistant Attorney

21 General for Antitrust in the U.S. Department of Justice,

22 as well as in various capacities with the President's

23 Council of Economic Advisors, the National Academy of

24 Sciences, the Urban Institute, and the National Bureau of

25 Economic Research.


1 Professor Rubinfeld's major books include

2 "Econometric Models and Economic Forecasts" and

3 "Microeconomics." Recent publications include, "Antitrust

4 Enforcement in Dynamic Network Industries" in "The

5 Antitrust Bulletin," 1998; and "Empirical Methods in

6 Antitrust: Review and Evidence" in "American Law and

7 Economics Review."

8 He is President of the American Law and

9 Economics Association.

10 Professor Rubinfeld received his B.A. from

11 Princeton University in 1967; his M.S. and Ph.D. from the

12 Massachusetts Institute of Technology.

13 Dan.

14 MR. RUBINFELD: Thanks very much. I really,

15 like everyone else, appreciate the opportunity to appear

16 before you today. It's been about eight or nine years

17 since I left the Antitrust Division and I guess,

18 understandably I've aged about eight or nine years during

19 that time, and I find as one gets older one tends to

20 reflect back on the past, perhaps more than one should.

21 But what I'm going to do in my comments today is to really

22 do some reflection on what happened, and I might hit on

23 some of the previous commentators' issues, but see I can do it in

24 a way that will be constructive for the agencies as you

25 think about forming your policies.


1 So, the first point I want to make is why I

2 think it's really important to have an active Section 2

3 jurisprudence. And I want to look back and talk about the

4 legacy of "U.S. vs. Microsoft" for antitrust enforcement.

5 And, finally, I want to look at bundling and talk about

6 the legacy of "LePage's vs. 3M".

7 I should say, to make it clear, that I had an

8 interest in both of those cases. I helped to prosecute

9 the Microsoft case. And I have consulted for 3M with

10 respect to some of the issues that arose in its appellate

11 case. I was not involved in the LePage's case itself, but

12 I was involved in thinking about some of the appellate

13 issues. So, I have taken a pretty close look at the Third

14 Circuit opinion in that case.

15 If you're interested in some of the deeper

16 comments I am going to give today, they will appear in

17 two articles. One is an article that Doug Melamed and

18 myself are completing for our forthcoming volume in

19 which we are looking at the lessons of the Microsoft case.

20 And the second is an article I published a year or so ago,

21 looking at the bundling in the "LePaige's vs. 3M" case.

22 Before I go on to the cases, as far as the active

23 Section 2 jurisprudence is concerned, I guess history

24 affects how one views things, and I can be very quick, I

25 can just say, having been involved in actually bringing


1 both Microsoft and Dentsply, both of which I thought was

2 the right thing to do, and the D.C. Circuit and the Third

3 Circuit in both cases have written opinions that were

4 supportive of that decision, I'm proud to have been

5 involved in both of those cases, and I think that shows,

6 consistent with what Tim Bresnahan said, it shows the kind

7 of active Section 2 jurisprudence that I think makes

8 sense.

9 Both cases had a particular set of facts

10 associated with them that told a story that made them the

11 right cases to bring, viable cases. And I think the

12 agencies need to be careful because there is not going to

13 be a lot of good Section 2 cases. So, you need to be

14 careful and active and watchful for the appropriate

15 opportunities in the future.

16 So, having said that, let me go on and take a

17 look at "U.S. vs. Microsoft". And I am going focus now

18 really on sort of what we've learned from the case in a

19 very broad perspective. I'm not going to try to go into

20 some of the technical details unless we have discussion

21 later.

22 It's sometimes easy to forget, since this is

23 almost ten years ago when at least my version of Microsoft

24 was brought, that people were barely talking about network

25 effects. Now it's taken for granted that in high tech


1 it's common to face industries in which network effects

2 matters and that enters into the economics and to the law,

3 legal thinking about the cases.

4 I see one of the legacies of Microsoft is sort

5 of helping to bring us from the pre-network effect world

6 to a world where network effects are often the core of the

7 analysis.

8 Next important is people are thinking somewhat

9 differently now than they were before about barriers to

10 entry. When we originally think about investigating the

11 Microsoft case, obviously barriers to entry was something

12 that I paid a lot of attention to. We became convinced

13 that there was a significant barrier to entry, but it's

14 not the usual one you might imagine. It had to do with

15 the fact that in order to have a successful operating

16 system, you really needed to have successful applications.

17 There was what we called a two-level entry problem. And

18 we spent a lot of time developing the underlying economics

19 that describe this applications barrier to entry.

20 One of the things that people forget, actually I

21 almost forget myself, is that the term "application

22 barrier to entry" did not exist, at least to my

23 knowledge, prior to our work. We coined and reiterated it

24 every time we could at trial until the judge finally got it

25 into his mind.


1 And it was fun to watch the trial, by the way,

2 because at the beginning of the trial, Microsoft disavowed

3 the application "barrier to entry." By the end of the

4 trial it was being discussed by them as if it were a

5 common coin of the realm.

6 So, let's remember that that was one, for better

7 or worse, I think for better, one of the legacies of the

8 Microsoft case.

9 The other thing is, as you all know, the case

10 involved tying, but it was different than the classic kind

11 of tying case, which is usually thought of leveraging

12 market power from a market where a firm has substantial

13 market power to use some related power where it does not

14 necessarily have significant market power.

15 But this case did involve tying as well as

16 bundling. And it was a non-leveraged form of tying. And

17 now it's not, I think unusual to think about bundling in

18 that context in certain cases where it was probably quite

19 radical at the time.

20 The other thing is that the case brought to our

21 mind a different way, a different perspective of thinking

22 about market definitions. As Tim suggested earlier today,

23 there's always been a lot of talk about Schumpeterian

24 competition and certainly the agencies have been aware of

25 it for a long time.


1 In this case, to one degree or another,

2 Schumpeterian competition really came to the forefront

3 because, in the debate about market definition and market

4 power, Microsoft took the position that it was the threat

5 of entry by competitors that really not only restrained

6 this market definition, this market power, but also in

7 fact meant that the market should be defined very broadly.

8 Microsoft argued for an extremely broad market definition

9 that included almost all operating systems, from

10 hand-helds pretty much up through mainframe computers, and

11 argued that it had no market power over that relevant

12 market.

13 I still remember one particular trial exhibit

14 which Microsoft presented which sort of brought this issue

15 to the front. And the exhibit said that Microsoft faces

16 substantial competition from known and unknown

17 competition. And my view, which was borne out, by the

18 way, by the Circuit Court opinion, is that when you have

19 to defend your market power or lack of it by describing

20 competition that no one knows about yet, you really have a

21 fairly weak position.

22 And if you read the D.C. Circuit opinion, I

23 think the D.C. Circuit got it right, as they did in most

24 areas, they said, the nascent competition really could be

25 important but it really has to be competition which is


1 expected with reasonable certainty to actually be there in

2 the marketplace at some period in the future, thinking

3 about two years would be the relevant time period.

4 But the fact that someone might come along and

5 take away your market power isn't sufficient. I think the

6 court was pretty clear about that. And it's basically the

7 right place to be.

8 As far as legal issues I see coming out of the

9 case, there are about five. I'd like to highlight, again,

10 without getting into the technical/legal side of the case,

11 the first thing which I think we now take for granted, or

12 at least I hope we do, which is that the same antitrust

13 principles apply in dynamic high tech industries as apply

14 in the other industries. The application of course might

15 be somewhat different, but the principals are the same.

16 And I quote Judge Posner, who really says what I

17 have in mind, which is that antitrust doctrine really is

18 pretty well situated to allow us to handle high tech

19 industries. We don't need to rewrite Section 2, in my

20 view.

21 Up until Rich started speaking earlier, I would

22 have said hardly anyone remembers that there are IP issues

23 raised in Microsoft. Rich laid them out pretty well.

24 And so, what I wanted to say is that the court

25 makes it pretty clear that the same general antitrust


1 principles that apply to conduct involving intellectual

2 property that apply to any other form of property underc

3 the antitrust laws.

4 Originally, at one point in the case, Microsoft

5 actually claimed that their IP rights covered the entire

6 desktop, at least with respect to the first boot up of

7 their operating system. The court made it very clear that

8 (a) that was too expansive an interpretation, and (b) that

9 it was appropriate for the Sherman Act and the courts to

10 really look at the IP issues. You did not get a free ride

11 just because you did in fact have some legitimate

12 intellectual property.

13 And Rich described in detail and correctly where

14 the court finally came out about these specific IP issues.

15 With respect to product design, as I interpret

16 the court opinion, it makes clear that the court is going

17 to give pretty wide deference to firms that are designing

18 new products, along the lines Rich described. But the

19 court also said this is an area that's open for viable

20 investigation. And where particular aspects of

21 Microsoft's product design excluded rivals, the court did

22 shift the burden to Microsoft to establish a

23 procompetitive justification for the design. There is no

24 safe harbor just because you're involved in innovation or

25 product design. And the removal of the Add/Remove utility


1 which Rich described was one good example of that. The

2 court was very clear that was problematic and there was no

3 procompetitive justification given that I can see in the

4 case.

