Due to the lapse in appropriations, Department of Justice websites will not be regularly updated. The Department’s essential law enforcement and national security functions will continue. Please refer to the Department of Justice’s contingency plan for more information.

Understanding Single-Firm Behavior: Tying Session

This document is available in three formats: this web page (for browsing content), PDF (comparable to original document formatting), and WordPerfect. To view the PDF you will need Acrobat Reader, which may be downloaded from the Adobe site. For an official signed copy, please contact the Antitrust Documents Group.





















209:00 A.M. TO 1:00 P.M.




24Reported and transcribed by:

25Susanne Bergling, RMR-CLR




3Director, Bureau of Economics

4Federal Trade Commission




8Antitrust Division, U.S. Department of Justice




12David Evans

13Robin Cooper Feldman

14Mark Popofsky

15Donald J. Russell

16Michael Waldman

17Robert D. Willig










1C O N T E N T S





6     David Evans

7     Robin Cooper Feldman

8     Mark Popofsky

9     Donald J. Russell

10     Michael Waldman

11     Robert D. Willig


13Moderated Discussion














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

2 - - - - -

3MR. SALINGER: Good morning. I am Michael

4Salinger. I am one of the moderators of this session.

5My co-moderator is June Lee from the Antitrust Division

6at DOJ.

7Before we start, I have a few housekeeping

8matters. First, please turn off your cell phones,

9BlackBerries and any other devices that might ring in

10the middle.

11Second, the men's room is immediately to the

12left through the double doors you just came through.

13The women's room is on the left on the far side of the

14elevator banks.

15Third, one safety tip, particularly for

16visitors. In the unlikely event the building alarms go

17off, please proceed calmly and quickly as instructed.

18If we must leave the building, take the stairway, which

19is to the right on the Pennsylvania Avenue side. After

20leaving the building, please follow the stream of FTC

21people, we have practiced this many times, and we will

22all go to the Sculpture Garden, which is across the

23intersection of Constitution Avenue and Seventh Street

24at the other end of the building.

25DR. WILLIG: And have lunch?


1MR. SALINGER: It is a very nice place to have a

2fire drill on a day like today.

3Finally, we request that you not make comments

4or ask questions during the session. Thank you.

5We are honored to have assembled a distinguished

6panel of practitioners and professors who are well

7versed in the issue we will tackle today involving tying

8and product design. Our panelists this morning are

9Michael Waldman, the Charles H. Dyson Professor of

10Management and Professor of Economics at Cornell; David

11Evans, who is the managing director of LECG's Global

12Competition Policy Practice and is Chairman of

13eSapience; Donald Russell, a partner at Robbins,

14Russell, Englert, Orseck & Untereiner; Mark Popofsky, an

15Adjunct Professor at Georgetown University Law Center

16and a partner at Kaye Scholer; Robin Cooper Feldman, an

17Associate Professor of Law at the Hastings College of

18Law at the University of California; and Robert Willig,

19Professor of Economics and Public Affairs at the Woodrow

20Wilson School at Princeton, Director of Competition

21Policy Associates, and a former Deputy Assistant

22Attorney General in DOJ's Antitrust Division.

23In Jefferson Parish, the Court argues, "It is

24far too late in the history of our antitrust

25jurisprudence to question the proposition that certain


1tying arrangements pose an unacceptable risk of stifling

2competition, and therefore, are unreasonable per se."

3That was in 1984. We are now even later in the

4history of our antitrust jurisprudence, and yet we find

5ourselves reconsidering that question. We are doing so

6I think because the tying doctrine has turned out to be

7such a central issue in many of the most important

8antitrust cases of recent years.

9I suspect, although I probably should not make

10forecasts of this sort, that the easy part of today will

11be to get agreement on the proposition that per se

12treatment is inappropriate. Indeed, I read the passage

13I just quoted as, in fact, an admission that if we were

14to start over, that the Court would not choose per se


16The harder task is to figure out how, if the

17Court moves to a rule of reason, as many people think it

18might, how to go about deciding whether a tie is

19reasonable; how, in principle, you distinguish a

20competitive from an anticompetitive tie; and what sort

21of evidence you need. Do you rely on company documents

22about the rationale behind a tie, or if you are

23skeptical of the ability to use company documents to

24determine intent, what objective factors would you look



1We have a really distinguished panel today to

2help us sort through those issues, and so I would like

3to thank them now, and I will probably do it again, but

4I wanted to take the time to do that.

5Now I will turn the microphone over to June to

6make some introductory remarks of her own and to give a

7more complete introduction of the speakers.

8MS. LEE: Welcome to the tying panel, part of an

9ongoing series of hearings into single-firm conduct.

10The Department of Justice's Antitrust Division and the

11Federal Trade Commission are jointly sponsoring these

12hearings to help the advancement of the development of

13the law of Section 2 of the Sherman Act. Transcripts

14and other materials from previous sessions can be found

15on the Department of Justice and Federal Trade

16Commission web sites. Upcoming panels include exclusive

17dealing on November 15th and bundled loyalty discounts

18on November 29th, so mark your calendars.

19Today's session concerns the law and economics

20of tying. As Michael has noted, the treatment of tying

21under the antitrust laws has shifted significantly over

22time. Courts are far less likely to condemn ties today

23than 50 years ago when Justice Felix Frankfurter stated

24in Standard Stations that tying arrangements serve

25hardly any purpose beyond the suppression of


1competition. While economists, some of whom are on this

2panel today, have identified situations where ties pose

3a threat to competition and situations where ties result

4in efficiencies, assessing likely competitive effects in

5a given situation remains a challenge.

6I look forward to learning more about this

7complex topic today. I would like to thank my

8colleagues at the FTC and DOJ for organizing this

9hearing. In particular, I thank Don O'Brien and Joe

10Matelis, and I again reiterate Michael's thanks to the

11panelists for participating in today's panel.

12The organization of the panel is as follows:

13The first four panelists will speak. We will then have

14a short break, followed by the final two panelists.

15Those speakers will then have an opportunity to respond

16to each other's presentations, and this will be followed

17by a moderated discussion.

18Let me now introduce the first speaker. More

19complete biographical descriptions can be found in the

20handout and also can be found on the Antitrust Division

21and FTC's web sites.

22Our first speaker is Michael Waldman, who holds

23the Charles H. Dyson Chair in Management and is a

24Professor of Economics at the Johnson Graduate School of

25Management at Cornell University. Professor Waldman's


1main research area is applied microeconomic theory, and

2his main fields of interest are industrial organization

3and organizational economics. In these areas, he is

4best known for his work on learning and signaling in

5labor markets, the operation of durable goods markets,

6and the strategic use of tying and bundling in product


8Professor Waldman's work has been published in

9many of the top journals in economics, and he is

10currently a co-editor at the Journal of Economic

11Perspectives and an associate editor at the quarterly

12Journal of Economics.


14DR. WALDMAN: Thank you.

15Sorry, I am used to using overheads, and they

16are not set up for that.

17So, I want to start just by saying that a lot of

18my work on or a lot of my thinking on tying comes out of

19discussions with Dennis Carlton, so although Dennis is

20not responsible for any mistakes I make in the

21discussion, he is responsible for lots of the smart

22things I say during the discussion.

23Okay, so basically tying behavior has become a

24lot more focused in the economic theory literature over

25the last, say, 10 or 15 years, and the rationale for


1that is that with the Microsoft case, there has been a

2lot more attention to it, and what has happened since

3the Microsoft case is there has been a lot of

4theoretical contributions trying to focus on getting a

5better understanding of tying. So, you know, as of 15

6years ago, there was this sort of Chicago School

7argument sitting out there, and then Mike Whinston came

8along and sort of tried to sort of get a better sense of

9the Chicago School argument, and then when the Microsoft

10case came out, there has been lots of theory, some by me

11and Dennis, Choi and Stefanides, Barry Nalebuff, to try

12and get a better understanding of the theory associated

13with tying behavior, and there has been a lot of

14progress in terms of that issue, in terms of getting a

15better understanding of tying.

16But in terms of antitrust, it is not so

17clear-cut. So, there is lots of progress on the theory

18side, less progress or less consensus, I should say, in

19terms of what the progress on the theory side tells us

20for what the right policies concerning antitrust should

21be given our advances in terms of the theory.

22So, what I am going to try to do in this

23presentation is use theory and to some extent the old

24theory and the new theory to use as a guide to think

25about, okay, now, if we want to think about


1reformulating optimal antitrust policy, which is what

2the panel is about, what does the theory tell us about


4So, in the talk, what I am going to do is I am

5going to review various theories concerning sort of

6theoretical perspectives concerning tying, efficiency,

7price discrimination, exclusionary motivations and other

8strategic motivations, and then use the lessons of the

9various theories to talk about what that means in terms

10of optimal antitrust policy, and basically kind of jump

11to the conclusion.

12Although Dennis and I have been involved in

13writing a number of papers talking about how tying can

14be used for exclusionary or other types of behaviors

15that lower social welfare, my sense is that, in general,

16one should be very hesitant in terms of intervening in

17terms of tying policies. Although there certainly are

18cases -- and my view is the Microsoft case would be a

19good example -- where tying was used in an

20anticompetitive way that lowered social welfare, it is

21very difficult, given the frequency with which ties

22either have a positive social welfare effect, say

23through efficiency rationales or ambiguous social

24welfare effect through price discrimination rationales,

25it is very hard to kind of have -- I think it is wrong


1to have a very interventionist policy, because on net,

2given the difficulty the courts have in trying to

3identify the relevant motivations, very aggressive

4interventionist policy is likely to lower social welfare

5more often than raise it.

6So, here is what I will go through. I will talk

7briefly about efficiency rationales, price

8discrimination rationales. I think everyone is pretty

9familiar with those. I will not spend too much time

10talking about them. Then I will talk some about where

11most of the new literature has appeared, which is the

12exclusionary tying, start with the Chicago School

13argument and then talk about some of the more recent

14literature which talks about, you know, sort of how

15robust or in some sense when doesn't the Chicago School

16argument hold, both in terms of monopolies and tying,

17the tied market, and monopolizing the tying market. I

18will talk about a few other strategic rationales

19associated with tying and then get back to kind of

20antitrust perspectives, which I just very briefly


22One of the reasons that it is hard to think

23about antitrust intervention in terms of tying is

24because there are so many efficiency reasons associated

25with tying. So, if I just think about it from a


1transactions costs standpoint, there are very many

2reasons to tie goods. So, you would have right shoes

3and left shoes. People do not want to go shopping for a

4right shoe and then go to a different box for a left

5shoe. You know, cars and radios, people typically want

6to have the radio put directly into the car. So, there

7are lots of efficiency rationales for tying, and in some

8sense, almost any good you can find, defined in some

9sense, is a tying of various goods. So, when I bought

10this shirt, clearly the buttons were in some sense tied

11on, both figuratively and literally, okay?

12So, other efficiency rationales are search and

13sorting, which goes back to the old Kenney and Klein

14argument, and then you have variable proportion. So,

15the variable proportions arguments says that, well,

16suppose you have two goods, one that is someone with

17power and one without, if the goods are not tied, then

18there is going to be this inefficient substitution that

19consumers are going to do trying to substitute away from

20the product with market power which has an above

21marginal cost price.

22There has been a fair amount of research on that

23idea, Malella and Nahata has an early paper talking

24about it, Tirole talks about that, in terms of extending

25to after-market monopolization, and I have a paper with


1Dennis and a paper with a Dr. Morita showing how you can

2sort of take that same idea and extend it to

3after-market monopolization by competitive selling.

4I am going to skip over the details of

5after-market monopolization and go straight to price

6discrimination. So, another important reason that one

7might tie is for price discrimination reasons. So,

8there are sort of basically two arguments there. The

9initial argument goes back to a paper by George Stigler,

101968, which talks about negative correlations of values,

11and in Stigler -- so, there is just a simple example.

12Suppose you have an individual one who has a valuation

13on product A of 10 and product B of 6, and individual

14two has the reverse, product A of six and product B of

15ten, well, if you try to sell just product A or if you

16try to sell just product B, you have these heterogenous

17valuations, and so you cannot extract all the consumer

18surplus. By tying them together, creating a bundle, you

19have homogenized the valuations, you are able to extract

20all the surplus.

21Since that initial paper, it has been pointed

22out by a number of authors, in particular McAfee,

23McMillan and Whinston, that, in fact, this negative

24correlation of values is not required to get their

25argument to go through, and so there, I just give an


1example where the valuations are actually independent of

2each other, equal probabilities, and if you worked out

3the profits associated with it, you will see the same

4basic result that Stigler found even though there is no

5negative correlation of values.

6The second price discrimination story is the

7classic metered sales story that goes back to the old

8IBM punch card case kind of concerning -- actually,

9before computers, concerning -- oh, what is the term --

10well, anyway, and basically the idea that you have punch

11cards and you have, let's say, computers -- it was not

12computers -- and what you are doing is you are trying to

13price discriminate. You are trying to give the higher

14price to the individuals who use the good more

15intensively. If the individuals who use the good more

16intensively use the variable commodity, in this case the

17punch cards, at a higher rate, what you do is then you

18can charge a higher price for the variable commodity,

19the punch cards, a lower price on the machine, and that

20allows you to price discriminate.

21Clearly there are social welfare implications.

22It is well known that price discrimination has ambiguous

23social welfare implications, so from the standpoint of

24tying behavior in terms of antitrust, it is not clear

25why you would want to eliminate the ability to use tying


1for price discrimination and allow price discrimination

2in lots of other types of activities. That is likely to

3cause distortions in terms of people trying to price

4discriminate in other ways and might create additional


6Okay, the more recent literature is focused on

7exclusionary tying, and it starts with the Chicago

8School arguments. So, the Chicago School argument says

9you would never tie to extend your market power from

10market A to market B if you are already a monopolist in

11 market A, and the standard example that is given is

12think about right shoes and left shoes, and there I just

13work through a little example of suppose P equals A

14minus bX as demand for pairs of shoes and there is a

15constant marginal cost for shoes, then by basically

16being a monopolist on right shoes, you can extract all

17the monopoly power into left shoes as being sold


19Mike Whinston, in a very important paper, shows

20that that argument is correct in some settings but is

21not completely robust. What he shows is that in a

22one-period setting, if the monopolist's primary good is

23essential, then that argument goes through, but if

24you -- for various reasons or in various ways, if you

25move away from that basic one-period essential setting,


1the argument breaks down. So, in Mike's initial paper,

2he says, well, suppose that the primary good is not

3essential, and so there are some uses for the

4complementary good that do not use the primary good,

5then in some cases, what you can do is you can tie, you

6can drive out the competitors in the complementary

7market, and that allows you to monopolize this part of

8the market that does not use the primary good.

9He and Barry Nalebuff also have arguments where

10the goods are independent and show that tying can

11sometimes be used to get the monopolist to become a more

12aggressive competitor, and that can cause exit, which

13again, is similar to his original argument, and then

14improve profitability.

15Dennis and I have a working paper where we move

16away from the one-period setting, and you still have

17this essential nature of the good, but by moving away

18from the one-period setting as we specifically do in

19terms of durable goods, we show that tying can be used

20to capture later profits given upgrades and switching

21costs, which are common in durable goods markets.

22So, just a very quick summary in terms of tied

23good markets. If it is a one-period setting and the

24product is essential, then tying cannot be used to

25improve profitability, to monopolize this other market.


1It is not going to be a profitable thing to do, but

2there are various reasons that that old Chicago result,

3classic Chicago result is going to go away as you move

4away. It is not as robust a finding as people have


6Another basic argument is monopolizing the tying

7market, and there are a number of papers looking at

8that. So, the arguments that I just talked about with

9saying I am going to use tying to take a monopoly in

10product A and in some sense move it to product B and

11increase my profitability this way, there are a number

12of papers. Whinston in his initial paper has an

13argument along these lines. Dennis and I have an

14argument in a Rand paper of 2002 basically saying that

15what you can sometimes use tying to do is increase or

16preserve your market power in that initial monopolized

17market. In some sense, the paper that Dennis and I have

18formalized the Justice Department argument in the

19Microsoft NetScape browser case, and Choi and Stefanides

20also has an article along those lines.

21There are other strategic rationales I will talk

22about somewhat briefly. There are a pair of nice papers

23by Carbajo, De Meza, Seidman and Chen in 1977, and they

24basically show how tying can sometimes be used as a

25product differentiation device, and the basic idea is if


1you have this alternative product where, say, Bertrand

2competition with identical products, then you know there

3is going to be zero profits in that market, and what

4they show is that by tying, you get away from that

5Bertrand competition/zero profit result, and that can

6actually improve profitability.

7The other one which I will just mention very

8briefly is Dennis and I, along with Joshua Gans from the

9University of Melbourne, are looking at an argument

10where tying is used to shift rents from an alternative

11producer to the monopolist. The sort of novel part of

12that argument is that what happens is actually you tie,

13and the consumers still use the alternative producer's

14product, but that you have changed the nature of the

15pricing game, and it moves some of the profits from the

16alternative producer to the monopolist, and that turns

17out to be, in general, not a good thing for social

18welfare, because the monopolist is spending resources

19producing this alternative product, in which stuff winds

20up not getting used. We are hoping to have a finished

21product in just a month or two.

22So, just in terms of summary, there are a number

23of different rationales for tying, and they have

24different social welfare implications. Efficiency

25rationales tend to increase social welfare when there is


1tying. Price discrimination results tend to be

2ambiguous. Exclusionary tying, social welfare tends to

3fall if you go through the details of these analyses,

4though it is not always guaranteed to do so, and the

5other strategic rationales, the product differentiation

6argument tends to have ambiguous welfare consequences,

7while the rent-shifting argument tends to lower social


9So, now let's turn to what this means in terms

10of antitrust policy. So, I think what it means in terms

11of antitrust policy is that for various types of tying,

12the tying should basically be allowed. So, if it looks

13like efficiency, then clearly there is no reason to

14intervene. If it looks like price discrimination,

15again, price discrimination could hurt, but it could

16also help. Price discrimination has ambiguous social

17welfare consequences, and generally, given that price

18discrimination is allowed in lots and lots of other

19types of activities, it seems odd and probably decreases

20social welfare to just rule this particular type of

21price discrimination illegal.

22Product differentiation, again, if you go

23through the details of those analyses, it tends to be

24ambiguous social welfare effects, and finally, our sense

25or my sense is if the motivation is unclear but the


1primary market is competitive, like in the 1992

2U.S.-Kodak case, it basically makes sense to allow the

3tying, because we know that competitive markets tend to

4maximize social welfare, and in particular, in that

5case, I think that the courts made a mistake, because

6sort of the theory for what was going on there had not

7been spelled out, and they went with some very

8speculative theories. I think the right theory was

9actually one where they were using it to increase


11When might courts think about intervening?

