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
5
1 |
P R O C E E D I N G S
|
2 |
* * * * *
|
3 |
MR. COHEN: Good morning. I'm Joe Cohen, Deputy
|
4 |
General Counsel for Policy Studies at the Federal Trade
|
5 |
Commission and I'm going to be one of the moderators at
|
6 |
this morning's session. My co-moderator is Joe Matelis,
|
7 |
an attorney in the Antitrust Division at the U.S.
|
8 |
Department of Justice.
|
9 |
Before I start I'd like to cover a couple of
|
10 |
housekeeping rules. First, as a courtesy to our speakers,
|
11 |
please turn off your cell phones, Blackberries, anything
|
12 |
that might ring or clang or make noise.
|
13 |
Second, because these are set up as in a hearing
|
14 |
structure, we request that the audience not make any
|
15 |
comments or ask any questions during the session. We have
|
16 |
to limit it to the moderators and the panelists.
|
17 |
Before introducing our speakers and starting our
|
18 |
panel discussion, I would again like to thank the
|
19 |
University of California at Berkeley for hosting the
|
20 |
FTC/DOJ Section 2 hearing sessions yesterday and today.
|
21 |
In particular I'd like to thank Howard Shelanski, once
|
22 |
again, Richard Gilbert and Carl Shapiro for offering us
|
23 |
the facilities and making the necessary arrangements.
|
24 |
I'd also like to thank the Berkeley Center for
|
25 |
Law & Technology and the Haas Business School for |
6
1 |
providing the facilities, videotaping, web casting, etc.
|
2 |
And those who have provided us with logistical support,
|
3 |
Bob Pardue and others, I thanked you all once already, but
|
4 |
thank you again.
|
5 |
We're honored to have assembled this morning a
|
6 |
distinguished group of the finest lawyers from the
|
7 |
University of California Berkeley to offer their testimony
|
8 |
in connection with these hearings. They will provide
|
9 |
their perspectives on various themes and issues related to
|
10 |
the complex area of Section 2 jurisprudence, including
|
11 |
some research and economic analysis.
|
12 |
We've gathered seven panelists for today's
|
13 |
sessions. Four will talk this afternoon and three will be
|
14 |
our morning panelists.
|
15 |
This morning's panelists are Aaron Edlin, the
|
16 |
Richard Jennings Professor of Law, University of
|
17 |
California Berkeley; Joseph Farrell, Professor of
|
18 |
Economics at -- right here at the University of Berkeley,
|
19 |
and Howard Shelanski, here, Associate Dean and Professor
|
20 |
of Law and Director of the Berkeley Center for Law and
|
21 |
Technology.
|
22 |
Our format this morning will be pretty simple.
|
23 |
Each speaker will make an opening presentation from twenty
|
24 |
to thirty minutes. After the presentations are finished,
|
25 |
we're going to take a break, probably for about fifteen |
7
1 |
minutes, and then we'll come back, reconvene, and have a
|
2 |
moderated discussion with our panelists.
|
3 |
We're scheduled to conclude this morning's
|
4 |
session at approximately noon. So, we look forward to
|
5 |
hearing from our panelists.
|
6 |
And before we begin, the last group that I want
|
7 |
to thank are the panelists themselves. We appreciate the
|
8 |
time and effort and your willingness to share your
|
9 |
insights with us to make this a successful hearing.
|
10 |
I'd now like to turn to my DOJ colleague, Joe
|
11 |
Matelis, our co-moderator, for any remarks he'd like to
|
12 |
add.
|
13 |
MR. MATELIS: Thank you, Bill.
|
14 |
The Department of Justice's Antitrust Division
|
15 |
is very pleased to participate in today's single-firm
|
16 |
conduct hearings. We are delighted that such esteemed
|
17 |
panelists have agreed to share their views with us today.
|
18 |
And the Antitrust Division takes particular
|
19 |
pride in noting that five of today's panelists have served
|
20 |
in the Antitrust Division as Deputy Assistant Attorneys
|
21 |
General for Economics.
|
22 |
We expect that today's panelists will discuss a
|
23 |
wide range of topics that arise in evaluating single-firm
|
24 |
conduct and antitrust laws and we look forward to the
|
25 |
presentations and the panel discussions that follow. |
8
1 |
On behalf of the Antitrust Division, I would
|
2 |
like to take this opportunity to thank the Berkeley Center
|
3 |
for Law and Technology and the Competition Policy Center
|
4 |
at the University of California Berkeley for hosting us
|
5 |
today.
|
6 |
Also on behalf of the Antitrust Division, I'd
|
7 |
like to thank Joe, Aaron and Howard for agreeing to
|
8 |
volunteer your time and share your insights with us. It's
|
9 |
a great public service that you're doing and we're very
|
10 |
appreciative.
|
11 |
Finally I'd like to thank Bill and his
|
12 |
colleagues at the FTC for all their hard work in
|
13 |
organizing today's panel and assembling the great speakers
|
14 |
that we have lined up today. Thank you.
|
15 |
MR. COHEN: Our first speaker is going to be
|
16 |
Aaron Edlin, who has taught at Berkeley since 1993. He
|
17 |
now holds the Richard Jennings Chair and professorships in
|
18 |
both the economic department and the law school. He's
|
19 |
served on the economic side as Senior Economist at the
|
20 |
Council of Economic Advisers during the years of the
|
21 |
Clinton Whitehouse. He is co-author with Professors
|
22 |
Areeda and Kaplow of one of the leading casebooks on
|
23 |
antitrust and he has published many articles dealing with
|
24 |
competition policy and antitrust law.
|
25 |
Aaron? |
9
1 |
MR. EDLIN: Thank you. Let's see how we get to
|
2 |
the slides.
|
3 |
MR. COHEN: And yesterday we had the
|
4 |
representative from Microsoft [laughter].
|
5 |
MR. EDLIN: Maybe we could switch speakers?
|
6 |
MR. COHEN: I am going to introduce Joe Farrell,
|
7 |
then.
|
8 |
Joe is Professor of Economics here at Berkeley.
|
9 |
He's a Fellow of the Econometric Society, former Editor of
|
10 |
The Journal of Industrial Economics, and former President
|
11 |
of the Industrial Organization Society
|
12 |
Professor Farrell was Chief Economist at the
|
13 |
Federal Communications Commission in 1996 to 1997 and was
|
14 |
Deputy Assistant Attorney General for Economics at the
|
15 |
Antitrust Division of the Department of Justice from 2000
|
16 |
to 2001. From 2001 to 2004, he served on the Computer
|
17 |
Science and Telecommunications Board of the National
|
18 |
Academies of Science.
|
19 |
Joe
|
20 |
MR. FARRELL: Thank you. So, who am I and why
|
21 |
am I here? We've just heard who I am. Why am I here?
|
22 |
Because I've drifted into antitrust from economics. I
|
23 |
think that's true of a lot of the people here. And one of
|
24 |
the things that's most striking is that the whole
|
25 |
unilateral conduct field seems to have drifted a long way |
10
1 |
from first principles. And it's unsatisfying to me and I
|
2 |
worry that it leads to bad policy.
|
3 |
So, what I'd like to do is to try to bring us
|
4 |
back to some first principles. Because the field has
|
5 |
drifted so far from first principles, it's not even
|
6 |
clearly I think understood exactly what those first
|
7 |
principles are. And I'm going to put forward a suggestion
|
8 |
about what they might be.
|
9 |
The suggestion I'm going to put forward is one
|
10 |
that distinguishes quite importantly between the final
|
11 |
goal of antitrust, which I think most of us agree is and
|
12 |
should be economic efficiency, and the protections and the
|
13 |
process involved in antitrust enforcement. And it does
|
14 |
not logically follow that, just because the final goal is
|
15 |
economic efficiency, each case should be analyzed or each
|
16 |
transaction should be analyzed along the lines of economic
|
17 |
efficiency.
|
18 |
Just to give you a simple example, if I go into
|
19 |
a store and take an iPod off the shelf and put it in my
|
20 |
pocket and walk out, that's typically illegal if I didn't
|
21 |
do more than that. And it's illegal even if I can show by
|
22 |
thoroughly convincing evidence that my economic value for
|
23 |
the iPod exceeds the store's replacement cost. In other
|
24 |
words, it was an efficient transaction for me to steal the
|
25 |
iPod. Well, that doesn't cut any ice in law enforcement |
11
1 |
as I understand it and probably shouldn't. And the
|
2 |
economic market system that we have operates by enforcing
|
3 |
the property rights of the iPod. And that enforcement
|
4 |
does not look directly at whether the enforcement is in
|
5 |
the instant efficient or not. And I'm going to claim that
|
6 |
antitrust often does something rather similar, okay?
|
7 |
So, before I get to the first substantive slide
|
8 |
with the provocative title "Analyze This," let me say
|
9 |
that, as I understand it, the fundamental of antitrust is
|
10 |
that you are not supposed to restrain trade. That doesn't
|
11 |
mean you are not supposed to restrain your own trade.
|
12 |
People often comment that it's all right to restrain your
|
13 |
own trade. What you're not meant to do is to restrain
|
14 |
other people's trade.
|
15 |
And you might ask, well, how can you possibly
|
16 |
restrain other people's trade unless you actually tie them
|
17 |
up or something. Well, it turns out that there are
|
18 |
techniques by which a firm might be able to restrain
|
19 |
others' trade. And those techniques it seems to me are
|
20 |
the core problems.
|
21 |
So, that's all setup. Let's come to my purely
|
22 |
hypothetical example, "Analyze This." So, let's think
|
23 |
about the airline market. An airline that I've called
|
24 |
Northeast Airlines offers a five hundred dollar fare. And
|
25 |
it's the only airline that's in that market, so consumers |
12
1 |
buy it. No better deal is available.
|
2 |
An entrant that I've called Sprite would happily
|
3 |
sell at three hundred dollars a similar product.
|
4 |
Consumers would prefer that deal. So, why doesn't it
|
5 |
happen? Well, it doesn't happen in this instance because
|
6 |
everybody recognizes that if Sprite enters and offers the
|
7 |
three hundred dollar deal, Northeast will cut its price to
|
8 |
two hundred dollars. And Sprite is unable to make a
|
9 |
profit competing against the two hundred dollar fare.
|
10 |
So, Sprite anticipates that, doesn't enter, and
|
11 |
consumers continue to pay five hundred dollars.
|
12 |
So, before we get into, well, what law might it
|
13 |
violate and what policies are there and so on, I'd like to
|
14 |
observe that something is clearly wrong there. And let's
|
15 |
delve a little bit in a first principle kind of way into
|
16 |
what it is that's wrong there.
|
17 |
What's wrong I would argue -- and this is based
|
18 |
on discussions that Aaron Edlin and I have been having
|
19 |
over a pretty protracted period of time. What's wrong is
|
20 |
that Sprite's willingness to sell at three hundred
|
21 |
dollars, which consumers would prefer to the status quo, ought
|
22 |
to block Northeast's ability to charge those consumers five
|
23 |
hundred dollars. In other words, Northeast ought not to
|
24 |
be able to extract five hundred dollars from consumers,
|
25 |
given Sprite is willing to sell them the product for three |
13
1 |
hundred dollars. Okay.
|
2 |
And you might think that normally in a
|
3 |
competitive process, whatever that means, not only ought
|
4 |
it to block it but it would. And here it doesn't. And
|
5 |
what are the mechanics of how it doesn't.
|
6 |
Well, the mechanics we just went through.
|
7 |
Northeast, intentionally or not, thwarts Sprite's and
|
8 |
consumers' joint wish, given Northeast's five hundred
|
9 |
dollar price, to trade at three hundred dollars. And the
|
10 |
way that that works is that if Sprite came in it would not
|
11 |
have to compete against five hundred but against two
|
12 |
hundred, and it can't compete against two hundred.
|
13 |
I am saying nothing yet about what's illegal.
|
14 |
I'm just saying this is an instance of something going
|
15 |
wrong in the competitive process.
|
16 |
So, stepping back, and here are some first
|
17 |
principles, okay. Economists study by and large two approaches
|
18 |
to economic efficiency. And there's a little bit of a
|
19 |
disconnect, I think, between the formal material that you
|
20 |
spend a lot of time banging into the undergraduates' heads
|
21 |
in the microeconomics classes and the way that
|
22 |
professional economists typically think about real world
|
23 |
problems.
|
24 |
What we spend the most time with undergraduates
|
25 |
on is that you can get to an economically efficient |
14
1 |
outcome via price-taking perfectly competitive
|
2 |
equilibrium. Okay. However, it's sort of obvious that
|
3 |
the price-taking equilibrium, whether it would be
|
4 |
efficient or not, is unrealistic and unobtainable in many
|
5 |
sectors of the economy that are of antitrust concern. If
|
6 |
nothing else, large economies of scale make that a
|
7 |
nonstarter.
|
8 |
And it's also interesting to note that antitrust
|
9 |
doesn't just move cautiously, but I would say proudly
|
10 |
eschews many opportunities to move toward price-taking
|
11 |
equilibrium. So, in particular, if you have a legitimate
|
12 |
monopoly, quote, unquote, there is no attempt to try to
|
13 |
force you to do anything that's closer to price-taking
|
14 |
behavior. And not only is that potentially difficult and
|
15 |
problematic to do, but antitrust seems to take the
|
16 |
attitude, it's difficult, but we wouldn't try even if we
|
17 |
thought we could do it. Now maybe that's a little
|
18 |
controversial, but that's my impression.
|
19 |
The second approach to economic efficiency, which is
|
20 |
less juicy material for teaching undergraduates because it
|
21 |
has less of the mid-level mathematics that seems to appeal
|
22 |
to those who teach undergraduate micro classes, but is
|
23 |
actually probably more important, is based a little bit on
|
24 |
the Coase theorem, that's kind of the extreme expression of
|
25 |
it, or in formal economic terms is often called the core |
15
1 |
of the economy. And that's the idea that if there is some
|
2 |
inefficiency, then there's some group of people, possibly
|
3 |
unmanageably large but possibly not, that would have an
|
4 |
incentive to contract around it. Okay. And therefore we
|
5 |
think about just how difficult would that be, and if it
|
6 |
wouldn't be all that difficult, then we predict that the
|
7 |
inefficiency will either go away or won't be all that big.
|
8 |
So, for example, it's not exactly an
|
9 |
inefficiency but it's a problem for the consumers that
|
10 |
Northeast is charging such a high fare, and there are
|
11 |
inefficiencies that go along with that.
|
12 |
So, Sprite and consumers jointly would like to
|
13 |
contract around that high fare. And the question is: Why
|
14 |
doesn't that happen?
|
15 |
So, just to give you a little bit of jargon so
|
16 |
as to make you feel that there's real substance to this
|
17 |
talk, what economists call the core of an economy is a set
|
18 |
of possible outcomes such that no group of consumers and
|
19 |
firms could find an alternative that's better for all of
|
20 |
them. Okay. And the core contains only outcomes that are
|
21 |
economically efficient, of course, because if you have an
|
22 |
outcome that's inefficient, then the grand coalition, as
|
23 |
we call it, that is, the set of all consumers and firms,
|
24 |
could all do better by doing something else.
|
25 |
Of course that's not a very realistic process to |
16
1 |
imagine everybody getting together. But, conditional on
|
2 |
knowing that something inefficient is not in the core, we
|
3 |
have a reasonable shot at finding a smaller and more
|
4 |
manageable blocking coalition.
|
5 |
What's a blocking coalition? A blocking
|
6 |
coalition is a group of consumers and firms that can all
|
7 |
do better than the status quo given their endowments and
|
8 |
abilities to trade and so on.
|
9 |
So, in parallel, if you like, with the
|
10 |
competitive equilibrium analysis, we have core analysis.
|
11 |
And it suggests a rather different process. Instead of
|
12 |
suggesting a process where we kind of hammer on the
|
13 |
economy until most firms are somewhere close to
|
14 |
price-taking, okay, and which, as I mentioned, is not
|
15 |
actually feasible in many important sectors of the
|
16 |
economy, it suggests a process where we protect the
|
17 |
ability of these blocking coalitions to work around any
|
18 |
inefficiencies.
|
19 |
So, a perspective on antitrust is this: That
|
20 |
antitrust protects the process of forming blocking
|
21 |
coalitions that block bad outcomes. And how does it
|
22 |
protect that? Antitrust is -- it says certain things are
|
23 |
illegal. What sorts of things are illegal? Well, at some
|
24 |
level, things that thwart the formation of blocking
|
25 |
coalitions that would otherwise prevent bad outcomes. |
17
1 |
VThat's three negatives, which is a very large number of
|
2 |
negatives, okay, but that's the way it is, okay.
|
3 |
So, the last bullet, just to remind you, not all
|
4 |
contracts of course are protected by antitrust. Some of
|
5 |
them are illegal, so there's a little bit of a thorny
|
6 |
issue there, but I'll just note that in passing.
|
7 |
So, back to the Northeast and Sprite example,
|
8 |
Northeast is getting five hundred. Sprite and consumers
|
9 |
would all be better off trading at three hundred. So,
|
10 |
that's a blocking coalition that tells us that the five
|
11 |
hundred dollar fare is not something that would survive in
|
12 |
the core. And, in particular, there's this particular
|
13 |
blocking coalition. And Northeast, and, again, I am not
|
14 |
saying whether they do it on purpose or it's a natural
|
15 |
outcome of the way the market works, but thwarts the
|
16 |
blocking coalition by making clear that if the blocking
|
17 |
coalition tries to form, Northeast will block that in turn
|
18 |
with the two hundred dollar fare.
|
19 |
So, how do we assess Northeast's price cut from
|
20 |
five hundred to two hundred dollars? It seems to me
|
21 |
there's a very difficult and fundamental tension here. In
|
22 |
the instant, that is, if Sprite has actually entered and
|
23 |
is charging three hundred, Northeast then does cut its
|
24 |
price to two hundred, and the two hundred kind of is then
|
25 |
the outcome that we're looking at, well, that seems like |
18
1 |
part of the competitive process as I've described it. We
|
2 |
had this three hundred dollar outcome. Northeast is
|
3 |
forming a blocking coalition with consumers to block it
|
4 |
with a two hundred fare.
|
5 |
However, in its ex ante impact, the prospect of
|
6 |
this two hundred dollars thwarts the formation of Sprite's
|
7 |
blocking coalition against Northeast's five hundred
|
8 |
dollars. So, depending on which way you look at this, it
|
9 |
genuinely is at some level somewhat part of the
|
10 |
competitive process and somewhat a fundamental undermining
|
11 |
thwarting blocking of the competitive process. Okay.
|
12 |
Well, that's a pretty fundamental tension. How
|
13 |
are we going to deal with it? I don't know exactly. I
|
14 |
don't even know approximately. But one thing that's
|
15 |
pretty clear I think out of this discussion, knowing what
|
16 |
Northeast's costs are doesn't tell you anything very
|
17 |
relevant. Knowing whether Northeast made in any sense a
|
18 |
sacrifice with this price cut in some actual or but-for
|
19 |
sense isn't really relevant or doesn't seem to be
|
20 |
relevant. Okay.
|
21 |
So, there's a difficult question here. And the
|
22 |
specific rules and policies that have come to dominate the
|
23 |
law on this kind of behavior don't look as if they're
|
24 |
going to be of any help because, of course, until we
|
25 |
actually work our way through and figure out what the |
19
1 |
right answer is, you don't quite know what will be of
|
2 |
help.
|
3 |
So, what does this suggest about predatory
|
4 |
pricing. It suggests most fundamentally that predatory is
|
5 |
an adjective that doesn't apply to the level of price. It
|
6 |
applies to a pattern of pricing. And, in particular, it
|
7 |
applies to a pattern of pricing such that the price that
|
8 |
the entrant expects to have to compete against is very
|
9 |
different from the price that consumers actually end up
|
10 |
paying.
|
11 |
So, is Northeast's price cut primarily a
|
12 |
blocking coalition to Sprite's three hundred that's the
|
13 |
essence of the competitive process, or an
|
14 |
out-of-equilibrium threat to thwart consumers and Sprite
|
15 |
from blocking the five hundred. That I think might be the
|
16 |
essence of an antitrust offense. Okay.
|
17 |
So, one way to answer this that is sensible
|
18 |
seeming but a little bit ad hoc, departing a bit perhaps
|
19 |
from first principles, but perhaps not, is to say, well,
|
20 |
you sort of want to look at how stable that two hundred
|
21 |
dollars is. If that's really what you've arrived at and
|
22 |
now you are there and you're going to sort of stay there,
|
23 |
then that's sort of how the process is meant to work. We
|
24 |
had originally five hundred, then three hundred, now we've
|
25 |
got two hundred, and we've got there and that's good. |
20
1 |
Certainly good for consumers.
|
2 |
If, on the other hand, what happens is mostly
|
3 |
that consumers really end up paying five hundred and they
|
4 |
only pay three hundred or two hundred in the rare and
|
5 |
short-lived cases where Sprite makes a mistake and enters,
|
6 |
then that seems like a failure of the process. And,
|
7 |
again, it doesn't seem to me that there's much prospect
|
8 |
that sacrifice tests or cost tests are going to be very
|
9 |
helpful here. So, we don't know until we sort of figure
|
10 |
it out.
|
11 |
So, this suggests to Aaron and me a principle we
|
12 |
call freedom to trade. It's a nice phrase, but we mean
|
13 |
it. The incumbent is restraining trade when given its
|
14 |
pricing, etc., etc., etc., and there's a blocking
|
15 |
coalition, a potential blocking coalition, that would make
|
16 |
all its, that is, the blocking coalition's, participants
|
17 |
better off, but the incumbent strategically thwarts the
|
18 |
formation of that blocking coalition.
|
19 |
So, we saw one possible way in which the
|
20 |
incumbent might thwart the formation of a blocking
|
21 |
coalition, threatening that if that coalition starts to
|
22 |
form, then the price it charges will change.
|
23 |
Another way you might do that is through some
|
24 |
kind of divide-and-conquer strategy that says, offer
|
25 |
particularly favorable deals to some pivotal members of |
21
1 |
this blocking coalition while expropriating others. I
|
2 |
don't want to get into the game theory of how it can work
|
3 |
and how it can fail. The fact is it can sometimes work,
|
4 |
but the point I really want to stress here is, when it can
|
5 |
work, it seems like that is really disrupting the
|
6 |
competitive process.
|
7 |
Now, notice that none of this, according to my
|
8 |
suggestion of what the competitive process is, none of
|
9 |
this asks, well, just how unpleasant is it for Northeast
|
10 |
if Sprite comes in and takes away its customers. And that
|
11 |
would be an important aspect of a direct inquiry into
|
12 |
economic efficiency. Right? Because if Northeast
|
13 |
actually has very low costs, and if demand is fairly
|
14 |
inelastic, then having Northeast charging five hundred
|
15 |
dollars might be more efficient than having Sprite come in
|
16 |
and serving customers.
|
17 |
And I claim that Northeast thwarting this entry
|
18 |
would be a thwarting of the competitive process without
|
19 |
asking about that. Okay? So, as I said in the beginning,
|
20 |
it seems to me that if we're looking at the formation of
|
21 |
blocking coalitions as the process whereby we move towards
|
22 |
the core and that's what's economic efficiency, when we
|
23 |
talk about the formation of blocking coalitions, we don't
|
24 |
insist in the interim that they actually have to increase
|
25 |
efficiency. Instead, we know that if you allow the |
22
1 |
formation of blocking coalitions without that inquiry,
|
2 |
that process, when it settles down, will get you to
|
3 |
something that's in the core and therefore really is
|
4 |
economically efficient
|
5 |
So, it seems to me that that captures a lot of
|
6 |
the spirit of the competitive process, that we're
|
7 |
protecting the process of forming blocking coalitions. We
|
8 |
believe that in the long run that will lead to economic
|
9 |
efficiency and it is not necessary and may actually be
|
10 |
counterproductive to ask about economic efficiency at each
|
11 |
step.
|
12 |
That does not mean that I'm advocating a
|
13 |
consumer surplus criterion. Instead, I'm assuming that the final
|
14 |
criterion is actually economic efficiency. At each step,
|
15 |
we do actually look at what consumers want because it's
|
16 |
presumed, I guess, that if an entrant is willing to offer
|
17 |
consumers a better deal, then the entrant likes the
|
18 |
formation of this blocking coalition. So, the question
|
19 |
becomes: Do consumers also like it. But the fact that
|
20 |
there's a sense in which we're looking at consumer
|
21 |
preferences at each step, does not at all imply that the
|
22 |
final goal is consumer surplus.
|
23 |
So, that freedom to trade principle is, we
|
24 |
think, an intriguing and promising way to understand
|
25 |
antitrust starting -- or a lot of antitrust, anyway, |
23
1 |
starting from first principles. How far does it get you?
|
2 |
It gets you to understand, or at least understand the
|
3 |
difficulties in some cases, like the hypothetical I was
|
4 |
talking about and some others. But there's a huge range
|
5 |
of unilateral conduct that gets challenged in antitrust
|
6 |
that it really doesn't directly help you to understand.
|
7 |
And let me sketch this out
|
8 |
And in order to help this, what we're going to
|
9 |
do is to introduce a different phrase, also a good phrase,
|
10 |
"level playing field." So, the observations is that
|
11 |
freedom to trade is potentially at risk where the entrant
|
12 |
has to compete against the low price, but consumers
|
13 |
actually pay a high price. That is the case in my
|
14 |
Northeast/Sprite hypothetical. And I am going to say that
|
15 |
the playing field is level if those prices are equal.
|
16 |
That helps us understand, perhaps, predation, divide and
|
17 |
conquer, exclusive dealing and so on.
|
18 |
But, in the case of many challenged practices,
|
19 |
if the incumbent were simply to go away, consumers would
|
20 |
not be better off. So, a frequent allegation involves the
|
21 |
incumbent being asked to stick around but just do
|
22 |
something different.
|
23 |
So, you can put a lot of unilateral conduct
|
24 |
complaints into the following framework. The incumbent is
|
25 |
offering two trades to consumers, not as alternatives |
24
1 |
typically. I'm going say a price of one hundred dollars
|
2 |
for Product A and a price of five dollars for Product B.
|
3 |
And the discrepancy there is meant to reduce confusion
|
4 |
about which is which. Okay?
|
5 |
And as a potential blocking coalition, sort of,
|
6 |
when entrant and consumers enter B at a price of three
|
7 |
dollars. In other words, there's somebody out there who
|
8 |
would love to supply B for three dollars, but the entrant
|
9 |
simply can't do A, so the incumbent is a monopolist in A.
|
10 |
And the incumbent says, using one technique or another, if
|
11 |
you want to buy my A, you have to buy my B, or more
|
12 |
generally links A to B. Okay.
|
13 |
So, the incumbent might refuse to trade in A if
|
14 |
the customer deals with the entrant in B, or it might
|
15 |
raise the price of A from a hundred to, let's say, a
|
16 |
hundred and ten, which would swamp, of course, any gains
|
17 |
from buying B at three instead of five. And given that
|
18 |
we're assuming that there's a monopoly in A, by the way,
|
19 |
that may well not involve a big profit penalty for the
|
20 |
incumbent.
|
21 |
Now, if you look in B, it should look like
|
22 |
freedom to trade is violated and certainly the playing
|
23 |
field is not level. But in A and B together, there isn't
|
24 |
a potential blocking coalition. Nobody but the incumbent,
|
25 |
I assume, can do A, and consumers don't want to just get |
25
1 |
the cheaper B and not get A. So, if you take the freedom
|
2 |
to trade criterion strictly, there is no potential
|
3 |
blocking coalition, so there can't be a risk that the
|
4 |
incumbent is thwarting a potential blocking coalition.
|
5 |
And really what this comes down to is: What's
|
6 |
the right unit of analysis. Should we be looking at A and
|
7 |
B together? Should we look at B separately? What should
|
8 |
we do?
|
9 |
By the way, I tried to avoid using the term
|
10 |
"market" in talking about A and B because there's no
|
11 |
particular reason to think that A and B will be defined in
|
12 |
the usual way of antitrust markets.
|
13 |
So, just to illustrate this, in case it's
|
14 |
getting a little too abstract, a few of the traditional
|
15 |
boxes, so if A is the tying good, B is the tied good, and
|
16 |
the incumbent is somehow linking trade of the tied good to
|
17 |
trade of the tying good.
|
18 |
Exclusive dealing, A is a bunch of widgets that
|
19 |
the consumer wants to buy, and B is other widgets, maybe
|
20 |
it's a different date or maybe just more of them today or
|
21 |
maybe a different place or something.
|
22 |
If you look at aftermarkets, A might be the
|
23 |
original equipment and B might be service to the
|
24 |
equipment.
|
25 |
So, in all of these cases, it's not uncommon for |
26
1 |
there to be someone who wants to make a better offer in B
|
2 |
and is stymied by some sort of linkage with A.
|
3 |
So, what have I learned from all this? The
|
4 |
setup and the going back to first principles has, at least
|
5 |
for me, clarified the goal and the technique of antitrust.
|
6 |
I've come to think that, although price-taking equilibrium
|
7 |
does conduce to economic efficiency and is typically a
|
8 |
good thing, and is certainly not inconsistent with
|
9 |
analysis of the kind that gets us towards the core,
|
10 |
nevertheless the latter is more fundamental to the ideas
|
11 |
of antitrust than is price-taking equilibrium.
|
12 |
I also think that it's important to understand,
|
13 |
and I have made some steps in my own mind at least to
|
14 |
understanding, that protecting competition as a process is
|
15 |
potentially, and I think actually very different from
|
16 |
imposing on each step of the process a requirement that
|
17 |
has to increase, let's say, economic efficiency, if you
|
18 |
think that that's the final goal.
|
19 |
Trying to go much beyond that, based closely on
|
20 |
first principles as I've been trying to do, turns out to
|
21 |
be quite thorny. Okay. And I think there's a lesson in
|
22 |
there, which is it reinforces what you might already have
|
23 |
known or believed, which is a lot of the rules of thumb,
|
24 |
rules of law and policies that govern unilateral conduct
|
25 |
in antitrust has emerged from the kind of slightly vague |
27
1 |
process that hasn't really linked them very tightly to
|
2 |
first principles.
|
3 |
So, to me, it reinforces that these are thorny
|
4 |
issues. The positive message is, at least for me, it
|
5 |
brings the thorns into sharper focus. And the particular
|
6 |
thorn that I think is pervasive here and is brought into
|
7 |
sharper focus is when, how, in what circumstances, in what
|
8 |
ways can one in some sense require the incumbent to hold
|
9 |
fixed its offer in A, and then we analyze level playing
|
10 |
field or freedom to trade in B.
