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	UNDERSTANDING SINGLE-FIRM BEHAVIOR:
9	MONOPOLY POWER SESSION
10	WEDNESDAY, MARCH 7, 2007
11
12
13
14
15	HELD AT:
16	UNITED STATES FEDERAL TRADE COMMISSION
17	601 NEW JERSEY AVENUE, N.W.
18	WASHINGTON, D.C.
19	9:30 A.M. TO 4:30 P.M.
20
21
22
23
24	Reported and transcribed by:
25	Susanne Bergling, RMR-CLR

2

1	MODERATORS:
2	DENNIS W. CARLTON
3	Deputy Assistant Attorney General for Economic Analysis
4	Antitrust Division, Department of Justice
5	and
6	JOEL L. SCHRAG
7	Economist
8	Bureau of Economics, Federal Trade Commission
9
10	PANELISTS:
11
12	Morning Session:
13	Andrew J. Gavil
14	Richard J. Gilbert
15	Michael L. Katz
16	Philip B. Nelson
17	Joseph J. Simon
18	Lawrence J. White
19
20	Afternoon Session:
21	Simon Bishop
22	Thomas G. Krattenmaker
23	Miguel de la Mano
24	Joe Sims
25	Irwin M. Stelzer

3

1	C O N T E N T S
2
3	MORNING SESSION:
4	Introduction
5	Presentations:
6	Andrew J. Gavil
7	Richard J. Gilbert
8	Michael L. Katz
9	Philip B. Nelson
10	Joseph J. Simons
11	Lawrence J. White
12	Moderated Discussion
13	Lunch Recess
14
15	AFTERNOON SESSION:
16	Introduction
17	Presentations:
18	Simon Bishop
19	Thomas G. Krattenmaker
20	Miguel de la Mano
21	Joe Sims
22	Irwin M. Stelzer
23	Moderated Discussion
24	Conclusion
25

4

1	P R O C E E D I N G S
2	- - - - -
3	MR. SCHRAG: Good morning. Sorry about the
4	technical issues. Welcome.
5	My name is Joel Schrag. I am an economist at
6	the Bureau of Economics here at the Federal Trade
7	Commission, and I am one of the moderators for this
8	panel. My co-moderator, standing next to me, is Dennis
9	Carlton, Deputy Assistant Attorney General for Economic
10	Analysis at the Antitrust Division of the Department of
11	Justice.
12	Before we get into the substance of the program,
13	on behalf of the FTC staff who have worked on this
14	session, I would like to take the opportunity to thank
15	all of our colleagues from DOJ for their hard work and
16	their efforts to jointly present this session.
17	In addition, after today's and tomorrow's
18	hearings on monopoly power, the hearings will next turn
19	to issues involving remedies later this month, and so I
20	urge you all to be sure to check our agencies'
21	respective web sites for updates on these future
22	hearings.
23	As the FTC representative, I do have just a few
24	housekeeping matters to cover before we begin. First of
25	all, please turn off all of your cell phones,

5

1	BlackBerries and other noise-making electronic devices.
2	Second, the restrooms are located out through the double
3	doors and across the lobby. If you need help to find
4	them, there are signs that should guide you.
5	Third, one safety tip, especially for visitors,
6	in the unlikely event that the building alarms go off,
7	please proceed calmly and quickly as instructed. If we
8	must leave the building, you exit out the New Jersey
9	Avenue doors by the guard station. Please follow the
10	stream of FTC employees to a gathering point across the
11	street and await further instruction, but hopefully that
12	won't be necessary.
13	Finally, we request that you please not make
14	comments or ask questions during the session. Thank
15	you.
16	Let me just say a few things about the session.
17	Many of the prior sessions of the hearings addressed
18	particular conduct that's been challenged under Section
19	2 of the Sherman Act. Today, the hearings turn to
20	issues of monopoly power and market definition, and
21	these issues we believe are very important.
22	In fact, if you were at the opening day of the
23	hearings back in June, both Herbert Hovenkamp and my
24	co-moderator, Dennis Carlton, were given the opportunity
25	to place the issues for the subsequent hearings in

6

1	context. Both identified monopoly power and market
2	definition as areas where there are difficult, uncertain
3	questions that must be addressed in many cases, and I
4	expect that today's panel will help to clarify, if not
5	completely resolve, these difficult questions.
6	The hearings will be organized as follows:
7	First, we'll hear an approximately 15-minute
8	presentation from each of our six distinguished
9	panelists. We'll probably take a break after the fourth
10	panelist and then come back from the break and hear from
11	the two remaining panelists. After that, the panelists
12	will have an opportunity to comment on each other's
13	presentations, and we'll have a moderated discussion.
14	So, I think I'd now like to turn things over to
15	my co-moderator, Dennis Carlton, who will introduce our
16	distinguished panelists.
17	Thank you very much.
18	DR. CARLTON: Okay, thank you. I am Dennis
19	Carlton. I am a Deputy Assistant Attorney General in
20	the Antitrust Division, and it is a pleasure to welcome
21	all of you to these joint FTC/DOJ hearings.
22	I had the privilege of participating in the
23	opening session of the hearings, and one of the topics I
24	said that needed clarification was precisely the topic
25	of the panels today and tomorrow, a focus on what we

7

1	mean by "market power" and "market definition" in
2	Section 2 cases I think is really important.
3	I am also a Commissioner on the Antitrust
4	Modernization Commission, and despite my attempting to
5	do so was not able to convince the Commission to study
6	in depth the definition of market power and market
7	definition in Section 2 cases and to report on it. So,
8	that, I think, emphasizes all the more how important
9	this session, this panel discussion, is today, and the
10	real question is, can we reach consensus on any of the
11	hard questions or at least can we reach a consensus that
12	there's a lot of ambiguity and arbitrariness in what is
13	going on?
14	I am honored to chair such a distinguished
15	panel. All of the members of the panel have extensive
16	experience, both academic and nonacademic, in antitrust
17	and have served both in the private sector and in the
18	government sector.
19	In the interest of saving time, I am going to
20	introduce them all at once and hopefully by that time
21	the computer will work. So, starting with Phil, Phil
22	Nelson is a principal at Economists, Inc., an economic
23	consulting firm. Previously, he served as the Assistant
24	Director for Competition Analysis at the FTC and as an
25	Adjunct Professor at Fordham Law School. He has written

8

1	numerous articles and two books on antitrust topics, and
2	he edited the ABA's antitrust section of market power --
3	The Market Power Handbook. He currently is the
4	vice-chair of the section's Healthcare and
5	Pharmaceuticals Committee.
6	Beside Phil is Joe Simons. Joe is a well-known
7	attorney. He's a partner and co-chair of the antitrust
8	group at Paul Weiss. Previous to that, Joe was the
9	chief antitrust enforcer at the Federal Trade
10	Commission, serving as the Director of the Bureau of
11	Competition from June 2001 until August of 2003. He has
12	the interesting characteristic of once being the tenth
13	largest wireless carrier in the country, because I
14	believe he was a trustee and had a lot of wireless
15	licenses, but in addition to that, he has achieved
16	something that's actually quite rare for attorneys to
17	do, and that is he's written an article that economists
18	cite all the time and is associated with critical loss
19	analysis.
20	Beside Joe, in a missing seat, is Larry White,
21	who I am sure is on his way. Larry is the Arthur
22	Imperatore Professor of Economics at NYU School of
23	Business. He's the Deputy Chair of the Department of
24	Economics. Previously, in the early eighties, Larry
25	served as the Director of the Economic Policy Office in

9

1	the Antitrust Division. Larry has written several books
2	and articles, one of which is well-known to antitrust
3	practitioners called The Antitrust Revolution:
4	Economics Competition and Policy. He's currently the
5	editor of The Review of Industrial Organization. Prior
6	to serving at the Justice Department, he did extensive
7	government service both for the Federal Home Loan Bank
8	Board and for the Council of Economic Advisers.
9	Andy Gavil is a Professor of Law at Howard
10	University where he not only teaches antitrust, but he
11	has also extensively written on antitrust many articles
12	and has a very well-known case book with Bill Kovacic
13	and Jonathan Baker, Antitrust Law in Perspective. He is
14	about to publish or co-author a book called Microsoft
15	and the Globalization of Competition Policy, which I am
16	sure has focused on Section 2 type behavior. He's
17	currently the articles editor of The Antitrust Magazine
18	and serves on the ABA Antitrust Section's Liaison Task
19	Force to the Antitrust Modernization Commission. He is
20	Of counsel to the Sonnenschein Law Firm.
21	To Andy's left is Rich Gilbert. Rich is a
22	Professor of Economics at the University of California
23	at Berkeley. He served as the Deputy Assistant Attorney
24	General in the Antitrust Division in the mid-nineties,
25	and at that time, he led the effort to write the