5 There's also an issue in this kind of Section 2

6 case as to whether you ought to kind of just describe the

7 case with kind of a broad brush or kind of go into the

8 practices with fine detail. My sense, my personal sense

9 during the trial was that there were times when the

10 defense seemed to say, we want to just talk very broadly

11 about the rights of a dominant firm to engage in certain

12 kinds of potentially procompetitive activities. And the

13 government, as I saw it, focused really in with apparent

14 detail about the details surrounding each of these kinds

15 of conduct.

16 And I read the D.C. Circuit as basically saying

17 that any aspect, the explicit, discrete aspect of

18 monopolist conduct that tends to exclude rivals may be

19 illegal, unless there's a legitimate procompetitive

20 justification for that particular conduct.

21 So, there is at least a burden-shifting aspect

22 to some of the illegal rules that flow from the Microsoft

23 case, which I think is appropriate.

24 There is an issue about whether you ought to

25 focus on rules or cases, specific facts. Here, as you


1 know, the court, the appellate court, on the time claims,

2 suggested that the per se rule didn't apply because of the

3 particular attributes of platform software. So, we're now

4 left in a somewhat unclear world that may apply mostly to

5 Section 1, but also has Section 2 implications as to how

6 to treat tying.

7 And I have to say here, as an economist, you may

8 not be surprised to hear that I'm pretty sympathetic with

9 the comments of the court. I think it's really hard to,

10 as an economist, come up with per se rules that would

11 apply in this kind of high tech context.

12 Of course we don't know quite where that would

13 have ended up because the Department of Justice chose not

14 to appeal that part of the D.C. Circuit's ruling.

15 With respect to causation, I see the case telling

16 us conduct that violates the antitrust laws only if it

17 injures competition. Causation can be inferred when

18 exclusionary conduct is aimed at producers of nascent

19 competitive technologies, as well as when it's aimed at

20 producers of established substitutes.

21 So, basically the court spelled out causation along

22 the lines Tim suggested, and I think the court makes it

23 pretty clear that that's necessary and that the government

24 succeeded in that effort.

25 What about profit sacrifice? Here we could


1 debate exactly how to characterize the case. I would say

2 that the case we put forward did really involve a profit

3 sacrifice test. My definition would be that conduct is

4 anticompetitive when it would not make business sense for

5 the defendant but for its tendency to exclude rivals and

6 create or maintain market power for the defendant.

7 This is kind of a crude paraphrase. If you go

8 back and read the details of the case, you'll see a more

9 formal definition. It is a variant on a profit sacrifice

10 test. I wouldn't say it's quite a no nonsense test, but

11 it's pretty close.

12 Now, that's not what the D.C. Circuit said.

13 What the D.C. Circuit said was quite close to what Rich

14 Gilbert said earlier. The court said that the conduct is

15 anticompetitive if it harms the competitive process and

16 either it's not shown to further efficiency or to have

17 some other procompetitive justification or the

18 anticompetitive harm outweighs its procompetitive benefit.

19 So, the D.C. Circuit was suggesting more of a balancing

20 test than a profit sacrifice test.

21 And this leaves us with the question of what we

22 should do if we find Section 2 type conduct that harms

23 competition and furthers a legitimate purpose should we

24 have a balancing test.

25 Now, I should say here, I am not entirely sure


1 of where I would end up, but I lean strongly towards the

2 profit sacrifice test, at least in most cases, because I

3 think it's easier to operationalize. We could debate

4 about how to exactly operationalize it, but I think Tim

5 suggested that, in most of these cases, it's just not

6 possible to sit down and do a fully complete balancing

7 rule of reason analysis. We don't have the time or the

8 information available. And the cost, by the way,

9 including the cost to the parties, would be tremendous.

10 And I think in most situations, a profit sacrifice test

11 would get us to the right place. I think you can try to

12 find some counter-examples, but I think you have to work

13 hard to do it. So, I am on the side of the folks who

14 think we ought to just refine the profit sacrifice test.

15 Okay, let me switch to my other case of

16 interest, "LePage's vs. 3M". You have heard about it a

17 little bit already. This was the case involving bundled

18 rebates offered by 3M in the market for transparent tape.

19 3M was facing substantial competition from LePage's, not a

20 new entrant, but an entrant that had become very

21 successful in the production and sale of private label

22 tape.

23 And the question was: Were 3M's programs,

24 specific bundling programs, anticompetitive and a

25 violation of Section 2.


1 Now, here I'm very critical of the Third Circuit

2 opinion generally for two reasons. One is that the

3 opinion itself does not, in my mind, in any way provide

4 any clear guidance as to how firms ought to behalf when

5 they do have a dominant position and they are deciding

6 what kind of business practice to engage in. And I think

7 any clear legal rule ought to do so.

8 And, secondly, I actually think that I have been

9 unable to come up with what I think is any coherent theory

10 of predation or any Section 2 theory which fits the facts

11 of the 3M case. In my view, the Third Circuit was a

12 little bit loose in how they actually borrowed and used

13 facts of the case. I actually went back and read most of

14 the record in the LePage's case and I cannot find a theory

15 that I find coherent that actually fits the facts of the

16 case.

17 And the thing to remember is that bundling

18 itself of course is quite ubiquitous and often is

19 procompetitive. So, if we generate a legal rule, we want

20 someone else to define those relatively few cases where

21 bundling is a problem and distinguish it from the majority

22 of cases where bundling is procompetitive. So, we're

23 looking for those particular situations. Lack of clarity

24 is a problem.

25 Let me briefly take a few minutes and just very


1 quickly tell you about 3M's programs. There were a whole

2 bunch of programs being attacked, but the two that

3 involved bundled rebates were, first, the executive growth

4 fund program. And the thing that's key about this program

5 was it was actually I think a one-year program and it was

6 a pilot program for a small number of customers.

7 Now, what it did do was it set up growth targets

8 for six different errant divisions of 3M, which would

9 cover a lot of office supply products. And firms actually

10 had to meet target goals in each of these divisions.

11 Now, my view is that the executive growth fund

12 program -- let me be clear that this is my view and not

13 3M's view. My view is that, had this program been

14 expansive and had it covered all customers rather than

15 just a few, and had it continued for a number of years, it

16 could well have been an anticompetitive program. I don't

17 think it was because it was too narrow. It had no ability

18 really to substantially exclude competitors because many

19 of the key competitors, Walmart being the most important,

20 were not covered by this program. But it had the

21 potential if it continued to actually be restrictive

22 because of the specific design of the program.

23 But for various reasons, which I think relate

24 partly to the demands of some customers, including

25 Walmart, 3M changed its program to a partnership growth


1 program, and this program did involve discounts in six

2 different areas, but there were no specific targets to

3 reach in each of the areas. Basically you got a rebate

4 based on the aggregate of all your purchases in all six

5 categories. So, this amounted to a somewhat complex

6 discount program, volume discount program.

7 And my view is that the PGF program, as it's

8 called, was not anticompetitive, even though the court

9 felt otherwise.

10 So, if you go back and look at the LePage's

11 trial and ask -- take a look at the trial and ask if the

12 trial helps to support some of those theories of

13 competition, I would say no. I didn't see any testimony

14 in the record about economies of scale or scope, which

15 would be important, particularly to get at the issue of

16 whether LePage's or any other competitor would remain

17 viable in the face of these practices.

18 There was no predatory pricing claim.

19 Plaintiffs agreed that LePage's was pricing above cost.

20 In fact, by my calculations, even if you took all of the

21 discount programs at 3M, no matter what the products were,

22 attribute all the discounts to tape, it would still be

23 pricing above cost.

24 I didn't see anything about profit sacrifice

25 that I could infer from the opinion. So, there was


1 nothing that fit my particular interest in pursuing these

2 kinds of Section 2 cases.

3 There was no time claim at all. It was a

4 bundling case, not a tying case. There was also no showing

5 of market power with respect to any product other than

6 transparent tape. So, the kind of leveraging theory you

7 might expect to see in a time case was not present either.

8 Now, the jury did find, interestingly, no

9 exclusion under Section 1, but they did find a violation

10 under Section 2. So, this leaves me with a puzzle of what

11 the legacy is of "LePage's vs. 3M". I think for a while

12 the Commission may have though this case was unusual, but

13 it's pretty clear now that the Third Circuit opinion has,

14 let's say, encouraged a lot of litigation surrounding

15 these kinds of practices.

16 So, I went back and asked myself, what should

17 the principles be here. And I would say, speaking very

18 broadly, if the rebates associated with bundling reduce

19 consumer welfare by impairing rivals' ability to make

20 competitive offers to potential customers, that's going to

21 be something generally that's going to give me concern. I

22 am not going to say it's necessarily anticompetitive, but

23 that would give me great pause.