12Well, they might think about intervening in cases of

13exclusion or rent shifting, although I think the

14rent-shifting argument, which Dennis and Joshua and I

15are working on, is one that is very difficult, because

16the details of that argument say that that only works

17when, in fact, there is an efficiency associated with

18the tie if the tie had actually been used. So, I think

19it is very hard in that case to sort of say that there

20was not an efficiency possibility in that.

21So, evidentiary hurdles should be high in these

22cases. Why should the evidentiary hurdle be high? They

23should be high because it is very difficult to judge

24motivation, and as I was just saying earlier on, in the

25absence of being able to judge motivation, if you try to


1intervene aggressively, you are going to wind up hurting

2social welfare more often than helping social welfare.

3I do believe that it makes more sense to intervene on

4contractual ties rather than product design ties,

5because in product design ties, you are getting into the

6kind of internal workings of the firm, and it is a very

7dangerous thing for firms to be doing.

8So, I know we do not have any time, so just to

9give a 15-second conclusion, there has been a lot of

10recent progress in terms of the theory of tying sort of

11going beyond the old Chicago School argument. Although

12we have identified various reasons for why tying could

13make sense from an exclusionary standpoint and we have a

14much better sense of that than before, I think at the

15end of the day, even with those extra things in the

16literature by Barry Melba (ph), myself, Mike Whinston,

17given the difficulty courts have in terms of judging

18motivation, there still should be a very high hurdle

19before intervening in a tying case.

20Okay, thank you very much.


22MS. LEE: Thank you.

23Our next speaker is David Evans, who is the

24Managing Director of LECG's Global Competition Policy

25Practice and Chairman of eSapience. The author of four


1books and over 70 journal articles, he is an authority

2on the economics of high technology and patent-based

3businesses, primarily as it relates to competition

4policy and intellectual property, both in the U.S. and

5the EU.

6He has served as an expert and testified before

7courts, arbitrators, regulatory authorities, and

8legislatures in the U.S. and Europe. In addition to his

9consulting practice, David is an Executive Director of

10the Institute for Competition Law and Economics at the

11University College, London, where he is a visiting



14DR. EVANS: Thanks a lot. I have to say that I

15loved Mike's talk, and I agree with most of it, so I

16could probably just start with a "ditto" and sit down,

17but since I have 15 minutes, I will talk.

18So, I would like to make two points today.

19First, the enforcement agencies really should take a

20leadership position in ending per se liability for

21tying, and they should abandon any form of per se

22analysis themselves, and they should advocate change in

23both Congress and the Supreme Court.

24My second point is that tying is a routine

25competitive practice, as you have heard, and the courts


1and competition authorities should presume that tying is

2efficient or at least benign in the absence of

3significant contrary evidence.

4So, what I would like to do is to turn to my

5first point. So, under Jefferson Parish versus Hyde, at

6least as it is widely understood, a firm that has market

7power in product A is liable under Section 1 of the

8Sherman Act for requiring consumers to take product B.

9Now, hardly anyone in the antitrust profession

10supports what we might call a conditional per se

11analysis. There are lots of articles on tying, many of

12which Michael has surveyed, but you are more likely to

13be hit by lightning than to find a paper by an economist

14that comes close to supporting the Jefferson Parish test

15or anything really like it. Hardly any legal scholars

16advocate that test either. There is just no significant

17economic or judicial learning that supports the view

18that tying should be an especially pernicious business

19practice for which there ought to be an especially high

20level of judicial scrutiny.

21Now, despite that consensus, per se tying cases

22keep on trucking. More than 30 private antitrust cases

23with a per se tying claim have been filed in the last

24five years. Recent ones, just taking a quick look,

25include Jenson versus Oldcastle, importantly, Broadcom


1versus Qualcom, which is a case not only in the U.S. but

2is pretty much worldwide, Munford versus GMNC

3Franchising, and so forth.

4Now, you might also recall that the biggest

5settlement in antitrust history came just three years

6ago after a District Court judge found that MasterCard

7and Visa failed the major elements of the Jefferson

8Parish test as a matter of law on summary judgment. He

9noted, the District Court judge noted, the possibility

10that the courts might require a showing of competitive

11harm, and he left that issue and essentially that issue

12alone for a jury trial. Not surprisingly, MasterCard

13and Visa settled very soon after that.

14Now, some commentators have suggested that

15Independent Ink shows that the Supreme Court has backed

16away from Jefferson Parish. I think there is a recent

17Seventh Circuit decision that suggests just that. Now,

18I really wish it were true in the sense that matters for

19lower courts and businesses, but Justice Stevens appears

20to have been quite careful, at least in my reading, in

21saying nothing whatsoever in his decision in Independent

22Ink that repudiates his decision in Jefferson Parish.

23We continue to have conditional per se liability for

24tying that follows really all too easily from having

25market power in the tying product.


1There are good vibes from Independent Ink, and

2like many, I am optimistic that the Court will

3eventually conclude that tying is a relic of a bygone

4era in antitrust when populist hostility toward business

5practices prevailed and economics had not pointed the

6way, but the U.S. Department of Justice and the Federal

7Trade Commission should not in my view just sit still

8and wait another five years or ten years or whatever for

9that to happen. So, I have, if you will, four

10recommendations for the agencies.

11First, the Justice Department should adopt a

12policy that it will not file claims that companies have

13committed a per se violation of Section 1 of the Sherman

14Act as a result of engaging in tying. Now, I am not

15suggesting that DOJ has, in fact, been trigger-happy.

16In fact, as far as I can tell, the Department has not

17filed any Section 1 tying cases in the last five years,

18although I also do not believe that it has filed any

19significant single-firm conduct cases of any strength in

20the last five years. Maybe I have not counted properly.

21Second, at the next opportunity, DOJ and the FTC

22should encourage the Supreme Court to overrule Jefferson

23Parish. Unfortunately, as far as I can tell, there is

24not anything in the pipeline -- again, at least as far

25as I know -- that would allow the Supreme Court to do



2The two enforcement agencies should also

3encourage Congress to modify or kill Section 3 of the

4Clayton Act. By the way, and maybe I am just not on top

5of what is going on, it is unfathomable to me that the

6Antitrust Modernization Commission has not considered

7tying as part of its agenda for reform. It seems to me

8that the antitrust laws for the 21st Century should not

9target tying as an especially pernicious practice, and I

10think from what we have heard thus far from Michael, I

11think there is a consensus in the profession on this.

12My third point for the agencies is there is a

13bill in Congress now to repeal certain exemptions that

14the insurance industry has from the antitrust laws.

15This is the McCarran-Ferguson Act. Now, that is a

16debate that I sure do not want to wade into today, but

17HR-2401 perpetuates the mistake of treating tying as a

18separate and presumably especially harmful antitrust

19offense, and in my view, the enforcement agencies should

20oppose that provision of the bill.

21Fourth, the Justice Department should embark on

22a global recall of American tying law, perhaps prodded

23 by the FTC's Bureau of Consumer Protection. Following

24our lead, the courts and competition authorities in many

25jurisdictions have subjected tying to some form of per


1se or conditional per se liability. We should let them

2know, and the Justice Department talks to the agencies

3around the world all the time, that there is no sound

4support for that approach.

5Of course, saying farewell to per se liability,

6on which I think we have a consensus, leaves open, as

7Michael suggested earlier, the question of what approach

8we should welcome in its place. That brings me to my

9second proposition. The antitrust laws should set a

10high bar for finding that tying is anticompetitive and

11proscribe a structure to guide that analysis. To

12explain why, let me take a brief detour.

13I hazard to say this, and I have been advised

14not to, but most of us I think are Bayesian at heart;

15that is, to make decisions, we combine prior experience

16with the knowledge at hand, we recognize that given the

17inherent uncertainty, we will surely make mistakes, and

18we consider the likelihood and costs of making the wrong

19decision, and the courts have adopted precisely that

20kind of reasoning implicitly. It really underlies the

21whole distinction between per se and the rule of reason.

22Moreover, the courts have adopted that kind of

23reasoning more or less explicitly. Brooke Group is the

24leading example in antitrust, and there are other recent

25cases in criminal law where the courts adopt more or


1less this kind of Bayesian or error cost kind of

2 analysis.

3When it comes to single-firm conduct, I think it

4is helpful then to think about what prior information

5tells us, what the likelihood of error is, and the cost

6of those errors, and with that I have three general

7observations on analyzing single-firm conduct.

8First and perhaps most importantly, when

9practices are common in pretty competitive markets, we

10have prior information that these practices are

11efficient. That does not mean that they could not be

12used to harm competition, but it does mean that there

13should be a presumption that these practices are

14procompetitive. They really could not survive otherwise

15in competitive markets. Will Baumol and Dan Swanson

16have made this point in their article on price

17discrimination, and the Supreme Court recognized it,

18precisely that point, in Independent Ink, citing their

19 paper.

20Second, juries have a lot of trouble deciding

21complex cases. I have testified before a lot of juries,

22and I have a great respect for the jury system, but

23let's face it, the single-firm cases require complex

24assessment of facts and legal nuances. The DOJ and FTC

25have had trouble agreeing on how to treat bundled


1rebates. Asking 12 average citizens to do so, to

2analyze single-firm conduct cases, I think really

3invites error, and this is a particular problem, of

4course, in private litigation and especially in treble

5damage class action litigation involving single-firm


7My third point, and I think I am in complete

8agreement with Michael Waldman, modern industrial

9organization economics, at least insofar as he has

10discussed it with respect to tying, really I think

11emphasizes the need for caution. We can define in the

12industrial organization literature that businesses have

13the incentive and ability to engage in anticompetitive

14conduct in fairly limited circumstances, and there is

15not a lot of empirical evidence that these circumstances

16hold in practice and not a lot of guidance on how to

17figure them out, and, of course, that varies between

18different practices. I want to be careful in not

19generalizing too much, but I generally think that the

20thrust of the IO literature really does need to suggest


22Now, I am absolutely, positively not arguing for

23the repeal of Section 2 or for gutting Section 2 in

24practice. It plays a very important role in

25disciplining businesses with significant market power.


1I also believe, as Michael pointed out, that as economic

2learning progresses, we may find that it is easier to

3separate bad business practices from good ones, but for

4now, we ought to be pretty cautious about letting the

5courts and ultimately jurors in private litigation

6embark on a rule of reason inquiry without some

7structure, some discipline on it, to reduce the

8likelihood and cost of errors.

9So, let me apply those considerations to tying,

10and at the risk of restating what everyone knows and

11what the courts have acknowledged in Fortner, Jefferson

12Parish and Independent Ink, tying is ubiquitous, it is

13utterly common. Firms make decisions all the time on

14how to design their products and what product lines to

15offer. They take into account consumer demand for

16different options. That demand depends, as Michael

17pointed out, on transactions costs and information

18costs, and those have critical implications for what

19consumers want and what firms ought to offer them to

20maximize profits, and firms take into account their own

21costs of offering different product offerings. As a

22practical matter, that results in product offerings that

23could be characterized as tying pretty much all over the

24 place.

25Mike and I, as I think many of you, have a


1series of papers that go into many of these

2considerations. Perhaps the most important observation

3from that line of papers is that there are fixed costs

4of offering different product combinations, and that

5necessarily limits the variants offered by firms and can

6result in pure bundling or tying.

7Now, the case law sometimes talks about tying

8denying consumers' choice. The fact of the matter is

9that a lot of times, consumers do not want choice. They

10want producers to make decisions for them, because the

11producers are in a better position to really do that,

12and consumer choice is not costless. It can raise

13prices for all consumers as the market gets fragmented.

14So, our prior explication, when we see tying, is

15it is probably efficient and as a result of market

16forces. As the D.C. Circuit noted in its unanimous

17decision in Microsoft, "Bundling by all competitive

18firms implies strong net efficiencies."

19Now, that does not end the analysis. One might

20imagine that economists have spent the last 20 years

21researching the subject of tying and concluded that, as

22a matter of theory, it was a highly plausible,

23anticompetitive strategy for firms with significant

24market power, and you might imagine that economists had

25actually discovered empirical evidence that supported


1those theories, but you would, indeed, be imagining

2this, as Michael's presentation really emphasizes. We

3have lots of insights, but it is very clear from the

4literature that lots of assumptions need to be true in

5order for us to find anticompetitive tying.

6So, how, then, should we analyze tying going

7forward? Well, I agree with Michael, where tying is

8simply a device to engage in price discrimination, I

9would make it per se unlawful. There is no strong

10economic basis, you can have price discrimination in

11common and competitive markets. Michael went through

12whether social welfare increases or decreases, but I

13think what he left out, I think many of us have strong

14priors that in a lot of cases, price discrimination is

15probably beneficial.

16Now, the law of patent misuse could still

17address whether we should limit the returns from

18intellectual property rights by prohibiting tying, but I

19do not think there is any basis a priori for allowing

20patent holders to engage in price discrimination in a

21primary market but not through mechanisms that involve a

22secondary market.

23Otherwise, we should leave open the possibility

24that under the rule of reason, tying practices could be

25found unlawful; however, there again, I agree with


1Michael that plaintiffs should have a high hurdle, and

2if I could have perhaps one extra minute, I will tell

3you what I think that hurdle should be.

4First, plaintiffs should, of course, as a

5starting matter have to show that the defendant has

6significant market power in the tying product that the

7plaintiff has posited, and that, in itself, is a

8movement away from Jefferson Parish, merely inserting

9the words "significant market power" or "monopoly


11Second, plaintiffs should have to show that the

12tying practice has the likely effect of excluding a

13significant amount of competition from the market for

14the tied product. Such exclusion, at least as I

15understand the literature, is really the source of

16competitive harm in really all the economic work or much

17of the economic work in this area.

18Third, plaintiffs should have to raise

19significant doubts that the tying practice is not just

20normal competitive practice that is explained by

21efficiencies for consumers or firms. That means

22plaintiffs should have to show that there are two

23separate products and that in the absence of an

24anticompetitive, exclusionary strategy, we would expect

25that consumers would be offered the tied product without


1the tying product. So, I would put that burden onto the

2plaintiff in the first instance.

3And fourth, plaintiffs should have to show by

4way of economic theory and empirical evidence that the

5defendant has, in fact, embarked on a plausible

6anticompetitive strategy, and we can leave for the

7discussion what that actually requires.

8Ultimately, of course, plaintiffs need to be

9able to demonstrate persuasively that tying will cause a

10net reduction in consumer welfare. I do not think that

11these are impossible hurdles by any means. Plaintiffs

12ought to be able to find evidence to support each of

13these tests if, in fact, a firm has engaged in tying to

14acquire a monopoly in a secondary market or maintain a

15monopoly in a primary market, as might be suggested by

16some of the Carlton/Waldman works.

17So, that is where I end up, all in all pretty

18consistent with Michael. Thank you very much.

19MS. LEE: Thank you.


21MS. LEE: Our next speaker is Don Russell, who

22is a partner at Robbins, Russell, Englert, Orseck &

23Untereiner. In 1977, he joined the Antitrust Division

24of the U.S. Department of Justice, where he served for

2524 years. He was Assistant Chief of the Communications


1and Finance Section from 1986 to 1992, lead attorney in

2the Division's 1994 monopolization case against

3Microsoft, and Chief of the Telecommunications Task

4Force from 1995 to 2001. He is a founding partner of

5his law firm, where he maintains an active antitrust



8MR. RUSSELL: Thank you. I am happy to be here

9this morning with five very smart panelists who are

10going to answer the hard questions, and I am going to

11address the easy one, to a large extent repeating and

12emphasizing, again, what you just heard from David

13Evans, with very small areas of disagreement.

14My basic proposition this morning -- the two

15basic propositions I want to assert are, number one, the

16single most important thing that the FTC and the

17Antitrust Division can do and the easiest thing for them

18to do in this area is to say publicly, clearly,

19frequently and to the Supreme Court, as soon as they get

20a chance to do so, get rid of the per se rule for tying,

21whatever is left of it. We all recognize that it is not

22a true per se rule, but as David explained, it is enough

23of a per se rule that it still causes substantial harm

24and confusion and harm to consumer welfare. So, we

25ought to get rid of it.


1The second point I want to make, and the one

2that I want to spend most of my time on, is the point

3that I think the Supreme Court has indicated very, very

4clearly they are ready to take this step. Certainly

5lower courts have recognized that it would be an

6appropriate step, and many other people have as well,

7and this is the area where I might have a slight

8disagreement with David's reading of the Independent Ink

9decision, which I will get to in a few minutes.

10Let's start with the Jefferson Parish decision

11in 1984. I think you are all probably familiar with the

12basic facts there. I will point out the holding of that

13case, which is that there was no violation of the

14antitrust laws, no tying violation, when the defendant

15did not have market power. That is the holding. Now,

16there are many other things that were said in the case

17that I would describe as dicta, the most famous part of

18that being the one that is up on the slide now and the

19one that Mike Salinger referred to earlier.

20In the opinion, the majority opinion by Justice

21Stevens, he said, "It is far too late in the history of

22our antitrust jurisprudence to question the proposition

23that certain tying arrangements pose an unacceptable

24risk of stifling competition and therefore are

25unreasonable per se." A couple of things I want to


1point out about this sentence, first, as you heard

2earlier, one very easy way to read this sentence is that

3Justice Stevens is saying, well, we really are not sure

4that this is right, but it is far too late to do

5anything about it.

6The second thing I want to point out, going to

7the underlined language on the screen, is the sentence

8is really fundamentally inconsistent with virtually

9everything else that the Supreme Court has said about

10per se rules, the proposition that certain tying

11arrangements, but not necessarily all, pose an

12unacceptable risk to competition. In every other

13context the Supreme Court has said the fact that certain

14do does not mean that you need to have a per se rule

15that encompasses all of them. Per se treatment is

16reserved only for those situations in which it is

17virtually always the case that there is harm to

18competition and virtually never the case that there is a

19substantial efficiency rationale. Therefore, just

20reading this sentence in that context, it makes no


22Going to one of the concurring opinions in

23Jefferson Parish signed by two of the justices, they,

24again, make this point very clearly, that whatever merit

25the policy arguments against the per se rule might have,


1Congress has not done anything about it, and again, this

2seems to me to be pretty clear even back then that these

3two Justices had substantial doubts that the rule made

4any sense, but for other reasons, they did not think it

5was appropriate at that time to do anything about it.