10

1	Antitrust Guidelines for the Licensing of Intellectual
2	Property. He has written widely on antitrust topics.
3	He is currently the Director of the Competition Policy
4	Center at Berkeley and is associated with the economic
5	consulting firm of COMPASS.
6	Finally, Mike Katz at the end of the table.
7	Mike is currently the holder of the Sarin Chair in
8	Strategy and Leadership at Berkeley, the Business
9	School, and also holds an appointment in the Economics
10	Department. Mike served as the Deputy Assistant
11	Attorney General in the early 2000s, and he also served
12	as the Chief Economist at the Federal Communications
13	Commission. He's written numerous articles on economics
14	and antitrust and has specialized in many topics,
15	including network industries.
16	So, with that introduction, I'll turn it over to
17	our first speaker, Phil, and just let me remind the
18	speakers, we're kind of running tight because we started
19	late, so if you could keep to the 15 minutes, that would
20	be good. The organization of this is going to be four
21	speakers will go, 15 minutes, we'll take a 10-minute
22	break, we'll have two more speakers. We will give the
23	speakers a brief opportunity to talk to each other, and
24	then I'll moderate a discussion for about an hour or so.
25	Thank you.

11

1	MR. NELSON: So, we have a -- are we moments
2	away or should I just proceed without slides?
3	DR. CARLTON: Is the computer still not working?
4	MR. NELSON: Well, okay, the reason they put me
5	first is the slides that you can't see are really sort
6	of a background deck that gives you the background on
7	market power. The first slide cites the definition of
8	market power that's at the front of the monograph that
9	the ABA published that was referred to earlier, which is
10	market power is the ability of a firm or a group of
11	firms within a market to profitably charge prices above
12	the competitive level for a sustained period of time,
13	and as you can't see on the screen, the word
14	"profitably" is in italics, and so one of the important
15	things in the definition is that a monopolist profit by
16	doing this.
17	If entry is easy, you may be able to raise
18	prices, but not profitably, because somebody will enter,
19	and if there are a lot of competitors, they can steal
20	customers away from you, so you can't profit. That may
21	become of importance in some of the discussion as to
22	what type of performance evidence one might use in
23	determining whether a firm has market power or not.
24	A price above the competitive level, the
25	"competitive level" was in italics, because people talk

12

1	about the standard monopoly raising prices, and if you
2	are not raising them above the competitive level,
3	usually people don't care.
4	Then a "sustained period of time" is in the
5	definition because you may be able to opportunistically
6	raise prices for a little bit, but again, entry or
7	something might undermine the ability to do that.
8	Now, in some of the legal cases, you see
9	reference to the ability to exclude competition, and I
10	will suggest that is something worth consideration,
11	because in some contexts -- and there were FTC hearings
12	many years ago about standard-setting organizations
13	where there might be a collection of, let's say, 10 or
14	more people making a particular product, and there might
15	be enough competitors that they compete and charge a
16	competitive price because there's so many people
17	operating under that standard.
18	Well, somebody may develop a new technology that
19	would come in and completely take the market away from
20	the incumbent competitors with the older technology.
21	Acting jointly in that case, they might be able to block
22	entry by controlling the standard-setting organization.
23	Are they raising prices above the competitive level?
24	They're excluding an entrant, somebody that would
25	dynamically help the market with a new technology that

13

1	might have better performance characteristics and be
2	able to be sold at a lower price. What they get out of
3	it is where their profit is, is that they get to earn
4	the competitive rate of return rather than being maybe
5	in bankruptcy court.
6	So, while I gave you the standard definition,
7	there are other things and other contexts, as you can
8	see from the get-go, that you have to worry about in
9	deciding whether a firm or a group of firms have market
10	power. And today, largely we'll focus on dominant
11	firms, but there are contexts where a group of firms
12	acting together might have trouble. And if a dominant
13	firm has control over a patent that's a blocking patent
14	that blocks a new technology, he might have an interest
15	in blocking the new technology just like the group of
16	firms that ran the standard-setting organization has an
17	incentive to block technology. So, that's one thing.
18	The other thing that I wanted to highlight at
19	the beginning is, some people talk about market power;
20	some people talk about monopoly power. Often,
21	economists mean the same thing, but in some contexts,
22	people have defined them differently. Greg Werden is
23	sitting there, and he's drawn a distinction in one of
24	his articles and alludes to other people that
25	distinguish market power and monopoly power perhaps in

14

1	terms of the time period over which people have the
2	ability to raise prices and the like.
3	There are articles out there that talk about
4	antitrust monopoly power, again, trying to make a
5	distinction. And there, often the thought is if you
6	have a differentiated product and thus have a
7	downward-sloping demand curve for your product, you
8	might have some degree of ability to raise prices above
9	costs and you might in that sense have market power, but
10	you might not have a substantial ability to do it.
11	Because there are a lot of products out there that are
12	roughly close substitutes, not exactly the same thing,
13	and you might in that context have some market power but
14	not antitrust monopoly power or antitrust market power,
15	because you don't really have substantial ability to
16	earn substantial profits and the like. So, some people
17	try to distinguish that downward-sloping demand curve
18	idea by talking about antitrust monopoly power.
19	I think with that background, we're talking
20	about antitrust market power. Something that's somewhat
21	significant. And then different panelists may have
22	different degrees of market power in mind when deciding
23	how you go about measuring whether it is significant
24	enough market power. So, with that sort of definition
25	of market power, the next slide was going to lay out

15

1	sort of the touchstones in a typical market power proof
2	that you sort of run through, and the first thing people
3	often define is product market definition. Then they go
4	to defining geographic market definition.
5	Once you have a relevant product/geographic
6	market combination, often it is standard to look at
7	market concentration in a monopoly case, and once you
8	clear that hurdle and see that maybe it is substantially
9	concentrated or a firm has a dominant market share, a
10	high-level market share, you then start looking at
11	things like entry conditions, other structural
12	characteristics of the market. Maybe you look at in
13	some contexts, you know, the structure of the buyer-side
14	of the market, and if it is a collusion case type of
15	monopoly power issue, maybe you look at the
16	characteristics of the market that make it easier or
17	harder for firms to collude in that market.
18	Then finally, in a lot of the monopolization
19	cases, you see a consideration of market performance
20	evidence, and that's where you start having things like
21	profit rates of return, profit margins, looking at
22	prices over time or across geographic areas. You look
23	at output patterns and how they vary with prices. And
24	you look at new product introductions. You can either
25	look at them in terms of formal econometric analysis or

16

1	often you look at events -- market events that allow you
2	to sort of control for some things -- and look at how if
3	the events give you insights either directly into the
4	market power or at some of the related issues like
5	market definition.
6	Now, increasingly, because of the success of the
7	Merger Guidelines, you see references to the approach
8	used in the Merger Guidelines of developing a relevant
9	market in the context of monopolization cases, and there
10	were a couple slides that sort of just quoted the
11	Guidelines. I suspect with this audience, there is no
12	reason to go through it, but it is the hypothetical
13	monopolist test. Can the monopolist raise prices above
14	the -- in the Guidelines, they talk more or less about
15	the current level as opposed to the competitive level
16	and see if that's profitable.
17	Now, one thing that is worth pointing out,
18	especially in transferring that concept, is that in the
19	Guidelines themselves, Section 1.11 says that while you
20	might look at prevailing prices in the Guidelines, there
21	is a caveat that says if pre-merger circumstances are
22	strongly suggestive of coordinated interaction, in that
23	situation, the agency will use a price more reflective
24	of the competitive price. So, there is a caveat in
25	there where they don't always use prevailing prices.