24 And that general rule takes into account

25 efficiencies and allows price increases by firms, as long


1 as they don't impair rivals' ability to compete. But that

2 general rule is really not very helpful from a process

3 point of view. It's really too broad to make applicable.

4 So, I would say the following. I'd say, there

5 are conditions under which one may be anticompetitive, but

6 none of them fit LePage's.

7 And, just quickly, because I think we're running

8 out of time, here's some examples of situations in which I

9 think bundling might be anticompetitive, none of which

10 fits the LePage's case.

11 The first would be traditional contractual tying

12 of the kind that we saw in Jefferson Parish. The second

13 would be predation through profit sacrifice of the kind

14 where bundling was used in the form it was in the

15 Microsoft case, and perhaps I'd include Dentsply there as

16 well. The third might be monopoly maintenance through the

17 creation of barriers to entry, which is, at least my

18 interpretation of "SmithKline versus Eli Lilly," a case I

19 was not involved in, where at least the court stated that

20 the sale of monopoly products were used to harm

21 competition in a non-monopoly market.

22 Now, where does this leave us? We need a

23 workable test. I wish I could come here and tell you I

24 figured out what that test is. I have read many papers

25 written by folks in the agencies and elsewhere suggesting


1 various tests.

2 I still not have seen one that I am entirely

3 happy with, but a couple things strike me as important

4 when and if we get such a test. One is that, weakening a

5 rival should not be sufficient to condemn a monopolist,

6 otherwise we will be discouraging firms from innovating

7 and growing and being successful, which I think would be

8 harmful to our competitive process.

9 Secondly, while it would be very nice to have an

10 incremental cost benefit test for certain kinds of

11 bundling, there are a lot of difficulties in putting that

12 test into play that I won't bore you with here. So, we

13 have more work to do there.

14 Third, we might say that for a bundled rebate

15 program to be anticompetitive, it at least necessarily

16 ought to be the case that the incremental costs associated

17 with the available discounts exceed the incremental

18 profits associated with the incremental sales that

19 generate. If you take that language, I think you can

20 create a viable safe harbor at least that would at least

21 give firms some comfort that certain practices would be

22 presumed to be legitimate.

23 And I actually believe, having done my work in

24 LePage's, that the behavior of 3M would actually satisfy

25 this safe harbor test. But you don't want to condemn


1 nondiscriminatory price cuts in single markets and you

2 want to be careful not to penalize policies that exclude

3 less efficient competitors.

4 That's a different issue because if you want a

5 test that's workable for a firm that's engaged in a

6 policy, it's very hard to say you shouldn't exclude a less

7 efficient competitor because the firm is not going to know

8 typically whether its competitors are more or less

9 efficient.

10 So, this test really is not going to be a

11 perfect test and probably never will be.

12 A workable rule should be one that's clear and

13 manageable. We don't want businesses to say what I hear a

14 lot in recent years, which is we have no idea which

15 practices we can engage in or not because anything that

16 seems to have any bundling aspect to it could lead to a

17 Third Circuit lawsuit.

18 Now, as far as the thoughts I have given you, I

19 just happened to go back and look on the web recently at

20 the AMC's tentative recommendations. I assume they're

21 still tentative. And I found myself in agreement with

22 their recommendations in the areas I am talking about.

23 There are some other areas I would disagree.

24 But I noticed that the AMC tentatively is

25 recommending no need to revise the antitrust laws to apply


1 to high tech industries. And I agree very strongly with

2 that.

3 The AMC is proposing no need for Congress to

4 amend Section 2. And I agree strongly with that as well.

5 And, finally, it looks like the AMC is thinking

6 of recommending additional clarity and improvement in

7 Section 2, particularly with respect to areas such as

8 bundling. And I agree strongly with that as well.

9 Thank you very much.

10 (Applause.)

11 MS. GRIMM: I'd like to thank all of our

12 panelists.

13 We are going to take a 15-minute break now.

14 We'll reconvene in 15 minute for our round-table

15 discussion.

16 (A brief recess was taken.)

17 MS. GRIMM: Before we get to our questions and

18 round-table discussion, I would like to introduce our

19 fourth panelist, who will discuss some of the ideas that

20 have been advanced by our other panelists this afternoon,

21 as well as some of his own ideas about Section 2.

22 Carl Shapiro, our fourth panelist, is the

23 Transamerica Professor of Business Strategy at the Haas

24 School of Business at the University of California at

25 Berkeley. He also is Director of the Institute of


1 Business and Economic Research and Professor of Economics

2 in the Economics Department at U.C. Berkeley.

3 He earned his Ph.D. in economics at MIT in 1981;

4 taught at Princeton University during the 1980s; and has

5 been at Berkeley since 1990.

6 He has been editor of the "Journal of Economic

7 Perspectives," and a Fellow for the Center for Advanced

8 Study in the Behavioral Sciences.

9 Professor Shapiro has published extensively and

10 his current research interests include antitrust

11 economics, intellectual property and licensing, product

12 standards and compatibility, and the economics of networks

13 and interconnection.

14 Professor Shapiro served as Deputy Assistant

15 Attorney General for Economics in the Antitrust Division

16 of the U.S. Department of Justice in 1995 and 1996.

17 Carl.

18 MR. SHAPIRO: Thank you very much. I don't have

19 any slides. I am going to cover some ideas I have and

20 then comment on and kind of get the discussion going about

21 each of the previous panelists.

22 You probably already picked up the theme here

23 is that we get up here and we reminisce about the cases

24 that were brought or investigated while we were at the

25 Antitrust Division. Okay? We really appreciate you


1 coming out here because we all have a love for the

2 Antitrust Division. FTC, too. And we sort of appreciate

3 your coming out here so we don't have to go again to D.C.

4 One of the themes that we've picked up here and

5 throughout many of these hearings is that Section 2 cases

6 are inherently really hard because it's a single-firm

7 conduct and it's not like a cartel case. They're really

8 hard and there's always elements and you have to be very

9 careful.

10 And I don't disagree with any of that, but I

11 want to focus on they seem to be harder than they need to

12 be in some cases. And it's one of my themes, intersecting

13 with the role of patents and plus innovation and

14 Section 2.

15 And I'm going to depart from the DOJ reminiscing

16 and actually talk about the Unocal case, which was brought

17 by the FTC, and which I served as an expert witness for

18 complaint counsel. And that was litigated at the -- by an

19 administrative law judge, before the administrative law

20 judge.

21 So, let me just quickly remind you of that case

22 or tell you about the case. So, Unocal had some patents

23 -- had patents -- came to have patents during the '90s on

24 reformulated gasoline. The State of California through

25 the California Air Resources Board, CARB, established


1 regulations for gasoline in order to make the

2 cleaner-burning reformulated gasoline.

3 And it came to pass that the regulations that

4 were adopted, that Unocal's patents, apparently or very

5 likely, many of the refineries would have to infringe

6 those patents for a large fraction of the gasoline they

7 would make if it would comply with the state regulations.

8 So, and the allegation was that Unocal had acted

9 deceptively by leading the industry members to believe

10 that its patents would be -- that either it did not have

11 patents or would make them available on a royalty-free

12 basis. That was the representation when the regulations

13 were being formulated and that Unocal then later sought to

14 get royalties. That was the allegation of deceptive

15 conduct.

16 So, you would think -- well, let's say I would

17 think, at least, maybe you would think, that this should

18 be the sort of Section 2 case, and I guess it was FTC

19 Section 5, and I'm not distinguishing those for my purpose

20 here, that it would be relatively straightforward.

21 Big factual question about whether Unocal acted

22 deceptively. They vigorously denied that they did so.

23 The FTC or certainly complaint counsel was arguing they

24 had. I simply assumed that they had for the purposes of

25 evaluating market power and competitive effect. That was


1 the fact question. If they had not engaged in any

2 deception, I believed there was nothing to the case. That

3 was my understanding, as I recall it.

4 So, if they acted deceptively, and let's take

5 the really cleanest version, they led people to believe

6 patents would be available on a royalty-free basis.

7 Regulations are selected. Literally billions of dollars

8 are invested by refiners to comply with these regulations,

9 made CARB gasoline, as it is called, and then they

10 asserted patents.

11 So, the reason I would say this should be, to my

12 way of viewing, a relatively simple case because the

13 conduct alleged and assumed by me, as an expert at least,

14 deception is not something that we have to wring our hands

15 over, oh, is that something that's procompetitive, is it

16 important that companies engage in that sometimes. It's

17 not like discounting. It's not like product innovation.