6There were four Justices in that case who, as

7you know, came out and said very plainly and

8straightforwardly, tying should not be regarded as per

9se illegal in any sense, it should be evaluated under

10the rule of reason, and the reason that they said that

11was stated very clearly. It incurs the cost of a rule

12of reason approach without achieving its benefits.

13The second quote there, "The legality of

14petitioners' conduct depends on its competitive

15consequences, not whether it can be labeled 'tying.' If

16the competitive consequences are not those to which the

17per se rule is addressed, then it should not be

18condemned irrespective of its label."

19Now, there may be a few people in the audience

20who have studied all of this history very carefully who

21will realize that what I have done here is played a late

22Halloween trick on you. The second quote there is

23actually from the majority opinion. It is in a footnote

24in Justice Stevens' opinion for the majority. So, even

25then, as he is saying this is per se illegal if the


1defendant has market power, he is saying in almost the

2same breath, well, of course, you really have to look at

3the competitive consequences, not labels, which sounds

4to me an awful lot like rule of reason.

5Looking more specifically at what Justice

6Stevens said were the competitive concerns with tying,

7he identified two. The first is that it would insulate

8the tied product from competitive pressures, and the

9second is that it might increase the social costs of

10market power by facilitating price discrimination, and

11those were the reasons that he advanced for the Court's

12historical hostility towards tying.

13So, let's fast forward to the case that the

14Supreme Court decided earlier this term, the Independent

15Ink case, and again, the basic pattern in the

16proceedings below were quite similar to what had

17happened in Jefferson Parish. The District Court had

18the good sense to rule in favor of the defendant. The

19Court of Appeals, thinking that it was bound by old

20Supreme Court precedence, said no, you cannot rule in

21favor of the defendant here. In Independent Ink, it was

22because of the statement that Justice Stevens had made

23in Jefferson Parish and that the Court had made in other

24cases, if the Government has granted the seller a

25patent, it is fair to presume that the inability to buy


1the product elsewhere gives the seller market power.

2So, when the Supreme Court got this case, which

3had been decided below based on what Justice Stevens had

4said in Jefferson Parish, the Supreme Court unanimously

5reversed in an opinion written by Justice Stevens,

6ironically enough. Why does it change here between what

7Stevens said in Jefferson Parish and what Stevens said

8in Independent Ink?

9The one area where I think I may disagree with

10David Evans is he looks at the Independent Ink decision

11and says Justice Stevens was very careful not to say

12anything that would undermine what he had said about per

13se illegality in Jefferson Parish. I think that is

14factually true. There is nothing that is flatly

15inconsistent between the two decisions, but as I read

16the Independent Ink decision, it is written the way that

17it is precisely because Justice Stevens and the rest of

18the unanimous Court are inviting a re-examination of

19this per se rule and signaling very clearly that they no

20longer believe that it makes any sense.

21Let me go through specifically the reasons why I

22believe that. First, if you look at the actual issue

23that was presented in Independent Ink, it was a very

24simple and very narrow issue. Should you presume market

25power from the fact that there is a patent? The issue


1that was presented in the case had absolutely nothing to

2do with assuming that there is market power, what is the

3appropriate mode of analysis of the antitrust issues?

4But when you look at the Independent Ink decision, the

5Court spends a great deal of time and devotes a great

6deal of attention to precisely that second issue which

7was not raised in this case, and I think it is

8significant that they did so.

9For those of you who are particularly fascinated

10by these issues, I will recommend to you an article that

11was written by Kevin MacDonald, "There's No Tying in

12Baseball," in which I think Kevin does a very, very good

13job of explaining why if you want to look at the narrow

14issue that was presented in Independent Ink, there are

15many, many, many ways the Court could have come out, as

16it did, addressing only the fact that all of its old

17precedence about patents and copyrights and presumptions

18were really being misread. People were relying on

19dicta, and the Court very easily could have

20distinguished those cases and said, you know, that is

21just wrong. When we look at this narrow issue, it has

22to come out the other way. But they went well beyond


24The first reason they gave for the way they came

25out was the presumption that a patent confers market


1power is a vestige of the Court's historical distrust of

2tying arrangements, which seems to me a very odd thing

3to say. It was not saying, you know, the Court's

4historical belief that patents confer market power. It

5was an historical distrust of tying arrangements

6generally, and they emphasized that is what we are

7addressing today. There are some specific quotes here.

8Over the years, this Court's strong disapproval

9of tying arrangements has substantially diminished. The

10dissenters' view in Fortner that tying arrangements may

11well be procompetitive ultimately prevailed. The

12assumption that tying arrangements serve hardly any

13purpose beyond the suppression of competition has not

14been endorsed in any opinion since. That seems to me to

15be very strong language supporting the rule of reason


17When you look at the specific concern that

18Justice Stevens had articulated as a rule in favor of a

19per se prohibition of tying, price discrimination, what

20the Court said in Independent Ink is, "While price

21discrimination may provide evidence of market power...it

22is generally recognized that it also occurs in fully

23competitive markets."

24The Court in Independent Ink gave a second

25reason for why they were coming out differently today


1than they had in the past. They emphasized over and

2over again that there was a very, very solid consensus

3among economists and legal scholars that the old rule

4made no sense, and I think what we have heard from this

5morning and what we probably all knew before we came in

6this morning is as to the per se rule against tying,

7there is a very substantial, very solid, very

8long-standing scholarly consensus that that rule makes

9no sense. In Independent Ink, the Supreme Court is

10saying that kind of a consensus is a very important

11consideration when we are deciding these cases.

12The third rule, which is particularly

13interesting, I think, is the Supreme Court talked about

14congressional action that kind of ratified this view

15that maybe tying arrangements are not so bad after all.

16Now, if you look at the legislation they were pointing

17to, they were actually pointing to legislation about,

18you know, this presumption of market power, but look

19again at the way Justice Stevens described this concept.

20"At the same time that our antitrust jurisprudence

21 continued to rely on the assumption" -- not about market

22power -- "the assumption that tying arrangements

23generally serve no legitimate purpose, Congress began

24chipping away at the assumption."

25So, again, I think this opinion in a way is


1misleading and misstating what actually happened but in

2a way that suggests to me that the Court is paving the

3way to get rid of the last vestige of the per se rule.

4And, of course, as to congressional action, they again

5emphasized in Independent Ink, as they have said in

6other recent cases, you know, even this assumption that

7we normally would take congressional acquiescence as

8some sign in favor of keeping our old precedents intact,

9in the antitrust area, it is different, because Congress

10has basically delegated to the courts this common law

11authority to change doctrine over time, and they

12repeated that observation in Independent Ink and

13emphasized it again. So, even if congressional action

14would be helpful to persuade them that they should

15overrule prior cases, they do not regard it as necessary

16in the antitrust arena.

17Reason number four is I think the most important

18reason for today's discussion. The Supreme Court said,

19well, the other thing that has changed is the

20Government's position, the position of the enforcement

21agencies, and again, they walked through a history,

22which some, including Kevin MacDonald, is kind of a

23creative rereading or rewriting of history, to say what

24we did in the past was because the Government was

25telling us to do it in the past. The Government today


1is telling us something very different, and we are going

2to follow the Government's advice, suggesting, again, to

3me that it would be very, very important for the

4Division, for the FTC, to offer that advice to the Court

5and that there is a very high likelihood that the Court

6will accept that advice.

7So, if you want to sum up what the Supreme Court

8said in Independent Ink to explain their decision there,

9almost the last sentence of the opinion says, "Congress,

10the antitrust enforcement agencies, and most economists

11have all reached this conclusion. Today, we reach the

12same conclusion."

13I think that is a very clear indication, you

14know, here is the road map, here are the things we will

15look at if this remaining per se rule comes before us,

16and I think when you look at the record, it is pretty

17clear how they would come out on that.

18Now, I will admit that I may be reading too much

19into this, and I will certainly agree with David,

20virtually every quotation I have put on the screen

21there, you can read it in a different context and you

22can say, well, it is not really inconsistent with the

23per se rule, it is not really inconsistent with

24Jefferson Parish, and they were really just talking

25about this narrow issue about patents and presumptions,


1but I do not really think that that is right, and one of

2the reasons that I do not think it is right, in addition

3to the things that the opinion itself says, are the

4questions and the comments that various Justices made

5during the argument in Independent Ink.

6Justice Stevens was the most active questioner

7and the most active participant in this argument, and

8time after time after time, the issue he focused on is,

9does this per se rule make sense? And if you want to

10get to what seems to be his tentative conclusion, the

11last quote on this screen, "It doesn't seem to me it

12makes any difference whether General Motors has a

13monopoly or not," that is, whether they have market

14power or not, "when it wants to sell two components as

15part of the same package." What he seems to be saying

16here, the question that he keeps asking is, you know,

17why shouldn't that be okay?

18Justice Roberts had an even stronger statement.

19"Much of the economic literature sort of sweeps away

20this question because it rejects the notion of tying as

21a problem in the first place."

22Justice Breyer, again, had many questions all

23devoted to the same point, and, among other things,

24focusing specifically on price discrimination, in which

25he says, "I think most economists, in fact, everyone I


1have read agrees with the notion that price

2discrimination is sometimes good and sometimes bad. The

3scholarly consensus that you see later on when the

4opinion comes out.

5And Justice Scalia, again, in a provocative way

6says, is there anything to this notion of tying as an

7anticompetitive practice at all?

8So, to focus here, I think the Supreme Court in

9the Independent Ink decision has laid out very clearly

10what arguments it needs to hear with respect to the

11remaining per se rule, and they have indicated, I think

12pretty clearly, how they will come out on that question

13if and when it is put in front of them. The first

14point, they point to the Supreme Court's prior

15recognition that tying is often a procompetitive

16practice, which is the way they are now reading that


18Second, they point to a scholarly consensus,

19which I think we will hear today and we have heard

20elsewhere is clearly in place with regard to the per se

21treatment of tying.

22Third, congressional action, the Supreme Court

23has already identified congressional action that they

24think is an indication that maybe tying is not so bad

25all the time anyway.


1The thing that is missing at the moment and the

2thing that I think is critical, which is why I focused

3 my remarks this morning on this, is support for a change

4in the rule from the antitrust agencies. There was an

5opportunity for the Government to do this in the

6Independent Ink case. The question was asked very

7clearly, what is your position on this? And the

8Government's lawyer said, well, Justice O'Connor, who

9argued for rule of reason treatment, made persuasive

10points, but we have not taken a position on that


12I want to make it clear I am not criticizing

13that answer. I think it was perfectly appropriate in

14the context of that case, but I also think it is very

15important, very critical, that the next time the

16question comes up that the Government does take a

17position, which is the per se rule makes no sense. This

18should be a rule of reason analysis.


20MS. LEE: Thank you.

21Our final speaker before we take a short break

22is Mark Popofsky, who has been a partner at Kaye Scholer

23since leaving the Antitrust Division of the Department

24of Justice in 1999, where he was senior counsel to the

25Assistant Attorney General. Mark works in the


1antitrust, intellectual property and technology practice

2groups at Kaye Scholer and chairs the firm's technology

3and competition practices.

4Mark is an Adjunct Professor at Georgetown

5University Law Center where for several years he has

6taught the Advanced Antitrust Law and Economics Seminar.


8MR. POPOFSKY: Thanks, June. It is a pleasure

9to be here today. I would like to thank both

10enforcement agencies for holding these hearings and for

11inviting me to participate in them, and it is nice to

12see so many familiar and well-respected faces here in

13this room, both in the audience and on the panel today.

14I approach this topic like Don Russell as a simple

15country practitioner, a formal federal enforcer, and a

16veteran of several rounds in the Microsoft jungle, a

17veteran of those wars.

18I think it is fair to say, to start with the

19issue that Don talked about and David Evans touched on,

20that if the Supreme Court today were hearing a case

21about whether Jefferson Parish should be overruled,

22there is no doubt in my mind there is a majority on the

23Court right now to overrule Jefferson Parish. I think

24 it is notable in my view that Justice Stevens is not

25among them, and my slight disagreement with Don will be


1I see the opinion in Independent Ink as very craftily

2written by Justice Stevens, who has had a 40-year agenda

3in this area, to say, well, what we are talking about

4today is not Jefferson Parish at all but a special per

5se rule that was applicable to intellectual property and

6perhaps even only to patent ties, and I am here today,

7Justice Stevens, writing for the Court, to address only

8the viability of that per se rule.

9To be sure, much in the decision and especially

10in his reasoning probably was prompted by many of his

11colleagues to get them all on board, and this suggests

12exactly what I said a few minutes ago, there is a

13majority out there to overrule Jefferson Parish, but I

14think it would indeed need a swift kick in the Supreme

15Court's rear by the enforcement agencies, among others,

16to get them to take that next step. I do not think it

17is inevitable.

18But why I think we are here today is to not talk

19about that next step, which may not be inevitable but

20perhaps is upon us soon, but to talk about what happens

21after that. After all, we are here in the Section 2

22single-firm conduct hearings. Whether or not Jefferson

23Parish remains or falls, tying will remain unlawful

24under Section 1 either under the strange presumptive per

25se rule of illegality, which is rebuttable in some


1senses, as Jefferson Parish articulated, or under a full

2or truncated rule of reason. Why are we here, in other

3words, to talk about tying under Section 2 of the

4Sherman Act? What does it accomplish?

5In my view, that question depends on answering

6two questions. The first is the conduct subject to

7Section 2 from a legal perspective. I am not one of

8these fancy guys with a Ph.D. or fancy gals with a Ph.D.

9In a legal sense, does Section 2 reach a broader range

10of conduct that can be labeled tying in Section 1? And

11two, and perhaps most importantly, regardless of the

12answer to that first question, should we have different

13rules of liability for Section 2 for tying-like conduct

14than Section 1? I will address each of these briefly in


16I believe it is fairly clear that Section 2 does

17reach a broader array of tying-like conduct than Section

181. Let me give you three examples. A conditioned

19refusal to deal, which is set up like a good old

20fashioned Colgate policy. The monopolist says to its

21customers, I will not deal with you in the future unless

22you take this tied good with the tying good. The

23 customer acquiesces.

24Suppose, like in a Colgate situation, we do not

25have enough of a basis to infer a Section 1 vertical


1agreement and all we have is, technically, unilateral

2conduct. That is something that Section 2 and, indeed,

3perhaps even Clayton Act Section 3 would reach that

4Sherman Act Section 1 does not, the conditional refusal

5to deal, which could, of course, ripen into an agreement

6but need not.

7The second and more intriguing and important

8example, which I gather we will discuss after the break,

9is technological tying and product design. Now, it is

10notable that the Microsoft case, which I lived, did

11treat technological tying and product design as conduct

12subject to both Section 1 and Section 2, but I think the

13Court really glossed over the issue there. If all you

14have is a monopolist or would-be monopolist designing a

15 product, it is not clear to me that every court is going

16to reach the conclusion that that is the functional

17 equivalent of an agreement or a contractual tie. I

18think it is an issue of great dispute in the case law,

19and that might be yet a second area where a Section 2

20liability rule used for tying makes a substantial


22The third and presently very hot area brought to

23us by one of Don Russell's partners in the LePage's

24cases is bundled discounts, which, of course, is a

25category of conduct that can achieve similar results to


1tying and exclusive dealing. Indeed, tying and

2exclusive dealing, of which there is, of course, going

3to be another forum and of which tying is but a form,

4are just extreme forms of bundled discount. There is a

5discrete rule here. There is law dating back at least

6to the Way and Means case in the Northern District of

7California as to when a bundled discount should be

8treated as an outright tie depending on what percentage

9of the tied item is purchased outside of the bundle, but

10that rule, as I just mentioned, is discreet. It would

11only capture some forms of bundled discounting under

12Section 1, and there will be a large number of bundled

13discounts reached only under Section 2 and not Section


15Bottom line, in my view, there very much is a

16difference between the coverage of the two provisions,

17Section 1 and Section 2, with respect to tying and

18tying-like conduct, and I think it is largely settled

19that there is a difference and it will remain.

20The second issue I wish to address today, the

21appropriate legal standard, is, by contrast, extremely

22unsettled. The issue, put brightly, is whether Section

23 2's legal test for liability for tying is different than

24Section 1's, even assuming here we have the Don Russell,

25David Evans, post-Jefferson Parish, halcion world of


1being under the full rule of reason. So, what I am

2about to say assumes that Don Russell and David Evans

3have, perhaps rightly, won the battle and we are

4confronted with a Section 1 rule of reason rule for

5tying, and the question is, what should we do under

6Section 2?

7Now, stepping back for a minute, I think it is

8critical that the answer to that question we observe to

9turn on what sort of conduct we are talking about. So,

10let me start with the brightest beacon in this area in

11the last ten years, and that is the Microsoft case, for

12that is an intriguing case I think for tying, despite

13the fact that the Government, of course, brought a per

14se claim, and I was in the room when that decision was


16There was, of course, a holding by the Court of

17Appeals that the tie-in in this case involving not just

18a technological tie-in but related conduct should be

19evaluated under the rule of reason, not the per se rule,

20number one, and two and more importantly, and I am sure

21the economists will start jumping up and down, the

22Microsoft Court held there is a difference in what you

23do depending on what market you are looking at.

24And what did the Microsoft Court hold? The

25Microsoft Court held that Section 1 tying law, under the


1rule of reason -- so this is presumably the

2post-Jefferson Parish world come a little sooner because

3the Microsoft Court created an exception to Jefferson

4Parish -- the Court said Section 1 is concerned

5exclusively with harms to competition in the tied

6product market. Look only to harms in the browser

7market, the Court said, ignore this monopoly of

8maintenance in the tying product market, operating

9systems. The Court said, we are, in other words,

10concerned only with how the tied market can be affected.

11Strikingly, the Court also said the standard of

12liability here is higher in some sense under Section 1

13when you are looking at a tied product market than

14Section 2, which, of course, the Court said had to do

15with in that case the tying product market. The Court

16said for a Section 1 rule of reason tying claim, we need

17actual harm to competition in the tied product market.

18The Government must define that market with precision,

19they must show a substantial likelihood of

20anticompetitive effects. Government, you have not even

21gotten past go on that issue, you are likely to lose.

22We are not willing to do what we did in the Section 2

23side of the case -- where the Court said the concern

24about tying under Section 2 requires looking at the

25upstream tying product market -- where the Court was


1willing to infer causation of anticompetitive harm

2merely from the fact that Microsoft engaged in a

3category of conduct which the Court said was likely to

4cause anticompetitive effects.

