Academic Testimony

1

1	UNITED STATES FEDERAL TRADE COMMISSION
2	and
3	UNITED STATES DEPARTMENT OF JUSTICE
4
5
6
7	SHERMAN ACT SECTION 2 JOINT HEARING
8	ACADEMIC TESTIMONY
9	WEDNESDAY, JANUARY 31, 2007
10
11
12
13
14	HELD AT:
15	UNIVERSITY OF CALIFORNIA AT BERKELEY
16	2220 PIEDMONT AVENUE
17	WELLS FARGO ROOM
18	BERKELEY, CALIFORNIA
19	9:30 A.M. TO 4:30 P.M.
20
21
22
23
24	Reported and transcribed by:
25	Kathleen Carr Meheen, CSR 8748

2

1	MODERATORS
2	Morning Session:
3	WILLIAM E. COHEN
4	Deputy General Counsel for Policy Studies
5	Federal Trade Commission
6	and
7	JOSEPH J. MATELIS
8	Attorney, Legal Policy Section
9	Antitrust Division, U.S. Department of Justice
10
11	PANELISTS
12	Morning Session:
13	Aaron Edlin
14	Joseph Farrell
15	Howard Shelanski
16
17
18
19
20
21
22
23
24
25

3

1	MODERATORS
2	Afternoon Session:
3	KAREN GRIMM
4	Assistant General Counsel for Policy Studies
5	Federal Trade Commission
6	and
7	JUNE K. LEE
8	Economist
9	Antitrust Division, U.S. Department of Justice
10
11	PANELISTS
12	Afternoon Session:
13	Timothy Bresnahan
14	Richard Gilbert
15	Daniel Rubinfeld
16	Carl Shapiro
17
18
19
20
21
22
23
24
25

4

1	C O N T E N T S
2	MORNING SESSION (ACADEMIC):
3	Introduction
4	Presentations:
5	Joseph Farrell
6	Aaron Edlin
7	Howard Shelanski
8	Moderated Discussion
9	Lunch Recess
10
11	AFTERNOON SESSION (ACADEMIC):
12	Introduction
13	Presentations:
14	Timothy Bresnahan
15	Richard Gilbert
16	Daniel Rubinfeld
17	Carl Shapiro
18	Moderated Discussion
19
20
21
22
23
24
25

5

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

6

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

7

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.

8

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?

9

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

10

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

11

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

12

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

13

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

14

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

15

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

16

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.

17

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

18

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

19

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.

20

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

21

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

22

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,

23

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

24

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

25

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

26

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

27

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.

28

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

29

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

30

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

31

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

32

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,

33

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

34

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.

35

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.

36

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.

37

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,

38

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

39

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

40

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

41

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

42

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

43

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

44

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

45

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

46

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

47

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.

48

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

49

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

50

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

51

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

52

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

53

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.

54

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

55

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.

56

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

57

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

58

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

59

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

60

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

61

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

62

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

63

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

64

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

65

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

66

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?

67

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

68

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.

69

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.

70

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

71

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

72

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.

73

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

74

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.

75

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

76

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

77

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.

78

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

79

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

80

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

81

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

82

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

83

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.)
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

84

1	AFTERNOON SESSION
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.

85

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

86

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.

87

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)

88

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

89

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

90

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

91

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

92

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

93

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

94

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

95

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

96

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

97

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

98

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

99

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,

100

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

101

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

102

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.

103

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

104

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

105

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

106

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.

107

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

108

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?

109

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

110

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

111

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

112

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

113

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

114

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

115

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

116

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.

117

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

118

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

119

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

120

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

121

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

122

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.

123

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.

124

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.

125

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

126

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

127

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.

128

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.

129

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

130

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

131

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

132

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

133

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

134

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

135

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.

136

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

137

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

138

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

139

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

140

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

141

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

142

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

143

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

144

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

145

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

146

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

147

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

148

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

149

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

150

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

151

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

152

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

153

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

154

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

155

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

156

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

157

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,

158

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.

159

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

160

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

161

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

162

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

163

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

164

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
16	comments.
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
25

165

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

166

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.

167

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

168

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

169

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

170

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.

171

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

172

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

173

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

174

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

175

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

176

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

177

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.

178

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

179

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

180

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

181

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

182

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.

183

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

184

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

185

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,

186

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.

187

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.)
23
24
25

188

1

C E R T I F I C A T I O N O F R E P O R T E R

2

3

DOCKET/FILE NUMBER: P062106

4

CASE TITLE: SECTION 2 HEARING, PREDATORY PRICING

5

DATE: JANUARY 31, 2007

6

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.

12

DATED: February 22, 2007

13

14

______________________________

15

KATHLEEN CARR MEHEEN, CSR 8748

16

17

18

19

20

21

22

23

24

25

Updated January 2, 2024