18 Deception.

19 So, now then the question is, okay, we don't

20 really have to worry about stifling deception, okay. So,

21 does it have a significant effect on prices, on market

22 power? And if they represented that the patents would be

23 available royalty-free and are later seeking something

24 like five cents a gallon, to throw out a number, for

25 pretty much the whole industry a very large fraction of


1 the gasoline that would be produced, well, that's a price

2 increase. There's very strong evidence that would be

3 passed to the final consumers, motorists. Not that that

4 matters so much because, even if not, it would be borne by

5 the direct customers of the technology, refiners, who

6 would be using the technology. And so you get right away

7 the competitive effects without any real business

8 justification for the conduct that's alleged or

9 challenged.

10 And yet, Unocal raised many, many arguments.

11 We do not know how the administrative law judge or the

12 commission or subsequent appeals court might have reacted

13 to these. We do know from other cases, the case of --

14 the Rambus case. There are a variety of Rambus

15 cases that also involve similar allegations regarding

16 standards and patents. And we know from other cases I

17 won't get into that the courts have tended to say, well,

18 wait a minute, you have a patent and so you get some

19 market power associated with the patent, and so we should

20 be very careful not to jump on -- not to conclude that,

21 just because there's market power, somehow it has to do

22 with anticompetitive conduct, because patents may very

23 well confer market power in a perfectly desirable way.

24 So, I guess I'm raising a concern that what

25 should be a simple case, there seems to be, in some


1 quarters at least, sort of a worship of patents that

2 therefore mixes up market power attributable to the

3 innovation versus market power -- additional market power

4 that comes about from conduct, just the sort of thing that

5 Tim was mentioning, actually, look at additional effects

6 of the conduct.

7 And the economic opportunities of hold up I

8 think are very clear, going back at least to Oliver

9 Williamson, my distinguished colleague here at Berkeley,

10 and yet these were denied essentially by Unocal and its

11 economic expert. That is to say, the notion that once

12 refiners had invested enormous sums in order to comply

13 with the regulations, that would necessarily put Unocal in

14 a stronger bargaining position to get royalties that they

15 could not have gotten earlier.

16 So, I would say it's relatively fundamental

17 economic principles, fairly clear fact pattern, and yet we

18 have -- and, for example, the whole Antitrust debate about

19 defining the relevant market. Defendants can often, in

20 this case at least, try to make that very complicated,

21 exactly which technologies are in the market and which

22 ones are substitutes, and what was the best alternative,

23 and how good was it, and how much -- they even argued, our

24 technology is so good that people would have picked it

25 anyhow and, therefore, even if we engaged in deception, it


1 wouldn't matter.

2 Well, I just don't think that's right because

3 there's additional market power that results from lock in.

4 So, sometimes the elements that we always think

5 of for Section 2 cases: defining the market, measuring

6 the market power; being cognizant of preexisting market

7 power, in this case because of patents, I think we need to

8 be careful not to lose sight of what may be a simple or

9 more direct argument that can get us to analysis without

10 doing -- without necessarily following some of these steps

11 and without getting tied up particularly in market

12 definition. And, again, Tim, I know, emphasized that he

13 really, as most economists, if we can, we want to get to

14 competitive effects. And market definition may or may not

15 be helpful in getting us there in market shares.

16 And if you think about the cases I've described

17 today, measuring exactly which share of how much of the

18 gasoline infringes or might infringe and what other

19 technologies are being used is a distraction,

20 fundamentally a distraction to what's being looked at

21 here.

22 And that came in in terms of remedy as well. My

23 testimony was, we should restore competition, which means

24 they should license these patents on a royalty-free basis,

25 as they had represented under my working assumption. And


1 yet Unocal argued that, well, our technology is so good

2 that we should be able to charge more than that, even if

3 we engaged in deception, because under competition somehow

4 they would have been able to charge a lot.

5 Then you ask, well, then why did you act

6 deceptively. And they say, well, we didn't. Well, what

7 if you have. So, you go back and forth. All right.

8 So, while I'm not expecting the DOJ or FTC to

9 suggest that we throw out market definition, for example,

10 in Section 2 cases. I do think looking for shortcuts that

11 are reliable is a good thing to do.

12 Let me go on to say something about the previous

13 speakers now that I've made some points about some of my

14 own thoughts about Unocal.

15 So, Tim first, Professor Bresnahan. Very

16 gracious of him to come up here to Berkeley and appreciate

17 his kind words about Berkeley. I will try to reciprocate

18 and I will make two trips to Stanford in the next week for

19 conferences there, and with pleasure.

20 I took some of what you said, Tim, to be

21 suggesting that we could think of screening cases based on

22 whether there's a theory of harm that the conduct would

23 lead to a significant increase in market power, or let's

24 put that differently, relax the constraints on pricing

25 that are facing the firm that's accused, or the defendant


1 firm.

2 And I think that's a really good way to go. So,

3 I support that.

4 One way I like to think about it is we could ask

5 if the conduct is directed at certain competitors or maybe

6 at certain distributors who then would be important for

7 certain other competitors in your Dentsply case, we could

8 ask, if the conduct was really effective and eliminated

9 those competitors, a certain class or group of

10 competitors, would the firm be able to significantly raise

11 price. Or, alternatively, if those competitors were fully

12 enabled, would that lead prices to fall significantly.

13 If that's true, then we need to proceed further

14 in the inquiry. If not, because the price is really

15 governed by some other set of dynamics, you know, in the

16 case of patented drugs, if you get rid of the generic

17 competition, that would usually lead to a higher price,

18 but it could be in some cases that competition from other

19 patented drugs is what's driving price or, in principle,

20 that sort of competition, and then we could stop that

21 inquiry if the targets were not really providing sufficient

22 competitive discipline. So, I am very supportive of that

23 line.

24 You said at some point, Tim, that it was very

25 hard to do some sort of balancing, you know, particularly


1 quantifying the balancing of net effects, harm to

2 consumers, benefits to consumers. And so I guess the

3 economic theorists, I guess that's going to include me

4 now, may like to measure all these things and do this in

5 our models, but in practice that balancing would be hard

6 to do. It is hard to do.

7 One thing we might do is then focus more on the

8 competitive process, rather than necessarily a particular

9 outcome.

10 But you also said the defendant could show that

11 the practices were efficient and that would be a defense.

12 So, if there was anticompetitive danger, the defense could

13 come back and say the practices were efficient. I don't

14 know what that means in practice. I guess I'd like to

15 hear more from you about that. Because there is typically

16 going to be some story about, oh, this has lower prices

17 for some customers so it's efficient, or this is going to

18 prevent free riding, so I need to have exclusive dealing

19 here. There's going to be some efficiency story and I

20 don't understand how you can avoid doing some balancing

21 after the efficiency flag is raised and now are we done.

22 I don't think you mean they're done just because the

23 defense raises the efficiency argument. So, what happens

24 next?

25 My last comment was on -- I don't want to get


1 into Microsoft. Believe me, I really don't want to get

2 into Microsoft. But you did mention -- I like your term,

3 the "remedy fizzle." I don't know if you coined that

4 term, but I like it. You took some responsibility, I

5 think --

6 MR. BRESNAHAN: I lived that term.

7 MR. SHAPIRO: For years, right? I just wanted

8 to share the responsibility because, having testified for

9 the states at the remedy phase, I want to share that

10 responsibility with you.

11 MR. SHAPIRO: Rich -- next, Rich Gilbert. I

12 really liked to hear what you had to say about interfaces,

13 Rich, because this seems to me -- I kept coming -- this

14 came up when I heard you talk about IBM and Microsoft and

15 other examples, it seems to me, going back to at least

16 IBM, and probably selling machines in the 19th century or

17 something, you've often got this pattern where, I have a

18 product and I innovate, I improve it and, as part of

19 improving it, I change the interface or I start producing

20 a complementary product that needs to be compatible and

21 it's innovative and very often intellectual property

22 rights are used to control or secure an interface. And

23 yet we know from the telecommunications, we know from

24 other network industries, that controlling interfaces can

25 lead to a certain octopus-like nature from what might be a


1 secure monopoly in one product initially.

2 And speaking for myself, I get really torn

3 because I feel like, well, fine, the monopolist, if you

4 want to call them, improved their product. Integration,

5 where different components are integrated together, is a

6 very important element of improved performance, and so how

7 are we going to draw these boundaries. You know, do we

8 want to treat interfaces differently, for example, either

9 under a copyright or patents or how does it intersect with

10 antitrust. I think these things are hard and I wonder if

11 you want to say more about that.

12 I was -- it was shocking to me, I have to say,

13 to have an economist tell lawyers to focus on the process

14 rather than the outcome. I just --

15 MR. GILBERT: Not the first today.

16 MR. SHAPIRO: I know, it's true. This is all

17 the more shocking because lawyers are very good at process

18 in my experience and economists are always thinking about

19 these outcomes and are often blind to the process. So, I

20 just -- I don't know, we might have to revoke your card.