5So, just to step back and summarize, we have a

6clear difference, the Court of Appeals says, for Section

72 tying and Section 1 tying. Section 1, give me actual

8effects in the tied product market. Section 2 tying,

9give me a reasonable likelihood that we have conduct

10likely to cause upstream monopolization for Section 2

11tying. The liability standard in a very discrete way is

12lower, ironically, under Section 2 after the Microsoft

13decision than Section 1, at least in terms of what is

14the nuance and the measure and the strength of the story

15you have to have as a plaintiff to infer competitive or

16show competitive harm.

17I think this was no accident in this unanimous

18per curiam en banc opinion. Tying, as we have heard, is

19ubiquitous in competitive markets. If you have a legal

20rule that it is very easy to show anticompetitive

21effects that satisfy the rule of reason under Section 1,

22you are potentially going to be condemning under Section

231 a broad swath of otherwise benign conduct. It is very

24easy to get to those jurors David Evans mentioned if you

25can have a Section 1 tying rule that says, basically,


1have any story of plausible anticompetitive effects and

2have a story of some market power. Differentiated

3products, we all know, is very easy to show some market

4power over.

5So, the Court is saying, higher standard for

6liability, at least under some categories of cases under

7Section 1, there -- technological tying. Perhaps a

8break to the plaintiff under Section 2, provided the

9plaintiff has a clear story of how the tie-in can

10actually lead to monopolization of the tying market, and

11it was a story of how NetScape's distribution of

12browsers would enable Microsoft to prevent NetScape from

13reaching certain economies of scale to grow into a

14threat for Microsoft.

15So, whether or not one agrees with what the

16Microsoft Court said about the concern of each provision

17of the Sherman Act, exclusively downstream for Section

181, exclusively upstream for Section 2, you have a court

19saying the rules are different depending on what you are

20looking at for tying, and this leads to my final major


22This says something more general, I think, about

23Section 2 tying, where we are going in this area, and

24importantly, what the enforcement agencies can

25contribute. As I have written recently in an Antitrust


1Law Journal article, there is a holy war raging over the

2appropriate liability standard under Section 2

3generally. Everything, at least almost everything, save

4perhaps very discrete areas like charging a monopoly

5price and after-Trinko refusals to deal, are up for


7In fact, I think this revolution in Section 2 is

8inherent in Trinko, where Trinko itself, often read as a

9very pro-defendant decision, says in designing Section 2

10legal standards, we should be Bayesians, as David Evans

11said. We should look at the risk of type one errors,

12the risk of false positives, type two errors, the risk

13of false negatives, the relative likelihood and the

14magnitude of the likely effects of each, and enforcement

15costs, and under that process, in a very common law

16fashion, courts will arrive at the appropriate Section 2

17doctrine or legal rule for the conduct at issue.

18I think that is where we really are with Section

192 law and tying. Much is up for grabs despite what

20Microsoft said about the difference and focus between

21Section 1 and Section 2, and I think what is yet to be

22written in the next ten years I think will show us is

23where the courts go applying many of the principles that

24Dr. Waldman, Dr. Evans, and I am sure Ms. Feldman will

25enlighten us of about the economic learning and


1translating that into concrete legal tests for discrete


3Now, there is no time today for me to lay out

4plausible stories of where this will take us and

5specific examples of what legal rules might emerge for

6Section 2 law in tying, but let me give you sort of

7three rules of thumb as I see it.

8First, I think as Dr. Waldman said, condemning

9tying through contracts likely poses fewer risks of

10false positives than condemning unilateral tying, true

11unilateral tying, like product design. This suggests

12that some forms of "unilateral tying" reached only under

13Section 2 might have applied to them a more lenient

14legal test for the defendant than Section 1. We might

15indeed have the courts leading to a higher standard of

16what the plaintiff has to show.

17Now, there have been some cases which have gone

18the other way recently. The Teva-Abbott decision, which

19some of you may be aware of, held that a monopolist

20product design decision should be analyzed under the

21rule of reason, did not really get into what that means.

22The next step will be deciding what that rule of reason

23entails under Section 2, whether it is a different

24standard than under Section 1 or the same, and there is

25a good argument it should be different.


1That said, how tying should be treated under

2Section 2 really should not depend on a game of

3formalisms, is it unilateral, is it contractual,

4although that can inform, as I just said, the analysis.

5What is important in this area is that related forms of

6conduct, related from an economic perspective, be

7treated similarly under the antitrust laws. The last

8thing we want is courts all over the country coming up

9with different legal rules that create incentives for

10firms to inefficiently substitute to different conduct

11to avoid the most plaintiff friendly doctrine, and let

12me give you an example of that.

13Suppose courts come out with a rule that

14exclusive dealing, if you have a contract, is under the

15full rule of reason, but exclusive dealing done in the

16form of a conditional refusal to deal, I will only deal

17with you if you deal with me exclusively or I will deal

18with you with bundled discounts and induce you to

19exclusivity, is determined under some different test.

20Courts should think very carefully before taking that

21step. The last thing we want is to induce firms to

22inefficiently substitute to perhaps less efficient

23conduct to avoid what they perceive as the most

24restrictive doctrine.

25The third factor I will mention is that some


1forms of tying present strong or unusual cases for

2 efficiencies. Certain bundles of IP rights, for

3example, may provide an insurance function that other

4tying arrangements lack. There may be special

5efficiencies for certain forms of bundled discounting or

6volume discounts, and those situations might argue for

7differently restructured analyses than the traditional

8general rule of reason, taking into account, as I said,

9you want to treat what the economists demonstrate to be

10economically similar arrangements similarly.

11Backing up in my final point, what does this

12suggest about the role of the enforcement agencies in

13this area? Putting aside the issue of whether the

14agencies should jump on the next opportunity to overrule

15Jefferson Parish v. Hyde, I think through their closing

16statements at the end of investigations, the Section 2

17cases they elect to bring, importantly, the amicus

18briefs they elect to file (a lot of the actions are

19private), the business review letters they issue, and

20the competition advocacy in which the agencies engage,

21particularly as regimes overseas decide what their

22Section 2-like rules of the road are going to be, the

23agencies can play an important role in shaping what

24Section 2's rule of reason looks like as applied to

25tying arrangements in the years to come.


1As I said, much is up for grabs, and this is the

2moment when the agencies should seize the initiative and

3set forth what their views should be of where these

4arrangements should and should not cross the line.

5Thank you very much.


7MS. LEE: We will now take a short break and

8reconvene at five after 11:00.

9(A brief recess was taken.)

10MR. SALINGER: Welcome back. Our next speaker

11is Robin Cooper Feldman. Professor Feldman is an

12Associate Professor of Law at the University of

13California, Hastings College of the Law. She

14specializes in law and bioscience and is Director of

15Hastings' Law and Bioscience Project. Professor Feldman

16also serves on the Executive Committee of the Antitrust

17Section of the American Association of Law Schools.

18Professor Feldman has produced many publications

19in the intellectual property, antitrust, biotechnology

20areas. She received her JD from Stanford, where she

21served in the Articles Department of the Stanford Law

22Review. After graduating from law school, Professor

23Feldman clerked for The Honorable Joseph Sneed at the

24U.S. Court of Appeals for the Ninth Circuit.

25Also, I understand that Professor Feldman is


1going to have to leave the session a little bit before

2we end, so I will take the opportunity now to thank you

3in advance for taking the time to be with us today. So,

4Professor Feldman.


6I agree with our moderators who said that you

7will probably find considerable consensus about moving

8away from a per se rule for tying, the notion that all

9tying is bad and it should be enough to just point out

10the behavior of tying, and maybe with a little more

11information, we can condemn it. I am a little bit word

12worried that in our rush to move away from that old

13position we are going to swing all the way in the other

14direction and end up saying, nothing to see here, folks,

15just move right along, all tying is good. I think there

16is a consensus in the legal, academic and the economic

17literature that all tying is not bad, but it is not true

18that the legal and economic literature believes that all

19tying is good. So, the question is, how do we find out

20how do we identify what it is that we are concerned

21about if we continue to acknowledge that there is

22something of concern?

23I want to talk about Section 2 as it relates to

24technology markets, both high-tech and biotech, and in

25particular, I want to highlight the fact that in my


1view, pharma and biotech are the next frontiers for

2antitrust enforcement in general and for Section 2 in

3particular, and I have chosen some of my examples with

4that in mind.

5I also want to frame my comments in terms of

6what is different about technology markets and what is

7not different about technology markets. In terms of

8what is different about technology markets, I want to

9talk about a particular kind of leveraging, and that is

10what I call defensive leveraging. For almost a century

11legal scholars and economists have struggled to

12understand leveraged behavior and determine when it is

13harmful. Most of that debate has centered on what I

14would call traditional leverage, in which a monopolist

15in one product tries to leverage its power in a

16complementary product. You can imagine an ice cream

17monopolist who bundles and says I will not sell my ice

18cream unless you buy cones as well. With the more

19traditional form of leverage, the economic debate

20concerns whether monopolists can get any profit out of

21that or cause any harm that. But there is another form

22of leveraging, and in this form of leveraging, the

23monopolist is not trying to reach into another market

24and grab more monopoly profits. The monopolist is

25trying to protect its original monopoly from the next


1generation of products that could serve as substitutes.

2It is using the power of multiple markets to maintain

3its original monopoly, and I call this defensive


5Now, technology markets are ripe for this form

6of leveraging, among other reasons, because of their

7tendencies towards network effects. That is, they tend

8to be industries in which there are advantages in doing

9what everyone else is doing. Where there are network

10effects, a monopolist who has the bulk of the customers

11can use its existing base to project into the market for

12new technologies that are threatening to erode its

13original monopoly. So, tech markets are different

14because of their strong potential for defensive


16They are also different because of product

17design challenges, and here, let me offer you a pharma

18example. A few years ago the FTC brought a successful

19enforcement case against a pharmaceutical house that

20sought to tie its dominant drug to a new monitoring

21product. Now, this monitoring product could have been

22used just as easily with all the competitors' drugs, but

23the pharmaceutical company wanted to say we will only

24sell our monitoring product if you will also buy our

25version of the drug. The concern was that the pharma


1house was trying to use its new monitoring product to

2protect its power in the drug market as its power

3started to wane.

4Now, if we would not allow a company in these

5circumstances to tie a drug together with a product that

6monitors the drug, why would we allow a product designed

7to do both, that is, to administer the drug and monitor

8it at the same time? Or from another perspective,

9should we allow two products to be bio-engineered so

10that they work only in combination with each other?

11That is an issue in agri-biotech. If we are not careful

12in the area of product design, what we are doing is

13simply inviting parties to design around the patent laws

14and the antitrust laws, and then the question of whether

15behavior violates the antitrust laws becomes a

16scientific question rather than an economic one, the

17question being, "Is it feasible to combine products

18technologically?" If so, you have no problem with

19enforcement agencies. It should not be that our legal

20decisions turn on questions like that.

21There are tremendous challenges in the areas of

22product design, but whatever benchmarks we develop in

23the law, I believe it is critically important not to be

24dazzled by the wonderful science involved in product

25design. Technology and biodesign are increasingly


1offering avenues for avoiding the appearance of tying

2and bundling simply by manipulating the product. These

3are wonderful products, and it is so easy to be swayed

4by how wonderful they look without asking what is

5happening behind the science. We still have to

6delineate, even if you are talking about biodesign and

7product design, what is reasonable and what is not


9And finally, technology markets are different

10because of patent groupings. Patents tend to travel in

11packs. Companies build or acquire portfolios, and they

12typically engage in defensive patenting; that is, trying

13to file patents for all of the space surrounding their

14key patent so nobody else can develop any substitutes to

15compete. And most importantly, tech products have

16multiple patents within them, which creates


18Now, patent-groupings can be and often are

19 perfectly procompetitive or they can create

20opportunities for strategic anticompetitive behavior.

21The key is, how are we going to find the difference

22between these?

23I talked a little bit about the fact that I

24think there are differences with technology markets.

25They operate differently from what we are accustomed to


1seeing in traditional markets, and they present

2interesting challenges for analyzing behavior. While

3technology markets are different, they are not sacred,

4and I am very concerned by language in some recent court

5decisions which suggest that markets that relate to

6intellectual property should be treated more gently

7under antitrust laws. It is an eerie throw-back to

8language in the early 1900s when courts were struggling

9with the question of whether antitrust laws could even

10be applied to patents or to other intellectual property


12Intellectual property rights are not sacred

13monopolies. They are not even monopolies at all, at

14least not in the antitrust sense of the word. They may

15be downright worthless, and I can discuss some of this

16in the question period. They are not even an exclusive

17right, again, not in the way that antitrust thinks about

18it. There are certainly challenges in understanding

19these rights, but they need to receive the same reasoned

20consideration as other types of products. I use the

21term "reasoned" carefully and also intentionally. It is

22certainly true, as all of the panelists have pointed

23out, that we have moved away from a strict per se rule

24in tying cases, and that we appear poised to move even

25closer to a rule of reason approach, if not completely


1to a rule of reason approach. I am going to jump to a

2world in which we have moved very close or completely to

3the rule of reason. I think the important part of this

4shift will be figuring out how to react when companies

5that engage in tying behavior claim to have very good,

6procompetitive reasons for the tie.

7How do we analyze what is a legitimate

8procompetitive reason and what is not? To do this, I

9want to suggest that we borrow from the experience of

10regulators at other agencies in different contexts, and

11I think there is a perfect example from Patent and

12Trademark Office experience. The PTO requires that

13parties who want to make certain types of claims must

14show that those claims are substantial, and credible.

15I would like to spin out how it works there and how I

16think it would work here.

17A few years back, researchers began fishing out

18little pieces of genes, not the whole gene, but some

19little pieces from a soup of genetic material, and they

20wanted to get a patent on that little piece that they

21found. Now, in order to get a patent, you have to tell

22the PTO how you can use the thing that you are

23patenting. When they fish this little piece out of the

24genetic soup, researchers had no idea what it was. They

25did not know what gene it came from, they did not know


1whether it promoted disease or whether it helped fight

2against disease. They just had a little snippet, and

3they did not have a use for it.

4They began to file patents using very general

5uses. They said, "These little snippets can also be

6used for fishing out other snippets or for doing

7research." This is when the PTO developed its test:

8Specific, substantial and credible. Don't just tell us

9something general that can be true of any of the

10category of things that you are talking about. Tell us

11something specific to what it is that you have found and

12what it is that you are doing.

13I think a test like that, specific, substantial

14and credible, is the essence of what courts and

15regulators are going to have to ask about procompetitive

16defenses offered in tying cases. Don't just give us

17general reasons that would apply to any tie or that

18would apply to any tie in your industry. Give us

19something that is specific to your product and to your


21So, in computers, for example, anyone can say it

22is easier for consumers if you put things together in an

23operating system. When different applications are

24together in an operating system, Ma and Pa do not have

25to worry about loading things together, they do not have


1to worry about interoperability. There are always

2consumer advantages when things are put together in

3computers, but it cannot be that any tie in the computer

4industry is always okay. You must tell us something

5about what it is that you are doing and why we should

6see this as procompetitive.

7If you think outside of computers to products in

8general, any company can say, "We can control quality

9better if we control all the parts you use with our

10equipment or all the pieces that might integrate

11together. Our customers do not suffer through people

12finger-pointing about which part is wrong. They only

13have to call one person when they need a repair." But

14again, that is true of any combination of things. If

15you want to claim a procompetitive benefit, I would say

16tell us something that is specific to your product and

17to your tie.

18I want to point out, again, the reason I am

19concerned is that there has been a swing in the

20pendulum. We needed to talk about what was

21procompetitive about tying in order to move away from

22the notion that all tying is bad. We want to be

23careful, once we have talked about ways in which ties

24can be good, that that does not blind us, and that now

25all we ever talk about are the good things in tying.


1Let me give you an example of something that I

2think would qualify as a specific, procompetitive

3defense for a tie. There was a pharmaceutical house

4that recently received a lot of criticism when it sought

5regulatory approval to combine its existing cholesterol

6drug, that was losing market share, with a new

7blockbuster heart drug and to sell them only as a single

8pill formation. They had a product that was losing

9market share, and they were going to combine it with a

10new kind of blockbuster as the only way consumers were

11going to be able to get it. The company only agreed to

12sell the two separately after a lot of public criticism.

13Imagine, instead, that the company's drug is

14about to be pulled from the market for dangerous side

15effects. You can fill in the name of a number of recent

16drugs that have gotten into trouble. Now, suppose the

17company sought regulatory approval to produce only a

18combined pill including another substance that would

19mitigate the dangerous side effects. That is a

20legitimate and specific procompetitive benefit for

21bundling a product. In other words, tell me something

22about your product and your tie that helps us understand

23why this is a good thing that you are doing.

24I suggested asking whether the claim is

25specific, substantial and credible, and in evaluating


1credibility, I would borrow a page from another agency,

2the SEC. The SEC looks very closely at stock

3transactions that occurred right before big news. They

4find these highly suspect. In the same vein, I believe

5we should look at the market timing of a company's

6decision to tie in order to test the credibility of its

7claims of procompetitive benefits.

8For example, I would be very wary when a company

9seems to find all kinds of procompetitive reasons for

10tying just before the patent on its blockbuster drug is

11about to expire or just when a fundamental market shift

12is taking place. Under those circumstances, one might

13have reason to doubt the sincerity of the company's

14procompetitive fervor.

15In short, what I want to say today is that

16markets related to high-tech and biotech present

17significant pressures and opportunities for

18anticompetitive behavior. We should be aware of those

19as we move forward in the new sets of tests. The

20challenge for law makers and for regulators is to be as

21intellectually creative as the emerging markets

22themselves in order to preserve competition without

23hampering the innovation that we have come to expect in

24technology, both biotech and high-tech.

25Thank you very much.



2MR. SALINGER: Our final speaker today before we

3begin our round table discussion is Robert Willig,

4Professor of Economics and Public Affairs at Princeton

5University, where he teaches in the Economics Department

6and also in the Woodrow Wilson School of Public and

7International Affairs, where he serves as the Faculty

8Chair of the Masters of Public Affairs Program.

9He served as Deputy Assistant Attorney General

10at the Department of Justice, Antitrust Division, from

111989 to 1991. Before joining the Princeton faculty in

121978, he was a supervisor in the Economics Research

13Department, Bell Laboratories. He received his Ph.D. in

14economics from Stanford University in 1973, an MS in

15Operations Research from Stanford in 1968, and an AB

16from Harvard in 1967.

17Bobby has written, lectured --

18DR. WILLIG: Have I been around that long?

19MR. SALINGER: Apparently.

20DR. WILLIG: Only in the eyes of some beholders.

21MR. SALINGER: It seems shorter because we have

22 been having such a good time with you.

23DR. WILLIG: Okay.