17

1	One sort of footnote is that the original
2	guidelines were focused on coordinated effects, and then
3	they later on added more information about unilateral
4	effects. I think there's a little glitch here, because
5	I think the Merger Guidelines actually should make a
6	reference not only to coordinated interaction, but also
7	if the dominant firms raise prices above the competitive
8	level, then you might want to look at the competitive
9	price level.
10	Why might you want to do that? Well, that is
11	because you get a different elasticity and different
12	substitutes depending on at what price level you measure
13	the substitution. And this is where the lack of slides
14	really hurt us the most, because I put together an
15	illustrative example of a demand curve with a concrete
16	slope and all the rest, calculated the marginal revenue
17	curve from that, showed where the competitive price
18	would be, where basically price equals marginal cost,
19	then showed where the monopolist would operate, which is
20	at a higher price, and then estimated the elasticities
21	of a couple of the different points along the demand
22	curve.
23	What you see is that even though a demand curve
24	is a straight line and thus the slope is constant over
25	the whole curve, the elasticity changes. And at the

18

1	higher prices, the demand is more elastic. And, the
2	reason that that makes sense is that a monopolist is
3	going to keep raising its price, you know, and find a
4	price that is more profitable. And in the monopolistic
5	equilibrium, he has got a high enough price that demand
6	becomes elastic and a further price increase would lose
7	a lot of customers to other products. That is called
8	the Cellophane fallacy -- that sets up the Cellophane
9	fallacy, which is if you measure the elasticities at the
10	monopoly price, you are going to run into problems
11	because there are a lot of substitutes out there that
12	are not substitutes at the competitive price. You can
13	do all the econometrics you want and estimate the
14	elasticities, but if you do not know whether you were at
15	a competitive price or a monopoly price, that elasticity
16	estimate does not tell you anything when you are doing a
17	monopolization case particularly.
18	So, then you get into this tautological
19	situation. If you think about the paradigm of starting
20	with a monopoly case and saying, "Well, do I have a
21	monopoly here?" And you have to define the market, and
22	you have to define a monopoly price to define the
23	market, then why bother defining the market? So, you
24	have got a couple of issues here that suggest, what do
25	you do about it? And the rest of my deck talks about

19

1	the sorts of things that one might look at. But, the
2	basic thing that I wanted to suggest is -- while I think
3	there are great problems with a simplistic analysis of
4	the standard paradigm I outlined -- I think there are
5	elements of it that, if you can go through it all, can
6	help you in many circumstances unravel this thing and
7	cross-check your conclusion.
8	So, it is a way of organizing your story.
9	Making sure that you look at your story or your analysis
10	as consistent, and that it gives you insights into what
11	you might look at. And where it leads you, I think, is
12	looking more and more at some of the performance
13	evidence. But you have got to be careful in looking at
14	the performance evidence, because as economists have
15	shown, things like profits and accounting data are
16	tricky.
17	Having said that, I also think that how
18	difficult a problem it is varies a lot from market
19	circumstance to market circumstance. I think it is
20	probably trickiest when you are dealing with
21	consumer-differentiated products, like Cellophane
22	wrapping paper or something. It may be less of a
23	problem when you are dealing with an input into an
24	industrial process, where you can look at substitutes in
25	a more maybe engineering approach type of way.

20

1	My time is basically up, so to keep us on
2	schedule, I would recommend -- they are going to post
3	the slides later on, and they are written in a way that
4	they are readable -- so I suggest you look at the slides
5	for the rest of the story.
6	Thanks.
7	(Applause.)
8	DR. CARLTON: Thank you.
9	Our next speaker is Joe Simons.
10	MR. SIMONS: Thanks, and good morning, everyone.
11	I would like to start out by complimenting the
12	FTC and the Department of Justice in holding these
13	hearings and doing a terrific job. I am really quite
14	encouraged that something really valuable will come out
15	of this.
16	So, one of the first things that happened this
17	morning is the audience was instructed not to ask any
18	questions or make any comments. So, I thought, well,
19	gee, I was planning to hear you violate that restriction
20	right away, but maybe we'll try something a little bit
21	different.
22	Perhaps by a show of hands, who in here would
23	say that the 1982 Department of Justice Merger
24	Guidelines market definition paradigm was the most
25	significant development in market definition in the last

21

1	30 years?
2	So, we have got most of the panelists and maybe
3	half of the audience. That is pretty good for one
4	thing. I would have expected it might have been a
5	little bit higher.
6	But in any case, what that showing would
7	demonstrate is an enormous amount of success for that
8	effort, I think by any standard, and why is that the
9	case? Why were those guidelines on the market
10	definition paradigm so successful?
11	In my view, it is because those guidelines
12	reflected an understanding that the tools of antitrust
13	analysis should be designed for a specific purpose.
14	Previously, you had market definition which was pulled
15	out of the economics literature and it was not designed
16	to do an antitrust analysis. The merger guidelines
17	market definitions was done specifically for that
18	purpose.
19	The other thing that was really important is
20	that the agency, the DOJ in that case, was willing to be
21	out in front of the case law. I think there was a
22	pretty good argument that those guidelines, the market
23	definition therein, did not really reflect the case law.
24	So, I thought it would be useful to do a little case
25	study and talk about first principles and market

22

1	definition.
2	The Guidelines, the Merger Guidelines, were
3	built around the goals defined right in the Guidelines
4	of preventing mergers from creating or increasing market
5	power -- initially through coordinated interaction and
6	then later unilateral effects. And as I said, they
7	geared this market definition specifically to this
8	overall goal of the Merger Guidelines. So, it was
9	designed to identify that universe of firms that were
10	necessary to profitably engage in coordinated
11	interaction or in unilateral effects. Then for the
12	unilateral effects, arguably the analysis could collapse
13	the market definition into the competitive effects
14	analysis. The market definition in the Guidelines is
15	rigorous, it is logical, and it is transparent.
16	Now, sitting here today, 25 years later, and
17	seeing what a success this was, you might forget what it
18	was like when these things were first issued. There
19	were hoots and howls from all sectors of the Antitrust
20	Bar and the academic community. These guidelines were
21	ivory tower nonsense; they were completely hypothetical;
22	they were totally inoperable and just downright
23	impractical; a complete waste of time. These were
24	comments that people made very regularly, and some
25	people even said it was a conspiracy to do away with the

23

1	antitrust laws.
2	There was a little bit of a kernel of truth to
3	some of those complaints, not the conspiracy stuff, but
4	to the practicality of this test. There were a lot of
5	people who saw the initial attempts to implement this by
6	the agencies in the following way. One of the staff
7	lawyers would have a conversation with the customers and
8	say, "Gee, do you think that the sellers in this market
9	could profitably raise price 5 or 10 percent?" You are
10	shaking your head, but I heard people do that, Greg, and
11	the customer has no concept of what it takes for it to
12	be profitable. There is no context to the question.
13	So, there were reasonable criticisms.
14	But what happened is that because the algorithm
15	was rigorous and logical and transparent, it enabled the
16	development of applications basically, tools, to
17	implement this approach, econometric tools. Examples
18	are Baker and Bresnahan and Scheffman and Spiller, Greg
19	Werden as well, something near and dear to me, critical
20	loss. These things did not exist when those guidelines
21	were first issued, and that really was an important
22	lesson to learn, that if you have the right structure,
23	then you have created a platform on which you can build
24	something that really works.
25	So, what does this translate into in terms of

24

1	what we should do for section 2? Well, what are the
2	goals of section 2? What are we trying to accomplish?
3	Is there a consensus? You know, there has been a lot of
4	ink been spilled in relation to the Trinko case, for
5	example. There are differences already between the way
6	the DOJ and the FTC look at this. There is the profit
7	sacrifice test, the no economic sense test; there is the
8	disproportionate harm relative to efficiencies test.
9	So, where does that leave us for market
10	definition? Does that create a problem? Can we rely on
11	what is in the case law? Reasonable interchangeability,
12	what does that mean? How much interchangeability is
13	reasonable? It is basically relying on
14	cross-elasticities of demand. How high does the
15	cross-elasticity have to be? Is that even something you
16	can look at? Can we rely on the Merger Guidelines
17	market definition? Does the hypothetical monopolist
18	paradigm, as applied in the Merger Guidelines, really
19	work for section 2? And one of the issues in section 2
20	is, are we focused on the same phenomenon that we are
21	for section 7?
22	The Merger Guidelines, the Horizontal Merger
23	Guidelines, are basically focused on collusion, an
24	extreme form of which is unilateral behavior. So you
25	are talking about situations in which a group of firms

25

1	is trying to restrict their own output, whereas in
2	section 2, what you are dealing with is a situation in
3	which one firm, the large firm, the dominant firm, is
4	trying to restrict the output of somebody else in most
5	cases and maybe sometimes themselves as well. So, what
6	do we do with all of that?
7	One possible thing to do -- and I am just
8	throwing this out -- would be to come up with a set of
9	goals for section 2, what is the purpose, what are we
10	trying to do, and then work through various scenarios as
11	to what the market definition would be under each of
12	those. So, one potential scenario is, we are going to
13	say that the goal of section 2 is to prevent unilateral
14	conduct that is reasonably likely to significantly raise
15	price or reduce quality. Reasonably significantly, you
16	can come up with other adjectives, number one.
17	Number two, and you are going to focus on
18	conduct that either, A, has no efficiencies, B, has
19	disproportionately low efficiencies relative to their
20	exclusionary effect, or C, would make no economic sense
21	in the absence of exclusionary effect, and potentially
22	D, permits recoupment of the exclusionary conduct. So,
23	kind of a menu from which to choose.
24	Well, one could argue that the first condition,
25	that the unilateral conduct be such that it is