21 I don't know.

22 And then -- well, I guess I was maybe not

23 shocked, but a little surprised that you said, well, the

24 courts have done fine because all of this is hard. If

25 it's sham innovation that's your standard at the end, that


1 seems very hard for plaintiffs. And maybe that's what you

2 want. I mean, what would it take -- what would count

3 as a sham? Could you give us an example? For example, to

4 say where, well, the product is a little better but they

5 didn't have to do it this way, for example. What would be

6 a sham? You know, I think it's sort of ironic when I

7 think about Microsoft -- I said I wouldn't talk about it

8 much -- but one of the things Microsoft really pushed

9 throughout the trial was freedom to design their product

10 the way they wanted to and the great benefit of

11 integrating different features, as opposed to more

12 components or modular.

13 Well, what is it now, eight, ten years later? I

14 think they're really having trouble because what the

15 computer science community always does know is, no, that's

16 not good design. Good design is modular and basically

17 people on the other side are telling Microsoft, you

18 wouldn't do this except for strategic reasons. And now in

19 a way that's sort of spaghetti code or the increasingly

20 complexity of Windows has made it very, very hard for them

21 to meet deadlines in terms of coming out with new versions

22 and a lot of other problems they've had.

23 So, what would you do in that case to say, well,

24 you don't have to design it this way, or maybe you don't

25 want to go there if it's not a sham. Any company can


1 choose how to design their product, even if it's not

2 something they would choose to do except for strategic or

3 exclusionary reasons. Or is that too intensive. I don't

4 know.

5 But maybe, and you can confirm this, Rich,

6 you're saying it's so hard to do these cases, that it's

7 true a sham innovation standard is very hard for a

8 plaintiff, but that's okay and we're just not going to get

9 many cases. And maybe that's where we're at. Is that

10 what you support?

11 Dan. I will finish soon here. Dan, there's a

12 lot to say, but I noticed you were emphasizing the

13 somewhat novel nature of network effects and the coining

14 of the application "barrier to entry" in the mid to late

15 '90s by you and Joel Klein, I guess.

16 I have to tell a little story. So, Mike Katz

17 and I did work on network effects going back to the '80s.

18 And so we're working -- (laughter). No, that's neither

19 here nor there. Academics can do anything, but until it

20 comes into practice... So -- but I just want to tell a

21 little story around that.

22 So, we're working in the early '80s and we're

23 working on the network effect. And actually personal

24 computers and computer software is a good example of

25 applications -- that was our example, actually,


1 applications that run on an operating system.

2 And Mike said to me -- and we're getting kind of

3 excited about this and I guess we got published in a top

4 journal, and Mike says, this is great, but I have to tell

5 you, I have a friend who is doing a lot more with this.

6 Not a friend. I should say, a former classmate. So, he

7 says, back when he was at Harvard, there was this guy and

8 he was making a lot of money on this. The guy's name was

9 Bill Gates.

10 So, we often think, oh, we work out these

11 theories, but often after somebody else puts them into

12 practice and understands them pretty well, then the law

13 can kind of catch up with that and maybe academics as

14 well.

15 Okay, I'll leave it at that.

16 MS. GRIMM: Tim, would you like to start off

17 here and respond?

18 MR. BRESNAHAN: Yes, I want to start off. I'm

19 not sure I want to respond. I really like Carl's

20 restatement of my screening idea. That was exactly what I

21 was trying to say.

22 Let me take on hard-to-balance because I don't

23 think I'm against balancing. And I want to use the

24 example of sham innovation because I think that's pretty

25 interesting.


1 The art of balancing, I'm against two things

2 that sounds like balancing. One is a burden-shifting

3 argument that suggests either an efficiency defense,

4 defendant has to show that one rule really is better than

5 the other quantitatively, or in a plaintiff's case where

6 some sort of efficiency defense has been raised, an

7 argument that plaintiff has to show that the world is

8 going to be better off without the market power.

9 I think that those procedures in which one party

10 or the other has to sort of calculate the counterattack

11 from the rule with precision are not going to go very far.

12 And I guess I wouldn't go all the way to saying

13 we should only like the competitive process. But, you

14 know, a courtroom is a hostile environment for numbers.

15 That's just a fact. There are things that courts are

16 better at than numbers. So, a quantitative balancing I

17 think is going to be very difficult.

18 If we were going to have something, for

19 example, bigger than sham innovation, what if a

20 court were going to say, you know, cutting off future

21 races to replace Office and Windows, cutting off the

22 widespread distribution of new innovations in the PC

23 business sounds like a lot of harm to competition to me.

24 There's maybe a lot of zeros at the end of the numbers.

25 Mixing the code between the early stage browser and


1 the operating system, you know, you really got to hold

2 your nose to call that innovation. Maybe there was

3 something innovative to it. Maybe there were some

4 benefits to integration, but it doesn't sound very

5 innovative to me. So in this case the balance is

6 pretty obvious.

7 At that level of a balancing test, I'd be very

8 comfortable, and I think I'd be comfortable with a broader

9 definition than just the innovation has to be literally a

10 sham. I guess I'd be comfortable with the view that the

11 court can feel that the efficiencies are either clearly

12 smaller or clearly -- not smaller in a quantitative sense,

13 but in a salient sense or in a quality of evidence sense

14 than the market power or vice versa. So, I'd be in favor

15 of balancing. I just don't want to do it first.

16 And I think the question that Rich raised

17 earlier about, all the traditional tests are going

18 to look pretty bad for innovation, I guess I would want

19 a balancing test in that area. There's a lot of things

20 that can get labeled as innovation. There's a lot of

21 things which may seem like "innovation" to the defendant

22 but which are dramatically less innovative than what

23 other firms in the industry can do. I think this is

24 one of the enduring lessons of the Microsoft case.

25 On one of my trips to Silicon Valley to discuss


1 the Microsoft case, I talked to a roomful of people and somebody

2 said, weren't they accused of "innovating too fast." And

3 somebody else said, they can't possibly be guilty of

4 innovating too fast; those guys (Microsoft) have never

5 innovated too fast in their lives; they never innovate fast

6 enough. And stuff like that will come out in a courtroom.

7 For this reason, I think that a standard that

8 innovation has to be a sham is too narrow.

9 MS. GRIMM: Professor Gilbert?

10 MR. GILBERT: Well, when I started this project

11 of looking at standards for innovation, I did a lot of

12 reading. And one of the papers I came across was the

13 paper by a Mark Popofsky. And Mark, in that paper,

14 advocated basically different standards for different

15 types of conduct, very much a process-oriented approach.

16 And my initial reaction when I read that paper

17 was I sort of reeled back and said, oh, this doesn't make

18 any sense at all where we're going to put everything that

19 goes on in the economy in a separate category and have a

20 different set of antitrust rules for it. I guess at that

21 point I still had my economist card.

22 But the more I looked at this area, the more I

23 started to think, how do we actually do this analysis and

24 what do you have to take into account to do the analysis

25 right, the more I was led to the conclusion that maybe


1 Mark got it right, that there were certain things that you

2 do and a lot of things you can't do, and that different

3 standards apply to different types of conduct.

4 I mean, certainly the failure to innovate is not

5 an antitrust violation, even though it's really what we're

6 concerned about or should be concerned about.

7 Other problems in this -- along this line, I

8 have a paper with Mike Reardon where we look at

9 technological tying. And the point of that paper is that

10 there are lots of different outcomes. And even if you had

11 really good information, you could do an analysis and you

12 really could examine the problem, you don't know which

13 equilibrium outcome is going to occur in the market. And

14 there could be good outcomes from technological tying and

15 there could be bad outcomes from technological tying. But

16 putting a court into the position of trying to figure out

17 which equilibrium the market is at and which one is

18 better, that's a tough place to be.

19 But I do understand that a lot of this conduct

20 can have very undesirable consequences. If there are less

21 restrictive alternatives, and you can identify them and

22 really carve them out from the conduct, well, that's

23 great. But unfortunately, lots of times the restriction

24 that goes along with an innovation is inherent in the

25 innovation. That's where it's difficult. I think, of


1 course, if you can separate it out, that's fine, it's a

2 lot easier.

3 You mentioned IP protection. Yeah, it would be

4 nice if we could -- it's hard to find an academic these

5 days who wouldn't like to see lesser IP protections, and

6 particularly for things that have network externalities,

7 the other barriers to entry like interface standards. But

8 that's a little bit out of our area.

9 Let me talk a little bit about sham innovation.

10 Again, I'm very sympathetic to the concept that just

11 calling it innovation should not be able to protect all

12 kinds of undesirable conduct and consequences. That just

13 seems pretty obvious.

14 But how you actually measure how discrete an

15 innovation has to be before it is not a sham brings you

16 right into the kind of numbers that Tim was saying are

17 very hard for a court or anybody else to do. What number

18 is big enough? And it's not just the innovation need,

19 it's when the innovation occurs and how it occurs. Is it

20 rolled out in every market, does that make it a sham or

21 not?