|
11 |
Is that always illegitimate? That would be a
|
12 |
strict interpretation of freedom to trade as the only
|
13 |
criterion. Is it always legitimate? That would be the
|
14 |
opposite, I guess. Or is there something in between?
|
15 |
Ideally, based firmly on these same first
|
16 |
principles. So, it's not a question of saying, well,
|
17 |
let's consider a hypothetical and figure out what we
|
18 |
intuitively think. But I'd like to work towards getting
|
19 |
there in a way that's closely linked to these first
|
20 |
principles.
|
21 |
Thank you.
|
22 |
(Applause.)
|
23 |
MR. COHEN: Where are we, Aaron?
|
24 |
MR. EDLIN: I will after the break, or any time
|
25 |
I think, be able to project the slides. |
28
1 |
MR. COHEN: Okay, should we then go on to
|
2 |
Howard?
|
3 |
MR. EDLIN: No. I am ready to present,
|
4 |
MR. COHEN: Fine. We're now going to turn to
|
5 |
Aaron Edlin.
|
6 |
MR. EDLIN: Look at that, okay. Great progress.
|
7 |
Let's do the show.
|
8 |
So, the title is, "Sacrifice, Extreme Sacrifice,
|
9 |
and No Economic Sense," three criteria that have been
|
10 |
bandied about a lot recently and increasingly over the
|
11 |
past two decades.
|
12 |
After the colon, the title is: "The case
|
13 |
against these necessary and sufficient tests for
|
14 |
monopolization."
|
15 |
So, of course the big question, the $64,000
|
16 |
question in Section 2 is: When is exclusion
|
17 |
anticompetitive and when is it not? The easy case that we
|
18 |
all understand, presumably, as to how to answer is, if a
|
19 |
monopoly excludes competitors by consistently charging low
|
20 |
prices, well that is anticompetitive. It's the essence of
|
21 |
the competitive process. It's good for consumers.
|
22 |
What that example goes to prove, however, is
|
23 |
that we need something other than exclusion to be
|
24 |
anticompetitive. So, the question is: What plus
|
25 |
exclusion is anticompetitive. The "what" is clearly not |
29
1 |
consistently low prices. The question, though, is what
|
2 |
the "what" is.
|
3 |
And three possible whats have been, as I said,
|
4 |
bandied about a lot of late. They all are basic
|
5 |
sacrifice tests. The basic sacrifice suggested in "Aspen"
|
6 |
and "Trinko" is foregoing profits now or in one line of
|
7 |
business to make more later or in another line of business
|
8 |
as a result of lessened competition.
|
9 |
There is of course another variant, which is
|
10 |
extreme sacrifice, which comes more directly out of
|
11 |
predatory pricing, and you see it applied in "Barry
|
12 |
Wright" and "American Airlines," which is that the test is
|
13 |
really about actually losing money, not just not making as
|
14 |
much as you could, pricing below cost and losing money to
|
15 |
make more later or in another line of business as a result
|
16 |
of lessened competition.
|
17 |
More recently, Greg Werden and Doug Melamed put
|
18 |
forward, and a DOJ "Trinko" brief puts forward a no
|
19 |
economic sense test, which is that the action makes no
|
20 |
economic sense but for a lessening of competition.
|
21 |
These sacrifice tests are on the move, or have
|
22 |
been on the move. In one sense from pricing cases to
|
23 |
non-pricing cases. My reading is that they began and were
|
24 |
first advocated in the predatory pricing context. Thanks
|
25 |
to "Areeda and Turner" and "Willig." And they later |
30
1 |
spread to non-pricing contexts. Thanks, for example, to
|
2 |
"Aspen," "Trinko" and "Covad."
|
3 |
They've also been on the move from sufficiency
|
4 |
once other elements are shown, which is to say, from
|
5 |
something that's helpful in making a case to something
|
6 |
that's necessary for the plaintiff to make a case. So, in
|
7 |
"Barry Wright," we see that there's been no violation,
|
8 |
where above cost, where the pricing is above cost, which
|
9 |
says that extreme sacrifice is necessary in pricing cases.
|
10 |
The DOJ "Trinko" brief advocates the no economic
|
11 |
sense test as necessary. "Covad" assumes that sacrifice
|
12 |
is necessary. Doug Ginsberg writes, "'Covad' will have to
|
13 |
prove Bell Atlantic's refusal to deal caused Bell
|
14 |
Atlantic's short-term economic losses."
|
15 |
Scalia's "Trinko" interpretation of "Aspen,"
|
16 |
which I think is a bit revisionist, is that Ski Company
|
17 |
sacrifice is necessary to violation. And Werden and
|
18 |
Melamed have quite explicitly argued that no economic
|
19 |
sense is the unifying principle of Section 2 violations.
|
20 |
My fundamental contention which I've been
|
21 |
arguing for years is that sacrifice is not needed for
|
22 |
anticompetitive effect and frequently not needed.
|
23 |
My "Yale Law Journal" article argues that this
|
24 |
is true for what I call above cost predatory pricing. And
|
25 |
if you think that below cost is part of the definition of |
31
1 |
predatory pricing, then what I mean is above cost pricing
|
2 |
that is exclusionary and anticompetitive. There I explain
|
3 |
how consumers can be hurt by threats to lower prices, much
|
4 |
as Joe Farrell explained, even though prices will remain
|
5 |
above cost, and perhaps even though prices may be profit
|
6 |
maximizing.
|
7 |
I ask rhetorically: If sacrifice is wrong
|
8 |
headed in the predatory pricing context, why are we
|
9 |
extending it to non-pricing cases? Consider "Aspen."
|
10 |
Now, suppose, as I think is likely, that Ski Company's
|
11 |
refusal to sell at retail prices to Highlands increased
|
12 |
Ski's retail sales to skiers. What I'm thinking there is
|
13 |
that it certainly is conceivable, perhaps even likely,
|
14 |
that when Ski Company refused to sell at retail to
|
15 |
Highlands, what that meant was that, sure, they sold a
|
16 |
couple less tickets as part of Highlands' adventure packs.
|
17 |
However, on the other hand, what likely happened was that
|
18 |
the consumer decided, or many of them did, that they would
|
19 |
buy a whole week of skiing at Ski Company. So, there may
|
20 |
have been no sacrifice there of profits, even though they
|
21 |
refused to sell at retail.
|
22 |
But would that mean that the refusal was any
|
23 |
less exclusionary or anticompetitive? I think not. The
|
24 |
"Aspen" court didn't just rest on what I think is a shaky
|
25 |
notion of Ski Company's sacrifice, but they also |
32
1 |
emphasized what they took to be consumer harm, the
|
2 |
revisionist claims of Trinko about "Aspen"
|
3 |
notwithstanding.
|
4 |
Another case or set of cases where I think it's
|
5 |
fairly clear that sacrifice is not necessary for
|
6 |
anticompetitive effect are submarine patents. If you seek
|
7 |
a patented process into an industry standard, that may not
|
8 |
involve sacrifice of any kind that I can see. But that
|
9 |
fact doesn't make it a good thing to do.
|
10 |
Many people have been talking about an extreme
|
11 |
case where Firm A blows up a competitor's plant. Now,
|
12 |
Werden and Melamed, and fellow travelers with them,
|
13 |
emphasize that this isn't a problem for them because the
|
14 |
cost of the dynamite triggers liability. There is a
|
15 |
sacrifice; you had to pay for the dynamite. And that is
|
16 |
what triggers liability and means that there's no economic
|
17 |
sense to blowing up your competitor's plant but for the
|
18 |
lessening of competition, which justifies the cost of
|
19 |
paying for dynamite.
|
20 |
Like Joe Farrell, I don't -- this reasoning
|
21 |
doesn't grab me and I feel a great suspicion that the cost
|
22 |
of the dynamite could really be important here. But one
|
23 |
way of saying that is to change the hypo. What if Firm A
|
24 |
is avoiding a dump fee by deposing of surplus dynamite in
|
25 |
this way. If they didn't blow up the competitor's plant, |
33
1 |
they would have had to pay a dump fee to dispose of the
|
2 |
dynamite.
|
3 |
Well, now I gather that the dynamite has a
|
4 |
negative cost. So, according to the no economic sense
|
5 |
test or the sacrifice test, there should be no liability.
|
6 |
Well, this just can't be. It can't be that it should
|
7 |
hinge on that. This suggests to me that the sacrifice
|
8 |
test is not looking at the right thing.
|
9 |
If the sacrifice test is not looking at the
|
10 |
right thing, neither is extreme sacrifice. Extreme
|
11 |
sacrifice, that is losses, are certainly not needed for
|
12 |
anticompetitive effect. Consider the American Airlines
|
13 |
case brought by the DOJ unsuccessfully. The judge thought
|
14 |
there that the extra plane was profitable if you ignore
|
15 |
effects on other planes. I suggest that everyone reread
|
16 |
footnote 13 of that case over and over and over again if
|
17 |
you think that the extreme sacrifice test might make
|
18 |
sense, as the judge did.
|
19 |
Marginal revenue, as every economist and econ 1
|
20 |
student knows, is less than price. For firms with lots of
|
21 |
market power, which you might think are one of the focuses
|
22 |
of Section 2, marginal revenue is much lower than price.
|
23 |
What that means is that monopolies with lots of market
|
24 |
power can sacrifice enormously without triggering the
|
25 |
extreme sacrifice test. I think, as I pointed out |
34
1 |
previously, it is very ironic to give such firms a
|
2 |
license, such a license, such a grand license to exclude.
|
3 |
Let's go back and consider the case of blowing
|
4 |
up your competitor's factory. Could it be a violation
|
5 |
only if the dynamite is so expensive that its cost exceeds
|
6 |
Firm A's operating profits? It seems outlandish to me on
|
7 |
its face, but the extreme sacrifice test says yes.
|
8 |
And I'll point out that in that case, firms with
|
9 |
large profits have a substantial and much larger license
|
10 |
to blow up their competitors than other firms.
|
11 |
Rhetorically I'll ask why.
|
12 |
Consider the no economic sense test. Does that
|
13 |
make sense? Well, apply it to limit pricing. Consider a
|
14 |
firm that could charge a high price and make lots of
|
15 |
money, for a while anyway, but this firm chooses a low
|
16 |
price, less profitable for now. Why? In order to delay
|
17 |
or prevent entry.
|
18 |
Suppose there is no economic sense in charging
|
19 |
this low price before there is entry, except that it
|
20 |
prevents others from entering. Well, the no economic
|
21 |
sense test condemns that limit pricing. But note that
|
22 |
that's the essence of competition. It's what I had as the
|
23 |
easy case on slide two.
|
24 |
Werden doesn't apply the test here. Instead he
|
25 |
grants a safe harbor for charging the low price. |
35
1 |
Now, if your test would condemn this case and so
|
2 |
you have to make an exception and grant a safe harbor
|
3 |
because it's so obvious that this is procompetitive, I'd
|
4 |
suggest that the test is not getting at the fundamentals.
|
5 |
This smells bad to me.
|
6 |
Back to blowing up the competitor's factory, a
|
7 |
la "Conwood" discussion, Werden, page 425. Proponents of
|
8 |
the no economic sense test emphasize again that the cost
|
9 |
of the dynamite makes it illegal. As I pointed out, costs
|
10 |
might be negative in the dump fee hypothetical.
|
11 |
My claim would be that blowing up your
|
12 |
competitor's factory is anticompetitive regardless of the
|
13 |
cost of the dynamite, regardless of whether it has a
|
14 |
negative cost, a small positive cost, or costs more than
|
15 |
the operating profits, regardless of whether you pass the
|
16 |
no economic sense test.
|
17 |
The fundamental problem in my view with all
|
18 |
these sacrifice tests is that these tests don't flow from
|
19 |
any kind of first principles that are attractive. They
|
20 |
don't flow from consumer welfare or from efficiency. They
|
21 |
also don't flow from a notion of how the competitive
|
22 |
process would work, for example, a process by which rivals
|
23 |
can offer consumers - by which rivals who can offer
|
24 |
consumers higher utility actually get to provide that
|
25 |
higher utility. |
36
1 |
The tests don't flow from any other principles
|
2 |
I've been able to discern from reading about them.
|
3 |
Now, when someone like me points out that there
|
4 |
are many cases where the tests are not satisfied but the
|
5 |
action is anticompetitive, what you quickly bump into,
|
6 |
both in the commentary and in the cases, is a refrain
|
7 |
about false positives. It's a chorus. Fears and claims
|
8 |
about these false positives abound. However, I'd suggest
|
9 |
a modern example that I can put forward are pretty scarce.
|
10 |
A common argument is that you need a hurdle to
|
11 |
avoid these false positives. So, sacrifice is not needed
|
12 |
for anticompetitive effect, but the plaintiffs should be
|
13 |
required to show it anyway, in order to prevent an avalanche
|
14 |
of cases from chilling legitimate competition.
|
15 |
To me, when I hear that, I wonder, why not just
|
16 |
tax plaintiffs, if that's the goal. Or, if you really
|
17 |
want to eliminate these false positives, you could
|
18 |
eliminate Section 2 entirely, or you could eliminate
|
19 |
Section 2 for any plaintiffs whose name begins with A
|
20 |
through M, then you get rid of half the false positives.
|
21 |
Erecting arbitrary hurdles because the right
|
22 |
test is difficult to administer properly is, I would
|
23 |
argue, wrong-headed. What commentators should do, and
|
24 |
ultimately courts, is seek, as best they can, the right
|
25 |
test. |
37
1 |
Now, once you've sought the right test, if
|
2 |
administrative difficulties truly make false positives a
|
3 |
bigger problem than false negatives, and there is not all
|
4 |
that much discussion by the refrainers about false
|
5 |
negatives, there is an answer which doesn't involve
|
6 |
arbitrary hurdles or abandoning the right test. You could
|
7 |
raise the standard of proof in that case. You could
|
8 |
improve jury instructions. You could create procedural
|
9 |
hurdles like "Dauber" to require rigorous evidence. We
|
10 |
have a number of those. And, again, I think you'll find
|
11 |
that modern examples of clear false positives are pretty
|
12 |
rare.
|
13 |
What are my conclusions? That patience is
|
14 |
needed. We should be searching for the right standard, or
|
15 |
at least better ones, and that administrative difficulties
|
16 |
don't justify arbitrary tests. And too often they have
|
17 |
been used to do so.
|
18 |
Thank you.
|
19 |
(Applause.)
|
20 |
MR. COHEN: Okay. Our last presenter this
|
21 |
morning is Howard Shelanski, Professor of Law at Berkeley
|
22 |
here, where he is also Associate Dean and the co-director
|
23 |
of the Berkeley Center for Law and Technology. His
|
24 |
research focuses on antitrust policy and regulation.
|
25 |
On the economic side, from 1999 to 2000, |
38
1 |
Professor Shelanski served as Chief Economist of the
|
2 |
Federal Communications Commission, and in 1998 to 1999, he
|
3 |
was a Senior Economist to the President's Council of
|
4 |
Economic Advisers at the White House.
|
5 |
On the law side, Professor Shelanski served as a
|
6 |
clerk to U.S. Supreme Court Justice Antonin Scalia.
|
7 |
We welcome your presentation.
|
8 |
MR. SHELANSKI: Thanks very much, Bill. I'm
|
9 |
really happy to be here. And I want to make a
|
10 |
presentation that at least in some aspects will connect to
|
11 |
what my colleague Aaron Edlin was just talking about, in
|
12 |
the sense that it may give some insights into how to
|
13 |
choose among different kinds of tests for enforcement
|
14 |
under Section 2.
|
15 |
And I want to speak specifically about
|
16 |
enforcement in the area of unilateral refusals to deal, an
|
17 |
area that has, I think, become particularly challenging in
|
18 |
the wake of the "Trinko" case.
|
19 |
And the broad point that I want to make is this:
|
20 |
That at the same time that the Department of Justice and
|
21 |
the Federal Trade Commission are reviewing enforcement
|
22 |
policy for Section 2 of the Sherman Act, there are
|
23 |
parallel efforts ongoing, indeed some undertaken in recent
|
24 |
years by the Federal Trade Commission, to rethink and
|
25 |
reform intellectual property rights, and particularly to |
39
1 |
reform it in a way that makes it harder for firms to use
|
2 |
intellectual property to foreclose competition with weak
|
3 |
or questionable IP rights.
|
4 |
And I think that the potential outcomes of IP
|
5 |
reform could matter for aspects of antitrust reforms, and
|
6 |
notably for policy toward unilateral refusals to deal.
|
7 |
So, my main point is that, in thinking about
|
8 |
Section 2 enforcement, and in particular thinking about
|
9 |
unilateral refusals to deal, antitrust reform efforts
|
10 |
should not ignore intellectual property reform processes
|
11 |
So, I have a general suggestion, which is that
|
12 |
antitrust authorities should keep an eye on IP reform and
|
13 |
take into account how it might affect enforcement policies
|
14 |
under Section 2. Not a terribly original idea in broad.
|
15 |
Louis Kaplow in 1984 wrote a very nice paper talking about
|
16 |
how antitrust and IP should be thought of as part of an
|
17 |
interactive system. But I also want to talk about
|
18 |
specific conjecture and, as we get further along, you'll
|
19 |
see why I refer to it as merely conjecture, which is, if
|
20 |
IP reform is likely to reduce the strength or availability
|
21 |
of intellectual property protections, antitrust
|
22 |
authorities might consider enforcing less strictly against
|
23 |
refusals to deal.
|
24 |
Now, let me try to explain why. Under "Trinko,"
|
25 |
there is a presumption against requiring a firm to deal |
40
1 |
with competitors. Now, there are many things one can read
|
2 |
into "Trinko". "Trinko" adopts a very strong line against
|
3 |
duties to deal for firms in the unilateral context. But
|
4 |
"Trinko" did preserve "Aspen". Very interestingly,
|
5 |
"Aspen", which is a hard case to teach to students and in
|
6 |
many ways a hard case to explain. "Aspen" is a case that
|
7 |
imposed a duty to deal.
|
8 |
I agree with Aaron Edlin that Justice Scalia
|
9 |
engaged in some revisionism by finding profit sacrifice in
|
10 |
that case, but inherently what "Aspen" says is, if there
|
11 |
is nothing that you gain by refusing to deal, then we are
|
12 |
going to assume that what you gained is a reduction in
|
13 |
competition that inures to your benefit. That's one way
|
14 |
of looking at it. But "Aspen" still exists after
|
15 |
"Trinko". We have a strong presumption articulated in the
|
16 |
"Trinko" decision against imposing duties to deal.
|
17 |
The question that's left for the antitrust
|
18 |
agencies is the following: Okay, where do we impose the
|
19 |
duty to deal or not. So, I want to talk a little bit
|
20 |
about some policy issues that might arise, some background
|
21 |
issues, and then talk about how IP reform might affect the
|
22 |
answer to that question of what standard to use in
|
23 |
imposing a duty to deal.
|
24 |
Well, the first thing that we need to keep in
|
25 |
mind of course is that only some refusals to deal cause |
41
1 |
anticompetitive harm. There are many cases where refusals
|
2 |
to deal will cause competitive supply to enter the market,
|
3 |
would cause a firm to invent around the refusal to deal or
|
4 |
to innovate or produce something itself.
|
5 |
Mandatory dealing in cases where there isn't
|
6 |
anticompetitive harm could impede investment and
|
7 |
innovation by the firms being forced to deal. So, that's
|
8 |
an argument one often hears. If you go back to some of
|
9 |
the previous rounds of these hearings, Former Assistant
|
10 |
Attorney General for Antitrust Eupate has some testimony
|
11 |
saying exactly this, if you force firms to deal, they're
|
12 |
not going to innovate. There's some interesting counter
|
13 |
argument by Professor Steven Fallon that suggests the
|
14 |
evidence for such innovation deterrence is thin. But we
|
15 |
have to at least keep in mind the possibility that
|
16 |
mandatory dealing could impede investment.
|
17 |
I think that one of the bigger concerns is that
|
18 |
enforcement of a duty to deal might reduce competitive
|
19 |
innovation and production not by the firms being forced to
|
20 |
deal, but by other firms in the marketplace or by the
|
21 |
would-be buyer, by creating a quasi-regulated purchase
|
22 |
alternative.
|
23 |
So, "Trinko" takes into account all of these
|
24 |
possibilities, that there isn't a lot of -- that there are
|
25 |
many refusals to deal that are not anticompetitive and |
42
1 |
imposing a duty to deal in fact may have consequences to
|
2 |
justify its presumption against the duty to deal. But
|
3 |
"Trinko" does not necessarily mean refusals to deal are
|
4 |
evil per se.
|
5 |
So, refusals to deal can have anticompetitive
|
6 |
harms. And we would not necessarily want to exempt those
|
7 |
refusals to deal from enforcement.
|
8 |
Now, I want to suggest that one necessary
|
9 |
condition for such harm is that competitors and third
|
10 |
parties face economic barriers to providing the goods at
|
11 |
issue or that competitors and third parties face legal
|
12 |
barriers to providing the goods at issue.
|
13 |
And I would suggest we should not impose duties
|
14 |
to deal in goods for which economic or legal barriers to
|
15 |
competitive supply do not exist. There you get very
|
16 |
little pay off and you may creat some deterrent effects to
|
17 |
innovation either on the supply or the demand side.
|
18 |
But what about refusals that could be
|
19 |
anticompetitive, for which there are economic barriers or
|
20 |
legal barriers. There are several standards that we could
|
21 |
use to identify those situations and to decide whether or
|
22 |
not to enforce a duty to deal.
|
23 |
So, one thing we could do is to say, listen, we
|
24 |
should have per se legality for refusals to deal. This is
|
25 |
in the spirit of "Trinko", it's a strong reading of |
43
1 |
"Trinko", but it's a very clean line and we avoid any risk
|
2 |
of deterring innovation on either the supply or the demand
|
3 |
side.
|
4 |
Alternatively, we have a range of rule of reason
|
5 |
approaches. And I'm just going to very simplistically
|
6 |
phrase them as potential consumer welfare tests, the kind
|
7 |
of tests that Professor Salop proposed in an earlier round
|
8 |
of these hearings; a business justification test, which
|
9 |
Kolasky suggested in that same round; and a profit
|
10 |
sacrifice test of various stringency, ranging right up to
|
11 |
a no business sense kind of test of the kind that Doug
|
12 |
Melamed has articulated.
|
13 |
Then we have the old line essential facilities
|
14 |
approach, which as Justice Scalia tells us, the Supreme
|
15 |
Court has never adopted. One could quibble about what
|
16 |
"Onertel" means, but there is some precedent certainly in
|
17 |
the Appellate Court for the essential facilities approach,
|
18 |
notably in the Seventh Circuit.
|
19 |
So, how should the Justice Department and the
|
20 |
Federal Trade Commission choose among these various
|
21 |
approaches? Well, I don't much like the per se legality
|
22 |
approach because per se legality fails to block cases
|
23 |
where the only effect is anticompetitive. And while often
|
24 |
justified on the grounds of preserving the refusing firm's
|
25 |
innovation and investment incentives, there isn't clear |
44
1 |
evidence that that is [unintelligible]. And I think
|
2 |
you're likely to have poor welfare effects here.
|
3 |
I don't much like the essential facilities
|
4 |
approach either because it does ignore some legitimate
|
5 |
business justifications. And I think that it may too
|
6 |
easily allow access and deter innovation and investment by
|
7 |
the buyer or the third parties. And more -- of great
|
8 |
concern is it requires a quasi-regulatory solution.
|
9 |
While I fully agree with my colleague Aaron that
|
10 |
we should not let administrative difficulties justify a
|
11 |
bad test, we shouldn't ignore administrative difficulties
|
12 |
in the test that we actually choose to administer. And
|
13 |
there's some hard pricing questions that emerge any time
|
14 |
that we follow the full essential facilities test as it's
|
15 |
been articulated in the appellate courts.
|
16 |
Well, this leads to the rule of reason
|
17 |
alternatives. And I'm not going to exactly say which rule
|
18 |
of reason alternative I think is best. I think we've
|
19 |
heard a lot of very interesting and provocative arguments
|
20 |
for the specific nature of the test.
|
21 |
I want to oversimplify by assuming that if you
|
22 |
took all of the rules of reason tests that are proposed
|
23 |
that you can differentiate them along a spectrum from
|
24 |
relatively strong enforcement to relatively weak
|
25 |
enforcement. In other words, they can be differentiated |
45
1 |
according to the likelihood that we'll find conduct to be
|
2 |
anticompetitive by how strictly they would enforce against
|
3 |
refusals to deal and how likely they would be to impose a
|
4 |
duty to deal.
|
5 |
So, the policy for the courts and the antitrust
|
6 |
agencies I think may be how stringent or generous the rule
|
7 |
of reason test to choose for judging refusals to deal. I
|
8 |
think that IP rights, intellectual property rights, might
|
9 |
affect the answer. And here's why.
|
10 |
Intellectual property rights are a primary
|
11 |
source of legal barriers to competitive provision of goods
|
12 |
that an incumbent refuses to sell to rivals. We heard in
|
13 |
the testimony yesterday from some of the company
|
14 |
witnesses, notably QUALCOMM and a couple of others, that
|
15 |
they're very concerned about any rule that might require
|
16 |
them to deal in particular ways with their intellectual
|
17 |
property. Intellectual property rights grant them a legal
|
18 |
ability to give them the ability to impose a legal barrier
|
19 |
to invent around to innovations that would replicate their
|
20 |
invention, and therefore gives power, creates an effect
|
21 |
out of their refusal to deal or refusal to deal on
|
22 |
particular terms.
|
23 |
But, logically, any reduction in the strength
|
24 |
and availability of IP protections could reduce the pool
|
25 |
of goods for which there are legal barriers to competitive |
46
1 |
supply. There is an empirical question buried in here
|
2 |
that I will return to at the end. But I think that IP
|
3 |
reform could therefore affect the frequency with which
|
4 |
refusals to deal weaken the conditions for being
|
5 |
anticompetitive, in turn affecting the likelihood that
|
6 |
enforcement of the duty to deal was warranted.
|
7 |
So, what's the benefit of a more discerning
|
8 |
intellectual property policy if IP reform reduces a firm's
|
9 |
ability to use IP protections to block competitive supply
|
10 |
and innovation, then IP reform can limit the need for rule
|
11 |
of reason exceptions to Trinko's presumption against
|
12 |
mandatory dealing with rivals.
|
13 |
Now, one might say, okay, fine, why not have
|
14 |
intellectual property reform and a fairly liberal duty to
|
15 |
deal. Won't that unblock lots anticompetitive refusals to
|
16 |
deal.
|
17 |
Well, both intellectual property reform and
|
18 |
duties to deal aim to reduce barriers to competitive
|
19 |
supply and innovation, but I think that their individual
|
20 |
welfare effects may not be additive if they're undertaken
|
21 |
together.
|
22 |
Suppose that we do not have IP reform and that
|
23 |
there is some good that is being used anticompetitively to
|
24 |
block competitive supply. The duty to deal can increase
|
25 |
welfare with no risk of deterring investment or innovation |
47
1 |
by the would-be buyer or third parties. The would-be
|
2 |
buyer or third parties could be blocked by an intellectual
|
3 |
property barrier to competitive supply or innovation, and
|
4 |
so requiring that the refusing to sell or deal doesn't
|
5 |
block any innovation on the demand side by the would-be
|
6 |
buyer or by third parties. It might deter innovation and
|
7 |
investment by the incumbent. That is something that we
|
8 |
need to think about.
|
9 |
With reduction of legal barriers through IP
|
10 |
reform, however, the duty to deal now can undermine new
|
11 |
competition and innovation, reducing welfare. So, the
|
12 |
firm that is refusing to deal and the good that is
|
13 |
protected by intellectual property, if they now have a
|
14 |
weaker intellectual property right, we might want to say,
|
15 |
well, let's not make them deal because now there's an
|
16 |
invent around or a replication that didn't exist before.
|
17 |
So, IP reform raises the likelihood, whether to
|
18 |
any significant level is another question, but it raises
|
19 |
the likelihood of false positives in antitrust enforcement
|
20 |
through imposition of a duty to deal where the conditions
|
21 |
for anticompetitive harm as a legal barrier do not hold.
|
22 |
So, let's take a little bit of a closer look at
|
23 |
the implications of IP reform for Section 2 reform. There
|
24 |
are several kinds of proposals for intellectual property
|
25 |
reform that could bear on the effects of refusals to deal. |
48
1 |
There's just some broad examples
|
2 |
There are proposals to raise the bar for
|
3 |
patentability: better pre and post grant opposition
|
4 |
procedures; more transparent review, both in initial grant
|
5 |
and post grant of patent grants or annuities.
|
6 |
There are also proposals to reduce consequences
|
7 |
of patentability: a narrowed patentable subject matter,
|
8 |
for example, cutting software out of patentable subject
|
9 |
matters; expanded research exceptions and reduced
|
10 |
presumptions of harm in injunction proceedings which might
|
11 |
push parties to the bargaining table; and limit refusals
|
12 |
to deal. And these are proposals that can be found in the
|
13 |
National Academy of Sciences' proposal, in the Federal
|
14 |
Trade Commission's report of a couple of years ago; in
|
15 |
draft statute that floated around in 2004; and in a
|
16 |
variety of ongoing documents one can find these proposals.
|
17 |
So, the effects of these proposals would likely
|
18 |
be to make fewer goods subject to IP protections and to
|
19 |
make those protections less expansive. Some of the most
|
20 |
prominently discussed IP reforms, and I think this is the
|
21 |
important point, would reduce the ability of incumbents to
|
22 |
foreclose competitive provision of goods through the
|
23 |
exercise of intellectual property rights.
|
24 |
Depending on circumstances, these refined IP
|
25 |
protections could have varying effects on incentives to |
49
1 |
deal. The reduced ability to foreclose competitive
|
2 |
innovation through the enforcement of an intellectual
|
3 |
property right might make an incumbent more eager to sell
|
4 |
to rivals because it would expect greater competitive
|
5 |
entry in the relevant property market than existed
|
6 |
pre-reform, and the incumbent may therefore want to take
|
7 |
the sales for itself for as long as it can.
|
8 |
Alternatively, an incumbent may be less eager to
|
9 |
deal if the sale to others would raise the speed or
|
10 |
likelihood of competitive entry compared to what would
|
11 |
occur if it keeps the good to itself.
|
12 |
And which of these incentive effects occurs
|
13 |
would depend very much on the nature of the good, the
|
14 |
degree to which the selling firm is vertically integrated.
|
15 |
There are a number of questions that are factored in.