24MR. SALINGER: Bobby has written, lectured and

25consulted widely on the subjects of industrial


1organization, the relationships between government and

2business and domestic and international microeconomic

3policy. He has served as a consultant and advisor for

4the FTC and DOJ on antitrust policy, for OE CD, the

5Inter-American Development Bank, and the World Bank on

6global trade, competition, regulatory and privatization

7policy, and for governments of several nations on

8microeconomic reforms, and so with no further

9introduction, Bobby.

10DR. WILLIG: I am going to tie my conception of

11my time slot to that which we have already experienced

12from some of the previous speakers, not the last one,

13but particularly the first one. Nice, long, lazy, but

14hopefully very illuminating.

15I have been asked to speak today, challenging

16subject, and that is not only to make it unanimous, I,

17too, am against per se treatment of tying under the

18antitrust laws. I, too, think there is no business or

19economic or indeed any logical justification for such a

20treatment by the courts. I, too, would have the

21agencies articulate that at every possible forum,

22including the high courts of the land. Okay, let's get

23down to the hard work.

24To really advance that position -- I am not sure

25how courts actually work, Don is obviously all over


1this -- but Don, do you think they will go ahead and do

2that without some clear idea about how to go forward

3under a rule of reason?

4MR. RUSSELL: We are going to give it to them.

5DR. WILLIG: So, we have got to work this out

6today, I would say, and we have got a half hour for me

7and then an extra hour and a half to see if we all

8agree, that will be great, but it is particularly

9challenging when it comes to a particular form of tying,

10namely, the kind that kind of underlay the Microsoft

11case, although only, I think, spiritually, and that is

12tying in a technological fashion, not tying by contract,

13not tying in such an obvious way that the weight of

14public opinion and the law would come down on the

15alleged perpetrator, but instead, tying in a much more

16subtle way of the kind Robin was just talking about,

17perhaps, but in other domains as well.

18We can call this tying via product innovation,

19product innovation being blessed in our society, and

20therefore, perhaps, more untouchable than other forms of

21tying, or technological tying, which is a very good pat

22phrase that I do not think I tried to invent, so I use

23it with humility.

24I would like to start with general thoughts

25about monopolization and then move swiftly to general


1thoughts about tying, and then, after a few minutes,

2specialize down to the subject of tying through

3technological design. In general, we all know that

4there is a problem, a challenge, in issues of

5monopolization, because the very same practices that

6have the potential to harm competition in the antitrust

7sense, frequently those very same practices also may be

8very good for consumers and, indeed, be an intrinsic

9part of competition, even though perhaps, like other

10forms of competition, if the succeeding firm undoes the

11market presence of the losers, then, in fact,

12competition can be weakened by the very process of

13competition, at least in the short run. So, we have

14this conflict between good and bad practices or

15practices that can be good or bad depending upon their

16setting, and so we have a tough decision process and the

17need for an analytic framework.

18I would suggest, and I think experience really

19does endorse this observation, that we do need to be

20 especially careful when the practices at issue do affect

21innovation, because after all, innovation we all know is

22particularly valuable to consumer welfare and to the

23course of social welfare. This has been amply studied

24by economists going back to Schumpeter and before, and

25also, the other side of the coin is that innovation is


1particularly vulnerable in its underlying incentives.

2It is really distressingly easy to stultify the

3incentives for innovation by misuse of antitrust or by

4any other form, a policy that tends to strip off some of

5the rewards to victory, because innovation is so

6intrinsically risky as an economic activity, so we need

7to be really careful with innovation generally.

8Big picture, how do we go about assessing

9monopolization? This is writ very large, but I would

10say there are two basic phases. The first involves

11asking the question whether the challenged practice has

12actually harmed competition, or on the come, is there a

13dangerous probability that it will? That, of course, is

14easy to say. It is not so easy to analyze, and lots and

15lots and lots of mistakes are made in judicial settings,

16and plaintiffs are crazy in terms of their allegations


18This involves causality. It involves

19understanding what is competition. It is not just

20market share, it is not just the number of competitors

21involved in a marketplace, it is something more subtle

22than that. We, in this room, probably all understand

23this very well. I need not preach to you on the

24subject. I will just post it up there as the first of

25the two phases.


1The second phase is, well, perhaps the

2challenged practice has, indeed, harmed competition.

3Things like that happen. Some competitors are more

4efficient than others, and they exercise their

5efficiency in the marketplace. They win, they knock out

6their less efficient rivals or rivals with less

7efficient products, and now there is only a few or even

8one left in that relevant market, at least for a while.

9What should we do about that? Has the practice been

10monopolizing or has it been successfully competitive?

11What is the framework for that inquiry?

12I list here five different articulations which

13are part of what Mark characterized as the blazing wars

14of Section 2 turf today, various articulations to me.

15For present purposes, I think they are all close enough.

16Is the practice part of competition? I like to

17put it that way. As DOJ says, does the practice make

18economic sense? The difference between those two -- I

19have parsed Greg Werden's writings, and it is tough to

20find them, but his writing is very smart. I am sure

21there is a difference, but for present purposes...

22Is there a sound business rationale? Courts

23used to say that. Is that really any different?

24Grinnell, is the harm to competition willful? Well, I

25am a little nervous about that language, because


1sometimes it is viewed as a directive for a

2psychological study of subjective intent, reading of

3locker room type business documents and trying to infer

4psychology from them, but as long as we understand

5willfulness to be revealed only by careful economic

6analysis, then I think that, too, is a nearly equivalent

7 articulation.

8And then my personal favorite, whether there is

9sacrifice of profit, turns out to be a very nuanced way

10to say it as well. Lots of issues about how to unpack

11that neat phrase, but I think for present purposes, we

12can perhaps all agree -- Mark Popofsky sometimes does

13not agree with this --

14MR. POPOFSKY: We are not to Q&A session here,


16DR. WILLIG: I am just trying to stick a little

17pin in for later, you have got that question in your

18presentation, but yeah, I think we can all agree that

19somewhere among those five articulations lies our

20consensus view.

21Turning to tying instead of monopolization

22generally, how do we see whether there is, indeed, harm

23to competition, the first leg of those two for the

24assessment? I think this is right. Maybe we can agree.

25The first question is whether consumers are really


1impelled, really strongly forced, to buy the tying good,

2the one that purportedly has this levering power, and

3thus, the tied good, because of the tie, by market power

4that surrounds either the tying good itself or the

5system, the combination of the tied good along with the

6tying good. Are the market forces so strong that,

7indeed, consumers are pushed very hard into that

8behavior? Because if not, where is the tie? It is just

9consumers making a choice. So, that is the first leg,

10at least to an economist, this economist, for labeling

11whether or not the tie has the potential of harming


13That is not enough, though. Consumers can be

14impelled to buy the system whether or not there is a

15resulting harm to the ability of rivals in some relevant

16market to compete in view of the fact that consumers are

17being pushed to buy the tying and the tied good

18together. So, does the unavailability of the tied

19sales, that unavailability created by the tie, is that

20harmful to rivals' ability to compete, and are those

21rivals so precious and so unreplaceable to competition

22in some relevant market that competition is truly harmed

23as a result?

24That question can go either way, but I think it

25is the right question, and I have seen a lot of cases


1where those two reductions of the issues have been

2missed, but I think they are pretty persuasive in terms

3of the underlying logic.

4I probably should not take the time to go

5through this slide, but here is how to think about those

6issues in a more organized way, sensitive to I think

7standard practice, at least among the practitioners in

8this room. We need a relevant market. We need to know

9who the participants are. We need to know whether the

10unavailability of the tied sales weakens rivals who are

11scarce, concentration, and irreplaceable. We need to

12look at potential entry into the relevant market as

13well. We know that loss of share is not enough to claim

14competitive harm.

15We should be looking at whether it is a good

16deal for consumers to buy the tied system. As a result

17of the rivals being purportedly weakened and not just

18harmed, but weakened, have prices gone up? Are things

19worse for consumers as a result? I like the scale

20economies test, the room to dance test myself, getting

21underneath to the underlying opportunities that

22competitors need to be strong.

23Is there enough of the market left, after the

24tied sales have been accounted for, to keep the other

25rivals? Although they are sad to have lost those


1opportunities, is there still enough room for them to

2function, to do what they need to do effectively to

3remain as important competitors? Does their R&D hold

4up? Do their selling efforts hold up, for example? Can

5they bounce back in due course?

6Maybe those rivals were inefficient anyway and

7weakening anyway and here they are complaining because

8their last best asset is the right to bring an antitrust

9case. These are the usual kinds of issues that come up

10in Section 2, and they are particularly important, I

11think, and to remind ourselves of their importance in

12this context of tying.

13Okay, so suppose there is a tie, suppose it does

14harm competition. That does not mean that we should

15come down on it with antitrust, because there is still

16the second important phase, and that is where the

17challenged tie is truly part of competition, which just

18sometimes happen to weaken competition because the

19successful firm is emerging in a more concentrated form,

20and, of course, I think we can agree that since the

21valid policy goal here is competition itself, business

22conduct which is really part of competition should not

23be condemned, and we should not be deterring competitive

24conduct through the sort of distortion of the use of



1In a more particular way, we have a challenged

2tie. Would that challenged tie be profitable without

3taking into account this harm to competition and its

4impact on monopoly power that has been found in the

5first phase? We have found, say, that tying has harmed

6competition. Would the tie have been profitable for the

7perpetrator even without that extra monopoly power? I

8think that is a good organizing question before moving


10So, let's move on. Here we are now finally in

11the setting of tying via technology, via product design,

12and let me paint what for me is the toughest scenario.

13It is the most interesting scenario, where we actually

14do have a plausible allegation of exclusion through the

15technological tie. So, we have got a new product design

16that has been launched, and it technologically ties two

17components together of a system, of a duo, that could

18conceivably otherwise be open without the technological


20If there are two pills tied together chemically,

21that is a great example. It is the old local phone

22system and long distance when the Bell System was in

23charge before antitrust. It is a much more lurid

24example of Microsoft. Imagine if Windows had little

25explosive devices where if you tried to plug NetScape


1into it, the computer would fry. I mean, some alleged

2that was the case, but usually they forgot to do some

3sequencing keypunches that allow it to happen, depending

4on which side of Microsoft you are on, but that would

5have been a much more telling example of a technological


7How about the iPod, which are said to be

8technologically tied to iTunes, through the protocol in

9which the music is encoded and now the video as well?

10That is certainly a technological tie, or at least it is

11alleged to be in some sense.

12In the good old days, remember mainframe

13computers? They had their plugs changed, allegedly, so

14only IBM peripherals could plug into the mainframes.

15That was surely a technological tie. To say nothing of

16the radios in GM cars and so on.

17Okay, so as a result of this product design, the

18two components, one of which at least has real

19potentially competitive marketplace forces bearing on

20it, these two components are tied because of the product

21design. So, what could possibly be anticompetitive

22about that?

23Well, suppose that they are rivals for at least

24one of those components. There is NetScape as a

25browser, there are other web sites where you can go to


1get music, but that music does not go into the iPod.

2There are other places to go for pills that have some of

3the same therapeutic functions, not exactly the same,

4but surely substitutes. So, these rivals of the other

5competitive entrants into this marketplace are shut out

6of the system by the technological tie.

7Now, there are two lead theories of how that

8might create market power. The sellers of these other

9potentially competitive components have a much reduced

10ability to sell in the bad story. They lose economies

11of scale, they lose the impetus for R&D, and so they

12have a harder time competing for other applications of

13those same kinds of components.

14One of the applications is the kind that is

15subject to the tie, but there are non-coincident

16markets, not implicated directly by the tie, in which

17the NetScape alike has been competitively harmed by the

18inability of NetScape to be appealing to those who are

19running Windows in the Microsoft story.

20The other version of that story is that there is

21the potential for harm to competition in the market for

22the bottleneck, for the tying good. In Microsoft, the

23story, the DOJ economist's story anyway, as I understood

24it, was that with NetScape together with Java could form

25a competitive threat to Windows itself, so that to


1preserve the power over the bottleneck, Microsoft is

2said to have needed to weaken its potential rival in the

3potentially competitive browser market to preserve its

4power in the market in which it has much of a

5bottleneck. So, there is a competitive threat at both

6levels which might be mitigated, protecting monopoly

7power, by the technological harm.

8Well, that is the bad story, but on the other

9hand, we are talking about product design. We are

10talking about innovation, and, of course, we might well

11have a welfare-increasing innovation in our hands, and

12how are we to sort out whether the innovation is largely

13welfare-enhancing as an innovation or whether, instead,

14it is just a ruse, it is just a business tactic to

15preserve or create monopoly power?

16I have got a theorem or two for you. It is set

17in this picture. This picture has a long heritage in my

18life, but I need not go into that. My introduction was

19embarrassing enough about dates and years. A1 is the

20bottleneck that belongs to firm 1. It is the lever off

21of which the tying might go. A2 is the component that

22serves the ancillary function, the browser as it were,

23made by the same firm. So, firm A has a 1 and a 2.

24B2 is the other firm's substitute for the

25product which is here tied. It is NetScape, it is the


1other browser. NetScape could work with Windows, if you

2take Windows to be A1, so the horizontal line between

3them shows that they interoperate. They both feed into

4the systems market, which is what consumers want. They

5want systems. They want combinations of the operating

6system and the browser.

7Meanwhile, C1 is lurking up there in the right,

8that is Java. When Java works together with NetScape it

9has the potential for actually performing the same

10functionality or maybe a degraded version, as would

11Windows with Explorer or Windows and NetScape. So, that

12is the story without the tie. Everybody interoperates,

13there may be some degradation of function, there are

14pricing issues, but that is the world without the

15technological tie.

16Now, in the bottom part of the picture, along

17comes a new version of the operating system, A1 prime, a

18new version of the browser, A2 prime, they work

19together, but you know what, there are no APIs at all.

20There is no way that your NetScape can interoperate with

21them. There is a true technological tie here depicted

22on the picture. As the bottleneck holder moves from the

23upper system to the lower one, it implements the perfect

24technological tie, thereby shutting out B2.

25The bad stories are that B2 has to go out of


1business, it is so weakened by the inability to sell,

2and so if it had any other uses, like on servers, forget

3it, it is going to have to leave the entire space, it

4loses the economies of scale and scope, and then Java,

5C1, has got no partner to play with, so it evolves in an

6entirely different direction. It is no longer a

7candidate for the central part of a desktop operating

8system. It also goes off into server land, and the new

9Windows survives as the undisputed champion, delivered

10into that throne by the technological tie. So, it is

11the same story, but now it is on this picture, where we

12can start putting symbols for pricing and costs and

13things like that.

14I need to define a thought for you, the

15compensatory price. Just imagine that the open design

16bottleneck persisted even when the new system came out.

17The new system comes out. It is technologically tied,

18but imagine that the old open design system is still out

19there. This is just a mental exercise. Imagine it is

20still out there, and it is made available to consumers

21as well as to competitors at a compensatory price. If

22it is just out there and priced at an infinitely high

23level, it is not really a competitive force.

24Some court might rule that it had to be given

25away, but that would not be a marketplace solution. A


1compensatory price, by definition, puts the same profit

2margin on the use of the open access bottleneck, the

3same profit margin as the new system earns. The new

4system is the one with the tie. So, your perpetrator

5comes out with a tie, charges a lot for it, and that

6margin is now built into a compensatory price for the

7old open design system.

8The theorem is that when the open design

9bottleneck system is still available in the market at

10this high compensatory price that builds in the same

11profit margin, then the technological tie, the new

12system's introduction, eliminates the competitors if and

13only if the new closed system is actually socially

14superior to the open one, and here I wrote, "Ex-post,

15the R&D costs," the next slide -- and I am running out

16of patience and so are you for this -- the next slide

17will also talk about the R&D costs and reach essentially

18the same result.

19So, what does this say? This says that if you

20had a world where the open design system were still

21there, priced in the same high-priced way as the new

22system, then the marketplace would work, that the

23competitors would be knocked out if and only if they

24deserved to be knocked out on grounds of true total

25social welfare, that the new system is worth the R&D


1costs, it has improved functionality, it has better

2 costs perhaps or some balance of all of those elements,

3sufficient to make it better for true social welfare as

4economists measure it than the old system, so that this

5innovation is not just a hokey thing designed just to

6knock out the competitors under the ruse of somehow

7coming out with something new.

8It is not newness for its own sake, it is not

9newness for the sake of monopolization, it is really a

10better system. That is true only if the standard is

11being held that the open system is still available at a

12compensatory price. Without that design of the theorem,

13you can knock out the competitors without having the new

14and better system. You can just technologically tie

15them to death.

16So, here the right standard is what would happen

17in the market if the open design system were there at a

18compensatory price, then market outcomes are telling of

19efficiency. So, it is a very powerful result I think.

20It dates back a long way. I will not even highlight

21that, it is Ord over a long time ago, but it has new

22significance today, I believe.

23What does that mean for antitrust? Well, in

24antitrust, if, indeed, the open system is available, the

25old one, at a compensatory price, and there is a


1technological tie and the competitors are knocked out,

2the theorem would say, you really should not be coming

3down on that kind of innovation, because according to

4the theorem, that is good innovation, as proven by the

5continued availability of the old system at a fair


7Now, oftentimes the old system cannot or will

8not be left in place, although this kind of raises the

9question of why not, and maybe if this were part of the

10antitrust standard, that would be an impetus for

11companies to take some pains to keep the old systems

12alive. Maybe not. It does tell us, though, what the

13right standard is for this economic framework. If the

14open system is not preserved, we still have a mental

15standard, a but-for test, which is well adapted to

16technological tying for assessing whether we should

17condemn or smile upon the win in the marketplace by the

18new system.

19That standard is whether the competitors would

20still be going down, still be losing, if, in the but-for

21world, they would not be successful, and here the

22but-for world is the continued availability of the open

23design system, the alternative, at this fairly high

24compensatory price that builds in the full profit margin

25earned by the new system, that if you want to know


1whether or not we have an offense here or not, ask

2yourself the question, would the competitors have been

3beat anyway even if they had access to an open design

4version at a compensatory price?

5This question was not asked in Microsoft. It is

6not asked in Microsoft today in Europe. I do not know

7what the answer would be, I am not a partisan in those

8debates, but the theorems say that is the right question

9to ask. That is a good standard. Just like marginal

10cost is a good standard for Areeda-Turner, this is a

11good standard when it comes to technological tying in

12the role of exclusion accomplished through that kind of

13a tie.