26

1	reasonably likely to significantly raise price and/or
2	reduce quality, may be a necessary condition. That
3	defines the universe in which something bad can happen.
4	If you do not have that condition, then you might be
5	able to say that nothing bad can really happen. So, you
6	can use market definition in that sense, to focus on
7	that aspect as a screen.
8	You then could ask, "Well, gee, would the market
9	definition need to change depending on your choice of 2A
10	through D?" And at least at a first cut, I would say
11	probably not, that these factors relate to what might be
12	considered defenses or separate prongs of the analysis.
13	They would not be necessary to worry about in the first
14	market power screen, where you use market power or
15	market definition as the screen.
16	All right, so what would be the relevant
17	context, then, for measuring profitability of a price
18	increase? Well, obviously the options are before,
19	during or after the execution of the alleged conduct.
20	Well, we are concerned with the price going up as a
21	result of this conduct, so it seems to me you want to
22	focus on whether there might be a significant price
23	increase, whether a significant price increase might be
24	profitable during or after this alleged conduct.
25	Then similarly, if the conduct is already in

27

1	place, so you cannot observe it over time, then the
2	question might be the reverse, which is, absent this
3	conduct, would the price be lower, right?
4	You see, I think there is the same problem here
5	that you have -- not really a problem, but an issue in
6	the Merger Guidelines -- where for the most part, you
7	are measuring the profitability of a price increase
8	going forward. You are not looking at the current
9	level. You are really looking at a change in the
10	current level that is brought about by the conduct that
11	you are worried about. So, in the merger case, it is
12	the merger; in this case, it would be the alleged
13	exclusionary conduct.
14	You know, one of the things that is near and
15	dear to me, critical loss, might be a tool to help in
16	this analysis, and it would not be exclusive by any
17	means. Just like in the Merger Guidelines you can use
18	critical loss, you can use all kinds of other estimation
19	techniques, and they are not exclusive.
20	So, one way to think about this would be that
21	the burden would be on the plaintiff to show the likely
22	extent to which the alleged conduct restrains
23	third-party producers; in other words, whatever the
24	conduct is, exclusive dealing, refusal to deal,
25	whatever, what is the likely impact on third-party

28

1	producers? How much restraint does this have on their
2	ability to supply the market?
3	Then the plaintiff would have to show that it
4	would be profitable for the monopolist to raise price
5	significantly -- whatever the number is, 5, 10 percent,
6	whatever -- as a result of that exclusionary conduct.
7	You could calculate a critical loss for the monopolist
8	that would be based on margins, and you could estimate
9	whether a 10 percent price increase after or during the
10	alleged conduct would leave sufficient residual supply
11	such that a monopolist would lose in excess of the
12	critical loss. So, that would get you the market
13	definition part of this. Then what do you do?
14	One strategy would be to not even bother with
15	shares, because you have basically concluded that the
16	single firm was able to engage in this alleged conduct
17	and get the price up, and in terms of that, one could
18	say, "Well, that's what we needed to know," and we will
19	now we go through the rest of the analysis and determine
20	what are the efficiencies, and maybe you want to talk
21	about recoupment as well. So, one could reasonably say,
22	"Well, we don't really need a market share threshold."
23	Other people could say, "Well, gee, it is in the case
24	law. We want to try to make it consistent. It is
25	really important. So, we need a market share

29

1	threshold." How would that work?
2	Well, one way to think about it in the context
3	that I have just outlined would be you could say, "Well,
4	the firms in the market would be obviously the alleged
5	predator, and then potentially also other firms that
6	have also benefitted from a price increase as a result
7	of this exclusionary conduct," and you might base their
8	share calculations on their sales of that product for
9	which the price increase was experienced.
10	But then you ask the question, "Well, why have a
11	share requirement? What does that do for you?" You
12	might say, "Well, it gives us some comfort because
13	predatory conduct is only likely to occur where the
14	shares are high." Well, there is an issue about that,
15	because some exclusionary conduct is really cheap, and
16	some exclusionary conduct is really expensive. So, if
17	you are going to engage in really expensive exclusionary
18	conduct, yes, then you probably want to have a big
19	share, because you need to recover that expense that you
20	laid out to execute the exclusionary conduct, but if you
21	are executing really cheap exclusion involving, a
22	Hatch-Waxman type of scenario or something like that,
23	which costs virtually nothing, well, then, what does the
24	market share do for you? So, that is unclear.
25	I have got about 30 seconds left, and I just

30

1	wanted to sum up by saying I think there are some really
2	important lessons to be learned from the Horizontal
3	Merger Guidelines market definition, and I am hopeful
4	that what will come out of this is we will get a bunch
5	of smart people in a room, maybe Greg and some of his
6	colleagues from the Antitrust Division and the FTC will
7	sit in a room, take all of this together, and come out
8	with an algorithm that is of similar significance to
9	what they did with the Merger Guidelines -- use the
10	first principles integrated approach, not worry about
11	the fact that what they might come out with is a
12	theoretic framework, theoretic algorithm that is not
13	immediately implementable, and then not be afraid to
14	consider a market definition guideline that deviates
15	from traditional case law, because what happened with
16	the Merger Guidelines is people originally said, "Oh,
17	this is nothing like the original case law," and now we
18	have been able to bring the two together, and the courts
19	have seemed to have adopted what is in the Guidelines.
20	Thanks very much.
21	(Applause.)
22	DR. CARLTON: Thank you, Joe.
23	Our next speaker is Larry White, who has arrived
24	in time. You have already been introduced, Larry.
25	DR. WHITE: Well, thank you.

31

1	DR. CARLTON: And the ground rules are we are
2	running a little late, so if you could keep to 15
3	minutes.
4	DR. WHITE: Right.
5	DR. CARLTON: Does the computer work?
6	UNIDENTIFIED SPEAKER: Yes.
7	DR. CARLTON: And you are the first person who
8	has the use of the computer.
9	DR. WHITE: All right, great. Well, thank you.
10	I am very pleased to be here this morning, and sorry for
11	the delay of my arrival. I flew down from New York this
12	morning, and every once in a while you get hit with a --
13	I do not know whether it is the right-hand tale or
14	left-hand tale on variance, but we were an hour late
15	taking off. So, here I am. I am very pleased to be
16	here.
17	I think this is a terrifically important issue,
18	and it is an issue where unfortunately too many mistakes
19	have been made, too many mistakes continue to be made,
20	and I want to walk you through what I consider to be
21	some important issues. I have got a few call it partial
22	answers. I do not have the complete answer. At the
23	end, I am going to be echoing Joe Simons' call. We need
24	a new paradigm; a paradigm is missing.
25	So, like any good business school professor, I

32

1	am going to tell you what I am going to say, and then I
2	am going to say it, and then I am going to tell you what
3	I said. I will frame the issue, I will remind you what
4	the standard monopoly model looks like, I will remind
5	you what the implications of that model are, I will
6	point out the loose language that has been used by
7	people who do know better or who ought to know better,
8	and I'll tell you about the danger of that loose
9	language. That will bring me to the Cellophane fallacy.
10	Everybody is going to talk -- you cannot not talk about
11	the Cellophane fallacy when we're addressing this topic,
12	remind you of an ongoing dilemma, put out some partial
13	suggestions, and wrap it up.
14	What's the issue? I am not going to get into
15	this market power versus monopoly power. The way I was
16	taught, it is all the same thing, and the exercise of
17	this thing, call it monopoly power or market power, is
18	the seller can sell at prices above marginal cost and
19	earn rents, and I should have added for a sustained
20	period of time, but I will go ahead with my story. That
21	is the picture that we carry around in our head of what
22	monopoly power, market power, is about, the sustained
23	charging of a price above marginal cost, maintaining --
24	I am going to use that word over and over again --
25	maintaining a price substantially above marginal cost.