|
16 |
But I think on the whole refined intellectual
|
17 |
property could reduce the incidence and the impact of
|
18 |
refusals to deal. It is true that refined IP protections
|
19 |
could reduce the willingness to deal with rivals by
|
20 |
reducing an incumbent's ability to block replication of or
|
21 |
innovative alternatives to its technology. But I think
|
22 |
this effect is most likely where the goods involved are
|
23 |
easy to reverse engineer and replicate. And these in
|
24 |
turn, I think, are the goods where refusals to deal would
|
25 |
be less harmful because the would be-buyer or others will |
50
1 |
eventually be able to market.
|
2 |
So, on the whole, I think we'll find
|
3 |
intellectual property protections should either reduce
|
4 |
incentives to refusals to deal, or reduce the long-term
|
5 |
effects of refusing to deal by opening the door to
|
6 |
competitive supply and innovation.
|
7 |
So, what are the implications for Section 2
|
8 |
reform? The latter effect, competitive reinvention or
|
9 |
replication of the goods at issue in a refusal case should
|
10 |
be preserved. Antitrust reform should not impede a
|
11 |
competitive reinvention because they should not provide an
|
12 |
alternative or option to competitive entry or invention or
|
13 |
innovation where it is feasible to occur.
|
14 |
So, I think that if intellectual property reform
|
15 |
reduced legal barriers to competitive production of the
|
16 |
relevant good, Section 2 should be less willing to require
|
17 |
the incumbent to deal. Broad exemptions to the "Trinko"
|
18 |
presumption against mandated dealing could create a
|
19 |
quasi-regulatory alternative to buyers that is unnecessary
|
20 |
and unhelpful to economic welfare.
|
21 |
So, that's some questions to investigate before
|
22 |
we know whether intellectual property reform is actually
|
23 |
going to matter.
|
24 |
Several key questions. First of all, how likely
|
25 |
is IP reform and to what extent will it refine the |
51
1 |
consequences of IP protections for competition. I think
|
2 |
to question these efforts are under way. They're very
|
3 |
political and very contentious. What will emerge from
|
4 |
them is unclear. I think something will, but I think it's
|
5 |
hard to know exactly what.
|
6 |
The next question is really an empirical one and
|
7 |
I think lies at the core of what I'm suggesting today:
|
8 |
How much of a problem with refusal to deal stems from IP
|
9 |
protected goods for which the barrier to competitive
|
10 |
supply is a legal one rather than an economic one that
|
11 |
stems from scale or something else. If not much, then the
|
12 |
considerations I'm suggesting can be put aside as
|
13 |
Section 2 reform proceeds. But if a lot, even if only in
|
14 |
particular industries or markets, then refusal to deal
|
15 |
policy should recognize the welfare and complexities that
|
16 |
intellectual property reform might introduce.
|
17 |
And the final question is: What effects will
|
18 |
applied intellectual property protections have on the
|
19 |
incentive of incumbent firms to deal with rivals. I think
|
20 |
that's an interesting question to investigate.
|
21 |
So, I have some tentative conclusions.
|
22 |
The rule of reason approach for refusals to deal
|
23 |
has potential advantages over either per se legality or
|
24 |
the essential facilities test.
|
25 |
The policy problem is to decide how strict a |
52
1 |
test the courts and agencies should apply in assessing the
|
2 |
reasonability of refusals to deal with rivals. And the
|
3 |
potential results of intellectual property reform may be a
|
4 |
relevant consideration in that choice, with more refined
|
5 |
intellectual property rights weighing in favor of less
|
6 |
strict enforcement against refusals to deal.
|
7 |
Thank you.
|
8 |
MR. COHEN: Thank you very much Howard we're now
|
9 |
going to take a break for roughly fifteen minutes.
|
10 |
(A brief recess was taken.)
|
11 |
MR. COHEN: Fine. Before we begin our questions
|
12 |
and round-table discussions, I think a way to start this
|
13 |
second session would be to give each of our speakers a few
|
14 |
minutes to respond to or comment upon some of the issues
|
15 |
that were raised by the other panelists.
|
16 |
You can go in whichever order you prefer. We do
|
17 |
ask as a reminder to speak into the microphone so we can
|
18 |
get this transcript.
|
19 |
MR. SHELANSKI: I'll start because I expect
|
20 |
collusion over here on the right.
|
21 |
So, I really enjoyed Aaron's and Joe's related
|
22 |
presentations and I think that they are both in the core
|
23 |
respects correct. I do have just a couple of observations
|
24 |
or comments.
|
25 |
So, one suggestion I would make is if you take |
53
1 |
Aaron's presentation and Joe's presentation and put them
|
2 |
together, you could take them as saying that, if a firm
|
3 |
cuts price in response to entry, one test is that it is
|
4 |
not acting anticompetitively, it's in a safe harbor if it
|
5 |
keeps its price low.
|
6 |
And I just wonder -- the question I would have
|
7 |
or the thing I would ask them to consider is whether their
|
8 |
proposals, as compared with other tests that are typically
|
9 |
used in this area, would increase the ability of
|
10 |
competitive firms already in the market to raise rivals'
|
11 |
costs by entering, for example, on the airline route that
|
12 |
was at five hundred, bringing it down to two hundred, and
|
13 |
then basically telling the five hundred dollar firm, you
|
14 |
either need to cut your price and keep it there or face
|
15 |
some kind of antitrust scrutiny that you will find
|
16 |
unpleasant.
|
17 |
Is the raising of rivals' cost prospect greater
|
18 |
under proposal than under others? I don't know. It's
|
19 |
just something that I think ought to be thought about
|
20 |
The other comment that I have is that I am not
|
21 |
fully persuaded that costs don't matter at all in the
|
22 |
consideration of whether or not the five hundred dollar
|
23 |
price is a problem or not. Obviously, as Aaron points
|
24 |
out, the monopolist has the greater ability to sacrifice
|
25 |
profits because it has obviously much higher net profits. |
54
1 |
But I wonder, again, and this may relate to the
|
2 |
competitive strategy angle here, if the five hundred
|
3 |
dollar price is not three hundred dollars above the
|
4 |
competitive equilibrium, but a hundred dollars over the
|
5 |
competitive equilibrium, we might worry a little bit less
|
6 |
about the five hundred dollar price being the one that
|
7 |
we're running into in the market because someone decides
|
8 |
not to enter at four hundred dollars. Don't we have to
|
9 |
look at costs to know how great a welfare loss there is to
|
10 |
the current test? And would that matter to your
|
11 |
recommendation of what do in in a particular case?
|
12 |
MR. FARRELL: Well, let me start with that last
|
13 |
one.
|
14 |
I think if we knew everything, then you're
|
15 |
probably right. I would take pretty strongly the
|
16 |
perspective that the competitive process is about having
|
17 |
policies that don't require us to know what the
|
18 |
competitive equilibrium price is likely to be, and that
|
19 |
therefore enforcement of competition policy and antitrust
|
20 |
should not depend upon on our being able to say we think
|
21 |
the competitive price would be X.
|
22 |
And that's part of why I think the competitive
|
23 |
process, as I understand it, operates through the
|
24 |
formation of a blocking coalition that make the
|
25 |
participants better off, without an inquiry into how much |
55
1 |
the incumbent loses from this entry.
|
2 |
So, if you look at the entry in the oligopoly
|
3 |
literature, the usual citation is the Mankiw and Whinston
|
4 |
article, 1986 or thereabouts. And if you think about the
|
5 |
way that regulation has traditionally treated
|
6 |
cream-skimming and loss of income and profits due to entry
|
7 |
and, think in terms of access pricing to control and deal
|
8 |
with that, all of that it seems to me is extremely foreign
|
9 |
to competition policy. And the reason it's foreign to
|
10 |
competition policy is I think that the competitive process
|
11 |
works precisely by ignoring the effects on the incumbent.
|
12 |
And obviously if you want to increase welfare in the
|
13 |
small, ignoring something like that that could be quite
|
14 |
important is a stupid thing to do. But I think as part of
|
15 |
an overall process, it's brilliant and seems to work
|
16 |
rather well.
|
17 |
And I think there are times, perhaps many times,
|
18 |
when many, perhaps all of us, get confused about that.
|
19 |
Because there's no doubt, I think there's a consensus that
|
20 |
the eventual goal of all of this is economic efficiency.
|
21 |
So, it's always very tempting to look at economic
|
22 |
efficiency in each instance, and perhaps often is right to
|
23 |
do so, but I think it's often wrong to do so.
|
24 |
MR. SHELANSKI: And just a comment here on legal
|
25 |
precedence. |
56
1 |
I actually think that you're on pretty good
|
2 |
ground with some recent legal precedent. I mean, if I
|
3 |
understood your comments about "Barry Wright" correctly,
|
4 |
that that case made the mistake of thinking that downward
|
5 |
pricing was more important than the competitive process.
|
6 |
Maybe that's a way of summarizing your critique. I don't
|
7 |
know if that's unfair or not.
|
8 |
And certainly in Arizona against the Maricopa
|
9 |
Medical Association case, even though that was a Section 1
|
10 |
case, the Supreme Court said fairly strongly that we don't
|
11 |
care about direction price level. What we care about is
|
12 |
the competitive process and making sure it works well.
|
13 |
So, there might be some legal standing for you
|
14 |
to argue that your proposal is more in keeping with modern
|
15 |
processor oriented thinking instead of the price oriented
|
16 |
thinking that polluted the predatory pricing process.
|
17 |
MR. FARRELL: I have something else to say, but
|
18 |
if you want to respond to that.
|
19 |
MR. EDLIN: Well, I wasn't going to respond to
|
20 |
that. I was going to respond to what he said previously,
|
21 |
which I suppose is not the rule as to how a conversation
|
22 |
goes.
|
23 |
But I think Joe is right that, to the extent we
|
24 |
can, we're certainly better off having an antitrust
|
25 |
jurisprudence that doesn't focus on things that we are not |
57
1 |
very apt to know, like costs.
|
2 |
And as to Howard's point, which is certainly
|
3 |
correct, that if price is close already to the competitive
|
4 |
equilibrium, then you shouldn't worry very much about what
|
5 |
happens no matter what. I agree with that. And one thing
|
6 |
that -- this gets to the last slide I had, which is, you
|
7 |
may want to only worry about firms thwarting rivals from
|
8 |
providing very substantial value increases to consumers,
|
9 |
and not worry about situations where they are only
|
10 |
providing minimal value increases. And if the prices are
|
11 |
already pretty close to the competitive level, then you
|
12 |
won't find rivals offering to provide very substantial
|
13 |
value increases to consumers, and so we won't find that
|
14 |
antitrust interferes very much in those circumstances.
|
15 |
But now you wanted to respond to what he just
|
16 |
said.
|
17 |
MR. FARRELL: Well, I wanted to say something
|
18 |
else about the role of costs in all of this.
|
19 |
There's no doubt that sacrifice tests and cost
|
20 |
tests can be illuminating concerning intent. And it's a
|
21 |
bit of a paradox, I think, or piquant at least, that
|
22 |
many of the same people who are very keen on sacrifice
|
23 |
tests are also the first ones to lay into any attempt to
|
24 |
use intent evidence in an antitrust case.
|
25 |
It seems to me that intent is what you can |
58
1 |
sometimes infer from sacrifice tests, and one needs to be
|
2 |
careful using intent evidence. Obviously there is the
|
3 |
pervasive problem of testosterone poisoned sales managers.
|
4 |
But thoughtful, high level intent may often be the best
|
5 |
available evidence as to contemporaneous estimates of
|
6 |
likely effects.
|
7 |
And so I don't think we should be either too
|
8 |
credulous or too rude about intent evidence. It's a kind
|
9 |
of evidence, and it seems to me it's the kind of evidence
|
10 |
that's most directly brought out by looking at sacrifice.
|
11 |
Let me say one other thing, though, about how
|
12 |
cost information might be useful.
|
13 |
If it's right, as I suggested at one point, that
|
14 |
you'd want to look at, in my hypothetical Northeast two
|
15 |
hundred dollar price, and in some sense try to gauge
|
16 |
whether that is where we've now got to, or whether it's
|
17 |
just a quick and short-lived fighting price that will
|
18 |
disappear as soon as the entrant has gone away and will be
|
19 |
back to five hundred, if that's an important question,
|
20 |
which it may well be, then it's perhaps somewhat
|
21 |
informative to look at Northeast's costs, because if two
|
22 |
hundred is below Northeast's cost, you might say, well,
|
23 |
that more or less rules out the possibility that it's now
|
24 |
the permanent price.
|
25 |
Of course, there's a lot of other evidence about |
59
1 |
what the permanent price must be, such as what actually
|
2 |
happened post exit versus what was happening pre-entry.
|
3 |
And so I certainly don't see that costs would play a
|
4 |
determinative role there, but it might be relevant to
|
5 |
thinking about that question.
|
6 |
MR. COHEN: Okay. I think we'll start things
|
7 |
off by building on some of Aaron's testimony.
|
8 |
I'll try the first question. Given the critique
|
9 |
that you supplied of some of the existing tests as to
|
10 |
whether conduct is exclusionary, what's your thinking as
|
11 |
to whether it's sensible to be looking for any single test
|
12 |
that captures all the elements of what we would want in
|
13 |
all the various situations to determine whether something
|
14 |
is exclusionary or not? Is this something that we could
|
15 |
hope for? Is this something beyond our ability?
|
16 |
MR. EDLIN: Well, I'd say it's always reasonable
|
17 |
to hope, and physicists will hope for the grand unified
|
18 |
theory and they may find it, and we should similarly hope
|
19 |
here.
|
20 |
Now, however, I think that what you should not
|
21 |
hope for is that you'll find the right unified test and it
|
22 |
will be easy to apply to the facts in any given
|
23 |
circumstance. Whatever test you think is right is going
|
24 |
to necessarily lead to huge factual disputes as to how the
|
25 |
test comes out under the circumstance. I think a lot of |
60
1 |
people are driven by a desire to get away from that
|
2 |
problem. And I think ultimately there are only two ways
|
3 |
to get away from that problem, and one is per se legality
|
4 |
and the other one is per se illegality, and both of them
|
5 |
are very convenient, but I think that both of them are the
|
6 |
wrong answer.
|
7 |
MR. COHEN: Anyone else?
|
8 |
Another way of trying to get at sort of the same
|
9 |
set of issues, I guess, do you have any principles in mind
|
10 |
that might help us determine areas in which any given test
|
11 |
is more likely to work in a given setting than another
|
12 |
setting? For example, are we more likely to have success
|
13 |
with one of these tests in any price or non-price context?
|
14 |
Are we more likely to have success with one of these tests
|
15 |
in a setting where the issue is tying up inputs rather
|
16 |
than settings which involve some of type of tortious
|
17 |
conduct? Are there generalities that might guide us?
|
18 |
MR. EDLIN: I think the main generality I would
|
19 |
have is that one is more likely to have success with the
|
20 |
test when it's seen from a sufficiency point of view than
|
21 |
from a necessity point of view. And it -- or viewed
|
22 |
differently, that these things are very -- can be very
|
23 |
helpful evidence, either, as Joe said of intent, or of
|
24 |
likely effect, which is to say, if you would not do it but
|
25 |
for substantial diminution in competition, well, that |
61
1 |
suggests substantial diminution in competition is likely.
|
2 |
So, the test can be very relevant from that
|
3 |
point of view. It's when you start to push the
|
4 |
implication sign the other way, which is what's been
|
5 |
happening, that I think there's real danger. And the
|
6 |
danger is across all of the categories that you listed.
|
7 |
MR. COHEN: I noticed when you went through some
|
8 |
of the variance of these tests, in a couple of the
|
9 |
instances, you included a temporal dimension. You
|
10 |
included short-term sacrifice for long-term profits.
|
11 |
Does anybody regard the short-term/long-term
|
12 |
distinction as something that's really needed here? Is it
|
13 |
just a sacrifice in general? And if short-term/long-term
|
14 |
matters, what are we talking about for time? Anybody want
|
15 |
to comment on those temporal formulations?
|
16 |
MR. FARRELL: Well, I'll make a perhaps slightly
|
17 |
rude comment. Usually when you don't know quite what
|
18 |
version of the test you mean, it's because you're not
|
19 |
really clear on the logic of why the test makes sense in
|
20 |
the first place.
|
21 |
So, I think, for example, if you're trying to
|
22 |
infer intent, then you'd want to ask yourself, all right,
|
23 |
what is it exactly that the argument here is saying and
|
24 |
what time scale you're looking over.
|
25 |
If you're wanting to say there's no possibility |
62
1 |
that this is a price you would charge in the long run,
|
2 |
that might tell you about something about what time scale
|
3 |
you're looking over.
|
4 |
So, I would go back to the underlying logic.
|
5 |
And if you don't know how to go back to the underlying
|
6 |
logic, that's a sign that there are deeper problems than
|
7 |
just not knowing for what time scale to evaluate things.
|
8 |
MR. MATELIS: This is a question about false
|
9 |
positives and false negatives, which you mentioned, Aaron,
|
10 |
and I'd be interested in all the panels' views.
|
11 |
I suppose a slightly more spirited defense of
|
12 |
the concept of false positives, which the Supreme Court
|
13 |
has mentioned in just about every Section 2 case in the
|
14 |
last twenty-five years, is that the competitive process is
|
15 |
likely to fix false positives, whereas false negatives
|
16 |
become ingrained in precedent and we're stuck with them
|
17 |
for many, many years, as we were for decades in predatory
|
18 |
pricing jurisprudence, where plaintiffs were winning cases
|
19 |
where today I think everyone would agree they might not.
|
20 |
Is this really a concern? Is the Supreme Court
|
21 |
wrong stressing the idea of false positives, or is the
|
22 |
concern overstated in general? How should this play a
|
23 |
role in devising antitrust policy?
|
24 |
MR. EDLIN: Well, I think you flipped the false
|
25 |
positives and false negatives there, so I'll try to answer |
63
1 |
the question as I think you intended.
|
2 |
MR. FARRELL: It's what statisticians know as
|
3 |
Type 3 error [laughter].
|
4 |
MR. EDLIN: So, as I see it, if you find what
|
5 |
you consider to be the right test, whether that is a final
|
6 |
results oriented test like efficiency or consumer welfare,
|
7 |
or whether it's a process type test such as the freedom to
|
8 |
trade that Joe and I are suggesting, I think the problem
|
9 |
of false positives is not so much one of legal precedents
|
10 |
but one of application, which is to say, if you've got the
|
11 |
right test, then the real fundamental problem is, in its
|
12 |
application you may get it wrong.
|
13 |
And the question is: Will people so fear that
|
14 |
when the test is applied to them that it will be gotten
|
15 |
wrong that they don't do many procompetitive things,
|
16 |
whether that's process or results interpreted.
|
17 |
And I think we are so far from such a situation
|
18 |
today that it just doesn't concern me very much. But if
|
19 |
we were in that situation, I again don't think the right
|
20 |
thing to do would be to say, well, let's find -- let's
|
21 |
apply something that substantively doesn't make much
|
22 |
sense. Rather, I think you should look at the source of
|
23 |
where the false positives are coming from. If they're
|
24 |
coming from bad jury instructions, make better jury
|
25 |
instructions. If they are coming from courts having an |
64
1 |
insignificant standard of proof where it seems sufficient
|
2 |
to allege that something bad happened rather than to
|
3 |
really prove it, then we should crank up the standard of
|
4 |
proof. And if -- and/or you say that you have to show
|
5 |
that something really very bad happened, rather than just
|
6 |
a little bad.
|
7 |
So, I see the problem of false positives as
|
8 |
being less in the precedents than in the applications of
|
9 |
the facts.
|
10 |
MR. SHELANSKI: I agree with Aaron. I would
|
11 |
just add that I think a lot of rules look bad from a false
|
12 |
positive standpoint. They look worse from the false
|
13 |
positive standpoint at the beginning when the rule is
|
14 |
articulated, then after there has been experience gained
|
15 |
in its application.
|
16 |
I think that, as an agency gains familiarity
|
17 |
with the application of a rule, understanding of what
|
18 |
certain fact patterns really mean, as courts get more
|
19 |
experiences with reviewing cases and get a body of
|
20 |
precedence and a body of jury instructions, some of the
|
21 |
more frightening aspects of the rule may be damped down
|
22 |
and you may get beneficial application.
|
23 |
I do think there's a difference with respect to
|
24 |
false positives between public enforcement and private
|
25 |
enforcement under Section 2. I have a lot of faith in the |
65
1 |
agencies' abilities to gain a body of knowledge and
|
2 |
understanding that they then bring to bear in their
|
3 |
enforcement discretion under any given rules.
|
4 |
I think with the courts, where there's a perhaps
|
5 |
much less coherent body of learning, you have to rely on
|
6 |
any particular district judge's reading perhaps outside of
|
7 |
its own circuit and perhaps outside of its own circuit,
|
8 |
and rely on a cohesive body of understanding. And this is
|
9 |
not -- I am not trying to bash the capability of judges.
|
10 |
I'm trying to just suggest you may get a less coherent
|
11 |
development of a body of precedence and knowledge in the
|
12 |
judiciaries than you get in the agencies.
|
13 |
So, I think false positive may be worse for
|
14 |
private enforcement than for public enforcement. But, on
|
15 |
the whole, I would agree with Aaron, I think the
|
16 |
application is the key issue. The deterrence effect is
|
17 |
probably overemphasized in a lot of what one reads, and I
|
18 |
think it can be offset in light of experience.
|
19 |
MR. MATELIS: Anything to add, Joe?
|
20 |
MR. FARRELL: No. I'll reserve my time.
|
21 |
MR. MATELIS: Okay. Again this is a general
|
22 |
question based off of something Aaron has mentioned twice
|
23 |
now.
|
24 |
What are better jury instructions that we should
|
25 |
be giving juries in Section 2 cases? This might be |
66
1 |
another way of saying, if we don't want to instruct them
|
2 |
on the no economic sense test, on what should we be
|
3 |
instructing them?
|
4 |
MR. EDLIN: Well, I think that the two best --
|
5 |
the two best candidates that I think we should be
|
6 |
instructing them clearly on, whatever we think the right
|
7 |
test is, and the two best candidates that I have are a
|
8 |
results oriented test, which is consumer welfare, or a
|
9 |
process oriented test, which is that someone is being
|
10 |
blocked from providing higher value to consumers, which is
|
11 |
a process oriented test.
|
12 |
And the instruction should of course distinguish
|
13 |
all of the standard worries that people have, such as that
|
14 |
it's not sufficient that rivals are losing money, and
|
15 |
that's not the issue.
|
16 |
What I'm really getting at there is, if you
|
17 |
really -- I think the first thing before suggesting
|
18 |
approving jury instructions is to come to a clear
|
19 |
understanding of what antitrust is trying to accomplish.
|
20 |
The second thing is to see if there really are a
|
21 |
lot of false positives, and I don't see them. Right now I
|
22 |
would say the improvement to jury instructions would be to
|
23 |
not focus on tests that I think are nonsensical, which is
|
24 |
the primary problem with them now.
|
25 |
MR. MATELIS: Howard, Joe? |
67
1 |
MR. FARRELL: Well, in the unlikely event that I
|
2 |
ever end up on an antitrust trust jury, I guess what I
|
3 |
would want to hear is: The following specific questions
|
4 |
have been given some prominence, but you the jury should
|
5 |
please interpret them to the extent possible in light of
|
6 |
the kind of fundamental things that Aaron was mentioning.
|
7 |
MR. COHEN: Okay, let's turn a few questions to
|
8 |
Joe's presentation.
|
9 |
I really started with three questions, but as I
|
10 |
think about it more, they come together into one. I'll
|
11 |
throw it out in various forms.
|
12 |
You talked some time early on about whether the
|
13 |
results of not being able to successfully form a blocking
|
14 |
coalition results from actions of the five hundred dollar
|
15 |
airline, whether it happened intentionally or not, I think
|
16 |
you said at one point, or another time you phrased it,
|
17 |
whether it's a natural outcome of the way the market
|
18 |
worked.
|
19 |
But then your rule you were trying to focus on
|
20 |
where there's really a problem, you talked about whether
|
21 |
the incumbent, the five hundred dollar incumbent,
|
22 |
strategically thwarts the coalition.
|
23 |
I'm going to ask you to try to give us some
|
24 |
content about what you mean about "strategically thwarts."
|
25 |
And maybe you can think about it in terms of a question of |
68
1 |
whether this approach would make it unlawful for a low
|
2 |
cost producer merely to develop the reputation as an
|
3 |
aggressive price competitor.
|
4 |
Sort of a third way of asking the same question:
|
5 |
What's happened to the bad conduct element of Section 2 in
|
6 |
this core analysis?
|
7 |
MR. FARRELL: Well, so first off, as I
|
8 |
understand it, where we're surrounded by lawyers here, I
|
9 |
don't think there is a bad conduct. There's an
|
10 |
anticompetitive component, anticompetitive conduct.
|
11 |
And if you accept the ideas that are being put
|
12 |
forward about what anticompetitiveness means, then there
|
13 |
can be conduct that is anticompetitive that is harmful for
|
14 |
competition that isn't necessarily bad in any sense other
|
15 |
than being harmful to competition.
|
16 |
Now, there certainly has been a body of thought
|
17 |
and especially shorthand that says you want it to be bad
|
18 |
as well in some other way. That I think -- I try to
|
19 |
interpret that in the following way. Let's suppose that
|
20 |
in the course of trial, imagine it takes place in this
|
21 |
order although it wouldn't have to, it's been shown that
|
22 |
the defendant did some things that harmed consumers by
|
23 |
excluding competition and were not, let's say, highly
|
24 |
efficient. And I'm pulling together ideas of various
|
25 |
sources here, I think. |
69
1 |
And now we ask, well, was it bad conduct? Well,
|
2 |
from an economist's point of view, it seems as if in the
|
3 |
instance it has just been shown to be bad conduct. So,
|
4 |
the question is what further requirement is being asked
|
5 |
for here.
|
6 |
I think the further requirement that's being
|
7 |
asked for here is the following: That this conduct -- if
|
8 |
this conduct is condemned, it will have some sort of
|
9 |
deterrent effect on conduct that sounds like this when
|
10 |
described. And that deterrent effect will extend of
|
11 |
course to other places where the competitive implications
|
12 |
of the conduct might be a little bit different.
|
13 |
And so what you want in addition to finding this
|
14 |
conduct was inefficiently anticompetitive and
|
15 |
anti-consumer here, you want some degree of confidence
|
16 |
that similar-sounding conduct is going to tend to be not
|
17 |
such a good thing or a bad thing, in other circumstances
|
18 |
where maybe it won't be inefficiently anti-consumer,
|
19 |
anticompetitive.
|
20 |
Well, that puts a lot of weight on the
|
21 |
psychological or even philosophical concept of conduct
|
22 |
that sounds like this. There's a philosopher named I
|
23 |
believe Grice, who really tested foundations of that kind
|
24 |
of thing by inventing a word, grue, g-r-u-e, which means
|
25 |
green up until this morning or blue after this morning. |
70
1 |
And so all of your past observations that trees are green
|
2 |
are also observations that trees are grue. What do you
|
3 |
predict the tree color will be this afternoon.
|
4 |
Obviously that's playing with words in the way
|
5 |
that philosophers love to do, but it does suffice to make
|
6 |
the point that, if what you are looking for in a, quote,
|
7 |
"bad conduct" problem is something along the lines of
|
8 |
similar conduct that is going to be bad in other
|
9 |
circumstances, you need a concept of what's similar. And
|
10 |
that's not really an economic concept, as far as I can
|
11 |
tell. It's some sort of intuitive or possibly legal
|
12 |
concept.
|
13 |
MR. COHEN: Anyone else?
|
14 |
I'll shift ahead because your comments invite
|
15 |
this.
|
16 |
What kind of difficulties would you expect
|
17 |
courts have in operationalizing something like this? I
|
18 |
would hate to go in and try to tell them that trees are
|
19 |
green in the morning but blue later.
|
20 |
MR. FARRELL: Well, just to be clear, at least
|
21 |
in my own mind, I would be delighted if judges were to
|
22 |
listen, and when we get around to writing, read this kind
|
23 |
of stuff. But I am not convinced that it's ready for
|
24 |
courts yet.
|
25 |
What I think I would like courts to do is put up |
71
1 |
a lot of resistance to the incorrect tests that are being
|
2 |
bandied about on the pretext of administrability, bright
|
3 |
line, sort of vaguely right, perhaps, maybe, although we
|
4 |
can't exactly tell you why.
|
5 |
And I would like to see courts, led by the
|
6 |
Supreme Court, say, look, we really have not sorted out
|
7 |
yet what administrable concrete tests we need to apply for
|
8 |
Section 2 liability. For the time being let's do
|
9 |
so-and-so, but that's not meant to be the final answer.
|
10 |
Because I think it's pretty clear that nobody is
|
11 |
in a position to say yet what the final answer should be.
|
12 |
And I think there's a huge danger, given the way courts
|
13 |
and lawyers tend to think and talk, that things are going
|
14 |
to congeal prematurely.
|
15 |
MR. COHEN: I'm wondering if you're at a point
|
16 |
yet where you could predict if there are particular types
|
17 |
of conduct where the analysis you're thinking of is really
|
18 |
likely to lead to different results than you've been
|
19 |
getting through viewing perfect competition as the goal?
|
20 |
You may go through a different process. Do you have any
|
21 |
idea where the results are likely to come up?
|
22 |
MR. FARRELL: No. I think the salient
|
23 |
differences are going to be based on the question of how
|
24 |
closely you try to examine direct efficiency consequences
|
25 |
versus trusting the competitive process to do that and not |
72
1 |
requiring it in the narrow instance.
|
2 |
You know, technically if there is a perfectly
|
3 |
competitive equilibrium in an economy, it is then in the
|
4 |
core. And so I don't think there is a substantive
|
5 |
tension between the two. I think it's more a question of
|
6 |
what process each one suggests to you.
|
7 |
It seems to me the core -- and let me stress,
|
8 |
I'm not suggesting ever examining an outcome to see
|
9 |
whether it is in the core. I'm suggesting the process
|
10 |
that is suggested by that, which is, make it relatively
|
11 |
easy, or don't allow it to be made artificially difficult
|
12 |
to form blocking coalitions.
|
13 |
Whether there is a similar process that is
|
14 |
suggested by thinking about perfect competition, I am not
|
15 |
quite so sure. You know, economists have talked for a
|
16 |
long time about the fact that perfect competition is
|
17 |
describable as an outcome, and we don't have a very good
|
18 |
story about how you get there. There's the infamous
|
19 |
Walrasian auctioneer. That's obviously not a process that
|
20 |
takes place in reality, let alone is protectable by
|
21 |
antitrust.