14There are a bunch of caveats. The first caveat

15is, how do you know whether the R&D costs that were

16expected at the time of the decision by the

17technological tyer, how do you know what those really

18were? If they were very low, then that makes the system

19look better in terms of the standard. If they were

20expected to be higher than the skies, then it goes the

21other way. Part of what the fact-finder needs to do is

22assess the expected R&D costs as we get deep into this

23phase of the antitrust analysis. Obviously a tough task

24for the fact-finder.

25How can the fact-finder do this but-for test?


1Well, at least it is an organized test, the theorem

2tells you what to look for, but this is not necessarily

3an easy job for a judge and a jury in an antitrust

4 court, to do this kind of but-for test. If you do not

5have this kind of a structured standard, how is the

6 fact-finder going to in some other way decide whether

7the new system is really good or not? Talk about

8keeping science out of the antitrust case, this is

9science and consumer preference rolled together. How

10good is the innovation? I would not trust a judge to

11make that answer without an economic framework.

12On the economic side, the theorems, which I

13think are really very powerful, they are in a very

14oversimplified setting, as usual, but maybe even more

15than normal. This setting, in which these theorems are

16proved, is a setting in which there are no other issues

17whatsoever for social welfare besides the ones that the

18theorems focus on, namely, the possibility of

19monopolization through the technological tie. All other

20economic imperfections have been ruled out by the design

21of the abstract marketplace. And we know from common

22sense and from economics that in marketplaces where

23innovation is important, there are typically all kinds

24of other things that can go wrong, ambiguous

25externalities, inappropriability of benefits of


1innovation on the one side of the ledger and negative

2externalities conveyed by the innovator on others who

3are competing with the innovator in the market, lost

4profits to other market participants.

5On the one hand, you get too little innovation

6because of inappropriability issues, or you get too much

7innovation because of negative profit externalities, and

8in most economic models, the ones that I teach in my

9classes, it is thoroughly ambiguous whether innovation

10comes out just right even without antitrust issues, and

11all of those kinds of complications must be ruled out to

12get these neat results that our theorems get. Which way

13that biases the answer is decades away from my students

14and yours being able to figure out, and maybe never is

15the right answer. I mean, in a model you can figure it

16out, but how the model corresponds to reality is far

17beyond the state of the art.

18So, what did we learn from all of this other

19than the fact that you are very kind and patient? One

20additional lesson is that as a matter of economic logic,

21technological tying is real. It is a real possibility

22on the blackboard, in the journals, and there they may

23be very genuine, even strong incentives to do

24technological tying for anticompetitive reasons, but

25also for a long list of procompetitive reasons, the same


1kinds of reasons we heard about from earlier panelists,

2as well as a host of other ones arising just because it

3is innovation, and so, yeah, you cannot just say, oh, it

4does not happen or it cannot happen as a matter of

5logic. It can happen, it may happen, and on the other

6hand, technological tying may be a very, very good thing

7in many settings.

8The second point, which is newer and I really

9hope that you believe a little bit, is that there are

10logical and intuitive tests and, indeed, standards for

11analysis that would allow us to assess product design

12for monopolization by a tie-in. This is the kind of

13test that I was just talking about, the but-for being

14 open standard with compensatory pricing. These are not

15easy to apply. They do organize the mind, but they are

16hard to apply empirically, especially in a litigation

17setting, and so great humility is certainly called for

18in this area.

19Well, if we combine humility, due humility, with

20how delicate and important innovation really is, we

21reach the same policy bottom line that everybody else

22has reached, certainly no per se treatment, my goodness,

23but even more so in the world of rule of reason, we need

24to protect innovation as a process from being stultified

25by litigation with very, very strict and very demanding


1hurdles in front of litigation which must impose a tough

2discipline on the use of antitrust in this area, both by

3private parties and by the agencies, and that goes

4largely, I think, to the first part of the test, that

5there really has to be demonstrated harm to competition

6in a relevant market through the technological tie. It

7has got to be causal, and taking that part of the test

8very seriously alone would knock down most of the cases

9that I have been exposed to.

10 So, that is my plea, and I thank you.


12MR. SALINGER: Well, we are now going to give

13each of the panelists a chance to respond to the others.

14I do not know how long Professor Feldman is going to be

15with us, but since there seems to be perhaps some

16disagreement between you and Bobby on your --

17DR. WILLIG: You think?

18MR. SALINGER: -- on your take on how to deal

19with technologically advanced markets, maybe we will

20start with you.

21PROFESSOR FELDMAN: Well, let me start with,

22again, what we agree on, which is that we knock out per

23se, and I would not disagree about the importance of the

24harm to competition element. I begin by assuming that

25we are in something like a rule of reason setting in


1which we have already looked for market power and we

2have already looked for market power and we have already

3looked for harm and then we are trying to analyze what

4the claims are. Given that you are a very big guy and

5given that what you are doing is harmful to competition

6as opposed to competitors, how do we evaluate the things

7that you have said are so good about what it is that you

8are doing? So, I would not disagree there.

9I might disagree on what we talk about in terms

10of the harm to competition, again, remembering that

11particularly for innovation markets, such as high-tech

12and biotech, that these markets evolve so rapidly that

13the harm to competition is happening in the future.

14That can be difficult to measure in an economic analysis

15in a courtroom.

16What we want when you have a monopoly is that

17the natural forces of competition will make that

18monopoly erode and you will get new products that will

19look better and you will not have monopolists. If you

20have settings in which the monopolist can project into a

21new area as soon as new things are discovered, you are

22going to have monopolists who can stay in an entrenched

23base for a while, and that is a problem.

24You will have to tell me whether we disagree

25strongly on technological ties. I suspect there is a


1fair amount of agreement here. I think technological

2ties can be useful. I am wary of them, and I think we

3have to be careful of them in certain settings that

4already look anticompetitive to begin with.

5DR. WILLIG: How could I disagree?

6We agree on the logical possibility of

7technological tying. We agree on the importance of

8technological advances and competition that drive them.

9I think we agree -- I do not really know much, and you

10obviously know a lot -- that in biotech, there are

11opportunities every bit as lurid as they were in old

12mainframe computer spaces changing the metaphorical plug

13on the mainframe. Here, sprinkling a new coating over a

14pill and bonding it with some other pill, I mean,

15apparently the pharmas can do this all the time, and --

16PROFESSOR FELDMAN: Not all the time, but enough

17that I would worry about it.

18DR. WILLIG: But they can, anyway, they can.


20DR. WILLIG: And that certainly raises the issue

21of whether that kind of "innovation" is genuine,

22socially useful by an economist's measure, or whether it

23is a ruse to extend monopoly power. So, I think we

24really have a bonding here ourselves.

25MR. SALINGER: Well, maybe we can call on some


1of the attorneys on the panel to see whether they have

2 heard enough agreement that they feel confident they can

3go into court with good arguments about how to

4distinguish procompetitive from anticompetitive ties.

5MR. RUSSELL: I would like to jump in with a

6question for Professor Feldman about this concept of

7specificity when it is applied to the procompetitive

8explanation, and I may have misunderstood what you were

9saying, but if I were a lawyer on the other side, the

10way I would characterize your position is the fact that

11a particular kind of efficiency is seen so often in so

12many products and is so powerful, which is the natural

13inference I draw from the fact that it is seen so often

14in so many products, for that reason, you are completely

15disregarding it.

16PROFESSOR FELDMAN: I understand your concern

17about that, and maybe I can frame it again by looking at

18the point at which this inquiry comes up. We are

19already at a point where we have a monopolized tying

20product. We already are at a point where we have

21established that there is harm to competition. Now we

22are looking at the reasons for that, and I think that

23the concerns you have can be taken care of in the first


25What I am concerned about is when we get to this


1point, there will be boilerplate language in which

2everyone will essentially be saying the same general

3things that can always be said about ties and about the

4right shoe and the left shoe and about why things in

5combination are appealing to consumers. If we credit

6that type of an argument, we will be unable ever to

7target things that are anticompetitive, because those

8defenses are always available.

9MR. SALINGER: David, I find it hard to believe

10that you do not want to chime in here, so...

11MR. EVANS: Well, I am puzzled about a couple of

12things, both with respect to some of the things Bobby

13said and also some of the things Robin said, especially

14in the last statement, so the first thing I have always

15been confused about, and it comes up in Bobby's talk, is

16this term "harm to competition," because maybe I just do

17not know enough economics, but I do not really know what

18that means.

19I know what it means to talk about reducing

20long-run consumer welfare and stuff like that, but I

21guess my experience in these cases is when I start

22hearing phrases like "harm to competition," it leads to

23theological discussions of what competition is or is

24not, and depending upon the market structure and so

25forth, you know, competition means different things,


1including competition for the market and ultimately

2having a monopoly and having a monopoly despite what you

3said, Robin, that we actually do not want to have its

4power eroded, at least so long as it is efficient.

5The second thing I get confused about and do not

6really understand is this sequence where we talk about

7harm to competition and then say, "Oh, gee, then let's

8take a look and see whether there are efficiency

9benefits that offset that harm to competition." I mean,

10it seems to me that ultimately the inquiry is whether

11there is a harm to long-run consumer welfare, and I do

12not really understand the unbundling of the efficiency

13explanation for the practice and this term "harm to


15I mean, if I think about markets, I would think

16that the whole issue of why one engages in a

17technological tie or any other kind of tying practice

18has to be sort of an integrated aspect of the whole

19discussion of whether there is "harm to competition,"

20whatever that means.

21And I guess just the final thing that I will say

22both with respect to Robin's talk and Bobby's talk is

23both of them do kind of lead to this unstructured --

24well, maybe I am being unfair to Bobby -- I am being

25unfair to Bobby.


1DR. WILLIG: Yes, indeed.

2MR. EVANS: But it does seem to lead to a

3relatively unstructured rule of reason inquiry, and I

4really do think, as I think many of the speakers have

5pointed out, that we need to start with a position on

6where we are in terms of priors concerning where the

7timing is bad and error cost and so forth, and we need

8to start with that, and maybe you disagree that -- that

9anticompetitive tying is uncommon, in which case you can

10state that as a prior and go forward, but it seems to me

11you need to start with a position before we can really

12get into conversations on who ought to bear the purpose

13and stuff like that.

14So, I do not see how at the end of the day we

15can impose the burden of proof on a defendant for

16establishing efficiencies, as Don says, for a practice

17that we know is presumptively efficient. It does not

18make any sense to me.

19MR. SALINGER: Michael, David in his talk talked

20about how he was largely agreeing with you. Is there

21complete agreement among the economists or is there more

22of a wedge there than just --

23DR. WILLIG: Not anymore.

24DR. WALDMAN: Well, listening to David's

25response, I basically agree with almost everything he


1said. I agree that if I am thinking -- I do not think

2the right thing to think about is harm to competition.

3I think the right thing to think about is social

4welfare. There are lots of examples that one could

5come -- sort of formal models that one can show where

6thinking about tying as eliminating competition is

7actually social welfare improving.

8So, if you wind up focusing too much on the harm

9to competition, you will wind up allowing or eliminating

10tying when, in fact, you really would not want that,

11because in some sense there is sort of a larger

12competition ex ante or something else which says that

13the competitive process, thought of more generally, that

14particular submarket where you are not allowing

15competition is actually a good thing rather than a bad


17Also, you know, I am not sure David exactly

18specified this, but, you know, so I think consistent

19with what he is saying, you know, when I think about

20kind of how do I judge these cases, I want to say let's

21think about the different theories in some of these

22situations you can automatically almost rule out as

23saying, well, that looks okay, it is efficiency or it is

24price discrimination, and at least as a first blush, and

25I do not do court cases, but I would have thought that


1the -- or at least the way I conceptualized it is to

2think about from a rule of reason standpoint, is there

3an exclusionary argument that typically one would think

4of from a theoretical perspective that will lower social

5welfare? Does it fit the facts of the case well? And

6then say, is there no efficiency argument that fits the

7facts of the case well?

8If those two things hold, then you are sort of

9in the ballpark to think that maybe this might be a case

10that you would want to intervene, but if those two

11things do not hold, then that seems like a dangerous

12type of case in which to intervene. Maybe there is some

13general rule that Bobby is talking about that one could

14apply sort of to oversee it, but at least my sense of

15the literature is that these different types of cases

16are sufficiently kind of nuanced and different that I am

17a little skeptical, but I do not know the specifics as

18well, so I am hesitating to say too much there.

19But again, I would want to see more before I

20thought that there was some really general rule that one

21could apply rather than just kind of fitting the facts

22of the case to a specific theory.

23MR. SALINGER: Bobby, before you jump back in, I

24want to give Mark an opportunity to comment on whatever

25it is he has heard that he wants to comment on.


1MR. POPOFSKY: Well, I want to go back to what

2David Evans was talking about and his observation about

3the debates between Professor Willig and Professor

4Feldman, which is this: Until we have a definition of

5what the target is for harm to competition, we are not

6going to be able to advance the ball a lot here. All

7the action is going to be there. It is to put the

8action -- the debate very precisely, will you for tying

9arrangements under Section 2 require something like a

10profit sacrifice, for the plaintiff to get to the next

11step and put the burden on the defendant to show

12justification? Is that going to be the test for

13identifying a presumptively anticompetitive tie?

14Will that be a universal rule applied across all

15ties, or will we have the other extreme, where we have

16some broad, vague, potentially innovation-deterring, as

17Bobby suggested, rule of reason even for technological

18ties where you are not making unbundled option

19available, to be precise about what a technological tie

20is, or will we be somewhere in the middle, as Michael

21just suggested, perhaps, where we can identify some

22discrete categories of ties, where we say for this

23category of tie, the plausibility of anticompetitive

24effects, i.e., long-run cost to consumers and harm to

25social welfare, is real enough that we are going to give


1a little leeway in the joints and have the rule of

2reason apply, which is in some sense less of a burden on

3the plaintiff, or is it going to be a category of ties

4where we think intervention potentially carries such

5high costs, and for some that is product design, I think

6there are some arguments there that would require more

7of a showing from the plaintiff to go forward, maybe a

8profit sacrifice, maybe something else, and, indeed,

9taking that to an extreme, might there be categories of

10tie-ins where you really have a safe harbor absent a

11very strong showing for the plaintiff? That seems to me

12the type of thinking that needs to occur.

13MR. SALINGER: Okay, well, now that we have

14found some daylight within us, as we organize these

15hearings, we have tried to see whether or not there are

16agreements on various propositions and disagreements on

17various propositions, and we have a set of these for the

18panelists to comment on, so I will turn the mike over to

19June to lead us in that discussion.

20MS. LEE: Before I start, let me give Bobby a

21chance to respond to some of the comments.

22DR. WILLIG: Oh, thank you.

23Well, first of all, I was only invited to

24comment on Robin, and I had no problem with Robin, but

25these other folks, I just... .


1MS. LEE: Please.

2DR. WILLIG: Well, first of all, I do not know

3if we can go off the record here or expunge the record,

4but if the Supreme Court ever heard the things that have

5been said in the last ten minutes, there is no way we

6are going to get off the per se standard. I mean, if

7all these learned people cannot figure out rule of

8reason or even what harm to competition is, then I think

9we are going to be stuck with the per se test for

10another generation. So, can we go into private session

11so the Justices cannot hear us? I am just kidding, of

12course. I think we actually know a lot more than the

13last ten minutes has suggested.

14Well, let me pose to Michael and David and I

15guess Mark, too -- and, Robin, you are free of this

16mistake, I would say --

17PROFESSOR FELDMAN: It is the only one I am free


19DR. WILLIG: No, that is okay.

20The hard case, I agree with all of us who have

21said that price discrimination ought to be very, very

22presumptively innocent for a wide variety of deep

23economic reasons as well as just commonplace

24observations that the most competitive of industries are

25full of instances of price discrimination, at least one


1of us has written that it is parador superior (ph) to

2have price discrimination and so forth. Price

3discrimination is basically a good thing. There are

4counter-examples, but we do not know how to spot them.

5So, we certainly ought to be allowing business the

6freedom to do price discrimination. And we all

7understand that a very important function of lots of

8tying practices is to permit firms better, more

9effectively, to do price discrimination.

10And so I agree with those who have said if we

11can spot that there is a tie which effectuates price

12discrimination, then we ought not to be overly

13suspicious of it, and there should be a huge burden of

14proof on the part of the enforcers or the plaintiffs to

15overturn the presumption that tying to effectuate price

16discrimination is basically probably a good thing. It

17is only presumptively a fine business practice. I agree

18with all of that.

19On the other hand, it is very easy to imagine a

20circumstance where the tying does effectuate price

21discrimination in a very real way that is important to

22the business, and at the very same time, the important

23rivals are shut out by that same tie. Think about razor

24blades. This is a cartoon version, but Gillette comes

25out with a new overpoweringly good system, gives away


1the razor dirt cheap, charges a fortune for the blades,

2and very neatly ties the two together with patents and

3with interoperating devices that make sure that rival

4blades cannot use the same razor. There have been cases

5like this.

6We all say, oh, that is fine, that is price

7discrimination, that is promotional pricing, that is a

8good thing, if you happen to like the razor, which I did

9for two blades but not for four, but that is another

10subject entirely. Suppose that all the branded rivals

11of Gillette go out of business -- this has not happened

12to my knowledge, but just imagine in the cartoon. We

13have got two things going on. We have got exclusion and

14 we have got product innovation inspired by the

15opportunity to do effective price discrimination. They

16are both running in the same case.

17I suggest there is a lot of this in the economy,

18certainly in the antitrust courts. I think it is really

19very overly easy to say, oh, tying for price

20discrimination is fine, tying for exclusion is bad.

21They both tend to run together, and certainly plaintiffs

22will feel that they do if they are an aggrieved

23competitor who has lost out from this innovation.

24I think you have got to address -- and Mark, you

25too, don't look so quiet over there -- what do we do


1with those cases? Do we say the jury or the judge ought

2to weigh the pluses and the minus and be a meter of

3consumer welfare? Is the innovation permitted and

4motivated by the price discrimination? Together with

5the benefits of price discrimination, together --

6sufficiently a plus that the harm to consumers in the

7longer run from the loss of these important competitors

8does not outweigh it? Do we have a consumer welfare

9meter? Do we know how to do that? Do we trust

10ourselves, no less judges and juries, to do that? That

11is one possibility, quote, "the consumer welfare

12standard," Mark.

13The other possibility is that we say, look,

14there is a legitimate rationale, namely, the price

15discrimination and the innovation. Yeah, you cannot

16make an omelet without breaking eggs, competition has

17losers, successful products do raise some legitimate

18monopoly power for a while, and we have got to let the

19competitive process work. Do we say that?

20That is the big issue of the day. That is what

21the wars are about in the journals, and I do not think

22we can be quiet about that in this forum. So, I put

23that in your laps, gentlemen.