22 And I come back to this unilateral refusal to

23 deal analogy. Without defending -- I don't want to defend

24 a "Trinko" approach, but I just find it very odd that

25 innovation that has similar consequences should be held to


1 a higher standard.

2 So, I still think there are things that are

3 unlawful. I don't think that innovation should be able to

4 protect all kinds of activity. But when you are looking

5 hard to do, it's almost impossibly hard to take into

6 account all of the incentive effects and the chilling

7 effects if you get it wrong.

8 And the bottom line, it seems to me, is that

9 most of the time we're not going to have a problem and you

10 should just be careful about chilling innovation by

11 intervening where there might be a problem unless you're

12 absolutely, absolutely sure that that's the case.

13 MS. GRIMM: Professor Rubinfeld?

14 MR. RUBINFELD: I don't have anything to offer

15 specifically on that debate. I just have a couple quick


17 First of all, most of my good ideas actually

18 come from Carl Shapiro one way or another. So, my only

19 intimation was trying to get the courts to see that as

20 well.

21 The other thing -- that actually was a serious

22 comment. But the other slightly more serious comment is

23 that there is an interesting theme I've noticed just at

24 least from this group, and that is, when we -- before we



1 went off to Washington in one extent or another, we were,

2 let's say each of us in our own way, somewhat more

3 theoretically inclined in thinking about some of these

4 issues. And the effect of the Washington experience I

5 think on all of us to one degree or another is really for

6 us to worry about finding something that's really

7 operational that will actually help the agencies and

8 others really resolve practical problems.

9 And so the emphasis on process, and I would put

10 it as sort of finding workable kind of second best

11 solutions, is the natural thing to think about. And I

12 think that's something I do a lot of.

13 In another context, for example, I was struck in

14 a lot of mergers I worked on that we had, I think, at the

15 division, and also probably at the FTC as well, some very

16 sophisticated simulation software, which only as far as I

17 could tell one or two people understood, and not all of

18 them were in the agency. If you know the folks I'm

19 talking about, you know what I mean.

20 And it would have taken in many cases something

21 like six to eight weeks to make it actually functional,

22 which is hard to do under a Hart-Scott-Rodino. So, after I

23 left, I actually, with my co-author, Roy Epstein, wrote

24 some new software and came up with a much simplified

25 procedure which, while greatly simplified, actually is


1 something you can do within the thirty-day period.

2 So, a lot of our work has been driven by that

3 common theme. And I think with respect to sham

4 litigation, that's sort of the same issue I think we're

5 all heading towards, which is, we see a problem and now we

6 have to sort of help to think about what would be a

7 workable solution for the courts.

8 MS. GRIMM: I'd like to give our -- all of you

9 panelists an opportunity to kind of question each other,

10 if you'd like to, as Carl did for all of you, or to

11 respond to any of the points made by each other. And then

12 we'll ask a couple questions on our own.

13 MR. BRESNAHAN: I'd like to take the bait that

14 Carl offered us in discussing the Unocal matter, because I

15 bet that most economists would agree with him that,

16 if there's some amount of market power or power to

17 exclude associated with a patent, and if some act,

18 deception is an extreme, but there might be others,

19 some act or deception to embed it into an interface

20 standard, or maybe even just embedding it in an interface

21 standard in a way that doesn't have any technical

22 benefits, there's some act that extends the coverage of

23 that patent and gives the firm that holds the patent a lot

24 more market power than it would otherwise have, that

25 that's very troubling.


1 And this is one of the disciplinary divides I

2 think you see between economists and attorneys.

3 Economists are more eager to take that position.

4 I suspect that one of the problems with that is

5 that, patent law hasn't been particularly successful --

6 forgetting antitrust law for a minute. Patent law hasn't

7 been particularly successful at delineating the power to

8 exclude in any particular patent conveys on its owner.

9 So, when you get into these cases in the

10 pharmaceutical industry where the patent on the original

11 molecule is running out but there's a new patent on, the

12 same molecule but packaged into a lozenge form or something

13 like that, that it's actually not completely transparent,

14 what's the right answer to the question, "how much

15 market power does the patent provide?" And when the

16 pharmaceutical firm starts playing Carom shots off the

17 enormous complexities of the regulatory process under

18 Hatch-Waxman, what is the answer to the question, "how

19 much market power was conveyed by the original patent?"

20 So that even if we're fairly comfortable with

21 the idea that creation of additional market power beyond

22 what the patent originally would have given that can be

23 a thing that can be very hard to determine in a legal

24 sense.

25 There probably is a near consensus among academic


1 economists that patent policy in the United States over

2 protects the patent holder. I think I agree with Carl on

3 that. There's this other problem that patent policy is

4 too vague, that patents simply don't look like property rights

5 here. You have to go to courts or to the regulatory

6 system to find out who owns what. And that -- the antitrust

7 doctrine, Carl quoted the traditional antitrust doctrine that,

8 intellectual property law if what it is and we ask

9 whether there's additional market power on top of that.

10 That may be more attractive in its economics than its law

11 because it's hard to determine how much market power there

12 would have been absent the anticompetitive acts.

13 MR. GILBERT: I'd kind of like to reinforce what

14 Tim said earlier, Carl, and I think also Dan as well.

15 While a lot of our discussions today might be

16 interpreted as suggesting that Section 2 analysis is very

17 hard to do and therefore we shouldn't do it, and there's a

18 lot of ways in which I think that's absolutely wrong, and

19 that is Section 2 analysis isn't that hard and should be

20 done, I do think that the law creates a road map to make

21 Section 2 analysis unnecessarily difficult. You've got to

22 have -- you know, you've got to identify the market, the

23 product market, the geographic market, you have standing,

24 you have all of these things. In all of these cases, I

25 know cases I have been involved in, I'm sure everybody


1 else, it seems like you never get to the question.

2 You know, the relevant question is: Does the

3 conduct really raise prices. And most of the time that's

4 pretty obvious whether it does or doesn't and you don't

5 have to do all this other stuff. And I think the law

6 often puts us in a position of having to go through this

7 kind of rogue set of steps that's in many ways very, very

8 counterproductive.

9 MR. SHAPIRO: Well, two things. The first one

10 is to emphasize my concerns about the fetish over patents

11 in intellectual property rights, therefore in some cases

12 being a little blind to the fact that they can be

13 leveraged, if you want to use that word, and you can get

14 more power than was granted with the patent, particularly

15 with patents that are very iffy. And there's a whole set

16 of these questions about that.

17 I mean, I guess it's outside of Section 2, but

18 these pharmaceutical settlements cases, like the Shering

19 case the FTC brought, and where the Second Circuit has

20 gone with those cases was the tamoxifen case and seeing

21 the patent as, oh, well, even if you paid off a competitor

22 to leave because you have a patent, somehow it's okay, it

23 doesn't mean you've stated an antitrust claim, that's

24 something the -- you know, even if that's outside

25 Section 2, that thinking is something that both agencies


1 should really head off.

2 And I guess there's an IP report still coming.

3 There's -- that seems to be a very important role to

4 delineate the importance of patents, yes, and the reward,

5 yes, but there's a limited power that is granted, and

6 beyond that, we can have abuses.

7 I would shift topics a little bit and actually

8 ask a question of Dan that I skipped when I was standing

9 up.

10 I'm curious, Dan, in your discussion of

11 LePage's, whether you -- I guess you favor a bright line

12 test of comparing price to marginal cost for additional

13 units sold in a bundle. Or maybe, what about comparing

14 marginal revenue to marginal cost to see whether the extra

15 sale and bundling was profitable or not, a kind of profit

16 sacrifice test.

17 So, would you favor either of those? I mean,

18 you're objecting to LePage's as being vague. So, here are

19 two potential standards that are a lot more specific. I

20 guess I'm talking about a safe harbor, either if the price

21 is above marginal cost or if the marginal revenue is above

22 marginal cost, then the bundling is okay. Of course, even

23 if it's not, we assume you want to look first back to

24 scope and so forth. So, there's two questions related to

25 scope.


1 If the program is limited, there's only a few

2 customers or a short period of time, if that's the case,

3 would you just wave it through? It just doesn't matter

4 what the structure of the program is to you because it

5 couldn't have anticompetitive effects or not?

6 And then related to that, I don't know if you're

7 familiar with the EU's approach to this, but they're

8 required to share methodology and calculating volume

9 discounts, multi-product or single product, and whether

10 you think that's something that the U.S. should pick up

11 on.

12 MR. RUBINFELD: Good questions, Carl. I

13 actually am not familiar with the EU side, so I am not

14 going to try to answer that.

15 With respect to the workable test, you're right,

16 I was suggesting just a safe harbor and I think I would

17 accept your clarification. I was looking for a profit

18 sacrifice kind of test, so I would compare marginal

19 revenue and marginal cost, that's if marginal revenue is

20 different from price, but only to get a safe harbor.