|
22 |
It seems to me that thinking about the coalition
|
23 |
formation model gives you a stronger suggestion about what
|
24 |
process to protect than thinking about perfect
|
25 |
competition. |
73
1 |
MR. EDLIN: I'll hazard a guess, which is, if
|
2 |
you thought about things a little more the way that Joe
|
3 |
and I think about things, then you would find that the
|
4 |
Department of Justice would probably have won the American
|
5 |
Airlines case; that entry would be easier in many
|
6 |
industries because monopoly or dominant firms would have
|
7 |
more limited ability to thwart entry; more attempts by
|
8 |
monopolies to prevent entry by tying goods together would
|
9 |
be illegal, but not all; and those would be the kinds of
|
10 |
things that you would see in terms of substantive outcome
|
11 |
differences.
|
12 |
MR. SHELANSKI: I will just add that I think the
|
13 |
process emphasis, while extremely important theoretically
|
14 |
and at some level is absolutely correct economically does
|
15 |
have some pragmatic difficulties.
|
16 |
I actually really worry about instructing juries
|
17 |
on the process as opposed to outcomes. And you can
|
18 |
combine the two to halve their inquiry, but I think the
|
19 |
confusion between competition and competitor is one very
|
20 |
easily sown in juries.
|
21 |
And connected to your question earlier about
|
22 |
false positives, I think that as a firm, faced
|
23 |
particularly with a private suit, knowing the instruction
|
24 |
is going to the jury about process, you're worried about
|
25 |
looking aggressive, worried about looking the bad guy, and |
74
1 |
you get a lot of hidden false positives through
|
2 |
settlement, particularly in the private cases.
|
3 |
So, I do think it's worth thinking a lot more
|
4 |
about the pragmatic implications of the process
|
5 |
instruction of going forward.
|
6 |
MR. COHEN: Finally, for Joe.
|
7 |
The theory that you've explained depends on the
|
8 |
formation of these blocking coalitions. There are
|
9 |
obviously impediments to this. You recognize them and
|
10 |
they may not always be formed, but at least there's an
|
11 |
incentive to do them.
|
12 |
Have you thought about how we should take into
|
13 |
account the fact that not all of these coalitions will
|
14 |
ever form in the first place, that there maybe information
|
15 |
problems or the cost that prevents them from happening?
|
16 |
How do we bridge from incentive to actual assumption that
|
17 |
they're there and therefore that their losses are
|
18 |
significant?
|
19 |
MR. FARRELL: I don't. I mean, I think, as I
|
20 |
think I mentioned, the way you prove that a competitive --
|
21 |
that everything in the core is Pareto efficient, is by
|
22 |
pointing to the so-called grand coalition of everybody, if
|
23 |
it was prey to inefficient, then in theory this grand
|
24 |
coalition could block. That's obviously not going to
|
25 |
happen. |
75
1 |
So, I think any policy, including antitrust, is
|
2 |
not going to be able to get us all the way to Pareto
|
3 |
efficiency, whether it thinks of it in terms of central
|
4 |
planning, price-taking equilibrium or the core.
|
5 |
Now, as related more directly on a practical
|
6 |
point, which is, well, what happens if -- this is I think
|
7 |
maybe what you were getting at with the bad act question.
|
8 |
What happens if we have a not very good outcome in the
|
9 |
status quo and the blocking coalition that, quote, ought
|
10 |
unquote, to form doesn't form, not because of anything
|
11 |
that the incumbent does, but just because it's really hard
|
12 |
to form.
|
13 |
Well, I think at some level that could be a
|
14 |
competition policy question. There might be changes that
|
15 |
could be made in the way the market works to make it more
|
16 |
likely that such coalitions would form.
|
17 |
If it were a competition policy question, it
|
18 |
wouldn't necessarily be an antitrust question. I think
|
19 |
they're potentially distinct areas. And it might be
|
20 |
neither. It might just be, well, that's too bad, that's
|
21 |
one of the imperfections of the world.
|
22 |
MR. MATELIS: At the beginning of these
|
23 |
hearings, both the Assistant Attorney General and the
|
24 |
Chairman of the FTC stressed the importance of safe
|
25 |
harbors for guiding businesses that are seeking to comply |
76
1 |
with the antitrust laws.
|
2 |
And, Joe, I have a question for you. The
|
3 |
examples in your presentation were responses of a firm to
|
4 |
new entry. Northeast's response to Sprite's entry and the
|
5 |
A and B product potential responses at the new entry.
|
6 |
Are there responses to new entry that, you know,
|
7 |
looking at things through the core, should be within a
|
8 |
safe harbor and something that firms should always feel
|
9 |
comfortable doing?
|
10 |
MR. FARRELL: Well, I'm sure there are, but just
|
11 |
as I don't know exactly what the right rules for
|
12 |
liabilities should be in a practical sense here, I also
|
13 |
don't know what the right rules for safe harbor should be.
|
14 |
I mean, one can give the following answer, which
|
15 |
is sort of in the spirit of something Tim Bresnahan has
|
16 |
said, and you will be hearing from him this afternoon,
|
17 |
that the safe harbor is to make your money by being nice
|
18 |
to consumers, not to make your money by being the other
|
19 |
stuff you can be. That's not quite the way Tim put it,
|
20 |
but he had a somewhat similar line which maybe you can get
|
21 |
out of him if you ask him.
|
22 |
MR. COHEN: Directing some questions to Howard
|
23 |
Shelanski's presentation.
|
24 |
You focused very much on intellectual property,
|
25 |
the effects of possible changes in that area, bleeding |
77
1 |
over into how we might look at Section 2 issues.
|
2 |
If we're looking at Section 2 issues, we're not
|
3 |
likely to have differential treatment of instances in
|
4 |
which there are lateral refusals for intellectual
|
5 |
properties versus others.
|
6 |
Would your rule somehow -- are you envisioning
|
7 |
somehow distinguishing between the two, or just a one size
|
8 |
fits all modification?
|
9 |
MR. SHELANSKI: One size fits all is what I'm
|
10 |
looking at. I'm actually not so much proposing a
|
11 |
particular rule, because I agree with you there should not
|
12 |
be two rules. Obviously the precedent is a little choppy
|
13 |
between the various circuit courts on the extent to which
|
14 |
you get special Section 2 protections for intellectual
|
15 |
property.
|
16 |
But my view is you should not have a separate
|
17 |
rule. And I was really looking at the macro level. If
|
18 |
you take the total pool of goods that firms refuse to deal
|
19 |
with, some of them are going to impose barriers because
|
20 |
they're legally protected, legally blocked by IP.
|
21 |
The smaller the pool of goods where there's an
|
22 |
anticompetitive refusal to deal, the less enforcement
|
23 |
minded you want to be against refusals to deal.
|
24 |
So, for me it's really an adjustment mechanism
|
25 |
about how permissive or strict a unitary rule you apply. |
78
1 |
I mean, if you were to look and see, boy, a lot of these
|
2 |
refusals to deal cases have at their core intellectual
|
3 |
property. Then I think intellectual property would not,
|
4 |
say, have a different rule for those cases versus others,
|
5 |
but it would say we can have a more permissive rule
|
6 |
towards refusals if we had intellectual property
|
7 |
enforcement.
|
8 |
MR. COHEN: One thing that you mentioned a
|
9 |
number of times in your talk was issues about the degree
|
10 |
to which imposing liability or not imposing liability for
|
11 |
refusals to deal might affect innovation, might affect
|
12 |
efforts invent around whatever problem there is.
|
13 |
It's a little unfair, I know you gave a
|
14 |
theoretical presentation, but of course we're very
|
15 |
interested in anything empirical.
|
16 |
Do you have any -- can you give any summary or
|
17 |
are there any indications of what there is out there in
|
18 |
the way of empirical evidence on this?
|
19 |
MR. SHELANSKI: If I can cheat a little bit, I
|
20 |
think I can. So, I did raise that issue of demand side
|
21 |
innovation and competitive supply because I feel that in
|
22 |
the discussion about duties to deals there's been
|
23 |
overemphasis on deterring the initial innovation by the
|
24 |
supplier. I think that's extremely important. And I
|
25 |
wouldn't want to see a situation where we punished |
79
1 |
innovation per se. So, I want to be very careful. But I
|
2 |
wanted to build into the demand side there's innovation on
|
3 |
both sides of the enforcement question.
|
4 |
So, here's a possible place to look for some
|
5 |
empirical support, and this is contentious. I would go to
|
6 |
the regulatory arena and I would look at the unbundling
|
7 |
obligations of the Telecommunications Act of 1996.
|
8 |
There are allegations that overly permissive
|
9 |
access for competitors to incumbent networks reduced the
|
10 |
degree to which these new entrants built their own
|
11 |
facilities and their own networks, therefore leading to
|
12 |
less vigorous competitive entry.
|
13 |
I think there's a lot of debate over the extent
|
14 |
to which this is true, but there is some empirical
|
15 |
evidence that after the FCC repealed a very permissive
|
16 |
access to the incumbent platform under what some would
|
17 |
argue were subsidized rates -- there is a legitimate
|
18 |
dispute over that -- that after they repealed that access,
|
19 |
there was a lot more facilities-based entry, a lot more
|
20 |
actual building and installment of competitive facilities.
|
21 |
This does suggest that a duty to deal, which
|
22 |
would then include some kinds of terms of dealing, runs
|
23 |
the risk of stopping entry of competitive assets into
|
24 |
other markets. And the telecommunications market might be
|
25 |
one place to look for such evidence. And there is some |
80
1 |
literature out there with competing arguments about
|
2 |
whether the essential facilities treatment or the duty to
|
3 |
deal imposed by the Telecommunications Act of 1996 on
|
4 |
incumbent networks deterred and chased out new competitive
|
5 |
essence.
|
6 |
MR. FARRELL: I think part of the reason why
|
7 |
people have focused on incentives of the original
|
8 |
invention or the original investment is that, of course,
|
9 |
that innovation or investment directly leads to social
|
10 |
benefits.
|
11 |
Duplicative investment is -- I want to avoid
|
12 |
taking too narrow a view here, but nevertheless, at some
|
13 |
level duplicative investment is wasteful. And while
|
14 |
having some of it may well be part of the process and
|
15 |
negotiating for voluntary access in the shadow of the
|
16 |
threat when you look at the investment is probably a
|
17 |
bigger part of the process, I think it's actually wrong to
|
18 |
treat reducing the incentive for duplicative investment as
|
19 |
a policy downside in itself.
|
20 |
Now, it might actually be a kind of shorthand or
|
21 |
a proxy for some other harms that you think come out of
|
22 |
more mandated sharing than other policies would give you.
|
23 |
But I think one wants to be wary of that shorthand.
|
24 |
MR. SHELANSKI: I'll disagree slightly. I think
|
25 |
you're right that that's something to be taken into |
81
1 |
account.
|
2 |
I think the market conditions under which that
|
3 |
duplicative entry would be welfare decreasing are fairly
|
4 |
specialized. I don't know how common they are. I think
|
5 |
it needs to be taken into account. But while it's a
|
6 |
consideration, I am not sure that it's a big enough
|
7 |
problem that I would discount -- I certainly wouldn't
|
8 |
discount the value of at least some competitive investment
|
9 |
or duplicative investment, especially where it's not
|
10 |
economically blocked. There's not some kind of natural
|
11 |
monopoly or scale kind of argument that would make that
|
12 |
investment a not be beneficial end, but where there's
|
13 |
simply a legal barrier to producing something that could
|
14 |
be produced fairly cheaply. Software would be an example.
|
15 |
MR. COHEN: Just one more. I'm going to return
|
16 |
to something that Joe just mentioned a couple answers ago.
|
17 |
You drew the distinction in a sense between a
|
18 |
competition issue and an antitrust issue. Another way of
|
19 |
phrasing some of the same points we've already been going
|
20 |
over.
|
21 |
To the panel just generally: Do you see a
|
22 |
difference in your analysis between a competition issue in
|
23 |
the sense of maximizing efficiency, and an antitrust issue
|
24 |
in the sense of what should be a legal violation?
|
25 |
MR. FARRELL: I'm certainly very open to that, I |
82
1 |
think. First of all, I would not phrase a competition
|
2 |
issue quite as maximizing efficiency, for all the reasons
|
3 |
we spent all morning talking about.
|
4 |
But I think it's perfectly possible for a
|
5 |
competition agency, let's say, to discover that
|
6 |
such-and-such a market would work a lot more competitively
|
7 |
with these ground rules than with those ground rules. And
|
8 |
to try to use its influence, perhaps even its legal
|
9 |
authority, to have the better rules rather than the less
|
10 |
good rules apply.
|
11 |
And that doesn't necessarily involve anybody
|
12 |
having, quote, done anything wrong. And so I think
|
13 |
there's potentially a difference between competition would
|
14 |
work better in such-and-such a way than with the status
|
15 |
quo, and saying so-and-so has committed an antitrust
|
16 |
offense.
|
17 |
So, yes, I think there's probably a big area
|
18 |
there, actually.
|
19 |
MR. COHEN: Okay. Do any of the panelist have
|
20 |
any final points they want to make?
|
21 |
MR. EDLIN: I'm in favor of lunch.
|
22 |
MR. COHEN: Okay, we vote for lunch here.
|
23 |
I again want to thank all of our panelists for
|
24 |
their thoughtful and insightful remarks. I ask the
|
25 |
audience to please join me in a round of applause for our |
83
1 |
speakers.
|
2 |
(Applause.)
|
3 |
MR. COHEN: And our afternoon session will begin
|
4 |
promptly at 1:30.
|
5 |
(Whereupon, at 11:59 a.m., a lunch recess was
|
6 |
taken.)
|
7 |
|
8 |
|
9 |
|
10 |
|
11 |
|
12 |
|
13 |
|
14 |
|
15 |
|
16 |
|
17 |
|
18 |
|
19 |
|
20 |
|
21 |
|
22 |
|
23 |
|
24 |
|
25 |
|
84
1 |
AFTERNOON SESSION
|
2 |
(1:30 P.M.)
|
3 |
MS. GRIMM: Good afternoon. I would like to
|
4 |
welcome everyone to our afternoon session. And I'm glad
|
5 |
that you all could be with us today.
|
6 |
I am Karen Grimm. I am Assistant General
|
7 |
Counsel for Policy Studies at the Federal Trade
|
8 |
Commission. I am going to be moderating the session this
|
9 |
afternoon, along with June Lee, who is an economist at the
|
10 |
Antitrust Division of the U.S. Department of Justice.
|
11 |
Before we start, I would like to just go through
|
12 |
two housekeeping details. First of all, as a courtesy to
|
13 |
our speakers, please turn off all your cell phones,
|
14 |
Blackberries, and other devices
|
15 |
And, secondly, because these are hearings, we
|
16 |
request that the audience not make any comments or ask any
|
17 |
questions during the presentation.
|
18 |
This afternoon we are honored to have another
|
19 |
group of distinguished economists from the University of
|
20 |
California at Berkeley and Stanford University to offer
|
21 |
their testimony in these series of Section 2 hearings.
|
22 |
Our afternoon panelists, like those this
|
23 |
morning, will provide their perspectives on various issues
|
24 |
related to the complex area of Section 2 jurisprudence and
|
25 |
enforcement. |
85
1 |
Our panelists this afternoon are Timothy
|
2 |
Bresnahan, who is the Landau Professor of Technology and
|
3 |
the Economy in the economics department at Stanford
|
4 |
University; Richard Gilbert, who is a professor of
|
5 |
economics at the University of California Berkeley and the
|
6 |
chair of the Berkeley Competition Policy Center; Daniel
|
7 |
Rubinfeld, who is the Robert L. Bridges Professor of Law
|
8 |
and Professor of Economics at the University of California
|
9 |
Berkeley; and Carl Shapiro, who is the TransAmerica
|
10 |
Professor of Business Strategy and Professor of Economics
|
11 |
and the Director of the Institute of Business and Economic
|
12 |
Research at the University of California Berkeley.
|
13 |
Our first three panelists will make
|
14 |
presentations, and Professor Shapiro will be participating
|
15 |
in the discussion with his fellow panelists.
|
16 |
Our format this afternoon is as follows: Each
|
17 |
speaker will make a 20 to 30 minute presentation. After
|
18 |
all the presentations have been completed, we will take
|
19 |
about a 15 minute break. And after that break we will
|
20 |
reconvene for a round-table discussion. We are scheduled
|
21 |
to conclude this session about 4:30.
|
22 |
I would like to thank all of you for being with
|
23 |
us here today. I want to thank all of our panelists for
|
24 |
coming and for their participation. We very much
|
25 |
appreciate the time and effort all of them have put into |
86
1 |
preparing their presentations and their willingness to
|
2 |
share their insights with us.
|
3 |
I would now like to turn the podium over to my
|
4 |
DOJ colleague and co-moderator, June Lee, for any remarks
|
5 |
she would like to make
|
6 |
Ms. Lee: The Antitrust Division of the
|
7 |
Department of Justice is pleased to co-sponsor today's
|
8 |
single-firm conduct hearing. As noted by Joe Matelis this
|
9 |
morning, five of today's panelists were Deputy Assistant
|
10 |
Attorneys General in the Antitrust Division. Four of the
|
11 |
five are in the panel. I thank them for participating
|
12 |
and, like Karen, for sharing their insights. I look
|
13 |
forward to their presentations in what I'm sure will be a
|
14 |
lively discussion.
|
15 |
I join Joe in thanking the Competition Policy
|
16 |
Center and the Berkeley Center For Law And Technology at
|
17 |
the University of California Berkeley for hosting these
|
18 |
hearings. And I thank Karen and her colleagues at the FTC
|
19 |
for their work in organizing today's hearing and
|
20 |
assembling the august panel we have today.
|
21 |
Karen.
|
22 |
MS. GRIMM: Our first speaker this afternoon is
|
23 |
Timothy Bresnahan, who is Landau Professor of Technology
|
24 |
and the Economy at Stanford University and Chair of the
|
25 |
department of economics. |
87
1 |
He is Director of the Center for Research in
|
2 |
Employment and Economic Growth in the Stanford Institute
|
3 |
for Economic Policy Research. He also has served as Chief
|
4 |
Economist of the Antitrust Division of the U.S. Department
|
5 |
of Justice.
|
6 |
His research interests lie in the economic of
|
7 |
industry, especially of high technology industry.
|
8 |
Professor Bresnahan received his B.A. from
|
9 |
Haverford College and his master's degree and Ph.D. in
|
10 |
economics from Princeton University.
|
11 |
Tim.
|
12 |
MR. BRESNAHAN: Thanks for that very nice
|
13 |
introduction. Let me see if I can find my slides.
|
14 |
While I'm finding my slides, let me confess that
|
15 |
in my role as department chair, I worked with the agencies
|
16 |
in a failed effort to bring these hearings to Stanford
|
17 |
rather than Berkeley. If you think of the reputations of
|
18 |
those two great universities, you might infer that signals
|
19 |
a leftward shift in the antitrust enforcement effort.
|
20 |
But I don't think that's what it signals. If
|
21 |
you either look over here to my left or at the brochure
|
22 |
from the Competition Policy Center, you can see why
|
23 |
Berkeley is an enormous center of academic influence in
|
24 |
this area. This was the right place to put it.
|
25 |
I want to talk about monopolization (Section 2) |
88
1 |
cases. And my real agenda is to normalize them, to
|
2 |
regularize them within antitrust analysis.
|
3 |
We have a tendency in talking about Section 2
|
4 |
matters to immediately leap to the most difficult part,
|
5 |
which is the part that's about alternative efficiency
|
6 |
theories of whatever business practice it is that's
|
7 |
challenged in the Section 2 matter.
|
8 |
I think that that makes Section 2 matters more
|
9 |
difficult than they need to be, and I'm going to propose a
|
10 |
different approach, not inconsistent with what we've
|
11 |
done in the past, and which we'll see in a minute, not
|
12 |
inconsistent with recent court decisions.
|
13 |
I'm going to suggest a different approach where
|
14 |
we look at competitive effects first. It's not very
|
15 |
surprising that I want to look at competitive effects
|
16 |
first since I'm an economist.
|
17 |
And then I think I'm going to argue it's going
|
18 |
to make thinking about whether a Section 2 case is
|
19 |
procompetitive much easier than starting from that very
|
20 |
difficult question of whether the challenged practices are
|
21 |
an act of competing rather than anticompetitive act. So,
|
22 |
I'm going to start with competitive effects.
|
23 |
There's been a good bit of action in the courts
|
24 |
in Section 2 lately. You know, I see three big topics
|
25 |
here: boundaries with other parts of the law, notably with |
89
1 |
patent and copyright law; predatory pricing is another area.
|
2 |
What I want to talk about are bundling and
|
3 |
related practices. So, vertical Section 2 cases where a
|
4 |
monopolist commits monopolization or is alleged to
|
5 |
commit monopolization through bundling its monopoly
|
6 |
product with something else, or through contractual
|
7 |
restrictions that amount to de facto bundling.
|
8 |
I'm also going to talk about Microsoft and
|
9 |
Dentsply in some detail, but Dentsply first. This partly
|
10 |
reflects the idea that I think that the folks who do
|
11 |
judicial decisions have the same economics in mind that
|
12 |
I'm going to talk about this afternoon. And it's partly
|
13 |
that there are three cases, two recent cases in this area.
|
14 |
And I found those two, again another confession, much
|
15 |
easier to read than I found the LePage's case, which was a
|
16 |
struggle for me, although I am not sure it's inconsistent
|
17 |
with what I'm going to say.
|
18 |
So, any Section 2 inquiry I think has at its
|
19 |
heart an economic structure if it's a rule of reason
|
20 |
inquiry. Any rule of reasoning inquiry has economics in
|
21 |
it. I think the economics enters at two distinct places.
|
22 |
It has to enter in market power. You need economists to
|
23 |
figure out market power. And I want to say, as I've been
|
24 |
saying for a quarter of a century, the thing about market
|
25 |
power is sometimes it's a useful shortcut in antitrust |
90
1 |
enforcement. We should be thinking about competitive
|
2 |
effects when we're thinking about market power, particularly
|
3 |
I would encourage the agencies, when picking cases in the
|
4 |
merger area or in the Section 2 area, to pick cases where
|
5 |
there's potentially a substantial change in the conditions
|
6 |
of competition in the market and significant impact on the
|
7 |
economy. That's not the same as market power. That's a
|
8 |
change in market power.
|
9 |
The other place where economics matters is in
|
10 |
thinking about the causal flow from the acts which are
|
11 |
alleged to be anticompetitive in a Section 2 case to the
|
12 |
changes in market power. And I'm going to argue, this is
|
13 |
my theme for the afternoon, you can gain a lot of clarity
|
14 |
about a Section 2 case by bringing the competitive effects
|
15 |
and causation arguments to the forefront. And I think that's
|
16 |
consistent with the three bundling cases I cited, bundling
|
17 |
or tying cases, I cited on the previous page.
|
18 |
Section 2 cases are never going to be easy.
|
19 |
Let's be real. There's a reason for that. This is I
|
20 |
think the hardest part. Almost all conduct which would be
|
21 |
exclusionary in some context would be an ordinary and
|
22 |
competitive business practice in some other industry. So,
|
23 |
it's necessarily context specific. That makes it
|
24 |
difficult I think for attorneys to get their heads around
|
25 |
Section 2 matters all the time because it seems like |
91
1 |
there's a fairly unstructured rule of reason analysis in a
|
2 |
Section 2 case.
|
3 |
I'm going to argue again that monopolization can
|
4 |
lead you to a fairly structured economic competitive
|
5 |
effects decision. Let me do that right away. I'll do it
|
6 |
in Dentsply first.
|
7 |
This is a Department of Justice case. I know a
|
8 |
part of the history of it. I believe it was brought when
|
9 |
Dan Rubinfeld was Chief Economist. It was litigated when
|
10 |
I was Chief Economist. And I just learned from Professor
|
11 |
Shapiro that it was under investigation on his watch.
|
12 |
MR. GILBERT: It was under investigation at the
|
13 |
FTC before I was at DOJ.
|
14 |
MR. BRESNAHAN: Exactly. We are lucky that
|
15 |
prefabricated artificial teeth is not a market which
|
16 |
changed quite so quickly as computer software. But I note
|
17 |
that the other case I am going to talk about, Microsoft,
|
18 |
has a similarly long, long series of investigations before
|
19 |
there was a serious enforcement action.
|
20 |
So, what's the story of Dentsply? Why did the
|
21 |
Department of Justice bring a Section 2 action?
|
22 |
So, part of it, there is a market definition,
|
23 |
there is monopoly power, and there is, in the current
|
24 |
market, a monopoly in prefabricated artificial teeth.
|
25 |
There are some small sellers, but there is one great big |
92
1 |
seller named Dentsply.
|
2 |
Now, here's the competitive effects part. And
|
3 |
this is something I think that's a reason that's going to
|
4 |
make cases fairly rare in monopolization. While there is
|
5 |
a monopoly in prefabricated artificial teeth, there could
|
6 |
be substantially more competition in the market from a
|
7 |
number of non-Dentsply like artificial teeth -- prefab
|
8 |
artificial teeth providers who are smaller, very small at
|
9 |
the time the case was brought, and typically lower priced.
|
10 |
And it's the difference between the competitive
|
11 |
regime there is, monopoly, and the competitive regime
|
12 |
there could be, much less monopoly, which is the
|
13 |
competitive effect that I think we should bring to the
|
14 |
forefront.
|
15 |
If it's inevitable, if Dentsply has a
|
16 |
monopoly that cannot be changed, if there is some barrier to
|
17 |
entry which cannot be lowered by any earthly force, there
|
18 |
can still be a monopoly but how can there be
|
19 |
monopolization? Monopolization I think needs to be
|
20 |
cause of a change in the competitive regime or prevention of
|
21 |
the change to a competitive regime that otherwise might arise.
|
22 |
Now, in this case, the mechanism, you need a bad
|
23 |
act as well as a competitive effect to have a Section 2
|
24 |
case. The mechanism by which Dentsply prevented the
|
25 |
emergence of competition from these other firms was |
93
1 |
exclusive contracts with dealers. They were dealers who
|
2 |
supply dental laboratories with all kinds of things, but
|
3 |
in particular with prefabricated artificial teeth. And
|
4 |
those contracts block the laboratory from sourcing another
|
5 |
firm's teeth, preventing the American consumer from
|
6 |
having an effective prefabricated tooth choice.
|
7 |
You know, there's a market in everything. Some
|
8 |
of it might be competitive. As you get older, you get
|
9 |
more serious about the importance of health care markets
|
10 |
for having a competitive organization. And, Lord knows,
|
11 |
there is not enough competition in most health care markets.
|
12 |
So, I want to bring to the forefront, the
|
13 |
horizontal competitive effects. Impact, if there's a
|
14 |
Section 2 case, the impact of the bad acts, the contracts
|
15 |
in this case, is to reduce competition in the market for
|
16 |
prefab artificial teeth. So, it's possible that there are
|
17 |
two competitive regimes, one with monopoly and the other
|
18 |
with competition.
|
19 |
And I want to push to the second, the vertical
|
20 |
restraints logic, that the economic effects of these
|
21 |
contracts, these exclusive contacts, is to change that
|
22 |
competitive regime.
|
23 |
You know, it seems to me that you can, in the
|
24 |
course of investigating an alleged Section 2 violation,
|
25 |
discard an enormous number of cases just by thinking about |
94
1 |
-- not about the efficiency theory of the supposed bad
|
2 |
act, but rather just thinking about the anticompetitive
|
3 |
theory. The inquiry would ask: is it possible that there
|
4 |
could be less competition and also there could be more
|
5 |
competition in this industry? Is it possible that if the
|
6 |
dealer contracts weren't exclusive that then there
|
7 |
could be competition? Without a "yes" to both, further
|
8 |
inquiry is not going to lead a Section 2 case. The second
|
9 |
question, the exclusivity of the dealer contracts having
|
10 |
sufficient impact to change the compeitive regime, that is
|
11 |
not a small inquiry. There is a lot of assumptions under
|
12 |
there.
|
13 |
There are at least two base assumptions. The
|
14 |
monopolist, Dentsply, is in a position to compel the
|
15 |
dealers to accept these exclusive contracts. That's not
|
16 |
going to be true in all industries. There can't, for example,
|
17 |
be the possibility of some other parallel distribution segment
|
18 |
which can grow up and distribute the competitive prefab
|
19 |
teeth. Furthermore, while the distribution channel firms must
|
20 |
not be in a position to resist Dentsply, Dentsply's competitors
|
21 |
must need the distribution channel. Thus, the distribution
|
22 |
channel must be dependant upon Dentsply but depended upon by
|
23 |
the competitors. Not all distribution channels will satisfy
|
24 |
both conditions. So, there's a reason that these exclusive
|
25 |
dealership contractors have bite. Bite, it was entirely |
95
1 |
accidental pun. I think if we could go down the path of the
|
2 |
Dentsply puns, they would be very unhappy for us.
|
3 |
But I mean to emphasize that there are two
|
4 |
dualities just in the competitive effects part of a
|
5 |
Section 2 case, which means, before you get to the hard thing
|
6 |
about efficiencies, you could throw a lot of cases out. It
|
7 |
has to be possible that there's two competitive regimes,
|
8 |
monopoly and more competitive, and it has to be possible
|
9 |
that the bad act works to move the market between them,
|
10 |
and that itself has two steps. The little guys, the
|
11 |
potential competitive providers of these competitive teeth,
|
12 |
have to need the distributors. The distributors need to be a
|
13 |
powerful hard-to-replace force. And the existing monopolist,
|
14 |
Dentsply, has to be able to kick around the distributors.
|
15 |
So, you've got two dualities, it's monopoly, but
|
16 |
it might be more competitive. And the distributors are
|
17 |
important, but the monopolist is in a position to either
|
18 |
bribe them or compel them to prevent the outbreak of
|
19 |
competition, competition which would be plausibly in their
|
20 |
interests.
|
21 |
Those two dual tests I think will weed out a lot
|
22 |
of cases before you begin this open-ended discussion of
|
23 |
whether these particular contracts are efficient. So,
|
24 |
here is how I graph it. You've got -- your centerpiece
|
25 |
should be the anticompetitive effects. So, in |
96
1 |
monopolization case, the effects are anticompetitive.
|
2 |
There is an exclusionary act, in this case the contracts,
|
3 |
which is keeping us in a higher market power monopoly, in
|
4 |
this case industry regime rather than a lesser market
|
5 |
power.
|
6 |
And, as I said, that's a lot for the plaintiff
|
7 |
to show. In the case of the agencies, that's a lot for
|
8 |
them to show. And I want to urge a review of whether we
|
9 |
can show these things early in a case. When I said to
|
10 |
kind of regularize Section 2 review, you know, it's just
|
11 |
like merger review, is there a competitive effect this merger
|
12 |
is going to do? Is there a competitive effect these are bad
|
13 |
practices are going to have, too? Is it really true that
|
14 |
there is more market power in the current regime but there
|
15 |
could be less market power? And that is the centerpiece,
|
16 |
that there is this causation, there's these
|
17 |
exclusive contracts, which exist because the existing
|
18 |
monopolist wants to maintain a monopoly, or what's keeping
|
19 |
us in the less competitive regime rather than the more
|
20 |
competitive regime.