24MR. POPOFSKY: Let me make one comment. I am

25glad to see, Bobby, you actually read my article.


1DR. WILLIG: No, just the first paragraph and

2the like. A hundred pages of footnotes, Mark, I cannot

3do it.

4MR. POPOFSKY: And none of them cited you, I

5think we have pointed out.

6DR. WILLIG: That was the point.

7MR. POPOFSKY: Nothing from 25 years ago. I

8think to try to answer your question, Bobby -- since you

9put the pitch right over the plate, let me see if I can

10hit it over second base.

11As the hypothetical in my article implied, which

12is close to yours, there is a very sympathetic case

13there that the Microsoft Court of Appeals vague rule of

14reason standard is the last thing you want courts and

15juries to be doing in a case like that in some vague

16way, and the way Professor Salop somewhat suggests in

17his articles, reckoning up the social costs today

18against the social benefits tomorrow, you take that

19logic to the extreme, you would have courts regulating

20significant aspects of the economy. That cannot be what

21the rule of reason is all about.

22So, in devising the right legal rule -- and I am

23not sure what it is, to be honest, to answer your

24precise hypothetical -- you want to perhaps take into

25account what would be the detrimental impact of


1innovation on intervention, and that might mean you

2structure the rule of reason differently, it might mean

3you go to the profit sacrifice test, but you certainly

4do not want what you painted as the boogeyman of juries

5just saying, what is the net contribution to social

6welfare of this conduct? That cannot be what we are


8DR. WILLIG: We can quote you on that?

9MR. POPOFSKY: Oh, yeah. It is on the record


11DR. WILLIG: Okay.

12PROFESSOR FELDMAN: May I point out what is one

13other point of agreement among the panelists. In

14addition to the notion that per se is not the way to go,

15an open-ended rule of reason also is not where we should

16go. There must be some type of structure in the rule of

17reason for the benefit of all the parties involved. Are

18we in general agreement with that?

19MR. SALINGER: Yes. Okay, well, we should let

20June get into her areas to nail down points of agreement

21or disagreement.

22MS. LEE: Indeed, just to clarify some of these

23things, let's start with the first one, I do not think

24there will be disagreement with this one, which is

25certain tying arrangements pose an unacceptable risk of


1stifling competition and therefore are unreasonable per

2se. I do not think anyone on the panel agrees with

3this, but please correct me if I am wrong.

4Okay, so let me flip this question a little bit.

5Does anyone on the panel think that tying should be per

6se legal?

7(No response.)

8Okay. Then let me just -- backing down from

9that a little bit, are there any tying arrangements that

10are always or nearly always procompetitive and thus

11appropriate candidates for a safe harbor?

12Bobby and some others discussed a little bit

13that tying for price discrimination reasons should not

14be illegal.

15MR. EVANS: But then he backed away from that.

16MS. LEE: Yes, so --

17DR. WILLIG: Yeah, because I think typically it

18is hard to separate.

19MS. LEE: Right.

20DR. WILLIG: -- the enabling of price

21discrimination from the exclusion. I penciled on my

22notepad that tying arrangements are nearly always

23procompetitive where there are ample choices available

24to consumers among alternatives, both at the level of

25the tying good and at the level of the entire system of


1tying tied to the tied good, i.e., if there are other

2operating systems and browsers or other MP3 players and

3MP3 formats, if there are system alternatives available

4in ample supply, then within that framework, I think we

5should have per se legality.

6MS. LEE: Okay. Does anyone else have

7categories for which they would say that tying should be

8per se legal? Don?

9MR. RUSSELL: I just want to ask a follow-up

10question for Bobby. When you say there are

11alternatives, are you saying there is no market power or

12is that different?

13DR. WILLIG: No, ample, ample alternatives.

14MR. RUSSELL: But is it basically a market power

15test that you are advocating there?

16DR. WILLIG: Well, we gave up perfect

17competition a long time ago, but, you know, workably

18competitive set of alternatives, if you will.

19MR. POPOFSKY: No power of antitrust concern,



22MR. POPOFSKY: Power of antitrust concern?

23DR. WILLIG: That is in the eye of the beholder,

24Mark, yeah.

25MS. LEE: David, did you have a comment?


1MR. EVANS: Yeah, well, I think what we have

2just -- I think what Bobby just said is that where there

3is not significant market power, that ought to be per se

4legal. I think that the debate in question, I think

5this is one of the questions you ask later, is what

6exactly does that mean?

7I am not exactly sure what the answer to that is

8from the state of the theory and empirical evidence at

9this point, but keep in mind that the starting point

10with Jefferson Parish I believe is some market power. I

11think the consensus here is it ought to be significant

12market power. Whether that corresponds to a share of 50

13percent or whether it has to be a hell of a lot more I

14think is an interesting question for the initial screen.

15Whether it has to be something that is closer to

16monopoly power given where we are in the theoretical

17literature, I am not sure I know the answer to that.

18MS. LEE: Okay. Let's move on to actual --

19actually let's skip the next proposition and move on to

20the third one. Tie-ins may entail economic benefits as

21well as economic harms. So, I think everyone on the

22panel agrees with this. Let me make sure that everyone

23has -- can opine on their priors, as David Evans

24suggested, which is something that we should do. I

25mean, lots of commentators have observed that most ties


1are procompetitive. Does everyone agree with that?

2PROFESSOR FELDMAN: I would not agree that most

3ties are procompetitive. I would not fall into that,

4certainly not -- not in the industries or areas that I

5have talked about. I certainly believe that there are

6many procompetitive ties, but I would never say most

7 ties are procompetitive.

8MR. POPOFSKY: Let me just make a comment. We

9really have to be careful what we are talking about here

10in distinguishing bundling from tying. Most bundles may

11be procompetitive in the sense of offering two things

12to -- two items together.

13What a tie-in is is not offering the consumer

14the choice of taking the tying good without the tied

15good. It is not offering the car without the radio.

16And, you know, and maybe we can think of many, many

17examples throughout the economy where that is

18commonplace, it is plainly efficient, but what I think

19Robin is suggesting is those that come under the

20antitrust microscope, it is not clear what you are going

21to count them up and say you have seen more good ones

22than bad ones.

23Certainly going back to my favorite poster

24child, the Microsoft case was certainly one the

25Department thought and the court agreed, at least under


1Section 2, was a bad tying arrangement, but there are

2other software ties that are similar which are good, and

3you really have to be careful what you are talking

4about. The problem in Microsoft was in not offering the

5unbundled option, so phrased that way, we might reach a

6different conclusion.

7MS. LEE: Let me give Robin a chance to clarify

8what she said. Would you sign onto what Mark said, that

9what you are talking about is ties that come under

10antitrust scrutiny, most of those are not

11procompetitive, or are you talking more generally?

12PROFESSOR FELDMAN: I do not think I would say

13that those ties that come under antitrust scrutiny are

14mostly anti-competitive and those ties that don't come

15under antitrust scrutiny are procompetitive. I would

16agree that if we were talking about a form of tie

17leverage that is not somehow forced, where you can, as

18Mark was just saying, get to the product other than

19through the tie, that is not a problem. I would not,

20however, make the sweeping statement that tying and

21leveraging are almost always acceptable without a lot

22more discussion of what we meant by that.

23MS. LEE: Okay, Bobby?

24DR. WILLIG: Maybe we can all agree on the

25following language that I penned: Most arrangements,


1both technological and contractual, in our economy that

2do impel purchasers to buy two products together are

3procompetitive. So, it is not just antitrust, and it

4is -- it does not comment on whether the tie is

5artificial or not, which some of this discussion has

6suggested, just empirically, looking out at all

7arrangements, both technological, things just put

8together, and contractual, that impel, not force, but do

9result in purchasers actually buying two products

10together, that in that domain we are apt to see

11procompetitive effects rather than anticompetitive ones.

12DR. WALDMAN: I would certainly agree with that.


14MR. EVANS: Yes.

15MS. LEE: Robin?

16PROFESSOR FELDMAN: I'm afraid I will stay as

17the stick in the mud here. I can follow all of that

18language with all of the caveats we put in place as we

19discuss it. I can imagine that language taken out of

20context in which suddenly the conclusion becomes that

21tying is always procompetitive. Then, if tying is a

22good thing, what are the antitrust agencies doing

23looking at tying at all? That is the pendulum swing

24that I am very worried about.

25So, when the economists are all here placing


1things in context and with caveats, everything is fine.

2The statement taken as general is one I have great

3concerns about, however. If courts hear "Tying is

4generally procompetitive," there will never be another

5successful typing case.

6DR. WILLIG: But that will be misuse of that


8PROFESSOR FELDMAN: ... and that never happens.

9MS. LEE: What significance, if any, should be

10given to evidence that a challenged tie is similar to a

11tie used in the competitive industry?

12David Evans in his talk suggested that that

13should be evidence of efficiencies. Would the other

14panelists agree with that?

15PROFESSOR FELDMAN: This is going to come back

16to me. Yes, I do see that as evidence of efficiencies,

17subject to timing questions. If you have a market in

18which you see a key patent about to expire, and the

19patent holder suddenly finds efficiencies pointing to

20everybody else around, I find that action and that

21timing suspect.

22MS. LEE: Anyone else?

23MR. RUSSELL: And I think there is a great deal

24of ambiguity when you talk about similar arrangements,

25because in my experience, the tying issues that come up


1often have very unique characteristics that make them

2very different from other arrangements, even at the same

3time that you could look at some aspects of them and say

4they are very similar. So, I think that is a very fuzzy

5concept for me at least.

6MS. LEE: Mike Waldman, do you have anything?

7DR. WALDMAN: Well, I think it is evidence, but

8I think it is not definitive evidence, so it is one

9thing that you could weigh in terms of trying to make a

10decision as to whether it is procompetitive or


12DR. WILLIG: I think it is useful evidence, but

13it needs to be probed for all the elements that might or

14might not make the two circumstances the same or


16MS. LEE: Okay, let's move on to the next one.

17The time has come to abandon the per se label

18and refocus the inquiry on the adverse economic effects,

19and the potential economic benefits, that the tie may


21And everyone I believe agrees with this, but

22please let me know if you do not.

23 (No response.)

24MS. LEE: Okay, I am going to take that as



1If we move to a rule of reason analysis on

2tying, does economics give us the tools needed to

3determine whether a tie is reasonable? Let me start

4with you, Mike Waldman?

5DR. WALDMAN: As I was saying before, I mean, I

6do not have as much experience with cases, but the cases

7that I have looked at in detail, there is typically a

8theory of exclusion, and then the question is, how well

9does the theory -- does the facts of the case match the

10theory, and at least my experience in sort of looking at

11these cases is they do not push it hard enough, but I

12think that is the right approach, which is the theories

13are sort of all over the place.

14There is not kind of one general theory that one

15can apply, and one has to say, okay, here is a theory

16that is well founded theory from an economic theory

17standpoint, let's really probe the facts of the case and

18see whether it matches or do the facts of the case say,

19no, there is some alternative efficiency argument that

20is really driving this. That is how I would think about


22MS. LEE: David?

23MR. EVANS: Yeah, I agree with that. I think we

24understate how much progress the economic literature has

25actually made in understanding tying practices, and I


1think the literature, including Michael's paper with

2Dennis, for example, you know, it is an example of a

3good theoretical framework that you can employ in cases.

4I have the same problem with the actual cases that

5Michael points to, which is oftentimes you basically get

6lip service regarding the economic literature.

7So, rather than the literature and the economics

8being taken seriously and people actually testing with

9evidence the assumptions of the theory and the

10implications of the theory, you know, too often it is,

11you know, so and so economists wrote a paper that says

12tying can be anticompetitive in these kind of

13circumstances, therefore, this is anticompetitive.

14And what I see lots of times in the cases is

15really not taking the theory seriously, and I think if

16we do go to a rule of reason analysis, we do need to

17take the economics of this a lot more -- a lot more

18seriously with evidence and so forth.

19MS. LEE: Anyone else? Go ahead, Don.

20MR. RUSSELL: I almost always presume that more

21information is better than less, and I think that

22economic analysis, economic theory, economic evidence is

23very, very helpful. It is not perfect. It will not

24give you the right answer all of the time, because of

25inherent limitations, but it is clearly very important


1and something that we need to use and need to use


3I would also, though, like to make a pitch,

4which some may disagree with, that it is sometimes

5equally useful to look at intent, not in a sense of,

6well, they wanted to take customers away from a

7competitor, which I think is completely meaningless in

8antitrust terms, but more in the situation, as an

9example that Robin has given, if you look at the timing

10when a tie was first introduced, if you look at the

11documents within the company explaining why they were

12adopting the tie at that point in time, I think that

13will often give you a very useful indicator whether they

14are doing this for beneficial reasons or whether they

15are doing it for anticompetitive reasons.

16MS. LEE: What about the situation in which we

17do not have a preexisting theory that nicely fits the

18facts? Do we have the economic tools necessary to

19determine whether or not a given situation is pro or


21DR. WILLIG: Oh, we could make up new theories

22at the drop of a hat. It is putting them to the facts

23that is trickier.

24PROFESSOR FELDMAN: I do not know whether this

25is where the question is going, but there are some


1suggestions in the legal literature that we have to take

2hands off approach because economics is not clear enough

3or does not give us tools that we can apply in the

4judicial setting. In other words, we should be doing

5nothing here because economics cannot help us, so hands


7I think economics actually has come a tremendous

8distance in the last decade in terms of analyzing tying,

9understanding what its procompetitive and

10anticompetitive. If economics does not have an answer

11for us, however, that does not mean that the law should

12simply sit on its hands and say we cannot do anything.

13This is not economics. These are legal decisions, and

14we have to act within the legal realm. Sometimes we may

15have to actually translate economics into intuitive

16arguments that others will understand. We cannot always

17just ask if economics already has a theory that fits

18what is in front of us.

19MS. LEE: David?

20MR. EVANS: So, first of all, it seems to me the

21fact that we do not have good theoretical reasons to

22generally think that anticompetitive tying is going to

23exist, that has to be a factor that the courts take into

24account in thinking about legal rules. So, I think one

25of my problems with the last series of questions is it


1does sort of presuppose that we are in this full-blown

2rule of reason analysis or asking the question, well,

3what can economics do? And it seems to me we need to

4take into account the prior information that we sort of

5know from the theory, that boy, tying, as Michael has

6pointed out, can be used anti competitively only in

7limited circumstances, and the ability of the economists

8to identify those limited circumstances is not all that


10Having said that, no, I do not think that for a

11rule of reason case you always have to have a

12preexisting economic theory. In fact, I think a lot of

13economic theories actually come as a result of theorists

14trying to fit the theory to whatever case they happen to

15be working on or have heard about. So, I think so long

16as the economists can come up with a logical story based

17on economic evidence that there is -- I keep saying

18long-run consumer harm, if there is a consensus that it

19ought to be long-run social welfare harm, you know, that

20is peachy by me. But yeah, I mean, I think the

21economists can do that in a case. Whether they should

22do that, I am less sure about.

23MR. POPOFSKY: One further comment there. You

24know, one of the most puzzling comments I have read in

25an antitrust case in the last 15 years is Justice Scalia


1dissenting in Kodak, a tying case in part, back in 1993,

2where he said practices normal or ubiquitous in

3competitive markets can take on an exclusionary hue when

4practiced by a monopolist, and that comment has always

5puzzled me, but what you said, David Evans, I think puts

6it in a new light, which is what you need as a Section 2

7plaintiff is you need a story of exclusion that makes

8some economic sense, whether or not it is theoretical


10MR. EVANS: Um-hum.

11MS. LEE: If the per se rule is abandoned, if

12the rule of reason standard yields a sufficiently clear

13and objective rule to determine when a tie is unlawful?

14Let the record note there was a lot of laughter.

15Don, why don't we start with you.

16MR. RUSSELL: Well, I think the first issue that

17any counselor would look at under a per se analysis, I

18think, is do you have market power, are there separate

19products, are you forcing somebody to take both of the

20products? Those, of course, are the kinds of questions

21that are currently asked in deciding whether a tie is

22illegal under the so-called per se rule that we have in

23place today.

24I think those questions will give you the right

25answer most of the time in the real world. There will


1undoubtedly be clients that would come to you who

2probably do have market power, who probably are trying

3to force customers to take two distinct products, and I

4think that the answer to your question -- that Bobby

5will forgive me for stating this out loud -- we do not

6have those answers today because we have been living

7under this bizarre per se rule of law for so many years.

8So, in terms of the legal answer to that

9question, I think at this point it is very hard to say

10other than the very general concept of the rule of

11reason that is out there and the kinds of factors that

12you would look at in any rule of reason case, but over

13time, quite likely, I think refinements of that will be

14developed and rules of thumb and maybe a more structured

15analysis will be adopted by the courts, but it is going

16to take a while to get there.

17DR. WILLIG: I would like to advance as a

18proposition that we really are very good as a community,

19even though after the per se rule in some sense we are

20in new waters, but I think old waters will be fully

21adequate for addressing the first part of the inquiry,

22namely, whether or not the tie, the alleged tie,

23actually does pose a threat or a harm to competition,

24where that phrase is understood in the usual way, as it

25has evolved in the merger domain and in other elements


1of Section 2 analysis.

2When it goes on to this next phase, namely,

3whether the good and the bad impacts of the tying

4practice should balance one way or the other, I think

5those are fresher waters, and as our colloquy suggests,

6we need to talk that through as a community more over

7the next few years.

8I would like to ask a subquestion on that

9proposition to the panel. Do we all agree that when it

10comes to assessing whether a tie does harm competition,

11do you all agree with me that the so-called diminution

12in consumer choice that is the result of the tie is not

13part of what we mean or should mean by "harm to

14competition"? I am talking about noncoincident markets.

15We are talking about in the Microsoft case,

16monopolization back at the level of the tying good. We

17are not talking about the fact that the consumer is

18being forced by the tie to choose the tied good that the

19owner of the tying good is imposing on the market. That

20is not part of the harm to competition. That is my

21position. I am ready to defend it, but I just wonder if

22we all agree on that.

23MR. EVANS: Your proposition is that the denial

24of consumer choice should not be what, under your

25terminology, is harm to competition?


1DR. WILLIG: Right, it is not an element of it.

2It may cause it indirectly, but it is not -- yes.

3MR. EVANS: Putting aside my previous

4qualification that I do not think you have adequately

5addressed on harm to competition, yes, I agree with


7MS. LEE: Anyone else?

8DR. WILLIG: Well, don't be silent, members of

9the panel. Let's all agree on this.

10MS. LEE: Mike, do you have anything to say?