33

1	All right, now, what also gets talked about,
2	especially in an antitrust context, is actions --
3	exclusionary, predatory actions -- that can create or
4	enhance market power. So, somebody who did not have it,
5	can create it. Somebody who has it through an
6	exclusionary or predatory action can enhance it, make
7	the demand curve yet less elastic or inelastic and earn
8	even higher rents.
9	If the seller is engaging in this kind of
10	activity, whether he is exercising the market power or
11	enhancing, a likely precondition is that the seller has
12	a large share of its market. So, that is not necessary.
13	You can come up with examples where if the overall
14	supply is limited, where other suppliers cannot expand
15	their output very much, where demand is quite inelastic,
16	even somebody with a relatively small share of a
17	commodity market by his unilateral actions can affect
18	the price, but more generally, a large share of
19	something called a market is going to be necessary. But
20	that then raises this threshold or safe harbor issue,
21	what is the market, and there is no standard paradigm
22	for that determination.
23	So, this is the picture we carry around in our
24	head, and the implications of that picture, the
25	monopolist maintains its price at a level above the

34

1	competitive price. He would not want to raise his price
2	any further unless demand changed or costs changed. He
3	is already where he wants to be. In trying to raise his
4	price, he would lose too many customers to sellers of
5	something else, and, of course, if the market changes
6	from a competitive structure to a monopoly -- because of
7	cartelization, because of exclusion -- then the price
8	changes, then the price increases, the seller, newly
9	feeling this market power, raises the price from the
10	competitive to the noncompetitive monopoly level, but as
11	a characterization of what is going on when we take a
12	snapshot of the market, he is maintaining the price at a
13	level above the competitive level. That is clear in
14	this standard model.
15	About 40 years ago, George Stigler developed an
16	expanded version of this, the dominant firm and the
17	inverted price umbrella, where he described a firm that
18	was not strictly a monopolist, he faced a reactive
19	fringe of smaller firms that were limited in their
20	supply response, and he showed basically you get a
21	similar type of outcome. The dominant firm is able to
22	charge, maintain a price above competitive levels, but
23	he doesn't want to go any higher because -- and there,
24	in the Stigler model, it is implicit -- he would lose
25	too many sales to that competitive fringe.

35

1	Okay, why am I making such a big deal out of
2	this? Because there has been loose language out there,
3	first by my colleagues, all of whom do know better, and
4	they describe the phenomenon of monopoly power, market
5	power, in terms of the ability of the firm to raise
6	prices. In other words, I have put in italics over and
7	over again, this language of "raise prices," or in the
8	context of the Microsoft case, Fisher and Rubinfeld
9	making this claim that, "Gee, Microsoft could have
10	raised its price substantially and wouldn't have lost
11	customers," and you have got to scratch your head, how
12	come they didn't? Then Evans and Schmalensee on the
13	other side, again, talking the language of "raise."
14	Even earlier, as I walked in the door, I heard
15	Phil Nelson talking about the monopolist "raising" the
16	price. Maintaining is what we're talking about, but I
17	am sure I in my looser moments fall into this "raising."
18	It is an easy thing to do, but I am going to show you
19	the dangers of it in just a minute.
20	I'll go over to some noted legal cases and legal
21	opinions, and again, you have got the same -- oh, did
22	I -- no, I forgot to put the italics in there, but you
23	can see the word "raise" in each of those -- in each of
24	those quotations from those cases.
25	All right, what is the danger? The danger in

36

1	the "raise" terminology is that if we think market power
2	and monopoly power are the ability to raise the price,
3	then it is easy to then think, "Ah, well, the test of
4	whether somebody has market power or not is whether the
5	seller can raise prices above currently observed
6	levels." Remember, that is what Fisher and Rubinfeld
7	were talking about there.
8	Conversely, if the seller is constrained from
9	raising prices because of its fears of losing too many
10	customers, then does that imply that it does not have
11	market power? The trouble is, even in the standard
12	paradigm where the monopolist is maintaining a price
13	above competitive levels, it cannot profitably raise its
14	price because it would lose too many customers to
15	sellers of something else.
16	That, of course, then leads us to the Cellophane
17	fallacy, the U.S. v. Dupont case, where the issue was,
18	was the market a narrow market of cellophane, in which
19	case it is clear, Dupont had market power. There was
20	one other seller of cellophane, Sylvania. It was under
21	license from Dupont, and so, effectively, no question.
22	If the market was cellophane, Dupont had market power.
23	Or was it, as Dupont claimed, flexible wrapping
24	materials, in which case Dupont only had a 17.9 percent
25	share and didn't have market power?

37

1	The Supreme Court majority said it was
2	interchangeability that carried the day, that cellophane
3	was interchangeable with other materials mentioned --
4	there was wax paper and brown wrapping paper and
5	aluminum foil and glassine and lots of other things --
6	and the majority said, "Ah, look, it is interchangeable.
7	Dupont can't raise its price. So, it must be part of
8	that larger market."
9	The minority pointed out the fallacy of that
10	reasoning and also pointed out the comparison with
11	rayon, where Dupont also faced 15 to 18 other producers,
12	also had a market share that was below 20 percent, and
13	made much less profits. They also pointed out that
14	Dupont's price of cellophane did not move around when
15	those other flexible materials' prices changed.
16	So, we have this ongoing dilemma. Profit data
17	nowadays are relied on a whole lot less than was the
18	case back in the fifties when Stocking and Mueller were
19	writing, when the Supreme Court minority relied on those
20	profit data. The Horizontal Merger Guidelines cannot be
21	used, because they are a forward look, as you have heard
22	already, they are a forward-looking test.
23	The one exception, which Greg Werden has pointed
24	out, is that if we are talking about a practice that is
25	not yet in place, say an exclusive dealing plan that is

38

1	going to be put in place. A plaintiff comes in, asks
2	for an injunction. We are talking about something where
3	it is a prospective practice. Then the prospective,
4	forward-looking paradigm of the Merger Guidelines will
5	work. To the extent that that is what we are looking
6	at, fine, we have got an answer, but lots of instances
7	are not of that kind.
8	As Phil remarked earlier, elasticities do not
9	help us very much. You cannot tell the difference
10	between a true monopolist and just a different -- a
11	seller of a differentiated product, a Chamberlin/
12	Robinson monopolistic competitor.
13	Okay, what to do? Well, sometimes a complaint
14	will involve a prospective practice, and then we have
15	got the Merger Guidelines. Sometimes there will be
16	cross-sectional or time-series evidence involving prices
17	where we can tell that concentration matters, and when
18	concentration matters, you have got a market, and retail
19	services are an area where cross-sectional data may be
20	available.
21	I harken back now ten years to the Staples case,
22	where cross-section data showed that prices were
23	different, higher where only Staples or Office Depot was
24	present in the market, lower when both were there, yet
25	lower when they and a third office superstore were

39

1	there. That evidence carried the day, and I think
2	correctly, that there was a problem -- there would be a
3	problem if the two firms merged, and it told us office
4	superstores were a market.
5	Think of the American Airlines predatory
6	behavior case. Why do we think that city pairs are a
7	market, city pairs airline transportation? Because
8	there is lots of cross-sectional evidence that shows
9	that, controlling for other things, prices matter and
10	prices are related to concentration. Sometimes profit
11	data will be useful.
12	I mean, if you think the Microsoft case was a
13	good case, if you thought that Microsoft's behavior was
14	a problem, why did you think that? And I think at least
15	part of the story was those profits. They were so large
16	that even with all the problems that we know about
17	profits, they were telling us something. But what if
18	none of these possibilities are available?
19	Well, Phil Nelson and I a few years ago made a
20	proposal. It turns out similar language can be found in
21	a 20-year-old article by Tom Krattenmaker. Greg had a
22	version of this proposal in an article he wrote in 2000,
23	where basically it is asking in the presence of an
24	allegation of exclusion, what would have been the
25	consequences of the absence of exclusion? It requires a

40

1	two-step investigation.
2	First you have got to ask, in the absence of
3	exclusion, what would the plaintiff's sales have been?
4	And then you have got to ask, what would the price
5	consequences of those additional sales have been as
6	well?
7	Now, as was indicated earlier, this would focus
8	directly on effect, and it implicitly delineates a
9	market, but if you think about what the unilateral
10	effects analysis under the Horizontal Merger Guidelines
11	does, it is basically doing the same thing. It is
12	looking for an effect, and then, if somebody goes ahead
13	and then tries to delineate a market, that is sort of
14	redundant. You have already found the effect.
15	Implicitly, you have said there must be a market there,
16	and that is basically what the Nelson and White proposal
17	does as well.
18	But I think the best approach would be let's try
19	to develop -- you know, I have thought hard about it.
20	The best I could come up with was this joint proposal
21	with Phil. It may not be good enough. Can the world
22	come up -- can the Division, can the FTC, can a bunch of
23	smart people out there -- come up with a paradigm that
24	will have the power and eventual universality of the
25	Horizontal Merger Guidelines?