|
21 |
And I think if you do both that causation
|
22 |
carefully and that competitive effects carefully that
|
23 |
would make Section 2 cases look a lot more like ordinary
|
24 |
antitrust analysis.
|
25 |
So, I said a number of times that that's a lot |
97
1 |
to show. It has to be possible that the competitive
|
2 |
regime could change; it has to be possible that the bad
|
3 |
acts are what's preventing the competitive regime from
|
4 |
changing; there has to not be another explanation of why
|
5 |
the competitive regime is not changing.
|
6 |
We spend so much time in Section 2. Here's my
|
7 |
one slide. I think I only have one slide and it's sort of
|
8 |
ordinary analysis. We spend so much time thinking about
|
9 |
whether there's an efficiency theory of the
|
10 |
anticompetitive acts. And that is important. But, you
|
11 |
know, I guess I would say, solve the problem with whether
|
12 |
there's a harm to competition first and then worry about
|
13 |
if there's an efficiency theory.
|
14 |
A lot of this efficiency discussion -- and here
|
15 |
I'm echoing Professor Farrell's earlier remarks in these
|
16 |
hearings -- we're driving in the direction of that world
|
17 |
of pure economic theory where we can figure out in a
|
18 |
quantitatively precise and reliable way whether
|
19 |
the consumer of the industry is better off with the
|
20 |
existing industry structure, including its contracts,
|
21 |
versus some counterfactual regime where the contracts
|
22 |
would be gone and there would be less efficiency
|
23 |
presumably from the contracts, but also more competition.
|
24 |
In economic theory, the author of the model knows
|
25 |
everything and could calculate how well off consumers are in |
98
1 |
another world. In the real world, the ability of
|
2 |
empirical economics, even with the very high level of
|
3 |
inquisitory abilities of the enforcement agencies to figure
|
4 |
out what would happen in that but-for world in enough detail
|
5 |
to calculate social welfare seems to me to be a waste of
|
6 |
time.
|
7 |
So, I would say, plaintiff has to show that
|
8 |
there is an anticompetitive effect and that it's causal.
|
9 |
And defendant gets to rebut that. Defendant has to show
|
10 |
that their practices are efficient. Plaintiff gets to
|
11 |
rebut that.
|
12 |
If the world is not tired of hearing from me
|
13 |
about the Microsoft case, let me talk about that one too.
|
14 |
I mostly want to emphasize its parallels to Dentsply.
|
15 |
Again, my competitive effects story is in the
|
16 |
graph here, I think I'm very close to the D.C. Circuit's
|
17 |
logic here. The competitive story is slightly different
|
18 |
because the industries are slightly different. And this is
|
19 |
one of the inevitable costs of Section 2. Section 2 cases
|
20 |
are rare. They arise in those industries where there is the
|
21 |
possibility of a big change in competitive circumstances.
|
22 |
That's not most industries and that's probably
|
23 |
idiosyncratic industries. Certainly these two, the teeth
|
24 |
and the software, are both idiosyncratic.
|
25 |
So, what's the state of the market? There is a |
99
1 |
Windows monopoly in operating systems on PCs. That was
|
2 |
true when the case was brought. I got the year wrong. It
|
3 |
was tendered to the Department of Justice in 1997 or so.
|
4 |
There could have been dynamic competition for the operating
|
5 |
system market if the mass use of the Internet led to new
|
6 |
standards in new markets.
|
7 |
So, here -- well, in the case of Dentsply, there
|
8 |
was a monopoly and could have been competition in the
|
9 |
market for prefab artificial teeth. In the case of
|
10 |
operating system software, there is a monopoly and the
|
11 |
industry in the past had had dynamic competition where
|
12 |
entrants in many important software products had replaced
|
13 |
incumbents. And in other important software markets, they
|
14 |
had given incumbents a terrible scare and created
|
15 |
incentives to get some real innovation out of them. In
|
16 |
these software markets, there is persistent static
|
17 |
monopoly, but there could be the prospect of
|
18 |
Schumpeterian competition. So, that's the two competitive
|
19 |
regimes that you get the competitive effects on.
|
20 |
The other part of Microsoft is really quite
|
21 |
similar to Dentsply. What kept the world in the monopoly
|
22 |
regime rather than in the potentially more competitive
|
23 |
regime, a regime where say a Linux might have taken a run
|
24 |
at the position of Microsoft Windows on the desktop?
|
25 |
It was a distribution case just like Dentsply, |
100
1 |
how could actual distributors, and a wide number of
|
2 |
different kinds of complementors, with other third parties
|
3 |
that would have worked with something like Linux on the
|
4 |
desktop prevented a market test for the Internet entrepreneurs,
|
5 |
and thereby ultimately prevented Schumpeterian competition in
|
6 |
the operating systems market.
|
7 |
So, again, this is the stuff that the Antitrust
|
8 |
Division had to prove in Microsoft. That's why it's such a
|
9 |
long case. Two potential competitive regimes. One, the
|
10 |
present one in operating systems and other infrastructure
|
11 |
software on the PC, which is about ten years you've had
|
12 |
very little competition in those industries, but in the
|
13 |
same industry in the previous twenty years before that you had
|
14 |
it all the time. Maybe there could have been, certainly the
|
15 |
Microsoft guys thought there could have been, dynamic
|
16 |
competition against some of those valuable position if the
|
17 |
Internet entrepreneurs had succeeded.
|
18 |
Some of this was more complicated and it's vertical
|
19 |
in more senses. The Internet entrepreneurs wVere not
|
20 |
horizontal competitors for Windows. The browser
|
21 |
was a complement. So, this was vertical restrictions to
|
22 |
prevent vertical disintegration. The vertical
|
23 |
disintegration would have permitted horizontal (dynamic)
|
24 |
competition in the operating systems market. So, it's a
|
25 |
good thing that the history of the industry had so much |
101
1 |
vertical disintegration causing horizontal competition for
|
2 |
the market, and that the Microsoft guys in their internal
|
3 |
documents were so clear about that that such a complex
|
4 |
case could be argued.
|
5 |
So, there's Andy Grove. Everybody has seen
|
6 |
Andy's slide a hundred times. The way you get competition
|
7 |
is you get a vertical disintegration. Andy was, when he
|
8 |
wrote this, the CEO of Intel. Mr. Gates of Microsoft has
|
9 |
said this many times as well.
|
10 |
This was the essence of the antitrust case,
|
11 |
that the internal documents, used that model of
|
12 |
vertical disintegration leading to horizontal competition,
|
13 |
provided evidence for the potential change in the competitive
|
14 |
regime.
|
15 |
Now, again, I want to say, these cases are going
|
16 |
to be rare. There's not a lot of industries where
|
17 |
vertical disintegration is the key trigger for horizontal
|
18 |
competition. It happens to be in infrastructural or mass
|
19 |
market software on your personal computer that that's
|
20 |
true, and it's been true since the industry was founded.
|
21 |
But, the cases where there can be causation from a
|
22 |
vertical restriction to horizontal competition are going
|
23 |
to be reasonably rare. This was one.
|
24 |
I would emphasize again, look for evidence of
|
25 |
that causal change before you go worrying about |
102
1 |
efficiencies.
|
2 |
This part is pretty much the same as
|
3 |
Dentsply in many ways. Microsoft is more complicated
|
4 |
because it's vertical in two senses: vertical
|
5 |
restrictions to prevent vertical disintegration, and
|
6 |
vertical disintegration in turn preventing horizontal
|
7 |
competition.
|
8 |
But what was really important in the competitive
|
9 |
effects in the case was that chain of causation did lead
|
10 |
to blocking of a threat which could have led to the kind
|
11 |
of dynamic and very valuable competition we had seen over
|
12 |
the previous twenty years in this industry.
|
13 |
Microsoft -- this other pragmatic, question about
|
14 |
when to bring a Section 2 case, it's helpful to have a
|
15 |
defendant that tries to prove entirely implausible things
|
16 |
like, there's no market power in Windows. It was a bad moment
|
17 |
for their economics expert witness, I think.
|
18 |
The other very unwise thing
|
19 |
that Microsoft chose to prove was that their reaction to
|
20 |
the widespread mass market use of the Internet wasn't
|
21 |
strategic, even though there were hundreds and hundreds
|
22 |
and hundreds of internal documents saying that it was
|
23 |
strategic. The CEO, whose memo I just quoted saying, this
|
24 |
is a terrible threat to us, chose to testify that he had
|
25 |
no idea what the threatening firm was doing at the time. |
103
1 |
So, defendent's trying to prove that it wasn't
|
2 |
strategic, trying to prove that there was market power,
|
3 |
made it somewhat easier for the government to prevail.
|
4 |
These are complicated cases. The agencies are not always
|
5 |
going to prove both dualities, that there could be a change
|
6 |
in market conditions and that the distribution system is
|
7 |
essential causally to keeping an out.
|
8 |
So, here's another one with a slide. The ultimate
|
9 |
remedy chosen in Microsoft was to require divestiture of
|
10 |
all applications, including the browser and Microsoft office.
|
11 |
This was not on Richard's, Carl's or Dan's watch.
|
12 |
This one is on my watch. And I have to say, I had to put
|
13 |
up this slide. There slide -- actually there is a long
|
14 |
history of this particular slide. When Dennis Yao, who
|
15 |
was my roommate in high school, was a Commissioner in the
|
16 |
FTC in 1989 or 1990, called me and said, you know, we
|
17 |
figured out we don't want to go after IBM and Microsoft
|
18 |
together, should we go after Microsoft. And the metaphor
|
19 |
immediately leapt to my mind, you're going to be like a
|
20 |
dog that's chasing a fire truck, you know, they're rolling
|
21 |
down a little street, noisy, illegal as hell,
|
22 |
anticompetitive as hell, but what are you going to do with
|
23 |
it when you catch it?
|
24 |
As it worked out, they didn't catch Microsoft.
|
25 |
I did. And the dog in this picture turned out in actual |
104
1 |
history to be me. What did we get? Not any remedy which
|
2 |
changed the conditions of competition. Ultimately, there
|
3 |
was an entirely ineffectual settlement in the United States
|
4 |
and a mildly effectual settlement in the EU. Certainly not
|
5 |
enough remotely to have the kind of competitive conditions
|
6 |
change that was possible from the widespread use of the
|
7 |
Internet.
|
8 |
So, there's another problem with the agencies,
|
9 |
bringing large, complicated antitrust cases. The
|
10 |
counter example here would be, of course, U.S. v.
|
11 |
AT&T. The United States was incredibly well served by
|
12 |
that case. During the long interval between the AT&T
|
13 |
breakup and the soon-to-happen reestablishment of the
|
14 |
Bell System, we were incredibly well served to have
|
15 |
vertical disintegration in telephony. The fact that we
|
16 |
had vertical disintegration in telephony at the moment in
|
17 |
history when, for example, technologists finally figured
|
18 |
out how to have mass market use of online services. That
|
19 |
was incredibly fortunate and that resulted from the antitrust
|
20 |
case. But they can also fizzle. And even if you win a
|
21 |
case, there can be severe problems in finding a remedy
|
22 |
that the antitrust system will undertake.
|
23 |
So, let me go to my bottom line. I really want us
|
24 |
to turn around. These cases are going to be hard to prove and
|
25 |
I want us to turn around and think about both the potential |
105
1 |
for a competitive effect, meaning there could be change in
|
2 |
the conditions of competition. The form of that change was
|
3 |
different with the two cases I talked about. Second, think
|
4 |
about a causal link between the alleged act and monopoly. I
|
5 |
would bring those to the fore. Those would be my framework
|
6 |
for thinking about a Section 2 case.
|
7 |
But of course that discussion is only about the
|
8 |
question of whether there is an antitrust case. This doesn't
|
9 |
remove from the agencies or any other plaintiff, but particularly
|
10 |
not for the agencies, the problem of thinking about whether
|
11 |
there's enough of a harm to competition at stake to justify
|
12 |
any intervention. I guess I would say that in an cases like
|
13 |
AT&T or Microsoft, where you've got a substantial impediment
|
14 |
to technical progress in an infrastructure industry, that matters
|
15 |
to the whole economy, arising from the lack of competition.
|
16 |
That one might get you over the hump. But there are other
|
17 |
metrics that can be used, such as the size of the differnce
|
18 |
between the two competitive regimes and the importance to
|
19 |
consumers.
|
20 |
And also to think through whether there might be
|
21 |
an efficiency defense, whether there might be more harm
|
22 |
than good done by the antitrust intervention. I don't
|
23 |
want to take that away, but I do want to say that I would
|
24 |
emphasize -- I would emphasize thinking through whether
|
25 |
there is an antitrust case in a perfectly ordinary |
106
1 |
antitrust analytical way, competitive effects and
|
2 |
causation.
|
3 |
Thank you very much.
|
4 |
MS. GRIMM: Thank you very much.
|
5 |
Our next speaker is Professor Rich Gilbert, who
|
6 |
is Professor of Economics of the University of California
|
7 |
at Berkeley.
|
8 |
From 1993 to 1995, he was Deputy Assistant
|
9 |
Attorney General in the Antitrust Division of the U.S.
|
10 |
Department of Justice, where he led the efforts that
|
11 |
developed joint Department of Justice and Federal Trade
|
12 |
Commission "Antitrust Guidelines for the Licensing of
|
13 |
Intellectual Property."
|
14 |
Professor Gilbert has served as an Associate
|
15 |
Editor of the "The Journal of Industrial Economics," "The
|
16 |
Journal of Economic Theory," and "The Review of Industrial
|
17 |
Organization."
|
18 |
Professor Gilbert research specialties include
|
19 |
antitrust economics, intellectual property, and research
|
20 |
and development.
|
21 |
He earned his Ph.D. from Stanford University in
|
22 |
1976. He received a Bachelor of Science degree in
|
23 |
Electrical Engineering in 1966 and a Master of Science
|
24 |
degree in 1967 both, from Cornell university.
|
25 |
Professor Gilbert. |
107
1 |
MR. GILBERT: Thank you very much, Karen.
|
2 |
While I figure out how to find my talk here, I
|
3 |
will thank you for bringing these hearings to Berkeley.
|
4 |
We're very glad we could be able to host these hearings.
|
5 |
And here we go.
|
6 |
I'm going to talk about a very narrow slice of
|
7 |
conduct that could invoke Section 2 liability, namely
|
8 |
innovation or product design, and ask the question of
|
9 |
whether innovation, certain types of innovations can be a
|
10 |
source of Section 2 or contribute to Section 2 liability.
|
11 |
Now, I don't think many people would argue that
|
12 |
innovation is great for the economy. Nevertheless, there
|
13 |
are quite a number of cases that have alleged that
|
14 |
innovation or product design has contributed to
|
15 |
monopolization. Of course, Microsoft, as we just heard,
|
16 |
is one. A slew of cases involving IBM and standardization
|
17 |
for complimentary products, the use of complimentary
|
18 |
products. There are some interesting cases on the horizon
|
19 |
in the prescription drug industry that raise innovation
|
20 |
issues in a Section 2 sort of context.
|
21 |
So, I'm going to be reviewing some of these
|
22 |
cases and asking whether we could have a standard, we've
|
23 |
heard a lot about standards this morning to evaluate
|
24 |
Section 2 type conduct, whether any of these standards is
|
25 |
useful for evaluating innovation. Maybe I will give you |
108
1 |
my punch line right away. I think the answer is no, and
|
2 |
try to tell you why.
|
3 |
I'll begin -- let's see. I'm going to begin
|
4 |
with a very simple model. I hope not to raise the fear
|
5 |
factor too much and talk about letters here. If you are
|
6 |
worried about this, you can replace any letters with
|
7 |
numbers. So, I want to talk about a very simple model of
|
8 |
innovation.
|
9 |
I have here an old technology. It has a social
|
10 |
value, v, zero, for each use. You could use, say, fifty
|
11 |
dollars for v, zero. A new technology could come along
|
12 |
with a higher social value, maybe a hundred dollars, for
|
13 |
each use. I'm going to strip away marginal cost to keep
|
14 |
things as simple as possible. There are a bunch of users,
|
15 |
say there's a thousand users, if you want. And there's
|
16 |
some R&D costs.
|
17 |
Now, in this simple model, the innovation is
|
18 |
socially desirable, I mean, it's still the small one can
|
19 |
be as simple as possible if the total incremental social
|
20 |
value exceeds the cost of the innovation.
|
21 |
So, we have our thousand consumers and they each
|
22 |
use this technology in one application, the extra value of
|
23 |
the innovation is fifty dollars, so that would be fifty
|
24 |
thousand dollars. The question is: Does that cover the
|
25 |
cost? |
109
1 |
Now, in terms of whether the innovation is
|
2 |
privately profitable, there's a price that the innovator
|
3 |
can collect for the new technology and it's profitable if
|
4 |
the price it can collect times the number of people who
|
5 |
buy it, assuming they all buy it, in fact covers the cost.
|
6 |
So, the first that I want to make, and there is
|
7 |
a paper that should be coming out in "Competition Policy
|
8 |
International" on this topic, the first point is to say,
|
9 |
innovations can be socially desirable but not privately
|
10 |
profitable, or you can have innovations that are privately
|
11 |
profitable but not socially desirable.
|
12 |
So, the first point is a very simple point:
|
13 |
That innovation can go any way -- there can be any order
|
14 |
in evaluating social and private profitableness. It's not
|
15 |
like a price -- innovation is like a price change in some
|
16 |
respects. If you come out with an innovation for a
|
17 |
product, it's like reducing its quality-adjusted price,
|
18 |
and you can make an analogy between innovation and, say,
|
19 |
predatory pricing. If you reduce the quality-adjusted
|
20 |
price, that leads to the exit of competitor, and then you
|
21 |
raise your price again, that has a sort of predatory
|
22 |
flavor to it.
|
23 |
But unlike pricing, where lower price certainly
|
24 |
lowers the price above marginal cost is a good thing, we
|
25 |
really don't know if more or less innovation is a good |
110
1 |
thing unless you do the whole analysis.
|
2 |
So, the standards I want to talk about, these
|
3 |
came up this morning, I want to talk about different rules
|
4 |
of reason which I interpret as either a total rule of
|
5 |
reason, which looks at all of the economic value
|
6 |
associated with some conduct, whether it's value to
|
7 |
consumers or value to producers.
|
8 |
And then there's probably the more popular
|
9 |
consumer rule of reason analysis which focuses on
|
10 |
consumers, and some people would say is at the heart of
|
11 |
antitrust analysis, at least according to, say, Steve
|
12 |
Sala, although others such as Joe Farrell and Mike Katz,
|
13 |
and Ken Hirers from the antitrust division, have advocated
|
14 |
a total rule of reason standard.
|
15 |
Then there's the profit sacrifice test in one of
|
16 |
its many forms. There's the no economic sense test.
|
17 |
We've heard a little bit about that this morning. And
|
18 |
then I'll talk a little bit about sham innovation.
|
19 |
So, a total real of reason analysis, in a sense
|
20 |
it's the right thing to do if you are, sort of by
|
21 |
definition, an economist, it's the right thing to do
|
22 |
because it ask whether total surplus is increased from
|
23 |
some activity. And even if that makes producers
|
24 |
relatively better off than consumers, at least there's the
|
25 |
possibility that those producer profits will flow |
111
1 |
eventually to consumer benefit, or that somehow producers
|
2 |
can bribe consumers to get it all right.
|
3 |
But the problem of course is that you can have
|
4 |
the price being either larger or smaller than the
|
5 |
incremental social benefit. And all of the analysis would
|
6 |
have to be done when the innovation decisions from the
|
7 |
perspective of the decisions that are actually made, which
|
8 |
means what we call an ex ante analysis. And this really I
|
9 |
think sets up innovation as being distinctly different
|
10 |
from other conduct. Because when you talk about
|
11 |
innovation, it's absolutely necessary to keep going
|
12 |
backwards and backwards to what are the incentive effects
|
13 |
of whatever rules or policies you have in place, what are
|
14 |
their incentive effects for innovation in the first place.
|
15 |
And now it's easy to say, well, of course that's
|
16 |
right, of course we're going to take that into account.
|
17 |
But I want to ask you, if you have been in these hearings,
|
18 |
how many times have people really gone backwards and said,
|
19 |
what are the implications of what we're doing for the
|
20 |
kinds of decisions that people are make that could have
|
21 |
developed and could develop new products or new processes
|
22 |
or whatever ten years from now. And I would say you
|
23 |
haven't heard it very many times.
|
24 |
So, it very easy to lose sight of these
|
25 |
incentive effects. And on top of that, if you did a total |
112
1 |
rule of reason analysis, the analysis that you would have
|
2 |
to do is hugely complex. You have to really take into
|
3 |
account all spillovers, how innovation affects consumers
|
4 |
and firms in other industries, and those we know can be
|
5 |
very, very large. And with the complexity, you can lead
|
6 |
easily to false positive and false negatives. I'm not
|
7 |
going to say Type 1 and Type 2 because I always forget
|
8 |
which one is which, so I will just say false positives and
|
9 |
false negatives, and you can figure out which one is a
|
10 |
positive and which one is a negative on your own.
|
11 |
Too much enforcement or too little enforcement.
|
12 |
Portion. It can go either way.
|
13 |
A consumer rule of reason analysis. Again, it's
|
14 |
very complex. The problems are similar to those that
|
15 |
arise in a total rule of reason analysis. Again, the
|
16 |
ex ante problems, the uncertainties, the spillover
|
17 |
effects, etc. And as well can lead to conclusions that
|
18 |
just simply don't make sense. This is particularly a
|
19 |
problem in innovation. You could have an innovation that
|
20 |
just saves millions of dollars in production cost, but
|
21 |
maybe it leads to a nickel increase in price, which
|
22 |
certainly could happen. And would you want to say that
|
23 |
this is an anticompetitive innovation because consumers
|
24 |
are slightly worse off, despite the fact that it's
|
25 |
generated enormous savings and efficiencies on the |
113
1 |
producer's side.
|
2 |
Well, I know that people can differ on that, but
|
3 |
my view is that it just doesn't make any sense to discount
|
4 |
all of those efficiencies. Now, you can say that you're
|
5 |
looking at a merger case or you're looking at other
|
6 |
conduct that doesn't involve product design, that those
|
7 |
kinds of efficiencies are not likely to be huge or have
|
8 |
not been demonstrated to be huge, but when you're talking
|
9 |
directly about innovations these efficiencies exist as
|
10 |
part of the innovation. So, you can't discount them.
|
11 |
A profit sacrifice test. There are, of course,
|
12 |
different versions of a profit sacrifice test. And I'm
|
13 |
going to quote Janusz Ordover's and Bobby Willig's
|
14 |
definition: "Predatory intentions are present if a
|
15 |
practice would be unprofitable without the exit that it
|
16 |
causes but profitable with the exit." Now, Ordover and
|
17 |
Willig also say this is just talking about predatory
|
18 |
intent not facts, they add a lot of other conditions in
|
19 |
their analysis that make this analysis considerably more
|
20 |
elaborate, and in many ways closer to a total rule of
|
21 |
reason analysis. So, this is just the basic idea of a
|
22 |
profit sacrifice test.
|
23 |
Now, the profit sacrifice test, I am not the
|
24 |
first to say this, it doesn't seem to me to make any sense
|
25 |
to innovation, even though it was in fact developed |
114
1 |
originally to talk about innovation as well as price --
|
2 |
predatory pricing. The problem of course first of all is
|
3 |
that innovation almost always involves a profit sacrifice.
|
4 |
It's called investing in research and development. That's
|
5 |
what you do.
|
6 |
It's also the case that innovation, if it really
|
7 |
works, probably excludes competitors. So, exclusion is
|
8 |
sometimes a direct result of producing a really good
|
9 |
mousetrap. The other mousetraps can't compete.
|
10 |
Now -- and furthermore, and this is absolutely
|
11 |
crucial, is that we need to know how much market power
|
12 |
after the innovation occurs is necessary to justify the
|
13 |
investment in innovation in the first place. And you can
|
14 |
make statements about whether innovation creates too much
|
15 |
or too little market power relative to its social value.
|
16 |
But the social value is very hard to calculate. And the
|
17 |
amount of power or pricing power that is necessary to
|
18 |
evoke the right amount of investment in research and
|
19 |
development is simply a very hard question. So, I
|
20 |
conclude, based on this, that a profit sacrifice test
|
21 |
really doesn't do very much to inform this analysis.
|
22 |
What about a no economic sense test. I am going
|
23 |
to use Greg Werden's version of this. He says: "Conduct
|
24 |
is not exclusionary or predatory unless it would make no
|
25 |
economic sense for the defendant but for the tendency to |
115
1 |
eliminate or lessen competition."
|
2 |
Now, you can see that, with all the negatives
|
3 |
again, the no economic sense test is really a test of the
|
4 |
absence of predation. So, if it makes sense to do this
|
5 |
activity, then it's not predatory.
|
6 |
Now, although it's not really clear in the no
|
7 |
economic sense test what no economic sense means, there
|
8 |
are two interpretations of this, certainly as applied to
|
9 |
innovation. One is that it's not profitable. No
|
10 |
reasonable firm would have dumped all of this money into a
|
11 |
new product design unless it had a purpose of excluding
|
12 |
competition. A second interpretation is that innovation
|
13 |
really always makes economic sense because it's just a
|
14 |
good thing that firms do.
|
15 |
Depending upon which one of these
|
16 |
interpretations you have, if it's the first one, then the
|
17 |
no economic sense test is very similar to the profit
|
18 |
sacrifice test. Now, if it's the second one, the no
|
19 |
economic sense test is similar to really whether
|
20 |
innovation is a sham, meaning whether it's a fraud or not.
|
21 |
I think it's the case, and I know that Werden has said
|
22 |
that his view of the no economic sense test as applied to
|
23 |
innovation is the second version, not the first version.
|
24 |
And I also know that he has views of conduct that do not
|
25 |
in fact involve a profit sacrifice, even though there was |
116
1 |
some discussion this morning that they're the same. His
|
2 |
example was a world with no arson laws and flush with
|
3 |
matches. So, you can go out there and burn down anybody
|
4 |
you want, including your competitors.
|
5 |
So, let me review a little bit of some cases
|
6 |
involving predatory innovation, particularly with respect
|
7 |
to complimentary products, products that interact with
|
8 |
other products. Those are almost totally -- well, they're
|
9 |
not entirely, but to a great extent they have to do with
|
10 |
changes to the interface stands. That was certainly the
|
11 |
case with the IBM peripherals litigation, a bunch of these
|
12 |
in the late 1970s, whether it's other people's disk drives
|
13 |
would hook into and work with IBM mainframe computers.
|
14 |
And the Microsoft case. And there's been a few others.
|
15 |
As a general conclusion in looking through these
|
16 |
cases, well, you can find a lot of lower court decisions,
|
17 |
a general conclusion is that in nearly all of these cases,
|
18 |
weak evidence of efficiencies was sufficient to avoid
|
19 |
liability for predatory innovation.
|
20 |
So, after there was lots of talk about whether
|
21 |
there was monopoly power or not, or whether or not there
|
22 |
was a monopoly of market effect, competitive effect. The
|
23 |
final analysis, they said -- these courts generally said,
|
24 |
well, we can think of it as an efficiency reason for this
|
25 |
conduct, therefore it's okay. |
117
1 |
To my knowledge, only the Microsoft case,
|
2 |
Microsoft 4 as it is sometimes affectionately called,
|
3 |
purported to apply a rule of reason analysis to
|
4 |
innovation. So, let's talk about Microsoft a little bit.
|
5 |
The Microsoft case actually came up with a road
|
6 |
map to kind of evaluate innovation. There actually five
|
7 |
steps to the road map. I'm going to condense them to
|
8 |
three.
|
9 |
The plaintiff first must demonstrate that the
|
10 |
conduct that harmed consumers had an economic
|
11 |
anticompetitive effect. Second, if a plaintiff
|
12 |
successfully demonstrates anticompetitive effect, then the
|
13 |
monopolist may prefer a procompetitive justification for
|
14 |
its conduct. So, the second step is the monopolist,
|
15 |
alleged monopolist can talk facts and say, we have a
|
16 |
reason for doing this. And then the third step says,
|
17 |
well, the plaintiff can now come back and rebut the
|
18 |
monopolist's justification. Or, if it can't actually
|
19 |
justifiably rebut it, it can demonstrate that the
|
20 |
anticompetitive effect was bigger than the procompetitive
|
21 |
benefit and outweighs it. So, it can do a rule of reason
|
22 |
analysis is what it says.
|
23 |
Well, let me just review what happened in the
|
24 |
Microsoft case. There were many allegations having to do
|
25 |
with Java standards and with various contracting policies |
118
1 |
with lots of different players in the industry.
|
2 |
There were really three design elements that
|
3 |
were challenged in the Microsoft case. One was not having
|
4 |
Internet Explorer in the Add/Remove programs utility. The
|
5 |
other was designing Windows so as in certain circumstances
|
6 |
to override the user's choice of a default browser other
|
7 |
than Internet Explorer. And the third one was commingling
|
8 |
browser and operating system code.
|
9 |
Now, interestingly, the court concluded that
|
10 |
Microsoft offered no procompetitive justifications for the
|
11 |
first and the third, and these were held by the court to
|
12 |
contribute to the Section 2 violation. But then the court
|
13 |
also concluded that plaintiffs -- that Microsoft did offer
|
14 |
a justification for the second element of its conduct,
|
15 |
that is, the overriding of user's choice of the browser,
|
16 |
which the plaintiff did not rebut, and therefore in fact
|
17 |
the Microsoft court never got to the third step. So, the
|
18 |
court never got to the rule of reason balancing in the
|
19 |
third step because either it was anticompetitive with no
|
20 |
efficiencies or there were efficiencies and the plaintiff
|
21 |
didn't come back. So, maybe Tim will explain or Dan will
|
22 |
elaborate on this, but this is my reading of what happened
|
23 |
with the court.
|
24 |
So, the practical effect of what happened in the
|
25 |
Microsoft court's analysis was really, I think, similar to |
119
1 |
the no economic sense test of the first variety. That is,
|
2 |
was there some reason for this conduct. If there was,
|
3 |
it's okay.
|
4 |
Now I want to turn to another area that I find
|
5 |
quite interesting. As they say, this is emerging
|
6 |
antitrust. This is that drug patents may delay generic
|
7 |
competition. So, the innovation that contributes to these
|
8 |
drug patents can have competitive effect. It can have
|
9 |
competitive effect both through the nature of generic
|
10 |
substitution and also because of the specific elements of
|
11 |
the Hatch-Waxman Act, which impose a 30-month stay on
|
12 |
generic competition if you have a patent.