21 The problem in extending that test is that,

22 while I think there's some bundling cases which I think

23 are appropriately seen as really being an extension of a

24 predatory pricing case and probably ought to come under

25 Brooke Group, I think there are other kinds of bundling


1 practices which probably are not seen that way. So, the

2 safe harbor I don't think ought to be seen as

3 characterizing all, all types of bundling. Other types of

4 bundling might seem more smart with respect to other kinds

5 of exclusionary conduct of the kind we talked about

6 earlier today.

7 The other thing that you asked me about my point

8 about the effect of this initial program being very

9 limited. To me that is quite important because -- I may

10 hear something to the contrary in a second -- but it seems

11 to me that if there's a practice that cannot be shown to

12 either have the effect and be sufficiently exclusionary

13 that it makes a competitor not viable or perhaps even has

14 no effect on its ability to operate at an efficient scale.

15 I don't see how that practice ought to be considered

16 anticompetitive.

17 So, I think you do -- in my opinion, you do have

18 to show that if there's exclusion, it's substantial enough

19 to really matter from the point of view of the potential

20 competitiveness of the firm that's being affected.

21 We can debate whether we should focus on volume

22 scale or efficient scale, but certainly there ought to be

23 some measurable effect.

24 MS. LEE: Dan, you had said in your presentation

25 that a variant on the profit sacrifice test would be


1 appropriate to use as a general standard for all Section 2

2 conduct.

3 I was hoping that you could refine that a little

4 bit, in particular, you know, how is this different from

5 the traditional profit sacrifice test, whatever that may

6 be, and how does it differ from the no economic sense

7 test?

8 MR. RUBINFELD: That's a great question. I

9 think I really can't -- without going back to my drawing

10 board for maybe a few years, I don't think I can answer

11 that very well.

12 The reason why I was saying a variant in my

13 comments is that I have been trying to follow some of the

14 debate in the literature among the folks who prefer more

15 of a balancing test to a profit sacrifice test. And it's

16 not that hard to come up with hypotheticals that would

17 defeat almost any version of a profit sacrifice test under

18 certain circumstances.

19 And so what I was imagining was that one would

20 be able to come up with either a more robust rule that was

21 not subject to too many of these hypotheticals, or maybe a

22 complex rule that said under certain circumstances we do

23 the test one way and under other circumstances another.

24 But, unfortunately, I don't really have an

25 answer to that question. I am hoping, Jim, that you and


1 others at the Division will work hard to give me an

2 answer.

3 MS. LEE: Okay. That was a good way to deflect

4 the question.

5 MR. RUBINFELD: Others here may have an answer.

6 MS. LEE: Let me also get you to react to Tim's

7 proposal in terms of how we should evaluate Section 2

8 cases, I would call it a step-wise rule of reason. Tim,

9 please feel free to disagree with me if you don't think I

10 am charactering that appropriately.

11 MR. RUBINFELD: You are asking me that question?

12 MS. LEE: Yes. How would it be different from a

13 variant of the profit sacrifice test that you think would

14 be appropriate.

15 MR. RUBINFELD: Well, I guess without being too

16 specific, I have some of the same reactions I guess others

17 on the panel have expressed based upon my own experience

18 both in the Division and working on private cases, and

19 that is the cases often get bogged down in complex debates

20 about issues like market definition, without really

21 talking about competitive effects.

22 So, I'm actually -- at the level Tim is talking

23 about, I'm very symptomatic with his suggestion. I think

24 the pharmaceutical cases for me are really an excellent

25 example of that. I have been involved in a number of


1 these where there's a huge battle about market definition,

2 which can be a very tricky issue in pharma cases for a lot

3 of reasons, and yet I thought that -- the answer to the

4 question, how you define the relevant market, at least if

5 you are using the guidelines, really has almost no impact

6 on whether there's a competitive effect.

7 If you think that a generic would have entered

8 earlier, and the generic most of the time is going to

9 enter at a substantial discount off the price of the brand

10 product, there is likely to be an effect. It's going to

11 be the rare case where competition is driven just by other

12 branded products.

13 Now, if you think that's the case, then the real

14 battle is going to be on issues such as causation, whether

15 the practice itself had procompetitive benefits, and so

16 on. So, there will still be a lot to debate, but the

17 debate will be about whether this competitive effect A and

18 B, whether there are justifications that say that that

19 procompetitive effect was worth it.

20 Rather than debate, which can get pretty far off

21 the subject, or market power -- certainly most, if not

22 all, successful brand products generate a lot of market

23 power. That's the point of Hatch-Waxman to some extent,

24 or the point of patent laws generally. And -- but the

25 point of Hatch-Waxman in part is to encourage entry to


1 benefit consumers. And the effect of that entry is going

2 to be to reduce some of that market power.

3 And I don't think any of that should be very

4 controversial and yet I have seen a number of cases where

5 the battles over whether firms have market power seem to

6 take prominence. And so a process that in my view would

7 move us more quickly to the heart of the cases would be a

8 constructive process.

9 Ms. Lee: Tim, let me ask you to clarify

10 something that I didn't quite understand about your

11 proposed way of analysis.

12 In particular you had suggested that, well, if

13 you look at -- if you first establish a causal effect

14 between the act and then the effect, this gets you around

15 the whole complex processes of trying to figure out what

16 the appropriate but-for world is when you do the

17 traditional sort of economic efficiency analysis.

18 I don't quite see that in terms of, to establish

19 causality, don't you have to establish in some sense what

20 the world would have been absent the exclusionary act?

21 MR. BRESNAHAN: That's a good question.

22 I agree that to establish causality you need to

23 say what the world would have been like in a competitive

24 sense absent the anticompetitive act.

25 I think the force of my argument is to -- is


1 really procedural. It's to move the things which are

2 going to be most difficult for courts to do back in this

3 sequence. So, I mean, you heard us all economists say,

4 it's often easier to see whether there's a competitive

5 effect than to get market power right. I think that's

6 probably going to be true.

7 Certainly if there's a Section 2 case there,

8 it's going to be easy to see what the competitive effect

9 is. And then if you can't see it, there's no Section 2

10 case there.

11 Similarly, that the challenged conduct causes

12 the market to be less competitive, that's an inquiry that

13 can be undertaken within the four walls of what causes

14 competition, without any balancing against the efficiency

15 of the challenged conduct. Does it change the conditions

16 of the competition? And I bet a lot of cases will follow

17 thereto, and that's within the four walls of ordinary

18 antitrust analysis. Is the reason that the market is less

19 competitive because the challenged conduct raises entry

20 barriers, raises them in a way that, you know, the

21 entrants and third parties can't get around to the

22 relevant time frame. Those are all difficult tests to

23 pass.

24 So, most Section 2 inquiries should fall by the

25 wayside. I just want them to fall by the wayside cheaply.


1 And then you come to the last thing, which as

2 we've all said is really, really hard, you know, you've

3 got causation, there's some challenged conduct which is

4 changing the conditions of competition, but there's also

5 something good about it. You know, it's innovative or

6 it's a price cut so it's especially good for customers,

7 and now we've got to do this balancing, which I think is a

8 very, very difficult thing to do.

9 So, I just want to reduce the incidence of the

10 balancing. Rather than leaping to that right away, go

11 through other things first and discard cases. And I

12 think that the causation -- the causation inquiry which

13 says, is the challenged conduct holding entry barriers

14 high is an easier counter-factual inquiry than, is the

15 extent to which it's holding entry barriers high worse

16 than its countervailing efficiency. It's got one less

17 difficulty.

18 So that would be how I would proceed. And the

19 basic idea is to save wear and tear on the system, which

20 is potentially the result.

21 MS. LEE: Thank you for the clarification.

22 Rich, I wanted to ask you, you had said you have

23 became more sympathetic to the idea that in different

24 Section 2 matters different standards should apply.

25 How would one go about determining the best


1 standards to apply in each situation?

2 MR. GILBERT: Again, a very good question.

3 Certainly what sets innovation apart is the

4 temporal linkage and very complicated linkage between the

5 conduct at issue and the investment research and

6 development that create the innovation and the prospects

7 that any antitrust venture that would show that kind of

8 very beneficial investment. And suppose you had a case

9 where you didn't think that linkage was all that

10 important, so you intervene in that case. But then if you

11 do that, that also creates a precedence for there being

12 other cases the linkage could be very important, and you

13 definitely don't want to chill innovation in those other

14 cases.

15 If you think about how some of those early cases

16 -- if some of those early cases came out differently,

17 because almost all the cases that I can see ultimately

18 basically are pretty close to a sham innovation test. If

19 they had done something very different from that, what the

20 implications would be for people actually involved in

21 product design could be kind of interesting.

22 Now, there is a lot of conduct where I don't

23 think those issue are at all significant. You know, they

24 may be present to some extent, but they're just not

25 significant. And so if you're talking about ordinary


1 exclusive dealing or bundling or whatever, I think in many

2 of those cases you can if not forget about, certainly

3 discount, the more complicated intertemporal effects. And

4 the analysis I think becomes much easier. And the sort of

5 rule of reason analysis becomes much more possible.