41

1	I urge you, remember what the world looked like
2	before 1982. Remember what 1981, 1980 and 1979 looked
3	like. We did not have a paradigm. We had
4	Elzinga-Hogarty. We had Ira Horowitz's suggestion.
5	There were other ideas out there. George Hay was going
6	around talking about how the Division defined markets,
7	and he would say, "Well, we would look for whether there
8	was a specialized trade journal that the sellers in a
9	marketplace all submitted their data to." Those were
10	the kinds of indicia that people looked to. The Merger
11	Guidelines brushed all that stuff away, and we have now
12	got a powerful paradigm. I hope that some smart people
13	out there somewhere will be able to develop something
14	with similar power.
15	So, winding up, we have got an unsatisfactory
16	state for market definition. I would hope we are in
17	1981, and next year, somebody is going to come up with
18	something that will have the same kind of power as the
19	Horizontal Merger Guidelines. I have shown you some
20	partial remedies, but the best remedy would be a new
21	paradigm.
22	Thank you very much. I am very pleased to have
23	this opportunity today.
24	(Applause.)
25	DR. CARLTON: Okay, thank you, Larry.

42

1	Our next speaker is Andy Gavil.
2	DR. GAVIL: Good morning, everyone. Thank you
3	to the organizers for inviting me to join everyone
4	today. I am delighted to be here and agree with
5	everyone else that these are some very important --
6	indeed, fundamental -- issues to how we go about
7	analyzing antitrust cases, and in truth, they are not at
8	all unique to section 2. Questions of power and effects
9	really cut across all kinds of cases today. So,
10	resolving one area clearly is going to influence and
11	affect the others just as the Merger Guidelines has
12	affected many areas.
13	So, I start with my first slide in talking about
14	it is all about anticompetitive effects, and I think I
15	would add to that, and legal process. At the end of the
16	day -- that is a great phrase, "At the end of the
17	day" -- "At the end of the day, in the final
18	analysis" -- but at the end of the day, in the final
19	analysis, whatever we conclude as a matter of economics
20	is the right approach, we have to translate that into a
21	legal system of decision-making. It has to work in
22	courts. It has to work in a context where we have
23	burdens of pleading and burdens of production and
24	burdens of proof. It has to work in a context where we
25	have various methods for discovery of evidence, where we

43

1	have a role for expert witnesses, where we have judges
2	and juries, and if it cannot work in that context, then
3	perhaps there is a problem with what we have come up
4	with as a theoretical matter.
5	I forget who it was, I think it was Joe talking
6	earlier about how the Merger Guidelines were originally
7	received. Well, part of the problem in how they were
8	received is that they were received by a legal community
9	accustomed to looking at cases in one particular way.
10	They suggested that we needed to look at those cases in
11	a very different way, and it was very unclear in 1982
12	how you would translate, how you would take something
13	like SSNIP and what evidence would you need?
14	The lawyers that were asking the questions of,
15	what witness am I going to need to do this? What
16	evidence will I need from my client, from the other
17	parties? How will I assemble it? How will I present
18	it? There can be no doubt at all I think in anybody's
19	mind that the Merger Guidelines and subsequent
20	developments have been an economist's full employment
21	act, and certainly that has been evidenced in the
22	antitrust area. It is hard to imagine today proving any
23	kind of case, plaintiff or defense, without the role of
24	economists, and that is a result of the writing into our
25	substantive standards various economic ideas.

44

1	So, as I go through these slides, I want you to
2	sort of keep that in mind. The focus I have tried to
3	bring to my comments today is, how do we make it work in
4	this legal system? Well, common issues in antitrust are
5	effects, and we have certain ways that we go about
6	establishing them. We have irrebuttable presumptions --
7	that is what the per se rule is all about -- and we have
8	rebuttable presumptions; whether we are using direct
9	evidence or circumstantial evidence -- and that is going
10	to be an important issue that I am going to look at
11	today -- we have different ways that we go about trying
12	to establish effects.
13	Direct evidence, defined here, is the actual
14	exercise of market power. It may come out in
15	performance evidence. It may come out in before and
16	after studies of price. It is reflected to some degree
17	in our use of "quick look." The "inherently suspect"
18	formulation is also a way of looking at things that are
19	obvious, and a question I will be asking today is, do we
20	have equivalents for section 2 and would it make sense
21	to use them in section 2?
22	On the circumstantial evidence side, we have
23	something that I have called a "double inference." We
24	define a market, we calculate market shares from a
25	certain level of market share, we infer market power,

45

1	and in truth, from that, we then infer the capacity for
2	anticompetitive effect. In litigating terms, we are
3	dealing with two very standard paradigms of how to go
4	about proving something.
5	Well, power, of course, is a condition precedent
6	of effects, but if you look in the cases, there is a lot
7	of confusion -- again, loose language -- about how it is
8	used. Some cases say, "Well, what we need is market
9	power," and even in cases like NCAA and Indiana
10	Federation of Dentists that really were out in the
11	forefront in this quick look idea and the use of direct
12	evidence of actual effects, there is confusing language
13	about what "market power" means.
14	Well, power is the condition precedent of
15	effects. If you have the effects, the power is there.
16	So, part of the point of Indiana Federation, talking
17	about market definition and market power as surrogates,
18	was to make the point that when you have the actual
19	effects evidence, going sort of back around the
20	circumstantial evidence route, trying to define a market
21	and determine whether there are large market shares, may
22	be beside the point. Those things are surrogates for
23	direct evidence.
24	Well, as in many areas of antitrust, that leads
25	us to a point where we can identify easy cases and hard

46

1	cases. A good example I think of the easy cases, when
2	the direct and circumstantial evidence are aligned, when
3	they are pointing in the same direction, when you have
4	evidence of actual effects and you have high market
5	shares, those are easy cases. We do not argue about
6	those very much. The D.C. Circuit in Microsoft actually
7	structured its discussion of monopoly power that way,
8	looking at both direct evidence, circumstantial
9	evidence, they are both pointing in the same direction,
10	easy case.
11	On the other hand, for safe harbor ideas, if you
12	have de minimus evidence and no effects and you have low
13	market shares, again, pointing in the same direction,
14	and I would make this point -- I'll raise it a little
15	bit later -- in terms of safe harbors, I do not think
16	you can rely just on market shares alone. It has to be
17	market shares plus certain other factors, and I will
18	also suggest that if we are going to have safe harbors,
19	we need some danger zones, and again, it might be market
20	share plus some other characteristics.
21	But evidence and power effects are interrelated,
22	and I think this is what makes part of our current
23	framework very difficult to think about. Courts do
24	think, because of years and years of case law, first
25	monopoly power, then willful acquisition or maintenance,

47

1	when in truth, the evidence of conduct and effects in
2	the evidence of power is going to be very interrelated.
3	Well, again, thinking about direct and
4	circumstantial evidence, the benchmark for
5	circumstantial evidence is clearly the Horizontal Merger
6	Guidelines. They really did advance the science of
7	thinking in terms of circumstantial evidence. Recall,
8	though, that Cellophane was a section 2 case, and maybe
9	there are some different problems that come up when we
10	are doing prospective predictions about likely market
11	power versus retrospective methods when we have, you
12	know, the before and after ability to actually look at
13	the effect of conducts, but the Merger Guidelines in any
14	paradigm we come up with are probably going to have some
15	continuing significance. They have been cited by courts
16	outside of section 7. They are cited in section 1 cases
17	and section 2 cases. Basic ideas and concepts are
18	clearly interrelated.
19	So, my suggestion at this stage of our
20	development is we need something of a similar to the
21	Merger Guidelines to refine "actual exercise" standards
22	and to harmonize those standards across different
23	offenses. A critical question, I think, is how much and
24	what kinds of effects evidence should be sufficient to
25	shift a burden? And here I remind, again, that outside

48

1	the area of exercising prosecutorial discretion, outside
2	the walls of the agencies when they are deciding whether
3	to bring a case, if the decision to bring a case is made
4	and the economists agree, the next question the lawyers
5	are going to have is, "Well, how do we meet our burden
6	of production? What evidence are we going to assemble?
7	What is going to make us win this case?"
8	I think when you are thinking about direct
9	effects evidence, and market share as well, a critical
10	question in section 2 is, what does it take to shift a
11	burden? Frequently what you see defendants arguing in
12	cases is the burden didn't shift, the burden didn't
13	shift, the burden didn't shift. Well, what does that
14	mean?
15	It means that there is no requirement on the
16	part of the defendants to actually justify their
17	conduct. If they claim there are efficiencies, where is
18	the evidence of efficiencies? That does not happen
19	until the burden shift takes place. That is a critical
20	stage. It is a critical stage that has to be focused
21	on, and I have given some examples here of various cases
22	that raise some of those questions.
23	I think we are also feeling the weight of the
24	Alcoa paradigm. In looking back at Alcoa and the cases
25	that preceded it, all Judge Hand did was he surveyed the