|
13 |
So, one of these cases is called Tricor, which
|
14 |
is actually a drug called phenofibrate. It's used to
|
15 |
control triglyceride and cholesterol levels. And I should
|
16 |
acknowledge I have been a consultant in this case. A
|
17 |
second case is Prilosec and Nexium, which Prilosec is a
|
18 |
common drug prescribed for heartburn, gastric reflux, and
|
19 |
then your more serious conditions like esophageal and
|
20 |
duodenal ulcers. It turns out that Nexium is what is
|
21 |
called an isomer of the chemical that's in Prilosec. It's
|
22 |
basically the same molecules. It's been rearranged a
|
23 |
little bit. And it's supposed to have some advantages for
|
24 |
the esophageal and duodenal ulcers, but not for heartburn.
|
25 |
The allegations that came up in both of these |
120
1 |
cases is that the innovations are costly but minor
|
2 |
improvements, that they're contrary to the intent of the
|
3 |
Hatch-Waxman legislation to promote generic competition,
|
4 |
and they have large adverse competitive effects by
|
5 |
delaying generic competition.
|
6 |
Now, I think certainly if you just take a
|
7 |
snapshot of competition, once these drugs exist, anything
|
8 |
that delays generic competition has at lease the
|
9 |
possibility of a competitive effect. But it's important
|
10 |
to recognize that the Hatch-Waxman legislation was a trade
|
11 |
off between more generic competition and more protection
|
12 |
for patented drugs. In fact, the first three letters of
|
13 |
the Hatch-Waxman Act are patent term restorations. I
|
14 |
think it was designed to protect pioneer drugs, as well as
|
15 |
promote generic competition.
|
16 |
Product line extensions certainly increase
|
17 |
incentives for drug innovation. If you actually look at
|
18 |
the respective patent terms for prescription drugs,
|
19 |
patented prescription drugs, it's actually quite short.
|
20 |
It's one of the shortest of all industries because of all
|
21 |
the FDA delays and regulations required to actually
|
22 |
produce the drugs. And it's very hard to assess these
|
23 |
benefits from these innovations.
|
24 |
So, I think instead of looking at any of these
|
25 |
standards to inform a Section 2 analysis for innovation, I |
121
1 |
find all of them seriously lacking. I think instead you
|
2 |
can turn to consistency with other rules.
|
3 |
And let's talk about something we've heard
|
4 |
before, talk about the process rather than the outcome.
|
5 |
That was discussion was featured in this morning's session
|
6 |
to great extent by members of the panel talking about the
|
7 |
process rather than the outcome.
|
8 |
So, the quote that I'm quoting here is by a
|
9 |
distinguished economist, but not anyone from our group.
|
10 |
It's from an economist who works for the Oakland Athletics
|
11 |
who was quoted by Michael Lewis in "Moneyball," and he was
|
12 |
actually talking about how to hire baseball players, but I
|
13 |
think his insight here is equally applicable to antitrust
|
14 |
policy, "We have to look at process, not outcomes."
|
15 |
So, if we think about making an analogy between
|
16 |
innovation effects, and the effects and rules that are
|
17 |
applied to other conduct, I want to argue that, in many
|
18 |
innovation cases, the effects of the innovation are very
|
19 |
similar to the effects of a unilateral refusal to deal.
|
20 |
When you're talking about, say, if IBM refuses to make
|
21 |
mainframes compatible able with third parties' components,
|
22 |
it's a lot like saying, well, one day Microsoft gets up
|
23 |
and says, I don't want to work with these third party
|
24 |
people anymore, I want to build computers just for myself.
|
25 |
Microsoft refuses to make Windows compatible with other |
122
1 |
browsers. Or a generic drug manufacturer refuses to
|
2 |
supply a drug that generics can copy.
|
3 |
In effect, this conduct looks a lot like a
|
4 |
unilateral refusal to deal. Now, these days, after
|
5 |
"Verizon v. Trinko", seems like unilateral refusals to
|
6 |
deal have a long way to go before they can generate
|
7 |
antitrust liability.
|
8 |
Now, I don't want to state that as a categorical
|
9 |
fact, or that "Verizon v. Trinko," that all the words in
|
10 |
"Verizon v. Trinko" were necessarily the greatest words
|
11 |
that have ever been uttered in all of antitrust policy. I
|
12 |
am not sure it's the greatest policy.
|
13 |
But my only point is that if you are going to
|
14 |
have a policy that gives considerable deference to a
|
15 |
decision by a single firm about who that firm will deal
|
16 |
with or supply, it just seems odd that one wouldn't have a
|
17 |
more strict policy, more intervention policy with respect
|
18 |
to innovations that have very similar effects.
|
19 |
So, I'm not saying -- again I want to emphasize
|
20 |
that I'm not saying that we should have policies that say
|
21 |
that unilateral refusals to deal with per se legal, I
|
22 |
don't think that's necessarily the right thing. But if we
|
23 |
are going to have such a policy, then consistency seems to
|
24 |
say that if you unilateral innovations that have similar
|
25 |
effects should not be treated more severely. |
123
1 |
So, one of my conclusions here is that all of
|
2 |
the rule of reason and profit sacrifice tests have limited
|
3 |
value to evaluate what is sometimes called predatory
|
4 |
innovation. It's hard to do; likely to get the wrong
|
5 |
answer; very hard to look al at the incentive effects that
|
6 |
are necessary to really thinking about innovation.
|
7 |
The no economic sense test is better, but only
|
8 |
if it's interpreted as a test of sham innovation because
|
9 |
otherwise it comes out just like or very similar to a
|
10 |
profit sacrifice test.
|
11 |
And my other conclusion is that this is what
|
12 |
courts in fact almost always have done with very few
|
13 |
exceptions in the way they've treated these cases and it's
|
14 |
probably as reasonable an approach as any.
|
15 |
MS. GRIMM: Our third presenter this afternoon
|
16 |
is Daniel Rubinfeld, who is the Robert L. Bridges
|
17 |
Professor of Law and Professor of Economics at the
|
18 |
University of California at Berkeley, where he has taught
|
19 |
since 1983
|
20 |
He has also served as Deputy Assistant Attorney
|
21 |
General for Antitrust in the U.S. Department of Justice,
|
22 |
as well as in various capacities with the President's
|
23 |
Council of Economic Advisors, the National Academy of
|
24 |
Sciences, the Urban Institute, and the National Bureau of
|
25 |
Economic Research. |
124
1 |
Professor Rubinfeld's major books include
|
2 |
"Econometric Models and Economic Forecasts" and
|
3 |
"Microeconomics." Recent publications include, "Antitrust
|
4 |
Enforcement in Dynamic Network Industries" in "The
|
5 |
Antitrust Bulletin," 1998; and "Empirical Methods in
|
6 |
Antitrust: Review and Evidence" in "American Law and
|
7 |
Economics Review."
|
8 |
He is President of the American Law and
|
9 |
Economics Association.
|
10 |
Professor Rubinfeld received his B.A. from
|
11 |
Princeton University in 1967; his M.S. and Ph.D. from the
|
12 |
Massachusetts Institute of Technology.
|
13 |
Dan.
|
14 |
MR. RUBINFELD: Thanks very much. I really,
|
15 |
like everyone else, appreciate the opportunity to appear
|
16 |
before you today. It's been about eight or nine years
|
17 |
since I left the Antitrust Division and I guess,
|
18 |
understandably I've aged about eight or nine years during
|
19 |
that time, and I find as one gets older one tends to
|
20 |
reflect back on the past, perhaps more than one should.
|
21 |
But what I'm going to do in my comments today is to really
|
22 |
do some reflection on what happened, and I might hit on
|
23 |
some of the previous commentators' issues, but see I can do it in
|
24 |
a way that will be constructive for the agencies as you
|
25 |
think about forming your policies. |
125
1 |
So, the first point I want to make is why I
|
2 |
think it's really important to have an active Section 2
|
3 |
jurisprudence. And I want to look back and talk about the
|
4 |
legacy of "U.S. vs. Microsoft" for antitrust enforcement.
|
5 |
And, finally, I want to look at bundling and talk about
|
6 |
the legacy of "LePage's vs. 3M".
|
7 |
I should say, to make it clear, that I had an
|
8 |
interest in both of those cases. I helped to prosecute
|
9 |
the Microsoft case. And I have consulted for 3M with
|
10 |
respect to some of the issues that arose in its appellate
|
11 |
case. I was not involved in the LePage's case itself, but
|
12 |
I was involved in thinking about some of the appellate
|
13 |
issues. So, I have taken a pretty close look at the Third
|
14 |
Circuit opinion in that case.
|
15 |
If you're interested in some of the deeper
|
16 |
comments I am going to give today, they will appear in
|
17 |
two articles. One is an article that Doug Melamed and
|
18 |
myself are completing for our forthcoming volume in
|
19 |
which we are looking at the lessons of the Microsoft case.
|
20 |
And the second is an article I published a year or so ago,
|
21 |
looking at the bundling in the "LePaige's vs. 3M" case.
|
22 |
Before I go on to the cases, as far as the active
|
23 |
Section 2 jurisprudence is concerned, I guess history
|
24 |
affects how one views things, and I can be very quick, I
|
25 |
can just say, having been involved in actually bringing |
126
1 |
both Microsoft and Dentsply, both of which I thought was
|
2 |
the right thing to do, and the D.C. Circuit and the Third
|
3 |
Circuit in both cases have written opinions that were
|
4 |
supportive of that decision, I'm proud to have been
|
5 |
involved in both of those cases, and I think that shows,
|
6 |
consistent with what Tim Bresnahan said, it shows the kind
|
7 |
of active Section 2 jurisprudence that I think makes
|
8 |
sense.
|
9 |
Both cases had a particular set of facts
|
10 |
associated with them that told a story that made them the
|
11 |
right cases to bring, viable cases. And I think the
|
12 |
agencies need to be careful because there is not going to
|
13 |
be a lot of good Section 2 cases. So, you need to be
|
14 |
careful and active and watchful for the appropriate
|
15 |
opportunities in the future.
|
16 |
So, having said that, let me go on and take a
|
17 |
look at "U.S. vs. Microsoft". And I am going focus now
|
18 |
really on sort of what we've learned from the case in a
|
19 |
very broad perspective. I'm not going to try to go into
|
20 |
some of the technical details unless we have discussion
|
21 |
later.
|
22 |
It's sometimes easy to forget, since this is
|
23 |
almost ten years ago when at least my version of Microsoft
|
24 |
was brought, that people were barely talking about network
|
25 |
effects. Now it's taken for granted that in high tech |
127
1 |
it's common to face industries in which network effects
|
2 |
matters and that enters into the economics and to the law,
|
3 |
legal thinking about the cases.
|
4 |
I see one of the legacies of Microsoft is sort
|
5 |
of helping to bring us from the pre-network effect world
|
6 |
to a world where network effects are often the core of the
|
7 |
analysis.
|
8 |
Next important is people are thinking somewhat
|
9 |
differently now than they were before about barriers to
|
10 |
entry. When we originally think about investigating the
|
11 |
Microsoft case, obviously barriers to entry was something
|
12 |
that I paid a lot of attention to. We became convinced
|
13 |
that there was a significant barrier to entry, but it's
|
14 |
not the usual one you might imagine. It had to do with
|
15 |
the fact that in order to have a successful operating
|
16 |
system, you really needed to have successful applications.
|
17 |
There was what we called a two-level entry problem. And
|
18 |
we spent a lot of time developing the underlying economics
|
19 |
that describe this applications barrier to entry.
|
20 |
One of the things that people forget, actually I
|
21 |
almost forget myself, is that the term "application
|
22 |
barrier to entry" did not exist, at least to my
|
23 |
knowledge, prior to our work. We coined and reiterated it
|
24 |
every time we could at trial until the judge finally got it
|
25 |
into his mind. |
128
1 |
And it was fun to watch the trial, by the way,
|
2 |
because at the beginning of the trial, Microsoft disavowed
|
3 |
the application "barrier to entry." By the end of the
|
4 |
trial it was being discussed by them as if it were a
|
5 |
common coin of the realm.
|
6 |
So, let's remember that that was one, for better
|
7 |
or worse, I think for better, one of the legacies of the
|
8 |
Microsoft case.
|
9 |
The other thing is, as you all know, the case
|
10 |
involved tying, but it was different than the classic kind
|
11 |
of tying case, which is usually thought of leveraging
|
12 |
market power from a market where a firm has substantial
|
13 |
market power to use some related power where it does not
|
14 |
necessarily have significant market power.
|
15 |
But this case did involve tying as well as
|
16 |
bundling. And it was a non-leveraged form of tying. And
|
17 |
now it's not, I think unusual to think about bundling in
|
18 |
that context in certain cases where it was probably quite
|
19 |
radical at the time.
|
20 |
The other thing is that the case brought to our
|
21 |
mind a different way, a different perspective of thinking
|
22 |
about market definitions. As Tim suggested earlier today,
|
23 |
there's always been a lot of talk about Schumpeterian
|
24 |
competition and certainly the agencies have been aware of
|
25 |
it for a long time. |
129
1 |
In this case, to one degree or another,
|
2 |
Schumpeterian competition really came to the forefront
|
3 |
because, in the debate about market definition and market
|
4 |
power, Microsoft took the position that it was the threat
|
5 |
of entry by competitors that really not only restrained
|
6 |
this market definition, this market power, but also in
|
7 |
fact meant that the market should be defined very broadly.
|
8 |
Microsoft argued for an extremely broad market definition
|
9 |
that included almost all operating systems, from
|
10 |
hand-helds pretty much up through mainframe computers, and
|
11 |
argued that it had no market power over that relevant
|
12 |
market.
|
13 |
I still remember one particular trial exhibit
|
14 |
which Microsoft presented which sort of brought this issue
|
15 |
to the front. And the exhibit said that Microsoft faces
|
16 |
substantial competition from known and unknown
|
17 |
competition. And my view, which was borne out, by the
|
18 |
way, by the Circuit Court opinion, is that when you have
|
19 |
to defend your market power or lack of it by describing
|
20 |
competition that no one knows about yet, you really have a
|
21 |
fairly weak position.
|
22 |
And if you read the D.C. Circuit opinion, I
|
23 |
think the D.C. Circuit got it right, as they did in most
|
24 |
areas, they said, the nascent competition really could be
|
25 |
important but it really has to be competition which is |
130
1 |
expected with reasonable certainty to actually be there in
|
2 |
the marketplace at some period in the future, thinking
|
3 |
about two years would be the relevant time period.
|
4 |
But the fact that someone might come along and
|
5 |
take away your market power isn't sufficient. I think the
|
6 |
court was pretty clear about that. And it's basically the
|
7 |
right place to be.
|
8 |
As far as legal issues I see coming out of the
|
9 |
case, there are about five. I'd like to highlight, again,
|
10 |
without getting into the technical/legal side of the case,
|
11 |
the first thing which I think we now take for granted, or
|
12 |
at least I hope we do, which is that the same antitrust
|
13 |
principles apply in dynamic high tech industries as apply
|
14 |
in the other industries. The application of course might
|
15 |
be somewhat different, but the principals are the same.
|
16 |
And I quote Judge Posner, who really says what I
|
17 |
have in mind, which is that antitrust doctrine really is
|
18 |
pretty well situated to allow us to handle high tech
|
19 |
industries. We don't need to rewrite Section 2, in my
|
20 |
view.
|
21 |
Up until Rich started speaking earlier, I would
|
22 |
have said hardly anyone remembers that there are IP issues
|
23 |
raised in Microsoft. Rich laid them out pretty well.
|
24 |
And so, what I wanted to say is that the court
|
25 |
makes it pretty clear that the same general antitrust |
131
1 |
principles that apply to conduct involving intellectual
|
2 |
property that apply to any other form of property underc
|
3 |
the antitrust laws.
|
4 |
Originally, at one point in the case, Microsoft
|
5 |
actually claimed that their IP rights covered the entire
|
6 |
desktop, at least with respect to the first boot up of
|
7 |
their operating system. The court made it very clear that
|
8 |
(a) that was too expansive an interpretation, and (b) that
|
9 |
it was appropriate for the Sherman Act and the courts to
|
10 |
really look at the IP issues. You did not get a free ride
|
11 |
just because you did in fact have some legitimate
|
12 |
intellectual property.
|
13 |
And Rich described in detail and correctly where
|
14 |
the court finally came out about these specific IP issues.
|
15 |
With respect to product design, as I interpret
|
16 |
the court opinion, it makes clear that the court is going
|
17 |
to give pretty wide deference to firms that are designing
|
18 |
new products, along the lines Rich described. But the
|
19 |
court also said this is an area that's open for viable
|
20 |
investigation. And where particular aspects of
|
21 |
Microsoft's product design excluded rivals, the court did
|
22 |
shift the burden to Microsoft to establish a
|
23 |
procompetitive justification for the design. There is no
|
24 |
safe harbor just because you're involved in innovation or
|
25 |
product design. And the removal of the Add/Remove utility |
132
1 |
which Rich described was one good example of that. The
|
2 |
court was very clear that was problematic and there was no
|
3 |
procompetitive justification given that I can see in the
|
4 |
case.
|
5 |
There's also an issue in this kind of Section 2
|
6 |
case as to whether you ought to kind of just describe the
|
7 |
case with kind of a broad brush or kind of go into the
|
8 |
practices with fine detail. My sense, my personal sense
|
9 |
during the trial was that there were times when the
|
10 |
defense seemed to say, we want to just talk very broadly
|
11 |
about the rights of a dominant firm to engage in certain
|
12 |
kinds of potentially procompetitive activities. And the
|
13 |
government, as I saw it, focused really in with apparent
|
14 |
detail about the details surrounding each of these kinds
|
15 |
of conduct.
|
16 |
And I read the D.C. Circuit as basically saying
|
17 |
that any aspect, the explicit, discrete aspect of
|
18 |
monopolist conduct that tends to exclude rivals may be
|
19 |
illegal, unless there's a legitimate procompetitive
|
20 |
justification for that particular conduct.
|
21 |
So, there is at least a burden-shifting aspect
|
22 |
to some of the illegal rules that flow from the Microsoft
|
23 |
case, which I think is appropriate.
|
24 |
There is an issue about whether you ought to
|
25 |
focus on rules or cases, specific facts. Here, as you |
133
1 |
know, the court, the appellate court, on the time claims,
|
2 |
suggested that the per se rule didn't apply because of the
|
3 |
particular attributes of platform software. So, we're now
|
4 |
left in a somewhat unclear world that may apply mostly to
|
5 |
Section 1, but also has Section 2 implications as to how
|
6 |
to treat tying.
|
7 |
And I have to say here, as an economist, you may
|
8 |
not be surprised to hear that I'm pretty sympathetic with
|
9 |
the comments of the court. I think it's really hard to,
|
10 |
as an economist, come up with per se rules that would
|
11 |
apply in this kind of high tech context.
|
12 |
Of course we don't know quite where that would
|
13 |
have ended up because the Department of Justice chose not
|
14 |
to appeal that part of the D.C. Circuit's ruling.
|
15 |
With respect to causation, I see the case telling
|
16 |
us conduct that violates the antitrust laws only if it
|
17 |
injures competition. Causation can be inferred when
|
18 |
exclusionary conduct is aimed at producers of nascent
|
19 |
competitive technologies, as well as when it's aimed at
|
20 |
producers of established substitutes.
|
21 |
So, basically the court spelled out causation along
|
22 |
the lines Tim suggested, and I think the court makes it
|
23 |
pretty clear that that's necessary and that the government
|
24 |
succeeded in that effort.
|
25 |
What about profit sacrifice? Here we could |
134
1 |
debate exactly how to characterize the case. I would say
|
2 |
that the case we put forward did really involve a profit
|
3 |
sacrifice test. My definition would be that conduct is
|
4 |
anticompetitive when it would not make business sense for
|
5 |
the defendant but for its tendency to exclude rivals and
|
6 |
create or maintain market power for the defendant.
|
7 |
This is kind of a crude paraphrase. If you go
|
8 |
back and read the details of the case, you'll see a more
|
9 |
formal definition. It is a variant on a profit sacrifice
|
10 |
test. I wouldn't say it's quite a no nonsense test, but
|
11 |
it's pretty close.
|
12 |
Now, that's not what the D.C. Circuit said.
|
13 |
What the D.C. Circuit said was quite close to what Rich
|
14 |
Gilbert said earlier. The court said that the conduct is
|
15 |
anticompetitive if it harms the competitive process and
|
16 |
either it's not shown to further efficiency or to have
|
17 |
some other procompetitive justification or the
|
18 |
anticompetitive harm outweighs its procompetitive benefit.
|
19 |
So, the D.C. Circuit was suggesting more of a balancing
|
20 |
test than a profit sacrifice test.
|
21 |
And this leaves us with the question of what we
|
22 |
should do if we find Section 2 type conduct that harms
|
23 |
competition and furthers a legitimate purpose should we
|
24 |
have a balancing test.
|
25 |
Now, I should say here, I am not entirely sure |
135
1 |
of where I would end up, but I lean strongly towards the
|
2 |
profit sacrifice test, at least in most cases, because I
|
3 |
think it's easier to operationalize. We could debate
|
4 |
about how to exactly operationalize it, but I think Tim
|
5 |
suggested that, in most of these cases, it's just not
|
6 |
possible to sit down and do a fully complete balancing
|
7 |
rule of reason analysis. We don't have the time or the
|
8 |
information available. And the cost, by the way,
|
9 |
including the cost to the parties, would be tremendous.
|
10 |
And I think in most situations, a profit sacrifice test
|
11 |
would get us to the right place. I think you can try to
|
12 |
find some counter-examples, but I think you have to work
|
13 |
hard to do it. So, I am on the side of the folks who
|
14 |
think we ought to just refine the profit sacrifice test.
|
15 |
Okay, let me switch to my other case of
|
16 |
interest, "LePage's vs. 3M". You have heard about it a
|
17 |
little bit already. This was the case involving bundled
|
18 |
rebates offered by 3M in the market for transparent tape.
|
19 |
3M was facing substantial competition from LePage's, not a
|
20 |
new entrant, but an entrant that had become very
|
21 |
successful in the production and sale of private label
|
22 |
tape.
|
23 |
And the question was: Were 3M's programs,
|
24 |
specific bundling programs, anticompetitive and a
|
25 |
violation of Section 2. |
136
1 |
Now, here I'm very critical of the Third Circuit
|
2 |
opinion generally for two reasons. One is that the
|
3 |
opinion itself does not, in my mind, in any way provide
|
4 |
any clear guidance as to how firms ought to behalf when
|
5 |
they do have a dominant position and they are deciding
|
6 |
what kind of business practice to engage in. And I think
|
7 |
any clear legal rule ought to do so.
|
8 |
And, secondly, I actually think that I have been
|
9 |
unable to come up with what I think is any coherent theory
|
10 |
of predation or any Section 2 theory which fits the facts
|
11 |
of the 3M case. In my view, the Third Circuit was a
|
12 |
little bit loose in how they actually borrowed and used
|
13 |
facts of the case. I actually went back and read most of
|
14 |
the record in the LePage's case and I cannot find a theory
|
15 |
that I find coherent that actually fits the facts of the
|
16 |
case.
|
17 |
And the thing to remember is that bundling
|
18 |
itself of course is quite ubiquitous and often is
|
19 |
procompetitive. So, if we generate a legal rule, we want
|
20 |
someone else to define those relatively few cases where
|
21 |
bundling is a problem and distinguish it from the majority
|
22 |
of cases where bundling is procompetitive. So, we're
|
23 |
looking for those particular situations. Lack of clarity
|
24 |
is a problem.
|
25 |
Let me briefly take a few minutes and just very |
137
1 |
quickly tell you about 3M's programs. There were a whole
|
2 |
bunch of programs being attacked, but the two that
|
3 |
involved bundled rebates were, first, the executive growth
|
4 |
fund program. And the thing that's key about this program
|
5 |
was it was actually I think a one-year program and it was
|
6 |
a pilot program for a small number of customers.
|
7 |
Now, what it did do was it set up growth targets
|
8 |
for six different errant divisions of 3M, which would
|
9 |
cover a lot of office supply products. And firms actually
|
10 |
had to meet target goals in each of these divisions.
|
11 |
Now, my view is that the executive growth fund
|
12 |
program -- let me be clear that this is my view and not
|
13 |
3M's view. My view is that, had this program been
|
14 |
expansive and had it covered all customers rather than
|
15 |
just a few, and had it continued for a number of years, it
|
16 |
could well have been an anticompetitive program. I don't
|
17 |
think it was because it was too narrow. It had no ability
|
18 |
really to substantially exclude competitors because many
|
19 |
of the key competitors, Walmart being the most important,
|
20 |
were not covered by this program. But it had the
|
21 |
potential if it continued to actually be restrictive
|
22 |
because of the specific design of the program.
|
23 |
But for various reasons, which I think relate
|
24 |
partly to the demands of some customers, including
|
25 |
Walmart, 3M changed its program to a partnership growth |
138
1 |
program, and this program did involve discounts in six
|
2 |
different areas, but there were no specific targets to
|
3 |
reach in each of the areas. Basically you got a rebate
|
4 |
based on the aggregate of all your purchases in all six
|
5 |
categories. So, this amounted to a somewhat complex
|
6 |
discount program, volume discount program.
|
7 |
And my view is that the PGF program, as it's
|
8 |
called, was not anticompetitive, even though the court
|
9 |
felt otherwise.
|
10 |
So, if you go back and look at the LePage's
|
11 |
trial and ask -- take a look at the trial and ask if the
|
12 |
trial helps to support some of those theories of
|
13 |
competition, I would say no. I didn't see any testimony
|
14 |
in the record about economies of scale or scope, which
|
15 |
would be important, particularly to get at the issue of
|
16 |
whether LePage's or any other competitor would remain
|
17 |
viable in the face of these practices.
|
18 |
There was no predatory pricing claim.
|
19 |
Plaintiffs agreed that LePage's was pricing above cost.
|
20 |
In fact, by my calculations, even if you took all of the
|
21 |
discount programs at 3M, no matter what the products were,
|
22 |
attribute all the discounts to tape, it would still be
|
23 |
pricing above cost.
|
24 |
I didn't see anything about profit sacrifice
|
25 |
that I could infer from the opinion. So, there was |
139
1 |
nothing that fit my particular interest in pursuing these
|
2 |
kinds of Section 2 cases.
|
3 |
There was no time claim at all. It was a
|
4 |
bundling case, not a tying case. There was also no showing
|
5 |
of market power with respect to any product other than
|
6 |
transparent tape. So, the kind of leveraging theory you
|
7 |
might expect to see in a time case was not present either.
|
8 |
Now, the jury did find, interestingly, no
|
9 |
exclusion under Section 1, but they did find a violation
|
10 |
under Section 2. So, this leaves me with a puzzle of what
|
11 |
the legacy is of "LePage's vs. 3M". I think for a while
|
12 |
the Commission may have though this case was unusual, but
|
13 |
it's pretty clear now that the Third Circuit opinion has,
|
14 |
let's say, encouraged a lot of litigation surrounding
|
15 |
these kinds of practices.
|
16 |
So, I went back and asked myself, what should
|
17 |
the principles be here. And I would say, speaking very
|
18 |
broadly, if the rebates associated with bundling reduce
|
19 |
consumer welfare by impairing rivals' ability to make
|
20 |
competitive offers to potential customers, that's going to
|
21 |
be something generally that's going to give me concern. I
|
22 |
am not going to say it's necessarily anticompetitive, but
|
23 |
that would give me great pause.
|
24 |
And that general rule takes into account
|
25 |
efficiencies and allows price increases by firms, as long |
140
1 |
as they don't impair rivals' ability to compete. But that
|
2 |
general rule is really not very helpful from a process
|
3 |
point of view. It's really too broad to make applicable.
|
4 |
So, I would say the following. I'd say, there
|
5 |
are conditions under which one may be anticompetitive, but
|
6 |
none of them fit LePage's.
|
7 |
And, just quickly, because I think we're running
|
8 |
out of time, here's some examples of situations in which I
|
9 |
think bundling might be anticompetitive, none of which
|
10 |
fits the LePage's case.
|
11 |
The first would be traditional contractual tying
|
12 |
of the kind that we saw in Jefferson Parish. The second
|
13 |
would be predation through profit sacrifice of the kind
|
14 |
where bundling was used in the form it was in the
|
15 |
Microsoft case, and perhaps I'd include Dentsply there as
|
16 |
well. The third might be monopoly maintenance through the
|
17 |
creation of barriers to entry, which is, at least my
|
18 |
interpretation of "SmithKline versus Eli Lilly," a case I
|
19 |
was not involved in, where at least the court stated that
|
20 |
the sale of monopoly products were used to harm
|
21 |
competition in a non-monopoly market.
|
22 |
Now, where does this leave us? We need a
|
23 |
workable test. I wish I could come here and tell you I
|
24 |
figured out what that test is. I have read many papers
|
25 |
written by folks in the agencies and elsewhere suggesting |
141
1 |
various tests.
|
2 |
I still not have seen one that I am entirely
|
3 |
happy with, but a couple things strike me as important
|
4 |
when and if we get such a test. One is that, weakening a
|
5 |
rival should not be sufficient to condemn a monopolist,
|
6 |
otherwise we will be discouraging firms from innovating
|
7 |
and growing and being successful, which I think would be
|
8 |
harmful to our competitive process.
|
9 |
Secondly, while it would be very nice to have an
|
10 |
incremental cost benefit test for certain kinds of
|
11 |
bundling, there are a lot of difficulties in putting that
|
12 |
test into play that I won't bore you with here. So, we
|
13 |
have more work to do there.
|
14 |
Third, we might say that for a bundled rebate
|
15 |
program to be anticompetitive, it at least necessarily
|
16 |
ought to be the case that the incremental costs associated
|
17 |
with the available discounts exceed the incremental
|
18 |
profits associated with the incremental sales that
|
19 |
generate. If you take that language, I think you can
|
20 |
create a viable safe harbor at least that would at least
|
21 |
give firms some comfort that certain practices would be
|
22 |
presumed to be legitimate.
|
23 |
And I actually believe, having done my work in
|
24 |
LePage's, that the behavior of 3M would actually satisfy
|
25 |
this safe harbor test. But you don't want to condemn |
142
1 |
nondiscriminatory price cuts in single markets and you
|
2 |
want to be careful not to penalize policies that exclude
|
3 |
less efficient competitors.
|
4 |
That's a different issue because if you want a
|
5 |
test that's workable for a firm that's engaged in a
|
6 |
policy, it's very hard to say you shouldn't exclude a less
|
7 |
efficient competitor because the firm is not going to know
|
8 |
typically whether its competitors are more or less
|
9 |
efficient.