6 Weighing of benefits and costs becomes more reasonable.

7 MS. LEE: Carl, do you have anything you want to

8 say in addition to what you said already about general

9 standards? You had said in your comments that you were

10 very sympathetic to a standards approach.

11 Is there anything else you would like to add?

12 MR. SHAPIRO: Well, you called it a structured

13 -- what did you call it?

14 MS. LEE: No, I called it a step-wise.

15 MR. SHAPIRO: Good, that's the ticket.

16 MS. LEE: I think that's what it was.

17 MR. SHAPIRO: So, I think of it in terms of

18 screens. Traditionally, the monopoly power screen. You

19 have a lot of power, and if you don't, then Section 2

20 doesn't apply.

21 I think I would push for: Does the conduct hold

22 up the prospect to leading to significant increase in

23 market power, okay, as actually a better question to use

24 as a screen.

25 Now, the reason I think the traditional screen


1 has been applied, it's been assumed if you don't have any

2 power to start you, you can't manufacture something from

3 nothing. And that may be true in a lot of cases, although

4 not always. Maybe deception turns up.

5 Furthermore, even if you have power to begin

6 with, if the conduct couldn't add much to it, maybe you

7 have a patent, then we can dismiss that case, we don't

8 have to go anywhere. So, you would get something knocked

9 out on this increment screen that you wouldn't get knocked

10 out based on a preexisting traditional power screen.

11 So, I think it's a lot more closely tied to what

12 Tim was saying at the top of the program here about

13 looking at effects and increment. And there are ways to

14 do that, implement that, and I have written about that and

15 other people have, too. So, that's a general concept I

16 think that cuts across a lot of cases.

17 At the same time, I agree with Rich that -- and

18 I think Dan -- well, profit sacrifice may apply in some

19 cases but not others, so then you have to be more nuanced.

20 You know, profit sacrifice would not apply in the Unocal

21 case.

22 MS. LEE: So, let me ask you the same question I

23 asked Rich.

24 Do you have a suggestion about the methodology

25 of figuring out, well, which is the best approach in each


1 type of matter?

2 MR. SHAPIRO: It would be very unwise for me to

3 get into that at this late hour.

4 MS. GRIMM: I just have one question on

5 remedies, and this is for Tim. Again, on the Microsoft

6 remedy which you labeled a fizzle and you said the remedy

7 in AT&T from your point of view was successful.

8 I was wondering if could share any views with us

9 on appropriate remedies in Section 2 cases, perhaps

10 structural versus the conduct remedies.

11 MR. BRESNAHAN: I'm almost certain there's no

12 general law of remedies in Section 2 cases because

13 Section 2 cases are so context specific and so fact dense.

14 You know, in the structural remedy that was

15 negotiated rather than imposed by a court in AT&T, I think

16 the logic of that was caused by an attempt to minimize the

17 harm to competition and innovation by walling off the rest

18 of the industry (by vertical disintegration) from the

19 necessarily regulated sector of telephony, local phones.

20 And that's just a very specific argument.

21 So, some principle that has remedies that are

22 reasonably proportional to the harm to competition that's

23 been proved, I think it's going to -- I think it's going

24 to be very hard to go farther than that to a broader abstract

25 statement.


1 MR. SHAPIRO: If I could just make a quick

2 comment. I thought about Microsoft remedies in context

3 here. At one end you have, sin no more, don't do what you

4 did before, narrowly defined, maybe defined to reflect the

5 market changing. And, you know, that doesn't seem to me

6 that does much to restore competition if there's been real

7 damage with some lasting effect, okay, if the case was

8 significant to begin with.

9 One of the things that was interesting in that

10 case 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.

12 So, you can't say, ah, we're trying to engineer the market

13 to return to a certain state and that's what we mean by

14 restoring competition.

15 So, again, in that context, really the case was

16 about raising entry barriers, as Tim put. My view was,

17 you should have a remedy that lowered entry barriers and

18 then come what may. Maybe entry will occur, maybe it

19 won't.

20 But sin no more seems to me it's probably going

21 to be too weak in most cases where the case was worth

22 bringing to begin with.

23 MS. LEE: Let me ask a follow-up to that.

24 If the only suitable remedy is a sin no more

25 remedy, do you think the agency should bring a Section 2


1 case in that instance?

2 MR. SHAPIRO: Well, there still could be some

3 deterrent effects. And there are private cases that

4 follow on, for example, that could have a major role. And

5 there were private cases in the Microsoft case that

6 involved a lot of money.

7 So, it could well be. I guess I would hope if

8 it's a major case that either agency could come up with

9 something a little more effective and maybe even creative.

10 But, at the same time, partly from the Microsoft

11 experience, it's very hard for a court to impose a remedy

12 when the company says this is crazy, it won't work, you'll

13 destroy all sorts of good things, and the government

14 agency, you know, yeah, there's information but it's hard

15 to know. So, I think it's very hard. And so if you are

16 stuck with sin no more, it could still be worth bringing,

17 sure.

18 MS. LEE: Let me just solicit the other

19 panelists about that. Anything different or anything to

20 add?

21 MR. BRESNAHAN: Yes, I guess I'd be more

22 conservative on this ground than Carl. It's hard to get a

23 lot in deterrence in this area of antitrust law because

24 it's so hard to -- you know, we're never going to have a

25 doctrine that says these specific practices are


1 anticompetitive. I mean, guys will just know not to do

2 those particular practices. It's much more complex than

3 that.

4 So, other than generally wanting to keep the

5 idea that there might be this prosecution of particularly

6 egregious anticompetitive acts, this is not a great area

7 where you can get an awful lot of deterrence out of --

8 you know, out of a case where there's a remedy that

9 doesn't do anything.

10 So, I'd be less -- I'd put less emphasis on

11 deterrence and, more emphasis on the view that it

12 should really be looking for cases where you can make

13 a big difference for the American consumer.

14 I mean, before I was in government, in

15 connection with Microsoft, I took the position, don't

16 bring it unless you're going to do something really

17 big, which I went on to say, probably meant don't bring

18 it, although that turned out to be wrong. The government

19 did ask for a remedy that would have changed the

20 conditions of competition.

21 I think these experiences are rare, important

22 and efficacious in the first instance, and seeking

23 deterrence only, you know, only perhaps in flagrant

24 examples.

25 MR. GILBERT: Sometimes, not always of course,


1 the case that dominance leads to conduct that is

2 persistent and durable, that companies in dominant

3 positions tend to do the same sort of anticompetitive

4 things. And it's also the case that that dominance is

5 persistent, that even if you try to break it up, forces

6 are going to tend to recreate it. And I wouldn't say

7 that's always true, but that's sometimes true.

8 But I also say that, even in those cases where

9 you cannot have a real structural remedy, that structural

10 remedies wouldn't be very effective, a big case like this

11 brought by DOJ or FTC has a lot of consequences for these

12 companies. And I think you have a significant deterrence

13 effect.

14 MR. RUBINFELD: The only thing that I was going

15 to add is, these remedies come out of course not in just

16 in court decisions we're talking about, but also in

17 consent decrees that are reached. And I think it makes a

18 big difference how you craft a consent decree. You know,

19 I can think of some cases which I was involved in where we

20 literally got a promise never to do A again and nothing

21 more. There were other cases where the consent decree

22 really laid out fairly carefully what we meant by not

23 doing it again, not only for this company, but also the

24 consent decree sent a clear message since the consent

25 decree can be part of the public record.


1 So, you can get some deterrence even in a

2 situation where the structural remedy doesn't work if you

3 craft the right consent decree. And, obviously, it

4 depends on every case, but I think obviously the agencies

5 should and I am sure do think hard about exactly how to

6 did that. And that's an important exercise.

7 MR. SHAPIRO: Let me just clarify. There was

8 kind of a sin no more at one extreme and then I heard a

9 couple of people talking about structural remedies.

10 There's a lot of running room in between.

11 MS. LEE: Agreed.

12 MS. GRIMM: Well, my watch says it is 4:30. I

13 would like to thank all of our panelists for being here

14 this afternoon and sharing with us their very insightful

15 ideas.

16 I would also like to thank again the University

17 of California at Berkeley for their hospitality.

18 Would everyone please join me in giving our

19 panelists a round of applause.

20 (Applause.)

21 (Whereupon, at 4:30 p.m., the hearing was

22 concluded.)



1 C E R T I F I C A T I O N O F R E P O R T E R



5 DATE: JANUARY 31, 2007

7 I HEREBY CERTIFY that the transcript contained

8 herein is a full and accurate transcript of the notes

9 taken by me at the hearing on the above cause before the

10 FEDERAL TRADE COMMISSION to the best of my knowledge and

11 belief.

  DATED: February 22, 2007

Updated June 25, 2015