49

1	previous cases and looked at winners and losers to come
2	up with his three famous sort of -- you know, 33, not
3	enough; 66, maybe; over 90, definitely. Well, where did
4	he get that from?
5	If you look at the prior Supreme Court cases,
6	you will see that there were cases falling into each of
7	those categories. He sort of synthesized them and came
8	up with this benchmark. I think an important question
9	for us is, are we ready to move beyond the total
10	reliance on market shares, which sends us off in this
11	direction of conflicting evidence, plaintiffs and
12	defendants having experts -- the market is big, the
13	market is small -- and is that really where we want to
14	be? What can the role of direct evidence be?
15	The Re/Max case was an example of a court
16	relying on direct evidence, actual price effects
17	evidence in a section 2 case. The 7th Circuit in
18	Republic Tobacco rejected such an approach, said that
19	Indiana Federation did not apply and NCAA did not apply
20	to a vertical case. Is Staples -- and the unilateral
21	effects that people have alluded to already -- is it
22	related? I think it is. It is a way of trying to more
23	directly gauge. We have talked about the monopoly
24	versus market power being kind of a silly distinction.
25	So, I will move on.

50

1	I think there is an important role here for
2	decision theory, which obviously has begun to influence
3	our thinking. The emphasis tends to be on fear of error
4	costs, and often that motivates calls for more and
5	better evidence. We need more before that burden
6	shifts. One point I would like to walk away with today
7	is urging that we also consider the second half of
8	decision theory, which is process and information costs.
9	Is more evidence really always better?
10	In that regard, I sort of suggest -- and it is
11	not really new, there is a lot of general literature out
12	there on the economics of evidence. Richard Posner has
13	a long article on an economic analysis of evidence, and
14	I put forward the question, "When does the marginal
15	value of additional evidence in terms of economic
16	certainty (minimizing error costs) outweigh the costs of
17	obtaining and processing that evidence, taking into
18	account whether it is reasonably accessible to the party
19	bearing the risk of non-persuasion?" What I tried to do
20	in that question is integrate some economic ideas and
21	some legal process ideas from both the rules of
22	procedure and the rules of evidence. It is always easy
23	to demand more. It is always easy to pursue some kind
24	of level of absolute certainty and minimal error costs.
25	The question is, as a legal standard, when we take that

51

1	into court, is that really going to strike the right
2	balance for us in resolving cases?
3	Antitrust is not always rocket science, and I
4	think we need to get over the idea that it always is.
5	Yes, we need safe harbors to guard against false
6	positives. I think we also should be emphasizing
7	equally defining danger zones where we might be running
8	into false negatives.
9	Is monopoly power all that puzzling? I would
10	point out to everyone that neither 3M nor U.S. Tobacco,
11	in two U.S. Courts of Appeals monopolization cases, even
12	contested that they had monopoly power. In the
13	Microsoft case, they contested it, but rather
14	unpersuasively, and every agency and every court to look
15	at it has concluded that yes, indeed, Microsoft had
16	monopoly power.
17	We could go on with a couple other examples,
18	American Airlines, Dentsply. Were these really such
19	difficult cases? If they were not, then why were they
20	so difficult? Why would parties not even litigate the
21	point about their power? There must be, there must be
22	cases where -- again, market share plus -- where there
23	must be additional factors, information on entry
24	barriers. Entry barriers will always, for example, be
25	important.

52

1	Finally on this slide, sliding scales, not all
2	burden shifts are created equally. You see this in
3	cases like Baker Hughes and Heinz in the merger area,
4	the realization that sometimes when a burden shifts, it
5	really shifts, and the presumption is very strong, and
6	other times, it kind of is just enough to shift. Well,
7	in responding to those sorts of cases, we might want to
8	respond in different ways by considering what is
9	required to shift and what is required to shift back a
10	burden in different ways.
11	On legal standards and decision-making, I think
12	that the balancing of effects idea is a straw man. We
13	could cite, as Larry White did, we could put up lots of
14	slides with courts saying, "Anticompetitive effects; the
15	burden shifts. Efficiencies; then we balance one
16	against the other." We do not really do that. I have
17	looked; you can all look. If you can find me a Section
18	1 litigated case in which the case was actually decided
19	on balancing effects versus efficiency effects, consumer
20	surplus diminution versus increased producer surplus,
21	find me such a case. I would like to see it. It is not
22	what we do.
23	What we do is weigh evidence. What juries do is
24	they compare the evidence of anticompetitive effects
25	with the evidence of efficiencies, and they make a

53

1	decision about where the weight of the evidence is.
2	That has to do with credibility; it has to do with
3	persuasiveness. It does not have to do with $10 of
4	anticompetitive effect and $11 of efficiency.
5	Finally, a word about caricatures and corrosion
6	of the rule of law. The level of discourse and the
7	level of criticism of antitrust, as we all know, has
8	continued for quite some time. It has continued despite
9	the fact that in the last 40 years, we have seen some
10	pretty major corrections to antitrust.
11	I say caricature -- and this is not my
12	caricature -- but this is what you see in a lot of the
13	criticisms of antitrust, and I think it is a caricature
14	that ignores this last period of adjustment over the
15	last 30 years. Incompetence -- judges, just
16	incompetent. They can do habeus corpus, they can do
17	environmental, they can do securities law, but antitrust
18	is rocket science, keep them away.
19	The same thing with juries. They just do not
20	know the difference between somebody who is full of it
21	and somebody who really knows what they are doing. They
22	cannot tell the difference between economists in this
23	case and, of course, neither can they decide any other
24	possible case.
25	And, of course, enforcers. I have the asterisk

54

1	there just to remind me to say that. Typically it is
2	enforcers themselves who make this argument, God, we are
3	so stupid. You shouldn't really trust us to make any
4	decisions, and although we may -- and it gets very
5	personal -- we may be able to make the decision, but
6	other enforcers are really stupid, especially those guys
7	at the offices of the states.
8	Who are the untrustworthy self-interesteds, the
9	self-interesteds who are untrustworthy? Rivals, oh,
10	they are always full of it. They are always complaining
11	about more competition. Dealers, yeah, what's that
12	freedom of dealer stuff? You know, manufacturers,
13	consumers, aligned; dealers, out in left field. And
14	plaintiffs pretty much all are full of it, especially
15	class action reps.
16	Ah, but who can we trust? Dominant firms.
17	Dominant firms articulating efficiencies. Fear of error
18	cost? That's truth. We need to put a lot of weight in
19	that. We need to be concerned about it. Other
20	defendants, generally yeah, and especially efficiencies.
21	Two problems I have with this sort of
22	caricaturing of antitrust. One is, I don't think it is
23	true. I would like to see the list of false positives
24	in the last 25 years. We have been moving towards
25	reduced error costs, and here I think it would be

55

1	helpful to have the economists really define what they
2	mean as "false positive." It is not a case on which
3	reasonable people can differ. It is a case that sort
4	of -- again, borrowing from procedure -- no reasonable
5	party could have come out that way. To me, that would
6	be a false positive or a false negative. It is not a
7	case that we simply disagree about.
8	LePage's has, you know, been frequently used as
9	sort of this example of a false positive. Be reminded
10	that 3M did not contest its market power, and if it did
11	offer any evidence of efficiencies, nobody who looked at
12	it found it very convincing. Did the Court of Appeals
13	give us a useful standard for bundled pricing? No, but
14	neither has anybody else yet. So, to call that a false
15	positive and say, "This is an example of how we're going
16	to inhibit all kinds of other cases," I am not sure that
17	that is justified.
18	The final point and I will sit down. As I said
19	at the start, Larry said at the end, you say what you
20	said at the beginning. At the end of the day, these
21	cases have to go to court sometimes, and this kind of
22	rhetoric of criticism ultimately is corrosive of the
23	rule of law. I think it is heard in curious ways
24	outside the United States. These criticisms really go
25	to the heart of whether we are willing, at the end of