|
10 |
So, this test really is not going to be a
|
11 |
perfect test and probably never will be.
|
12 |
A workable rule should be one that's clear and
|
13 |
manageable. We don't want businesses to say what I hear a
|
14 |
lot in recent years, which is we have no idea which
|
15 |
practices we can engage in or not because anything that
|
16 |
seems to have any bundling aspect to it could lead to a
|
17 |
Third Circuit lawsuit.
|
18 |
Now, as far as the thoughts I have given you, I
|
19 |
just happened to go back and look on the web recently at
|
20 |
the AMC's tentative recommendations. I assume they're
|
21 |
still tentative. And I found myself in agreement with
|
22 |
their recommendations in the areas I am talking about.
|
23 |
There are some other areas I would disagree.
|
24 |
But I noticed that the AMC tentatively is
|
25 |
recommending no need to revise the antitrust laws to apply |
143
1 |
to high tech industries. And I agree very strongly with
|
2 |
that.
|
3 |
The AMC is proposing no need for Congress to
|
4 |
amend Section 2. And I agree strongly with that as well.
|
5 |
And, finally, it looks like the AMC is thinking
|
6 |
of recommending additional clarity and improvement in
|
7 |
Section 2, particularly with respect to areas such as
|
8 |
bundling. And I agree strongly with that as well.
|
9 |
Thank you very much.
|
10 |
(Applause.)
|
11 |
MS. GRIMM: I'd like to thank all of our
|
12 |
panelists.
|
13 |
We are going to take a 15-minute break now.
|
14 |
We'll reconvene in 15 minute for our round-table
|
15 |
discussion.
|
16 |
(A brief recess was taken.)
|
17 |
MS. GRIMM: Before we get to our questions and
|
18 |
round-table discussion, I would like to introduce our
|
19 |
fourth panelist, who will discuss some of the ideas that
|
20 |
have been advanced by our other panelists this afternoon,
|
21 |
as well as some of his own ideas about Section 2.
|
22 |
Carl Shapiro, our fourth panelist, is the
|
23 |
Transamerica Professor of Business Strategy at the Haas
|
24 |
School of Business at the University of California at
|
25 |
Berkeley. He also is Director of the Institute of |
144
1 |
Business and Economic Research and Professor of Economics
|
2 |
in the Economics Department at U.C. Berkeley.
|
3 |
He earned his Ph.D. in economics at MIT in 1981;
|
4 |
taught at Princeton University during the 1980s; and has
|
5 |
been at Berkeley since 1990.
|
6 |
He has been editor of the "Journal of Economic
|
7 |
Perspectives," and a Fellow for the Center for Advanced
|
8 |
Study in the Behavioral Sciences.
|
9 |
Professor Shapiro has published extensively and
|
10 |
his current research interests include antitrust
|
11 |
economics, intellectual property and licensing, product
|
12 |
standards and compatibility, and the economics of networks
|
13 |
and interconnection.
|
14 |
Professor Shapiro served as Deputy Assistant
|
15 |
Attorney General for Economics in the Antitrust Division
|
16 |
of the U.S. Department of Justice in 1995 and 1996.
|
17 |
Carl.
|
18 |
MR. SHAPIRO: Thank you very much. I don't have
|
19 |
any slides. I am going to cover some ideas I have and
|
20 |
then comment on and kind of get the discussion going about
|
21 |
each of the previous panelists.
|
22 |
You probably already picked up the theme here
|
23 |
is that we get up here and we reminisce about the cases
|
24 |
that were brought or investigated while we were at the
|
25 |
Antitrust Division. Okay? We really appreciate you |
145
1 |
coming out here because we all have a love for the
|
2 |
Antitrust Division. FTC, too. And we sort of appreciate
|
3 |
your coming out here so we don't have to go again to D.C.
|
4 |
One of the themes that we've picked up here and
|
5 |
throughout many of these hearings is that Section 2 cases
|
6 |
are inherently really hard because it's a single-firm
|
7 |
conduct and it's not like a cartel case. They're really
|
8 |
hard and there's always elements and you have to be very
|
9 |
careful.
|
10 |
And I don't disagree with any of that, but I
|
11 |
want to focus on they seem to be harder than they need to
|
12 |
be in some cases. And it's one of my themes, intersecting
|
13 |
with the role of patents and plus innovation and
|
14 |
Section 2.
|
15 |
And I'm going to depart from the DOJ reminiscing
|
16 |
and actually talk about the Unocal case, which was brought
|
17 |
by the FTC, and which I served as an expert witness for
|
18 |
complaint counsel. And that was litigated at the -- by an
|
19 |
administrative law judge, before the administrative law
|
20 |
judge.
|
21 |
So, let me just quickly remind you of that case
|
22 |
or tell you about the case. So, Unocal had some patents
|
23 |
-- had patents -- came to have patents during the '90s on
|
24 |
reformulated gasoline. The State of California through
|
25 |
the California Air Resources Board, CARB, established |
146
1 |
regulations for gasoline in order to make the
|
2 |
cleaner-burning reformulated gasoline.
|
3 |
And it came to pass that the regulations that
|
4 |
were adopted, that Unocal's patents, apparently or very
|
5 |
likely, many of the refineries would have to infringe
|
6 |
those patents for a large fraction of the gasoline they
|
7 |
would make if it would comply with the state regulations.
|
8 |
So, and the allegation was that Unocal had acted
|
9 |
deceptively by leading the industry members to believe
|
10 |
that its patents would be -- that either it did not have
|
11 |
patents or would make them available on a royalty-free
|
12 |
basis. That was the representation when the regulations
|
13 |
were being formulated and that Unocal then later sought to
|
14 |
get royalties. That was the allegation of deceptive
|
15 |
conduct.
|
16 |
So, you would think -- well, let's say I would
|
17 |
think, at least, maybe you would think, that this should
|
18 |
be the sort of Section 2 case, and I guess it was FTC
|
19 |
Section 5, and I'm not distinguishing those for my purpose
|
20 |
here, that it would be relatively straightforward.
|
21 |
Big factual question about whether Unocal acted
|
22 |
deceptively. They vigorously denied that they did so.
|
23 |
The FTC or certainly complaint counsel was arguing they
|
24 |
had. I simply assumed that they had for the purposes of
|
25 |
evaluating market power and competitive effect. That was |
147
1 |
the fact question. If they had not engaged in any
|
2 |
deception, I believed there was nothing to the case. That
|
3 |
was my understanding, as I recall it.
|
4 |
So, if they acted deceptively, and let's take
|
5 |
the really cleanest version, they led people to believe
|
6 |
patents would be available on a royalty-free basis.
|
7 |
Regulations are selected. Literally billions of dollars
|
8 |
are invested by refiners to comply with these regulations,
|
9 |
made CARB gasoline, as it is called, and then they
|
10 |
asserted patents.
|
11 |
So, the reason I would say this should be, to my
|
12 |
way of viewing, a relatively simple case because the
|
13 |
conduct alleged and assumed by me, as an expert at least,
|
14 |
deception is not something that we have to wring our hands
|
15 |
over, oh, is that something that's procompetitive, is it
|
16 |
important that companies engage in that sometimes. It's
|
17 |
not like discounting. It's not like product innovation.
|
18 |
Deception.
|
19 |
So, now then the question is, okay, we don't
|
20 |
really have to worry about stifling deception, okay. So,
|
21 |
does it have a significant effect on prices, on market
|
22 |
power? And if they represented that the patents would be
|
23 |
available royalty-free and are later seeking something
|
24 |
like five cents a gallon, to throw out a number, for
|
25 |
pretty much the whole industry a very large fraction of |
148
1 |
the gasoline that would be produced, well, that's a price
|
2 |
increase. There's very strong evidence that would be
|
3 |
passed to the final consumers, motorists. Not that that
|
4 |
matters so much because, even if not, it would be borne by
|
5 |
the direct customers of the technology, refiners, who
|
6 |
would be using the technology. And so you get right away
|
7 |
the competitive effects without any real business
|
8 |
justification for the conduct that's alleged or
|
9 |
challenged.
|
10 |
And yet, Unocal raised many, many arguments.
|
11 |
We do not know how the administrative law judge or the
|
12 |
commission or subsequent appeals court might have reacted
|
13 |
to these. We do know from other cases, the case of --
|
14 |
the Rambus case. There are a variety of Rambus
|
15 |
cases that also involve similar allegations regarding
|
16 |
standards and patents. And we know from other cases I
|
17 |
won't get into that the courts have tended to say, well,
|
18 |
wait a minute, you have a patent and so you get some
|
19 |
market power associated with the patent, and so we should
|
20 |
be very careful not to jump on -- not to conclude that,
|
21 |
just because there's market power, somehow it has to do
|
22 |
with anticompetitive conduct, because patents may very
|
23 |
well confer market power in a perfectly desirable way.
|
24 |
So, I guess I'm raising a concern that what
|
25 |
should be a simple case, there seems to be, in some |
149
1 |
quarters at least, sort of a worship of patents that
|
2 |
therefore mixes up market power attributable to the
|
3 |
innovation versus market power -- additional market power
|
4 |
that comes about from conduct, just the sort of thing that
|
5 |
Tim was mentioning, actually, look at additional effects
|
6 |
of the conduct.
|
7 |
And the economic opportunities of hold up I
|
8 |
think are very clear, going back at least to Oliver
|
9 |
Williamson, my distinguished colleague here at Berkeley,
|
10 |
and yet these were denied essentially by Unocal and its
|
11 |
economic expert. That is to say, the notion that once
|
12 |
refiners had invested enormous sums in order to comply
|
13 |
with the regulations, that would necessarily put Unocal in
|
14 |
a stronger bargaining position to get royalties that they
|
15 |
could not have gotten earlier.
|
16 |
So, I would say it's relatively fundamental
|
17 |
economic principles, fairly clear fact pattern, and yet we
|
18 |
have -- and, for example, the whole Antitrust debate about
|
19 |
defining the relevant market. Defendants can often, in
|
20 |
this case at least, try to make that very complicated,
|
21 |
exactly which technologies are in the market and which
|
22 |
ones are substitutes, and what was the best alternative,
|
23 |
and how good was it, and how much -- they even argued, our
|
24 |
technology is so good that people would have picked it
|
25 |
anyhow and, therefore, even if we engaged in deception, it |
150
1 |
wouldn't matter.
|
2 |
Well, I just don't think that's right because
|
3 |
there's additional market power that results from lock in.
|
4 |
So, sometimes the elements that we always think
|
5 |
of for Section 2 cases: defining the market, measuring
|
6 |
the market power; being cognizant of preexisting market
|
7 |
power, in this case because of patents, I think we need to
|
8 |
be careful not to lose sight of what may be a simple or
|
9 |
more direct argument that can get us to analysis without
|
10 |
doing -- without necessarily following some of these steps
|
11 |
and without getting tied up particularly in market
|
12 |
definition. And, again, Tim, I know, emphasized that he
|
13 |
really, as most economists, if we can, we want to get to
|
14 |
competitive effects. And market definition may or may not
|
15 |
be helpful in getting us there in market shares.
|
16 |
And if you think about the cases I've described
|
17 |
today, measuring exactly which share of how much of the
|
18 |
gasoline infringes or might infringe and what other
|
19 |
technologies are being used is a distraction,
|
20 |
fundamentally a distraction to what's being looked at
|
21 |
here.
|
22 |
And that came in in terms of remedy as well. My
|
23 |
testimony was, we should restore competition, which means
|
24 |
they should license these patents on a royalty-free basis,
|
25 |
as they had represented under my working assumption. And |
151
1 |
yet Unocal argued that, well, our technology is so good
|
2 |
that we should be able to charge more than that, even if
|
3 |
we engaged in deception, because under competition somehow
|
4 |
they would have been able to charge a lot.
|
5 |
Then you ask, well, then why did you act
|
6 |
deceptively. And they say, well, we didn't. Well, what
|
7 |
if you have. So, you go back and forth. All right.
|
8 |
So, while I'm not expecting the DOJ or FTC to
|
9 |
suggest that we throw out market definition, for example,
|
10 |
in Section 2 cases. I do think looking for shortcuts that
|
11 |
are reliable is a good thing to do.
|
12 |
Let me go on to say something about the previous
|
13 |
speakers now that I've made some points about some of my
|
14 |
own thoughts about Unocal.
|
15 |
So, Tim first, Professor Bresnahan. Very
|
16 |
gracious of him to come up here to Berkeley and appreciate
|
17 |
his kind words about Berkeley. I will try to reciprocate
|
18 |
and I will make two trips to Stanford in the next week for
|
19 |
conferences there, and with pleasure.
|
20 |
I took some of what you said, Tim, to be
|
21 |
suggesting that we could think of screening cases based on
|
22 |
whether there's a theory of harm that the conduct would
|
23 |
lead to a significant increase in market power, or let's
|
24 |
put that differently, relax the constraints on pricing
|
25 |
that are facing the firm that's accused, or the defendant |
152
1 |
firm.
|
2 |
And I think that's a really good way to go. So,
|
3 |
I support that.
|
4 |
One way I like to think about it is we could ask
|
5 |
if the conduct is directed at certain competitors or maybe
|
6 |
at certain distributors who then would be important for
|
7 |
certain other competitors in your Dentsply case, we could
|
8 |
ask, if the conduct was really effective and eliminated
|
9 |
those competitors, a certain class or group of
|
10 |
competitors, would the firm be able to significantly raise
|
11 |
price. Or, alternatively, if those competitors were fully
|
12 |
enabled, would that lead prices to fall significantly.
|
13 |
If that's true, then we need to proceed further
|
14 |
in the inquiry. If not, because the price is really
|
15 |
governed by some other set of dynamics, you know, in the
|
16 |
case of patented drugs, if you get rid of the generic
|
17 |
competition, that would usually lead to a higher price,
|
18 |
but it could be in some cases that competition from other
|
19 |
patented drugs is what's driving price or, in principle,
|
20 |
that sort of competition, and then we could stop that
|
21 |
inquiry if the targets were not really providing sufficient
|
22 |
competitive discipline. So, I am very supportive of that
|
23 |
line.
|
24 |
You said at some point, Tim, that it was very
|
25 |
hard to do some sort of balancing, you know, particularly |
153
1 |
quantifying the balancing of net effects, harm to
|
2 |
consumers, benefits to consumers. And so I guess the
|
3 |
economic theorists, I guess that's going to include me
|
4 |
now, may like to measure all these things and do this in
|
5 |
our models, but in practice that balancing would be hard
|
6 |
to do. It is hard to do.
|
7 |
One thing we might do is then focus more on the
|
8 |
competitive process, rather than necessarily a particular
|
9 |
outcome.
|
10 |
But you also said the defendant could show that
|
11 |
the practices were efficient and that would be a defense.
|
12 |
So, if there was anticompetitive danger, the defense could
|
13 |
come back and say the practices were efficient. I don't
|
14 |
know what that means in practice. I guess I'd like to
|
15 |
hear more from you about that. Because there is typically
|
16 |
going to be some story about, oh, this has lower prices
|
17 |
for some customers so it's efficient, or this is going to
|
18 |
prevent free riding, so I need to have exclusive dealing
|
19 |
here. There's going to be some efficiency story and I
|
20 |
don't understand how you can avoid doing some balancing
|
21 |
after the efficiency flag is raised and now are we done.
|
22 |
I don't think you mean they're done just because the
|
23 |
defense raises the efficiency argument. So, what happens
|
24 |
next?
|
25 |
My last comment was on -- I don't want to get |
154
1 |
into Microsoft. Believe me, I really don't want to get
|
2 |
into Microsoft. But you did mention -- I like your term,
|
3 |
the "remedy fizzle." I don't know if you coined that
|
4 |
term, but I like it. You took some responsibility, I
|
5 |
think --
|
6 |
MR. BRESNAHAN: I lived that term.
|
7 |
MR. SHAPIRO: For years, right? I just wanted
|
8 |
to share the responsibility because, having testified for
|
9 |
the states at the remedy phase, I want to share that
|
10 |
responsibility with you.
|
11 |
MR. SHAPIRO: Rich -- next, Rich Gilbert. I
|
12 |
really liked to hear what you had to say about interfaces,
|
13 |
Rich, because this seems to me -- I kept coming -- this
|
14 |
came up when I heard you talk about IBM and Microsoft and
|
15 |
other examples, it seems to me, going back to at least
|
16 |
IBM, and probably selling machines in the 19th century or
|
17 |
something, you've often got this pattern where, I have a
|
18 |
product and I innovate, I improve it and, as part of
|
19 |
improving it, I change the interface or I start producing
|
20 |
a complementary product that needs to be compatible and
|
21 |
it's innovative and very often intellectual property
|
22 |
rights are used to control or secure an interface. And
|
23 |
yet we know from the telecommunications, we know from
|
24 |
other network industries, that controlling interfaces can
|
25 |
lead to a certain octopus-like nature from what might be a |
155
1 |
secure monopoly in one product initially.
|
2 |
And speaking for myself, I get really torn
|
3 |
because I feel like, well, fine, the monopolist, if you
|
4 |
want to call them, improved their product. Integration,
|
5 |
where different components are integrated together, is a
|
6 |
very important element of improved performance, and so how
|
7 |
are we going to draw these boundaries. You know, do we
|
8 |
want to treat interfaces differently, for example, either
|
9 |
under a copyright or patents or how does it intersect with
|
10 |
antitrust. I think these things are hard and I wonder if
|
11 |
you want to say more about that.
|
12 |
I was -- it was shocking to me, I have to say,
|
13 |
to have an economist tell lawyers to focus on the process
|
14 |
rather than the outcome. I just --
|
15 |
MR. GILBERT: Not the first today.
|
16 |
MR. SHAPIRO: I know, it's true. This is all
|
17 |
the more shocking because lawyers are very good at process
|
18 |
in my experience and economists are always thinking about
|
19 |
these outcomes and are often blind to the process. So, I
|
20 |
just -- I don't know, we might have to revoke your card.
|
21 |
I don't know.
|
22 |
And then -- well, I guess I was maybe not
|
23 |
shocked, but a little surprised that you said, well, the
|
24 |
courts have done fine because all of this is hard. If
|
25 |
it's sham innovation that's your standard at the end, that |
156
1 |
seems very hard for plaintiffs. And maybe that's what you
|
2 |
want. I mean, what would it take -- what would count
|
3 |
as a sham? Could you give us an example? For example, to
|
4 |
say where, well, the product is a little better but they
|
5 |
didn't have to do it this way, for example. What would be
|
6 |
a sham? You know, I think it's sort of ironic when I
|
7 |
think about Microsoft -- I said I wouldn't talk about it
|
8 |
much -- but one of the things Microsoft really pushed
|
9 |
throughout the trial was freedom to design their product
|
10 |
the way they wanted to and the great benefit of
|
11 |
integrating different features, as opposed to more
|
12 |
components or modular.
|
13 |
Well, what is it now, eight, ten years later? I
|
14 |
think they're really having trouble because what the
|
15 |
computer science community always does know is, no, that's
|
16 |
not good design. Good design is modular and basically
|
17 |
people on the other side are telling Microsoft, you
|
18 |
wouldn't do this except for strategic reasons. And now in
|
19 |
a way that's sort of spaghetti code or the increasingly
|
20 |
complexity of Windows has made it very, very hard for them
|
21 |
to meet deadlines in terms of coming out with new versions
|
22 |
and a lot of other problems they've had.
|
23 |
So, what would you do in that case to say, well,
|
24 |
you don't have to design it this way, or maybe you don't
|
25 |
want to go there if it's not a sham. Any company can |
157
1 |
choose how to design their product, even if it's not
|
2 |
something they would choose to do except for strategic or
|
3 |
exclusionary reasons. Or is that too intensive. I don't
|
4 |
know.
|
5 |
But maybe, and you can confirm this, Rich,
|
6 |
you're saying it's so hard to do these cases, that it's
|
7 |
true a sham innovation standard is very hard for a
|
8 |
plaintiff, but that's okay and we're just not going to get
|
9 |
many cases. And maybe that's where we're at. Is that
|
10 |
what you support?
|
11 |
Dan. I will finish soon here. Dan, there's a
|
12 |
lot to say, but I noticed you were emphasizing the
|
13 |
somewhat novel nature of network effects and the coining
|
14 |
of the application "barrier to entry" in the mid to late
|
15 |
'90s by you and Joel Klein, I guess.
|
16 |
I have to tell a little story. So, Mike Katz
|
17 |
and I did work on network effects going back to the '80s.
|
18 |
And so we're working -- (laughter). No, that's neither
|
19 |
here nor there. Academics can do anything, but until it
|
20 |
comes into practice... So -- but I just want to tell a
|
21 |
little story around that.
|
22 |
So, we're working in the early '80s and we're
|
23 |
working on the network effect. And actually personal
|
24 |
computers and computer software is a good example of
|
25 |
applications -- that was our example, actually, |
158
1 |
applications that run on an operating system.
|
2 |
And Mike said to me -- and we're getting kind of
|
3 |
excited about this and I guess we got published in a top
|
4 |
journal, and Mike says, this is great, but I have to tell
|
5 |
you, I have a friend who is doing a lot more with this.
|
6 |
Not a friend. I should say, a former classmate. So, he
|
7 |
says, back when he was at Harvard, there was this guy and
|
8 |
he was making a lot of money on this. The guy's name was
|
9 |
Bill Gates.
|
10 |
So, we often think, oh, we work out these
|
11 |
theories, but often after somebody else puts them into
|
12 |
practice and understands them pretty well, then the law
|
13 |
can kind of catch up with that and maybe academics as
|
14 |
well.
|
15 |
Okay, I'll leave it at that.
|
16 |
MS. GRIMM: Tim, would you like to start off
|
17 |
here and respond?
|
18 |
MR. BRESNAHAN: Yes, I want to start off. I'm
|
19 |
not sure I want to respond. I really like Carl's
|
20 |
restatement of my screening idea. That was exactly what I
|
21 |
was trying to say.
|
22 |
Let me take on hard-to-balance because I don't
|
23 |
think I'm against balancing. And I want to use the
|
24 |
example of sham innovation because I think that's pretty
|
25 |
interesting. |
159
1 |
The art of balancing, I'm against two things
|
2 |
that sounds like balancing. One is a burden-shifting
|
3 |
argument that suggests either an efficiency defense,
|
4 |
defendant has to show that one rule really is better than
|
5 |
the other quantitatively, or in a plaintiff's case where
|
6 |
some sort of efficiency defense has been raised, an
|
7 |
argument that plaintiff has to show that the world is
|
8 |
going to be better off without the market power.
|
9 |
I think that those procedures in which one party
|
10 |
or the other has to sort of calculate the counterattack
|
11 |
from the rule with precision are not going to go very far.
|
12 |
And I guess I wouldn't go all the way to saying
|
13 |
we should only like the competitive process. But, you
|
14 |
know, a courtroom is a hostile environment for numbers.
|
15 |
That's just a fact. There are things that courts are
|
16 |
better at than numbers. So, a quantitative balancing I
|
17 |
think is going to be very difficult.
|
18 |
If we were going to have something, for
|
19 |
example, bigger than sham innovation, what if a
|
20 |
court were going to say, you know, cutting off future
|
21 |
races to replace Office and Windows, cutting off the
|
22 |
widespread distribution of new innovations in the PC
|
23 |
business sounds like a lot of harm to competition to me.
|
24 |
There's maybe a lot of zeros at the end of the numbers.
|
25 |
Mixing the code between the early stage browser and |
160
1 |
the operating system, you know, you really got to hold
|
2 |
your nose to call that innovation. Maybe there was
|
3 |
something innovative to it. Maybe there were some
|
4 |
benefits to integration, but it doesn't sound very
|
5 |
innovative to me. So in this case the balance is
|
6 |
pretty obvious.
|
7 |
At that level of a balancing test, I'd be very
|
8 |
comfortable, and I think I'd be comfortable with a broader
|
9 |
definition than just the innovation has to be literally a
|
10 |
sham. I guess I'd be comfortable with the view that the
|
11 |
court can feel that the efficiencies are either clearly
|
12 |
smaller or clearly -- not smaller in a quantitative sense,
|
13 |
but in a salient sense or in a quality of evidence sense
|
14 |
than the market power or vice versa. So, I'd be in favor
|
15 |
of balancing. I just don't want to do it first.
|
16 |
And I think the question that Rich raised
|
17 |
earlier about, all the traditional tests are going
|
18 |
to look pretty bad for innovation, I guess I would want
|
19 |
a balancing test in that area. There's a lot of things
|
20 |
that can get labeled as innovation. There's a lot of
|
21 |
things which may seem like "innovation" to the defendant
|
22 |
but which are dramatically less innovative than what
|
23 |
other firms in the industry can do. I think this is
|
24 |
one of the enduring lessons of the Microsoft case.
|
25 |
On one of my trips to Silicon Valley to discuss |
161
1 |
the Microsoft case, I talked to a roomful of people and somebody
|
2 |
said, weren't they accused of "innovating too fast." And
|
3 |
somebody else said, they can't possibly be guilty of
|
4 |
innovating too fast; those guys (Microsoft) have never
|
5 |
innovated too fast in their lives; they never innovate fast
|
6 |
enough. And stuff like that will come out in a courtroom.
|
7 |
For this reason, I think that a standard that
|
8 |
innovation has to be a sham is too narrow.
|
9 |
MS. GRIMM: Professor Gilbert?
|
10 |
MR. GILBERT: Well, when I started this project
|
11 |
of looking at standards for innovation, I did a lot of
|
12 |
reading. And one of the papers I came across was the
|
13 |
paper by a Mark Popofsky. And Mark, in that paper,
|
14 |
advocated basically different standards for different
|
15 |
types of conduct, very much a process-oriented approach.
|
16 |
And my initial reaction when I read that paper
|
17 |
was I sort of reeled back and said, oh, this doesn't make
|
18 |
any sense at all where we're going to put everything that
|
19 |
goes on in the economy in a separate category and have a
|
20 |
different set of antitrust rules for it. I guess at that
|
21 |
point I still had my economist card.
|
22 |
But the more I looked at this area, the more I
|
23 |
started to think, how do we actually do this analysis and
|
24 |
what do you have to take into account to do the analysis
|
25 |
right, the more I was led to the conclusion that maybe |
162
1 |
Mark got it right, that there were certain things that you
|
2 |
do and a lot of things you can't do, and that different
|
3 |
standards apply to different types of conduct.
|
4 |
I mean, certainly the failure to innovate is not
|
5 |
an antitrust violation, even though it's really what we're
|
6 |
concerned about or should be concerned about.
|
7 |
Other problems in this -- along this line, I
|
8 |
have a paper with Mike Reardon where we look at
|
9 |
technological tying. And the point of that paper is that
|
10 |
there are lots of different outcomes. And even if you had
|
11 |
really good information, you could do an analysis and you
|
12 |
really could examine the problem, you don't know which
|
13 |
equilibrium outcome is going to occur in the market. And
|
14 |
there could be good outcomes from technological tying and
|
15 |
there could be bad outcomes from technological tying. But
|
16 |
putting a court into the position of trying to figure out
|
17 |
which equilibrium the market is at and which one is
|
18 |
better, that's a tough place to be.
|
19 |
But I do understand that a lot of this conduct
|
20 |
can have very undesirable consequences. If there are less
|
21 |
restrictive alternatives, and you can identify them and
|
22 |
really carve them out from the conduct, well, that's
|
23 |
great. But unfortunately, lots of times the restriction
|
24 |
that goes along with an innovation is inherent in the
|
25 |
innovation. That's where it's difficult. I think, of |
163
1 |
course, if you can separate it out, that's fine, it's a
|
2 |
lot easier.
|
3 |
You mentioned IP protection. Yeah, it would be
|
4 |
nice if we could -- it's hard to find an academic these
|
5 |
days who wouldn't like to see lesser IP protections, and
|
6 |
particularly for things that have network externalities,
|
7 |
the other barriers to entry like interface standards. But
|
8 |
that's a little bit out of our area.
|
9 |
Let me talk a little bit about sham innovation.
|
10 |
Again, I'm very sympathetic to the concept that just
|
11 |
calling it innovation should not be able to protect all
|
12 |
kinds of undesirable conduct and consequences. That just
|
13 |
seems pretty obvious.
|
14 |
But how you actually measure how discrete an
|
15 |
innovation has to be before it is not a sham brings you
|
16 |
right into the kind of numbers that Tim was saying are
|
17 |
very hard for a court or anybody else to do. What number
|
18 |
is big enough? And it's not just the innovation need,
|
19 |
it's when the innovation occurs and how it occurs. Is it
|
20 |
rolled out in every market, does that make it a sham or
|
21 |
not?
|
22 |
And I come back to this unilateral refusal to
|
23 |
deal analogy. Without defending -- I don't want to defend
|
24 |
a "Trinko" approach, but I just find it very odd that
|
25 |
innovation that has similar consequences should be held to |
164
1 |
a higher standard.
|
2 |
So, I still think there are things that are
|
3 |
unlawful. I don't think that innovation should be able to
|
4 |
protect all kinds of activity. But when you are looking
|
5 |
hard to do, it's almost impossibly hard to take into
|
6 |
account all of the incentive effects and the chilling
|
7 |
effects if you get it wrong.
|
8 |
And the bottom line, it seems to me, is that
|
9 |
most of the time we're not going to have a problem and you
|
10 |
should just be careful about chilling innovation by
|
11 |
intervening where there might be a problem unless you're
|
12 |
absolutely, absolutely sure that that's the case.
|
13 |
MS. GRIMM: Professor Rubinfeld?
|
14 |
MR. RUBINFELD: I don't have anything to offer
|
15 |
specifically on that debate. I just have a couple quick
|
16 |
comments.
|
17 |
First of all, most of my good ideas actually
|
18 |
come from Carl Shapiro one way or another. So, my only
|
19 |
intimation was trying to get the courts to see that as
|
20 |
well.
|
21 |
The other thing -- that actually was a serious
|
22 |
comment. But the other slightly more serious comment is
|
23 |
that there is an interesting theme I've noticed just at
|
24 |
least from this group, and that is, when we -- before we
|
25 |
|
165
1 |
went off to Washington in one extent or another, we were,
|
2 |
let's say each of us in our own way, somewhat more
|
3 |
theoretically inclined in thinking about some of these
|
4 |
issues. And the effect of the Washington experience I
|
5 |
think on all of us to one degree or another is really for
|
6 |
us to worry about finding something that's really
|
7 |
operational that will actually help the agencies and
|
8 |
others really resolve practical problems.
|
9 |
And so the emphasis on process, and I would put
|
10 |
it as sort of finding workable kind of second best
|
11 |
solutions, is the natural thing to think about. And I
|
12 |
think that's something I do a lot of.
