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