11DR. WALDMAN: Despite my setting antitrust

12policy back ten years, I still think that harm to

13competition is not the right way to think about it, so I

14am a little fuzzy on an answer to which I do not think

15is a relevant question.

16MR. EVANS: And in terms of -- since Michael

17just teed that up, I did not take that as my mandate in

18answering your question, but since you have teed up, you

19know, the use of the merger guidelines framework for

20thinking about harm to competition, I do not actually

21think for Section 2 that is how the courts do or should

22think about things. I mean, we allow monopolies, we

23allow them to do things that raise prices, we want them

24to do all sorts of things, and I am not sure that I

25would want to import a merger guidelines framework into


1Section 2, but --

2DR. WILLIG: Well, we allow harm to competition.

3The question is, do we know it when we see it?

4MR. EVANS: Yeah, that is the question.

5DR. WILLIG: That is the question.

6MS. LEE: That is indeed the question.

7MR. EVANS: Yep.

8MS. LEE: Can we skip to page 9, Brandon?

9Antitrust law should treat ties where the tied

10product is used in variable proportions and ties where

11the tied product is used in fixed proportions with the

12tying product differently.

13Should the law make such a distinction? So,

14essentially when we are talking about tied products used

15in variable proportions, talking about instances such as

16metering, such as the issue in Independent Ink, examples

17of fixed proportions tying include Jefferson Parish and


19Mark, do you have any thoughts on this?

20MR. POPOFSKY: You know, I think we are still at

21a point where, you know, one could argue there is no

22reason for differentiating under either the rule of

23reason or the applicable Section 2 test between them,

24but plaintiff is going to need a story of that magic

25thing called harm to competition. It does not seem to


1me that whether the story makes sense is something that

2is cognizable, something that really sheds light on what

3is going to happen with the practice depends on what

4type of tie it is.

5As Bobby suggested, at the outset, you can

6imagine stories of variable proportion ties, where there

7is some anticompetitive aspect to it, and certainly you

8can imagine fixed proportion ties which are

9competitively benign.

10 MS. LEE: Robin, I know you have to go shortly.

11Do you have any comments?

12PROFESSOR FELDMAN: I do not have anything to

13add to what Mark said.

14 MS. LEE: Michael?

15DR. WALDMAN: I mean, I think there is a

16distinction in the sense that the set of theories that

17apply are different, and so one has to be careful in

18that sense. So, from a -- the variable proportions

19case, there is the efficiency issues concerning

20monopoly, something to competition, trying to use tying

21to avoid these inefficiencies, on the other hand, there

22is price discrimination arguments, and that is only

23going to apply in the variable proportions case, not the

24fixed proportions case.

25So, as long as there is a clear understanding


1that these two different types lead into different

2theories, and so you want to be sort of focusing on the

3relevant theory, then I think that is really the issue

4in terms of thinking about those two different types.

5DR. WILLIG: Yeah, I would much rather, if we

6are going to try to endorse the proposition, substitute

7for variable proportions the idea of price

8discrimination as a cause and motivation of the tie.

9Think about the radio, the prototypical radio in the

10automobile case. There is only one radio. You would be

11crazy to have two radios.

12But on the other hand, you could have a radio

13and CD player and MP3 player and super base speakers, or

14just the very simple stripped-down radio, with or

15without satellite. That is still economically variable

16proportions, but would the law recognize it if that were

17the phrase that we were to go with? So, I think the

18idea of price discrimination as a concomitant of the tie

19would be the right way to structure this sort of


21MR. SALINGER: If I can push you on that one, I

22think there is general agreement that the metering type

23of tying is often about price discrimination, but if you

24take the car and the radio example, that while the price

25discrimination might explain bundling, typically the


1opportunities for price discrimination are greatest with

2mixed bundling, which would not be tying from a legal

3standpoint, and so you would -- if you observe tying,

4then at least if you are not careful about it, you might

5use the Ordover Willig type of test to say, look,

6therefore, go on your profit opportunity, it must be


8DR. WILLIG: You are saying an important part of

9the whole stratagem would be offering the car without

10anything, a hole in the dashboard, at all, that would

11make it even more effective to price discriminate.

12MR. SALINGER: That is right.

13DR. WILLIG: Well, that is a possibility, but I

14think it is arguable whether that is actually true or


16MR. SALINGER: Well, Mike, do you disagree that

17in general the price discrimination argument pushes

18towards mixed bundling as distinct from tying?

19DR. WALDMAN: I think that is right, but I am

20not -- I would have to go back and think about it some

21more. That is my best memory, but that is not something

22I reviewed right beforehand.

23MS. LEE: Let's go to the next proposition.

24Antitrust law should treat contractual ties and

25technological ties differently.


1PROFESSOR FELDMAN: Well, since I am about to

2head out the door, and I have already commented on this,

3let me just add one thought. I think there is a real

4problem in doing that given the state of technology in

5many of our industries. You drive behavior towards

6technological ties, you just encourage people to change

7their products in order to avoid enforcement. So, you

8distort choices, and you are not effectively catching

9the behavior that you want to catch. So, I think it is

10a problem for that reason. There are product design

11issues you have to deal with when you are talking about

12technological ties, but I would be very wary of

13something that says we focus only on contractual ties

14and not technological ties.

15And as my last comment, I would like to point to

16the early 1900s. Treating contractual ties and

17technological ties differently is so close to the theory

18that the courts started out with, that is, antitrust

19enforcement only applies to contractually based

20behaviors and not to behaviors that are intellectual

21property based. That was such a disaster because

22suddenly everybody organized their affairs so that the

23anticompetitive behavior revolved around patents.

24Eventually the courts and Congress had to respond to

25that. I think we would be tempting the same kind of


1behavioral changes now, a hundred years later.

2Thank you for having me. I am so sorry that I

3have to leave, but I do need to get back to California,

4and I appreciate being included in this panel.

5MS. LEE: Thank you for coming.

6David, I under --

7MR. EVANS: Yeah, so three quick comments. If

8you adopted the kind of structured rule of reason

9approach that I suggested with a high hurdle for

10plaintiffs, then no, I would not make technological ties

11different from contractual ties. I would have the same

12high standard for both of them. So, that is point

13number one.

14Point number two, if you told me that the --

15 that it was going to be an unstructured rule of reason

16analysis but I had the possibility of making a

17distinction between technological ties and contractual

18ties, then yes, I think my prior would be that

19technological ties are even more likely to be

20anticompetitive and more likely to lead to errors than

21contractual ties, so then I would make a distinction.

22But third, and this would be my caveat to that,

23I have not looked at these cases for a long -- for a

24while, but my impression of the technological tying

25cases is that you basically have courts that really do


1not like the Jefferson Parish test and have tried to

2figure out ways out of it, and I swear that I have

3looked at some of these cases, and I cannot for the life

4of me figure out why it was a technological tie and not

5a contractual tie.

6MR. POPOFSKY: Let me make a couple of comments

7before Bobby hits them back over the plate, which are


9 You certainly, as Professor Feldman said, worry

10about inefficient substitution and other practices, if

11you condemn one thing under a higher standard than

12another, I mentioned that in my talk.

13On the other hand, to answer David's point, I

14have looked at the technological tying cases and what is

15really striking to me about them or you know aside from

16Microsoft saying we should have the rule of reason and

17not Jefferson Parish, is that those that were trying to

18deal with the issues universally condemned the

19technological tying only when there really was nothing

20on the other side to show any good in it.

21When you go back to the peripheral cases with

22the mainframes Bobby mentioned, the CalComs case, all

23the way through Microsoft, those courts have said, this

24is anticompetitive, have really concluded it is

25anticompetitive because we see nothing good there. We


1see only bad. And the cases where it has basically been

2mixed, the defendant has won. And whether or not the

3legal rule is going to be a profit sacrifice, a

4structured rule of reason, I think that is really

5telling as a descriptive matter of when those ties get


7MR. RUSSELL: My view is that what Mark just

8described is almost inevitable, because I think judges

9feel quite comfortable in saying we will not let you

10enforce this contract. They feel extraordinarily

11uncomfortable in saying you should have designed a

12product that would -- they feel perfectly qualified to

13do one and completely unqualified to do the other, and I

14think the difference that is perceived by most courts

15and judges is not so great in reality as what they are

16perceiving, but I think inevitably they will perceive

17that, and they will treat them differently, whether they

18articulate a formal rule for doing so or not.

19MS. LEE: Bobby?

20DR. WILLIG: Thank you.

21I think at bottom the intellectual framework for

22judging both can be the same, but I think the facts will

23inevitably come in somewhat differently, because in

24part, along with a technological tie comes a product

25design decision which is far more apt to have an


1efficiency rationale or excuse attached to it as opposed

2to lawyers saying, oh, I just had to write the contract

3that way, and inevitably there is more efficiencies that

4the court has to deal with, and I think that is part of

5what Mark was just saying.

6Also, from the point of view of social policy, I

7think there is more at stake, because I do think

8innovation is more delicate or more vulnerable to

9suppressing it than we are to a suppression of the

10writing of complex tying contracts, and so it is right

11to give more respect to the implementation of the tie

12through product design.

13But I do want to say that the right intellectual

14framework will give us the ability to avoid the abuse of

15the respect given to innovation, the false product

16design. It may be a little bit new, but still the main

17point is to exclude. In the situation like that, the

18test that I have suggested, and I think we are all

19pretty much on the same page with trying to uncover that

20kind of innovation, that we should proceed right to a

21real systematic look at the exclusion that takes place,

22even if it is driven technologically.

23MS. LEE: Did you have anything?


25MS. LEE: Okay. Can we go back to slide seven?


1Exclusive dealing is a rule of reason offense,

2requiring a plaintiff to show that the defendant has

3significant market power, the exclusivity arrangement

4serves to deny market access to one or more significant

5rivals, and that market output to consumers is lower (or

6prices higher) as a result. Perhaps the Supreme Court

7will see fit to put tying law on the same course.

8So, do the panelists agree with this statement

9as it applies to tying? I think this is very close to

10what David Evans suggested in a structured rule of


12David, do you want to start?

13MR. EVANS: Well, I do not know if that is a

14structured rule of reason, but --

15MS. LEE: No?

16MR. EVANS: -- but certainly it is a better rule

17of reason, I guess. So, I do not think I have anything

18more to say on that other than that there is a very

19interesting 1956 paper by Justice Stevens before he was

20Justice Stevens on precisely that topic that is

21interesting to read.

22MS. LEE: Bobby, what do you think?

23DR. WILLIG: I am a little worried about the

24middle of it, the one that --

25MS. LEE: Okay.


1DR. WILLIG: -- the part that says the

2exclusivity arrangement serves to deny market access to

3one or more significant rivals. As long as the second

4part of that sentence is really treated very seriously

5and endemically, then I am feeling somewhat comfortable

6about it, but just denying market access itself does not

7strike me as anticompetitive or as creating harm to

8competition, but if it does, then -- excuse the phrase,

9gentlemen -- but there is harm to competition, if as a

10result of the denial of access competition is harmed,

11the sign of that is output is lower and/or price is

12higher, and so we are definitely in the framework of

13having found that there is a problem.

14We are still, then, looking at the next step,

15which is to decide whether the process is essentially a

16competitive one or is it an anticompetitive one. So, we

17are not done. But I guess that is what Hovenkamp has in

18mind here.

19MS. LEE: Don, do you have any reaction to the


21MR. RUSSELL: I agree with the statement.

22MS. LEE: Okay. Anyone else?

23(No response.)

24MS. LEE: Okay.

25MR. SALINGER: I mean, just to follow up a


1little bit, I mean, what the statement seems to be

2saying is that tying should be treated comparably to

3exclusive dealing. One might argue that exclusive

4dealing is a more problematic practice from an antitrust

5standpoint. So, is there agreement here that tying is

6at least as problematic a practice as exclusive dealing?



9DR. WALDMAN: I do not necessarily see it that

10way. It is a question of is the evidence there, is the

11price going to be higher, is the output going to be

12lower? So, it could be the case that it is less

13problematic because it is less likely to cause the price

14to go up and supply to go down, but that the test is

15still the same. So, I think you want to be a little

16careful in terms of kind of that sort of analogy, the

17way you are flushing out the analogy.

18MR. POPOFSKY: One further comment on that,

19Michael. In all these vertical restraint cases, these

20labels, exclusive dealing, tying, bundled discounts,

21they are all imperfect ways of describing what Barry

22Nalebuff has described as a unitary phenomena where you

23are just changing it slightly. So, I think we want to

24be a little careful in saying one is inherently more

25problematic than the other, one is more benign than the


1other. As was just said, you have to look at what is

2going on in a particular segment.

3MS. LEE: David?

4MR. EVANS: Let me push back on that just a

5little bit. I think this is a view that Bill Kovacic

6and other people have as well, that we ought to get rid

7of these categories and recognize that there is

8substitution -- I think you are right about that,

9Mark -- that there is potential substitution between

10these practices, and if we have different legal

11standards, we will observe companies substituting

12between them, and I think you are quite right that that

13is a concern.

14I think as a practical matter, certainly for

15economists and I suspect the courts, I think there are

16sufficient differences between these different practices

17that it is actually useful to think about them

18differently, recognizing that they intersect in various

19places. So, when I think about the economics of tying,

20while I recognize that there are overlaps with bundled

21discounts, you know, they are different considerations,

22 and the way we think about the models and the way we

23think about efficiency effects and so forth, they are

24different, and they educate the analysis.

25I think my concern in just saying, well, there


1is just this stuff out there and we just need to look at

2competitive effects and that is what we should do, I

3think that is problematic because that kind of puts us

4back into this rule of reason stew where, you know,

5everything just goes into it, and we think that juries

6will come out with the best result.

7So, I think we actually do need to pay attention

8to the kinds of practices, make some progress with the

9economics, come up with some priors and some

10understanding of what the rules should be, recognizing

11that Mark is right, that there is going to be some

12substitution if we have different standards in different

13parts of Section 2, but I do not see losing the

14distinctions as being a practical thing to do either for

15economists or for the courts.

16MR. POPOFSKY: And let me just interject, I

17actually agree, David, with everything you have said.

18My only concern is --

19MR. EVANS: My God, I must have said something


21MR. POPOFSKY: No, for once everything is right.

22We just have to recognize, as you said, the linkage

23between these various practices. That is all.

24 MS. LEE: Okay, I want to give the panelists a

25last opportunity to say anything if they like before


1concluding. Anyone? Bobby, you do not want the last


3DR. WILLIG: Oh, I would like the last word. I

4am still worried about the Hovenkamp --

5MR. EVANS: Could I suggest you not go first if

6you want the last word?

7DR. WILLIG: Oh, I see what you mean. I would

8like to hear your reaction.

9It does sound in the Hovenkamp proposition like

10there is an engagement of a consumer welfare meter. It

11reminds me of the situation which is simpler but still

12maybe imponderable to us, a competitor innovates, is

13very successful, the innovation knocks out competitors,

14so a year later, the competitors are gone because they

15have been beat by the innovator, whereupon the

16monopolist really has the monopoly position, at least

17for a while, until the next generation of competitors

18come along.

19We honor the process. We like innovation. If

20we compare consumer welfare before the innovation to

21consumer welfare a year later, after the competitors are

22gone, it could be that prices are up and output is down,

23although that happened through a process that we

24basically honor and we expect another few years will go

25by and the world will be a better place. That is a very


1real sort of scenario, I think, and I think applying the

2consumer welfare meter to that situation would be

3telling us wrongly that innovation is destructive.

4I am kind of worried that when we are talking

5about Section 2 and all of these kinds of practices,

6exclusive dealing and/or tying, that the Hovenkamp

7formulation would be condemning the process, and I think

8in a way that would be unfortunate for antitrust.

9What do you think?

10MR. POPOFSKY: Well, I am going to go next,

11because one of the great things about hiring Bobby as an

12expert, which I have, is I can go after him and not give

13him the last word.

14DR. WILLIG: Redirect, recross?

15MR. POPOFSKY: Your concern is well founded,

16 Bobby, why don't courts condemn monopoly pricing? After

17all, a court could argue we are better off having lower

18prices today even if it deters innovation tomorrow.

19There are in the law safe harbors. There are in the law

20ways of structuring the analysis, whether it is

21structured rule of reason, Ordover-Willig or other

22things, that will filter out, at least in my view, the

23most troubling scenarios, such as designing the better

24mousetrap being found anticompetitive, something we

25should not have done, and the challenge is to really, in


1a particularized way, as David Evans was suggesting, to

2figure out what those are.

3DR. WILLIG: Well, let's do it.

4MR. POPOFSKY: The next panel.


6MS. LEE: Anyone else? Yes?

7DR. WALDMAN: I actually want to go back to

8something David was saying I think similar to what I

9have said, which is in terms of the case, I think what

10is very important is not to just have an existence group

11that some smart economist sat somewhere and came up with

12a theory that this sort of matches on the surface. I

13think that really, given the prevalence of efficient

14tying, I think you really want to make sure that the

15facts of the case fit the theory. Otherwise, you are

16likely to make lots of mistakes, and I think that when

17you go to a rule of reason approach, that is really

18something that needs to be emphasized.

19MR. EVANS: I will just make one sort of

20technical comment, which probably is not a good way to

21end my discussion, but we have kind of gone back and

22forth in the discussion between consumer welfare and

23total welfare, and probably for this area and lots of

24other areas in Section 2, I mean, it really makes a

25difference whether you are talking about consumer


1welfare or total welfare, and it also makes a difference

2in whether you are talking to economists, because,

3Michael, you are probably in a better position to tell

4me whether this is true or not, but my sense is that

5almost all the theories talked about social welfare, and

6the courts talk about consumer welfare, and the

7connection between the social welfare results and the

8theory and the consumer welfare results that the courts

9presumably care about are not quite as tight as we might

10like them.

11So, maybe another panel someday, another topic

12ought to be should there be a total welfare standard

13instead of a consumer welfare standard? It would make

14it easier for the economists.

15MS. LEE: Please join me in thanking our

16panelists for their presentations and our discussion.


18(Whereupon, at 12:56 p.m., the hearing was

19 concluded.)








1C E R T I F I C A T I O N O F R E P O R T E R





6I HEREBY CERTIFY that the transcript contained

7herein is a full and accurate transcript of the notes

8taken by me at the hearing on the above cause before the

9FEDERAL TRADE COMMISSION to the best of my knowledge and



 DATED: 11/21/2006






18C E R T I F I C A T I O N O F P R O O F R E A D E R


20I HEREBY CERTIFY that I proofread the transcript

21for accuracy in spelling, hyphenation, punctuation and




Updated June 25, 2015

Was this page helpful?

Was this page helpful?
Yes No