|
13 |
In another context, for example, I was struck in
|
14 |
a lot of mergers I worked on that we had, I think, at the
|
15 |
division, and also probably at the FTC as well, some very
|
16 |
sophisticated simulation software, which only as far as I
|
17 |
could tell one or two people understood, and not all of
|
18 |
them were in the agency. If you know the folks I'm
|
19 |
talking about, you know what I mean.
|
20 |
And it would have taken in many cases something
|
21 |
like six to eight weeks to make it actually functional,
|
22 |
which is hard to do under a Hart-Scott-Rodino. So, after I
|
23 |
left, I actually, with my co-author, Roy Epstein, wrote
|
24 |
some new software and came up with a much simplified
|
25 |
procedure which, while greatly simplified, actually is |
166
1 |
something you can do within the thirty-day period.
|
2 |
So, a lot of our work has been driven by that
|
3 |
common theme. And I think with respect to sham
|
4 |
litigation, that's sort of the same issue I think we're
|
5 |
all heading towards, which is, we see a problem and now we
|
6 |
have to sort of help to think about what would be a
|
7 |
workable solution for the courts.
|
8 |
MS. GRIMM: I'd like to give our -- all of you
|
9 |
panelists an opportunity to kind of question each other,
|
10 |
if you'd like to, as Carl did for all of you, or to
|
11 |
respond to any of the points made by each other. And then
|
12 |
we'll ask a couple questions on our own.
|
13 |
MR. BRESNAHAN: I'd like to take the bait that
|
14 |
Carl offered us in discussing the Unocal matter, because I
|
15 |
bet that most economists would agree with him that,
|
16 |
if there's some amount of market power or power to
|
17 |
exclude associated with a patent, and if some act,
|
18 |
deception is an extreme, but there might be others,
|
19 |
some act or deception to embed it into an interface
|
20 |
standard, or maybe even just embedding it in an interface
|
21 |
standard in a way that doesn't have any technical
|
22 |
benefits, there's some act that extends the coverage of
|
23 |
that patent and gives the firm that holds the patent a lot
|
24 |
more market power than it would otherwise have, that
|
25 |
that's very troubling. |
167
1 |
And this is one of the disciplinary divides I
|
2 |
think you see between economists and attorneys.
|
3 |
Economists are more eager to take that position.
|
4 |
I suspect that one of the problems with that is
|
5 |
that, patent law hasn't been particularly successful --
|
6 |
forgetting antitrust law for a minute. Patent law hasn't
|
7 |
been particularly successful at delineating the power to
|
8 |
exclude in any particular patent conveys on its owner.
|
9 |
So, when you get into these cases in the
|
10 |
pharmaceutical industry where the patent on the original
|
11 |
molecule is running out but there's a new patent on, the
|
12 |
same molecule but packaged into a lozenge form or something
|
13 |
like that, that it's actually not completely transparent,
|
14 |
what's the right answer to the question, "how much
|
15 |
market power does the patent provide?" And when the
|
16 |
pharmaceutical firm starts playing Carom shots off the
|
17 |
enormous complexities of the regulatory process under
|
18 |
Hatch-Waxman, what is the answer to the question, "how
|
19 |
much market power was conveyed by the original patent?"
|
20 |
So that even if we're fairly comfortable with
|
21 |
the idea that creation of additional market power beyond
|
22 |
what the patent originally would have given that can be
|
23 |
a thing that can be very hard to determine in a legal
|
24 |
sense.
|
25 |
There probably is a near consensus among academic |
168
1 |
economists that patent policy in the United States over
|
2 |
protects the patent holder. I think I agree with Carl on
|
3 |
that. There's this other problem that patent policy is
|
4 |
too vague, that patents simply don't look like property rights
|
5 |
here. You have to go to courts or to the regulatory
|
6 |
system to find out who owns what. And that -- the antitrust
|
7 |
doctrine, Carl quoted the traditional antitrust doctrine that,
|
8 |
intellectual property law if what it is and we ask
|
9 |
whether there's additional market power on top of that.
|
10 |
That may be more attractive in its economics than its law
|
11 |
because it's hard to determine how much market power there
|
12 |
would have been absent the anticompetitive acts.
|
13 |
MR. GILBERT: I'd kind of like to reinforce what
|
14 |
Tim said earlier, Carl, and I think also Dan as well.
|
15 |
While a lot of our discussions today might be
|
16 |
interpreted as suggesting that Section 2 analysis is very
|
17 |
hard to do and therefore we shouldn't do it, and there's a
|
18 |
lot of ways in which I think that's absolutely wrong, and
|
19 |
that is Section 2 analysis isn't that hard and should be
|
20 |
done, I do think that the law creates a road map to make
|
21 |
Section 2 analysis unnecessarily difficult. You've got to
|
22 |
have -- you know, you've got to identify the market, the
|
23 |
product market, the geographic market, you have standing,
|
24 |
you have all of these things. In all of these cases, I
|
25 |
know cases I have been involved in, I'm sure everybody |
169
1 |
else, it seems like you never get to the question.
|
2 |
You know, the relevant question is: Does the
|
3 |
conduct really raise prices. And most of the time that's
|
4 |
pretty obvious whether it does or doesn't and you don't
|
5 |
have to do all this other stuff. And I think the law
|
6 |
often puts us in a position of having to go through this
|
7 |
kind of rogue set of steps that's in many ways very, very
|
8 |
counterproductive.
|
9 |
MR. SHAPIRO: Well, two things. The first one
|
10 |
is to emphasize my concerns about the fetish over patents
|
11 |
in intellectual property rights, therefore in some cases
|
12 |
being a little blind to the fact that they can be
|
13 |
leveraged, if you want to use that word, and you can get
|
14 |
more power than was granted with the patent, particularly
|
15 |
with patents that are very iffy. And there's a whole set
|
16 |
of these questions about that.
|
17 |
I mean, I guess it's outside of Section 2, but
|
18 |
these pharmaceutical settlements cases, like the Shering
|
19 |
case the FTC brought, and where the Second Circuit has
|
20 |
gone with those cases was the tamoxifen case and seeing
|
21 |
the patent as, oh, well, even if you paid off a competitor
|
22 |
to leave because you have a patent, somehow it's okay, it
|
23 |
doesn't mean you've stated an antitrust claim, that's
|
24 |
something the -- you know, even if that's outside
|
25 |
Section 2, that thinking is something that both agencies |
170
1 |
should really head off.
|
2 |
And I guess there's an IP report still coming.
|
3 |
There's -- that seems to be a very important role to
|
4 |
delineate the importance of patents, yes, and the reward,
|
5 |
yes, but there's a limited power that is granted, and
|
6 |
beyond that, we can have abuses.
|
7 |
I would shift topics a little bit and actually
|
8 |
ask a question of Dan that I skipped when I was standing
|
9 |
up.
|
10 |
I'm curious, Dan, in your discussion of
|
11 |
LePage's, whether you -- I guess you favor a bright line
|
12 |
test of comparing price to marginal cost for additional
|
13 |
units sold in a bundle. Or maybe, what about comparing
|
14 |
marginal revenue to marginal cost to see whether the extra
|
15 |
sale and bundling was profitable or not, a kind of profit
|
16 |
sacrifice test.
|
17 |
So, would you favor either of those? I mean,
|
18 |
you're objecting to LePage's as being vague. So, here are
|
19 |
two potential standards that are a lot more specific. I
|
20 |
guess I'm talking about a safe harbor, either if the price
|
21 |
is above marginal cost or if the marginal revenue is above
|
22 |
marginal cost, then the bundling is okay. Of course, even
|
23 |
if it's not, we assume you want to look first back to
|
24 |
scope and so forth. So, there's two questions related to
|
25 |
scope. |
171
1 |
If the program is limited, there's only a few
|
2 |
customers or a short period of time, if that's the case,
|
3 |
would you just wave it through? It just doesn't matter
|
4 |
what the structure of the program is to you because it
|
5 |
couldn't have anticompetitive effects or not?
|
6 |
And then related to that, I don't know if you're
|
7 |
familiar with the EU's approach to this, but they're
|
8 |
required to share methodology and calculating volume
|
9 |
discounts, multi-product or single product, and whether
|
10 |
you think that's something that the U.S. should pick up
|
11 |
on.
|
12 |
MR. RUBINFELD: Good questions, Carl. I
|
13 |
actually am not familiar with the EU side, so I am not
|
14 |
going to try to answer that.
|
15 |
With respect to the workable test, you're right,
|
16 |
I was suggesting just a safe harbor and I think I would
|
17 |
accept your clarification. I was looking for a profit
|
18 |
sacrifice kind of test, so I would compare marginal
|
19 |
revenue and marginal cost, that's if marginal revenue is
|
20 |
different from price, but only to get a safe harbor.
|
21 |
The problem in extending that test is that,
|
22 |
while I think there's some bundling cases which I think
|
23 |
are appropriately seen as really being an extension of a
|
24 |
predatory pricing case and probably ought to come under
|
25 |
Brooke Group, I think there are other kinds of bundling |
172
1 |
practices which probably are not seen that way. So, the
|
2 |
safe harbor I don't think ought to be seen as
|
3 |
characterizing all, all types of bundling. Other types of
|
4 |
bundling might seem more smart with respect to other kinds
|
5 |
of exclusionary conduct of the kind we talked about
|
6 |
earlier today.
|
7 |
The other thing that you asked me about my point
|
8 |
about the effect of this initial program being very
|
9 |
limited. To me that is quite important because -- I may
|
10 |
hear something to the contrary in a second -- but it seems
|
11 |
to me that if there's a practice that cannot be shown to
|
12 |
either have the effect and be sufficiently exclusionary
|
13 |
that it makes a competitor not viable or perhaps even has
|
14 |
no effect on its ability to operate at an efficient scale.
|
15 |
I don't see how that practice ought to be considered
|
16 |
anticompetitive.
|
17 |
So, I think you do -- in my opinion, you do have
|
18 |
to show that if there's exclusion, it's substantial enough
|
19 |
to really matter from the point of view of the potential
|
20 |
competitiveness of the firm that's being affected.
|
21 |
We can debate whether we should focus on volume
|
22 |
scale or efficient scale, but certainly there ought to be
|
23 |
some measurable effect.
|
24 |
MS. LEE: Dan, you had said in your presentation
|
25 |
that a variant on the profit sacrifice test would be |
173
1 |
appropriate to use as a general standard for all Section 2
|
2 |
conduct.
|
3 |
I was hoping that you could refine that a little
|
4 |
bit, in particular, you know, how is this different from
|
5 |
the traditional profit sacrifice test, whatever that may
|
6 |
be, and how does it differ from the no economic sense
|
7 |
test?
|
8 |
MR. RUBINFELD: That's a great question. I
|
9 |
think I really can't -- without going back to my drawing
|
10 |
board for maybe a few years, I don't think I can answer
|
11 |
that very well.
|
12 |
The reason why I was saying a variant in my
|
13 |
comments is that I have been trying to follow some of the
|
14 |
debate in the literature among the folks who prefer more
|
15 |
of a balancing test to a profit sacrifice test. And it's
|
16 |
not that hard to come up with hypotheticals that would
|
17 |
defeat almost any version of a profit sacrifice test under
|
18 |
certain circumstances.
|
19 |
And so what I was imagining was that one would
|
20 |
be able to come up with either a more robust rule that was
|
21 |
not subject to too many of these hypotheticals, or maybe a
|
22 |
complex rule that said under certain circumstances we do
|
23 |
the test one way and under other circumstances another.
|
24 |
But, unfortunately, I don't really have an
|
25 |
answer to that question. I am hoping, Jim, that you and |
174
1 |
others at the Division will work hard to give me an
|
2 |
answer.
|
3 |
MS. LEE: Okay. That was a good way to deflect
|
4 |
the question.
|
5 |
MR. RUBINFELD: Others here may have an answer.
|
6 |
MS. LEE: Let me also get you to react to Tim's
|
7 |
proposal in terms of how we should evaluate Section 2
|
8 |
cases, I would call it a step-wise rule of reason. Tim,
|
9 |
please feel free to disagree with me if you don't think I
|
10 |
am charactering that appropriately.
|
11 |
MR. RUBINFELD: You are asking me that question?
|
12 |
MS. LEE: Yes. How would it be different from a
|
13 |
variant of the profit sacrifice test that you think would
|
14 |
be appropriate.
|
15 |
MR. RUBINFELD: Well, I guess without being too
|
16 |
specific, I have some of the same reactions I guess others
|
17 |
on the panel have expressed based upon my own experience
|
18 |
both in the Division and working on private cases, and
|
19 |
that is the cases often get bogged down in complex debates
|
20 |
about issues like market definition, without really
|
21 |
talking about competitive effects.
|
22 |
So, I'm actually -- at the level Tim is talking
|
23 |
about, I'm very symptomatic with his suggestion. I think
|
24 |
the pharmaceutical cases for me are really an excellent
|
25 |
example of that. I have been involved in a number of |
175
1 |
these where there's a huge battle about market definition,
|
2 |
which can be a very tricky issue in pharma cases for a lot
|
3 |
of reasons, and yet I thought that -- the answer to the
|
4 |
question, how you define the relevant market, at least if
|
5 |
you are using the guidelines, really has almost no impact
|
6 |
on whether there's a competitive effect.
|
7 |
If you think that a generic would have entered
|
8 |
earlier, and the generic most of the time is going to
|
9 |
enter at a substantial discount off the price of the brand
|
10 |
product, there is likely to be an effect. It's going to
|
11 |
be the rare case where competition is driven just by other
|
12 |
branded products.
|
13 |
Now, if you think that's the case, then the real
|
14 |
battle is going to be on issues such as causation, whether
|
15 |
the practice itself had procompetitive benefits, and so
|
16 |
on. So, there will still be a lot to debate, but the
|
17 |
debate will be about whether this competitive effect A and
|
18 |
B, whether there are justifications that say that that
|
19 |
procompetitive effect was worth it.
|
20 |
Rather than debate, which can get pretty far off
|
21 |
the subject, or market power -- certainly most, if not
|
22 |
all, successful brand products generate a lot of market
|
23 |
power. That's the point of Hatch-Waxman to some extent,
|
24 |
or the point of patent laws generally. And -- but the
|
25 |
point of Hatch-Waxman in part is to encourage entry to |
176
1 |
benefit consumers. And the effect of that entry is going
|
2 |
to be to reduce some of that market power.
|
3 |
And I don't think any of that should be very
|
4 |
controversial and yet I have seen a number of cases where
|
5 |
the battles over whether firms have market power seem to
|
6 |
take prominence. And so a process that in my view would
|
7 |
move us more quickly to the heart of the cases would be a
|
8 |
constructive process.
|
9 |
Ms. Lee: Tim, let me ask you to clarify
|
10 |
something that I didn't quite understand about your
|
11 |
proposed way of analysis.
|
12 |
In particular you had suggested that, well, if
|
13 |
you look at -- if you first establish a causal effect
|
14 |
between the act and then the effect, this gets you around
|
15 |
the whole complex processes of trying to figure out what
|
16 |
the appropriate but-for world is when you do the
|
17 |
traditional sort of economic efficiency analysis.
|
18 |
I don't quite see that in terms of, to establish
|
19 |
causality, don't you have to establish in some sense what
|
20 |
the world would have been absent the exclusionary act?
|
21 |
MR. BRESNAHAN: That's a good question.
|
22 |
I agree that to establish causality you need to
|
23 |
say what the world would have been like in a competitive
|
24 |
sense absent the anticompetitive act.
|
25 |
I think the force of my argument is to -- is |
177
1 |
really procedural. It's to move the things which are
|
2 |
going to be most difficult for courts to do back in this
|
3 |
sequence. So, I mean, you heard us all economists say,
|
4 |
it's often easier to see whether there's a competitive
|
5 |
effect than to get market power right. I think that's
|
6 |
probably going to be true.
|
7 |
Certainly if there's a Section 2 case there,
|
8 |
it's going to be easy to see what the competitive effect
|
9 |
is. And then if you can't see it, there's no Section 2
|
10 |
case there.
|
11 |
Similarly, that the challenged conduct causes
|
12 |
the market to be less competitive, that's an inquiry that
|
13 |
can be undertaken within the four walls of what causes
|
14 |
competition, without any balancing against the efficiency
|
15 |
of the challenged conduct. Does it change the conditions
|
16 |
of the competition? And I bet a lot of cases will follow
|
17 |
thereto, and that's within the four walls of ordinary
|
18 |
antitrust analysis. Is the reason that the market is less
|
19 |
competitive because the challenged conduct raises entry
|
20 |
barriers, raises them in a way that, you know, the
|
21 |
entrants and third parties can't get around to the
|
22 |
relevant time frame. Those are all difficult tests to
|
23 |
pass.
|
24 |
So, most Section 2 inquiries should fall by the
|
25 |
wayside. I just want them to fall by the wayside cheaply. |
178
1 |
And then you come to the last thing, which as
|
2 |
we've all said is really, really hard, you know, you've
|
3 |
got causation, there's some challenged conduct which is
|
4 |
changing the conditions of competition, but there's also
|
5 |
something good about it. You know, it's innovative or
|
6 |
it's a price cut so it's especially good for customers,
|
7 |
and now we've got to do this balancing, which I think is a
|
8 |
very, very difficult thing to do.
|
9 |
So, I just want to reduce the incidence of the
|
10 |
balancing. Rather than leaping to that right away, go
|
11 |
through other things first and discard cases. And I
|
12 |
think that the causation -- the causation inquiry which
|
13 |
says, is the challenged conduct holding entry barriers
|
14 |
high is an easier counter-factual inquiry than, is the
|
15 |
extent to which it's holding entry barriers high worse
|
16 |
than its countervailing efficiency. It's got one less
|
17 |
difficulty.
|
18 |
So that would be how I would proceed. And the
|
19 |
basic idea is to save wear and tear on the system, which
|
20 |
is potentially the result.
|
21 |
MS. LEE: Thank you for the clarification.
|
22 |
Rich, I wanted to ask you, you had said you have
|
23 |
became more sympathetic to the idea that in different
|
24 |
Section 2 matters different standards should apply.
|
25 |
How would one go about determining the best |
179
1 |
standards to apply in each situation?
|
2 |
MR. GILBERT: Again, a very good question.
|
3 |
Certainly what sets innovation apart is the
|
4 |
temporal linkage and very complicated linkage between the
|
5 |
conduct at issue and the investment research and
|
6 |
development that create the innovation and the prospects
|
7 |
that any antitrust venture that would show that kind of
|
8 |
very beneficial investment. And suppose you had a case
|
9 |
where you didn't think that linkage was all that
|
10 |
important, so you intervene in that case. But then if you
|
11 |
do that, that also creates a precedence for there being
|
12 |
other cases the linkage could be very important, and you
|
13 |
definitely don't want to chill innovation in those other
|
14 |
cases.
|
15 |
If you think about how some of those early cases
|
16 |
-- if some of those early cases came out differently,
|
17 |
because almost all the cases that I can see ultimately
|
18 |
basically are pretty close to a sham innovation test. If
|
19 |
they had done something very different from that, what the
|
20 |
implications would be for people actually involved in
|
21 |
product design could be kind of interesting.
|
22 |
Now, there is a lot of conduct where I don't
|
23 |
think those issue are at all significant. You know, they
|
24 |
may be present to some extent, but they're just not
|
25 |
significant. And so if you're talking about ordinary |
180
1 |
exclusive dealing or bundling or whatever, I think in many
|
2 |
of those cases you can if not forget about, certainly
|
3 |
discount, the more complicated intertemporal effects. And
|
4 |
the analysis I think becomes much easier. And the sort of
|
5 |
rule of reason analysis becomes much more possible.
|
6 |
Weighing of benefits and costs becomes more reasonable.
|
7 |
MS. LEE: Carl, do you have anything you want to
|
8 |
say in addition to what you said already about general
|
9 |
standards? You had said in your comments that you were
|
10 |
very sympathetic to a standards approach.
|
11 |
Is there anything else you would like to add?
|
12 |
MR. SHAPIRO: Well, you called it a structured
|
13 |
-- what did you call it?
|
14 |
MS. LEE: No, I called it a step-wise.
|
15 |
MR. SHAPIRO: Good, that's the ticket.
|
16 |
MS. LEE: I think that's what it was.
|
17 |
MR. SHAPIRO: So, I think of it in terms of
|
18 |
screens. Traditionally, the monopoly power screen. You
|
19 |
have a lot of power, and if you don't, then Section 2
|
20 |
doesn't apply.
|
21 |
I think I would push for: Does the conduct hold
|
22 |
up the prospect to leading to significant increase in
|
23 |
market power, okay, as actually a better question to use
|
24 |
as a screen.
|
25 |
Now, the reason I think the traditional screen |
181
1 |
has been applied, it's been assumed if you don't have any
|
2 |
power to start you, you can't manufacture something from
|
3 |
nothing. And that may be true in a lot of cases, although
|
4 |
not always. Maybe deception turns up.
|
5 |
Furthermore, even if you have power to begin
|
6 |
with, if the conduct couldn't add much to it, maybe you
|
7 |
have a patent, then we can dismiss that case, we don't
|
8 |
have to go anywhere. So, you would get something knocked
|
9 |
out on this increment screen that you wouldn't get knocked
|
10 |
out based on a preexisting traditional power screen.
|
11 |
So, I think it's a lot more closely tied to what
|
12 |
Tim was saying at the top of the program here about
|
13 |
looking at effects and increment. And there are ways to
|
14 |
do that, implement that, and I have written about that and
|
15 |
other people have, too. So, that's a general concept I
|
16 |
think that cuts across a lot of cases.
|
17 |
At the same time, I agree with Rich that -- and
|
18 |
I think Dan -- well, profit sacrifice may apply in some
|
19 |
cases but not others, so then you have to be more nuanced.
|
20 |
You know, profit sacrifice would not apply in the Unocal
|
21 |
case.
|
22 |
MS. LEE: So, let me ask you the same question I
|
23 |
asked Rich.
|
24 |
Do you have a suggestion about the methodology
|
25 |
of figuring out, well, which is the best approach in each |
182
1 |
type of matter?
|
2 |
MR. SHAPIRO: It would be very unwise for me to
|
3 |
get into that at this late hour.
|
4 |
MS. GRIMM: I just have one question on
|
5 |
remedies, and this is for Tim. Again, on the Microsoft
|
6 |
remedy which you labeled a fizzle and you said the remedy
|
7 |
in AT&T from your point of view was successful.
|
8 |
I was wondering if could share any views with us
|
9 |
on appropriate remedies in Section 2 cases, perhaps
|
10 |
structural versus the conduct remedies.
|
11 |
MR. BRESNAHAN: I'm almost certain there's no
|
12 |
general law of remedies in Section 2 cases because
|
13 |
Section 2 cases are so context specific and so fact dense.
|
14 |
You know, in the structural remedy that was
|
15 |
negotiated rather than imposed by a court in AT&T, I think
|
16 |
the logic of that was caused by an attempt to minimize the
|
17 |
harm to competition and innovation by walling off the rest
|
18 |
of the industry (by vertical disintegration) from the
|
19 |
necessarily regulated sector of telephony, local phones.
|
20 |
And that's just a very specific argument.
|
21 |
So, some principle that has remedies that are
|
22 |
reasonably proportional to the harm to competition that's
|
23 |
been proved, I think it's going to -- I think it's going
|
24 |
to be very hard to go farther than that to a broader abstract
|
25 |
statement. |
183
1 |
MR. SHAPIRO: If I could just make a quick
|
2 |
comment. I thought about Microsoft remedies in context
|
3 |
here. At one end you have, sin no more, don't do what you
|
4 |
did before, narrowly defined, maybe defined to reflect the
|
5 |
market changing. And, you know, that doesn't seem to me
|
6 |
that does much to restore competition if there's been real
|
7 |
damage with some lasting effect, okay, if the case was
|
8 |
significant to begin with.
|
9 |
One of the things that was interesting in that
|
10 |
case was that -- and I think it's true in a lot of cases
|
11 |
-- it's very hard to know exactly what the effects are.
|
12 |
So, you can't say, ah, we're trying to engineer the market
|
13 |
to return to a certain state and that's what we mean by
|
14 |
restoring competition.
|
15 |
So, again, in that context, really the case was
|
16 |
about raising entry barriers, as Tim put. My view was,
|
17 |
you should have a remedy that lowered entry barriers and
|
18 |
then come what may. Maybe entry will occur, maybe it
|
19 |
won't.
|
20 |
But sin no more seems to me it's probably going
|
21 |
to be too weak in most cases where the case was worth
|
22 |
bringing to begin with.
|
23 |
MS. LEE: Let me ask a follow-up to that.
|
24 |
If the only suitable remedy is a sin no more
|
25 |
remedy, do you think the agency should bring a Section 2 |
184
1 |
case in that instance?
|
2 |
MR. SHAPIRO: Well, there still could be some
|
3 |
deterrent effects. And there are private cases that
|
4 |
follow on, for example, that could have a major role. And
|
5 |
there were private cases in the Microsoft case that
|
6 |
involved a lot of money.
|
7 |
So, it could well be. I guess I would hope if
|
8 |
it's a major case that either agency could come up with
|
9 |
something a little more effective and maybe even creative.
|
10 |
But, at the same time, partly from the Microsoft
|
11 |
experience, it's very hard for a court to impose a remedy
|
12 |
when the company says this is crazy, it won't work, you'll
|
13 |
destroy all sorts of good things, and the government
|
14 |
agency, you know, yeah, there's information but it's hard
|
15 |
to know. So, I think it's very hard. And so if you are
|
16 |
stuck with sin no more, it could still be worth bringing,
|
17 |
sure.
|
18 |
MS. LEE: Let me just solicit the other
|
19 |
panelists about that. Anything different or anything to
|
20 |
add?
|
21 |
MR. BRESNAHAN: Yes, I guess I'd be more
|
22 |
conservative on this ground than Carl. It's hard to get a
|
23 |
lot in deterrence in this area of antitrust law because
|
24 |
it's so hard to -- you know, we're never going to have a
|
25 |
doctrine that says these specific practices are |
185
1 |
anticompetitive. I mean, guys will just know not to do
|
2 |
those particular practices. It's much more complex than
|
3 |
that.
|
4 |
So, other than generally wanting to keep the
|
5 |
idea that there might be this prosecution of particularly
|
6 |
egregious anticompetitive acts, this is not a great area
|
7 |
where you can get an awful lot of deterrence out of --
|
8 |
you know, out of a case where there's a remedy that
|
9 |
doesn't do anything.
|
10 |
So, I'd be less -- I'd put less emphasis on
|
11 |
deterrence and, more emphasis on the view that it
|
12 |
should really be looking for cases where you can make
|
13 |
a big difference for the American consumer.
|
14 |
I mean, before I was in government, in
|
15 |
connection with Microsoft, I took the position, don't
|
16 |
bring it unless you're going to do something really
|
17 |
big, which I went on to say, probably meant don't bring
|
18 |
it, although that turned out to be wrong. The government
|
19 |
did ask for a remedy that would have changed the
|
20 |
conditions of competition.
|
21 |
I think these experiences are rare, important
|
22 |
and efficacious in the first instance, and seeking
|
23 |
deterrence only, you know, only perhaps in flagrant
|
24 |
examples.
|
25 |
MR. GILBERT: Sometimes, not always of course, |
186
1 |
the case that dominance leads to conduct that is
|
2 |
persistent and durable, that companies in dominant
|
3 |
positions tend to do the same sort of anticompetitive
|
4 |
things. And it's also the case that that dominance is
|
5 |
persistent, that even if you try to break it up, forces
|
6 |
are going to tend to recreate it. And I wouldn't say
|
7 |
that's always true, but that's sometimes true.
|
8 |
But I also say that, even in those cases where
|
9 |
you cannot have a real structural remedy, that structural
|
10 |
remedies wouldn't be very effective, a big case like this
|
11 |
brought by DOJ or FTC has a lot of consequences for these
|
12 |
companies. And I think you have a significant deterrence
|
13 |
effect.
|
14 |
MR. RUBINFELD: The only thing that I was going
|
15 |
to add is, these remedies come out of course not in just
|
16 |
in court decisions we're talking about, but also in
|
17 |
consent decrees that are reached. And I think it makes a
|
18 |
big difference how you craft a consent decree. You know,
|
19 |
I can think of some cases which I was involved in where we
|
20 |
literally got a promise never to do A again and nothing
|
21 |
more. There were other cases where the consent decree
|
22 |
really laid out fairly carefully what we meant by not
|
23 |
doing it again, not only for this company, but also the
|
24 |
consent decree sent a clear message since the consent
|
25 |
decree can be part of the public record. |
187
1 |
So, you can get some deterrence even in a
|
2 |
situation where the structural remedy doesn't work if you
|
3 |
craft the right consent decree. And, obviously, it
|
4 |
depends on every case, but I think obviously the agencies
|
5 |
should and I am sure do think hard about exactly how to
|
6 |
did that. And that's an important exercise.
|
7 |
MR. SHAPIRO: Let me just clarify. There was
|
8 |
kind of a sin no more at one extreme and then I heard a
|
9 |
couple of people talking about structural remedies.
|
10 |
There's a lot of running room in between.
|
11 |
MS. LEE: Agreed.
|
12 |
MS. GRIMM: Well, my watch says it is 4:30. I
|
13 |
would like to thank all of our panelists for being here
|
14 |
this afternoon and sharing with us their very insightful
|
15 |
ideas.
|
16 |
I would also like to thank again the University
|
17 |
of California at Berkeley for their hospitality.
|
18 |
Would everyone please join me in giving our
|
19 |
panelists a round of applause.
|
20 |
(Applause.)
|
21 |
(Whereupon, at 4:30 p.m., the hearing was
|
22 |
concluded.)
|
23 |
|
24 |
|
25 |
|
188
1 |
C E R T I F I C A T I O N O F R E P O R T E R
|
2 |
|
3 |
DOCKET/FILE NUMBER: P062106
|
4 |
CASE TITLE: SECTION 2 HEARING, PREDATORY PRICING
|
5 |
DATE: JANUARY 31, 2007
|
6 |
|
7 |
I HEREBY CERTIFY that the transcript contained
|
8 |
herein is a full and accurate transcript of the notes
|
9 |
taken by me at the hearing on the above cause before the
|
10 |
FEDERAL TRADE COMMISSION to the best of my knowledge and
|
11 |
belief.
|
12 |
|
13 |
|
14 |
|
______________________________ |
|
15 |
|
KATHLEEN CARR MEHEEN, CSR 8748 |
|
16 |
|
17 |
|
18 |
|
19 |
|
20 |
|
21 |
|
22 |
|
23 |
|
24 |
|
25 |
|
|