56

1	the day, to rely on courts to make decisions.
2	We have numerous procedural devices, summary
3	judgment, judgment as a matter of law, Daubert
4	standards, appeal rights. If after all of that has
5	occurred a plaintiff actually wins a case, which does
6	not happen very often, I think we ought to be a little
7	bit more cautious about tossing the rhetoric around
8	about the incompetence and the untrustworthy
9	self-interesteds, all right?
10	Thanks very much.
11	(Applause.)
12	DR. CARLTON: Thank you very much, Andy. I was
13	pleased to hear I am not as incompetent as once
14	enforcers were thought to be, and to prove that I am
15	still competent, we are going to have a break, and it
16	will be a 10-minute break, and we will reconvene
17	promptly so that we can try and stay roughly on
18	schedule. Thank you.
19	(A brief recess was taken.)
20	DR. CARLTON: Why don't we try and start. Our
21	next speaker is Rich Gilbert.
22	DR. GILBERT: I would like to thank the
23	organizers for the opportunity to be here. I was
24	invited to talk about technology markets, so if any ink
25	gets spilled on this issue as a result of my comments,

57

1	you can be sure it will be Independent Ink, though I
2	will not talk about the presumption of market power for
3	patents. I thought we resolved that issue in the IP
4	Guidelines, although it is not the case that the Supreme
5	Court immediately adopts everything that the agencies
6	come up with.
7	Now, when we talk about market definition, there
8	is a real sense in which we are talking about either
9	guide posts or lamp posts. Now, lamp posts, as you
10	know, shed light on a subject but do not necessarily
11	shed truth about the subject. A lamp post might
12	illuminate the ground, but that does not mean that the
13	dollar that we are looking for is around the lamp post,
14	even though if it were, perhaps we could see it.
15	Guide posts, on the other hand, serve to focus
16	the analysis. The guide posts lead the way. The way
17	may be very foggy and very complicated and very
18	difficult, but can be very useful.
19	Now, my take, sort of in the spirit of Andy's
20	comments, the courts and defendants like the market
21	definition exercise, even though it is often used much
22	more as a lamp post than a guide post. They like the
23	exercise because, of course, for a defendant, if you can
24	show the market is very broad, chances are there is no
25	antitrust case there. For a court, they are all very

58

1	busy. They have full dockets. If you can show the
2	market is very broad, they do not have to worry about
3	it.
4	Plaintiffs also seem to like market definition
5	or many of them like market definition, because if you
6	can prove that or demonstrate or make a convincing case
7	that the market is narrow, well, chances are then there
8	will be an issue, but as I think everybody on this panel
9	is implying, none of those conditions, whether it is
10	broad or narrow, presumptive of a case or not
11	presumptive of a case, none of them are really relevant
12	directly to the analysis. We would rather have market
13	definition serve as the guide post to lead the way to
14	the right analysis rather than defining whether there is
15	or is not a case.
16	Now, so, if we talk about markets for
17	technology -- first I should distinguish, I am going to
18	focus more on technology markets than on markets for
19	technology. Markets for technology can be analyzed
20	using conventional goods markets, often using
21	conventional goods markets, which are sufficient for
22	analysis in many high technology industries, whereas
23	technology markets are useful when what is at issue is a
24	right or rights to a technology that are licensed rather
25	than embodied in a patent. So, I am focusing more on

59

1	technology markets than markets for technology, although
2	maybe in discussion, we will get to that distinction,
3	whether there should be a distinction.
4	Technology markets are defined in the IP
5	Guidelines. Technology markets consist of the
6	intellectual property that is licensed that are close
7	substitutes. Of course, here now, as in all market
8	definition exercises, the issue is, what are the close
9	substitutes? And when you are talking about technology
10	markets, the close substitutes are not only other
11	intellectual property rights, but also goods and
12	services that may substitute for those intellectual
13	property rights.
14	It adds another layer of difficulty and
15	complexity to the analysis, because just like in
16	conventional -- other section 2 goods market definition,
17	exactly what to sweep into that analysis and how, it
18	depends upon the prices, prevailing prices, and whether
19	the conduct is prospective or retrospective, these are
20	all challenging issues, which I am not going to entirely
21	resolve.
22	Now, technology markets also are -- there is an
23	upstream analysis for inputs which I think raises some
24	interesting questions by itself. Technology markets
25	have been used I think with some success to analyze the

60

1	licensing of technology to manufacture float glass, for
2	blending clean gasoline in the UNOCAL case, the float
3	glass with the Pilkington case, for designing fast
4	computer memory chips, as in the DRAM cases, perform
5	laser eye surgery, or to incorporate genetically
6	modified traits into agricultural seeds. These are all
7	some examples, I think, of markets that have been
8	analyzed using basically an upstream analysis for
9	licensed inputs.
10	Now, on the geographic market side, this is an
11	area where using technology markets in some cases
12	simplifies things. It is fair, I believe, to presume in
13	many cases -- not all, of course -- the geographic scope
14	for technology markets is very wide, because it is not
15	very difficult for a potential licensee to negotiate
16	with even quite distant licensors unless there are legal
17	or regulatory or some other restrictions that prevent
18	the use of licensed technology in different locations,
19	as there was, for example, with the UNOCAL case, but in
20	these other cases, the enforcement agencies I think have
21	correctly concluded that the technology markets are very
22	broad, U.S.-wide and sometimes worldwide.
23	Now, technology fees, should these be indicators
24	of market power? Interesting question which has not
25	been quite directly addressed. Marginal cost of

61

1	licensing is typically very low. It suggests that there
2	is market power if we define market power as the ability
3	to sustain prices above marginal cost, then looking at
4	technology fees, gives you an immediate read on whether
5	or not there is market power, not necessarily monopoly
6	power, but, as economists have said, that is a difficult
7	line to draw between market power and monopoly power.
8	Now, again, the relevant question is the ability
9	to increase or maintain technology fees significantly
10	above marginal cost for an extended time. In this
11	dispute about market power versus monopoly power, I am
12	certainly in the camp that says that monopoly power is a
13	lot of market power and that there is no clear dividing
14	line between the two, and the question is, the relevant
15	question is, is there conduct that leads to either
16	increasing or maintaining technology fees significantly
17	above marginal cost for an extended period of time and
18	whether it is prospective or retrospective?
19	If it is prospective, perhaps the ability is to
20	increase technology fees. If it is retrospective, then
21	the question is more has conduct contributed to the
22	ability to maintain technology fees significantly above
23	marginal cost?
24	This is now more in the spirit of what Larry
25	White was saying in his approach to section 2 market

62

1	definition. Also for technology fees, a related and
2	relevant question in a section 2 context is whether
3	competition, whether injecting more competition, would
4	result in a significantly lower technology fee if the
5	competition were not excluded.
6	I also agree that this opens up a lot of
7	interesting and unresolved issues, as in how much
8	competition should be enough to consider? What should
9	the price effect of that competition be in order to
10	define a relevant market? Is an elasticity of demand
11	faced at the existing prices for the fees and other
12	goods and services? Is an elasticity of demand minus
13	two, is that low enough, small enough in magnitude, or
14	does it have to be minus 1.1 or 1.5 or is minus 3
15	enough?
16	These are very important and serious questions
17	that need to be addressed if we are going to do this
18	kind of hypothetical decrease in price through a
19	hypothetical increase in output as a way to identify a
20	relevant market.
21	So, the focus on that additional competition and
22	whether it lowers the fee I do believe can get around
23	the Cellophane fallacy, and I think another important
24	aspect of that approach is that it focuses the analysis,
25	the definition of the market, on the analysis of the

63

1	competitive effects of the conduct. So, I think
2	sometimes it is a criticism of the hypothetical decrease
3	in price approach that it is too related to the conduct
4	that is being alleged as anticompetitive.
5	I turn it around and say that no, I think it is
6	an advantage of this approach, because it connects the
7	conduct at issue to the analysis of the impacts at
8	issue. Too often, I think many of us would agree that
9	the market definition exercise puts the cart in front of
10	the horse. We should be thinking about where are the
11	competitive effects, how significant can the competitive
12	effects be, and then let the market definition respond
13	to that rather than defining where the competitive
14	effects are. Again, this stems from the problem of the
15	Cellophane fallacy that a profit-maximizing firm has no
16	incentive to raise or lower its technology fee.
17	Another question about analysis of inputs, in
18	principle, the antitrust analysis for a technology input
19	is not qualitatively different from the analysis of any
20	other upstream good or service. The demand for the
21	input is derived from the demand for the final good or
22	service, and one thing to point out is the
23	Hicks-Marshall Law of derived demand, which says that
24	the elasticity of Derived Demand is proportional to the
25	cost share of the input. It is roughly the cost share