Understanding Single-Firm Behavior: Loyalty Discounts Session

This document is available in three formats: this web page (for browsing content), PDF (comparable to original document formatting), and WordPerfect.




















199:30 A.M. TO 4:00 P.M.





24Reported and transcribed by:

25Brenda Smonskey





4Federal Trade Commission



7Deputy Assistant Attorney General

8U.S. Department of Justice




12Morning Session:

13Joseph Kattan

14Thomas Lambert

15Barry Nalebuff

16David Sibley


18Afternoon Session:

19Daniel A. Crane

20Timothy J. Muris

21Janusz Ordover

22Willard K. Tom





1C O N T E N T S





6   Joseph Kattan

7   Thomas Lambert

8   Barry Nalebuff

9   David Sibley

10Moderated Discussion

11Lunch Recess





16   Daniel A. Crane

17   Timothy J. Muris

18   Janusz Ordover

19   Willard K. Tom

20Moderated Discussion







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

2- - - - -

3MR. DEGRABA: Good morning, and welcome to our

4first panel of the day on loyalty discounts which is

5part of an ongoing series of public hearings on

6single-firm conduct jointly sponsored by the Department

7of Justice Antitrust Division and the Federal Trade


9 This series is designed to help advance the

10development of the law concerning treatment of

11unilateral conduct under the antitrust laws.

12My name is Patrick DeGraba. I'm an economist

13here at the Federal Trade Commission Bureau of

14Economics, and I'm one of the moderators for this

15morning's session.

16My co-moderator is David Meyer, Deputy Assistant

17Attorney General of the U.S. Department of Justice.

18Before we start, I need to do a few housekeeping


20As a courtesy to the speakers, please turn off

21your cell phones, Blackberries and all other devices

22that will beep during the proceedings. Mine's off.

23Second, the restrooms are across the hall to the

24left of the guard desk where you came in. So ask a

25guard because that description won't help you get there.


1The third is in the unlikely event that the

2building's alarm goes off, please proceed calmly and

3quickly as instructed. If we must leave the building,

4exit through the main entrance. After leaving the

5building, please follow the stream of FTC people that

6are going to the staging area. They have practiced a

7number of times and some of them know where they are


9Also, we request that you not make comments or

10ask questions during the session. It is a moderated

11hearing. For the speakers, I'm going to ask you to

12please speak into the microphones. The sessions are

13being transcribed and videotaped and the microphones are

14the means by which the sound is captured.

15The transcripts and other materials from the

16session will be available on the DOJ and the FTC Web

17sites. And finally, our next hearing will be next

18Wednesday, December 6th, on misleading and deceptive


20Today's session, loyalty discounts include a

21host of related contracting practices. The simplest,

22often referred to as single-product loyalty discounts,

23involve the seller providing a discount on all units of

24a good sold to a buyer once that buyer has reached some

25purchasing threshold.


1More complicated practices, often called

2bundling loyalty discounts, involve the seller offering

3discounts or rebates when a buyer has reached a

4purchasing threshold on several possibly unrelated


6Such practices have raised antitrust concerns

7recently, and the appropriate antitrust treatment of

8such practices is clearly in a state of flux. We are

9honored to have this morning a distinguished panel of

10academists, economists, and private practitioners who

11will discuss the current thinking regarding the

12treatment of these loyalty discounts.

13Our panelists this morning will include Barry

14Nalebuff, a professor of economics and management at the

15Yale University; Tom Lambert, an associate professor at

16the University of Missouri Columbia School of Law; David

17Sibley, a professor of economics at the University of

18Texas at Austin; and Joe Kattan, a partner in Gibson,

19Dunn & Crutcher, LLP in Washington, D.C.

20The organization of the panel is as follows.

21The four panelists will give presentations of

22approximately 15 to 20 minutes. It will be timed by our

23staff here in the front row.

24We will then take a short break. And after we

25reconvene, the panelists will have a couple minutes to


1respond to each other's presentations, and then there

2will be a moderated discussion. We will end about noon.

3David, do you have any comments?

4MR. MEYER: Not at this point.

5MR. DEGRABA: All right. Let's get on with it.

6Our first speaker today is Barry Nalebuff, who

7is the Milton Steinbach professor of economics and

8management at the Yale School of Management.

9Professor Nalebuff has written extensively on

10applications of game theory to business strategy and has

11coauthored the first popular book on game theory, which

12is used in colleges and business schools throughout the


14His current academic research focuses on

15bundling and tying. He has provided expert testimony

16and seminars on antitrust matters to federal

17administrative agencies and courts in Australia and

18Europe and has extensive experience consulting with

19multinational firms.


21PROFESSOR NALEBUFF: Thanks. I'm going to be up

22there and control it?

23Greetings, good morning. What I'm going to try

24and do is give you my overall perspective in terms of

25the way I think about loyalty discounts and bundling.


1And I'm of the view to start with that unlike

2physics, where one is searching for a brand unification

3theory, you won't find that here.

4I still believe there is nothing so practical as

5a good theory. In this case it will be multiple

6theories. The reason is that it is different horses for

7different courses.

8 What matters is the nature of the competition.

9You care about whether the products in the bundle are

10substitutes with each other, as would be the case of

11branded and generic tape; where they are complements,

12such as aircraft engines and avionics; where they are

13used in some fixed proportions, like in a nail cartridge

14and a nail; whether or not one is essential to the

15other, such as Windows and a media player.

16Sometimes the goods are neither complements nor

17substitutes, in the sense of Aspen skiing. Before you

18go to Aspen, the different mountains are complements.

19Once you are there, they are substitutes.

20Sometimes there is no connection, substitutes or

21complements between them. For example, different blood

22tests are all essential but it is not that you use them


24The goods that are in the bundle might be

25positively correlated, negatively correlated or not


1correlated at all.

2All of these factors end up changing the

3motivations and the effects of bundling and you have to

4consider that when you are trying to understand the

5effects and what to do about it.

6The good news is that we are not in the desert

7here lost, that in fact in each case where when you

8understand where you are, we have the tools to analyze


10In my speed attempt to do 10 propositions in 10

11minutes, here we will go. I want you to know these are

12not bundled. You are free to accept any one of these

13individually. But there is a discount if you take more

14than three.

15The first point is that often bundled discounts

16or loyalty discounts lead to negative prices. The

17reason for that is the discount often goes back to the

18first unit that you buy. The end result of that is very

19peculiar prices, things that are hard to justify.

20This issue arises both with single and

21multiproduct rebates. Below, this is an example that is

22an amalgam of actual prices that I have seen from

23different cases where things have been normalized and


25But the way it works is your price for the first


131 units was 100. Your price for the 32nd unit was

2minus 6000. Your price for the next couple units is 100

3again. When you get to the 95th unit, your price is

4about minus 800. And then for units 96 through 100, it

5is 97.

6Now, if you thought about that as sort of a

7rational way of doing it, you would say what is going on

8here, does that really make any sense?

9Of course the customer should never be in a

10position of buying fewer than 31 items because in fact

11the first 32 are free. But then having bought 32, now

12they are okay until they get to 85 because once you get

13to 85, 85 through 95 is free.

14What that means is if a rival wants to come in

15and displace the firm entirely, it will not happen

16because 31 units are free. Moreover, a rival will never

17be able to sell between 85 and 95 or, in that case,

18between 5 and 15.

19The solution, in my view, to that is to still

20give out discounts but to give out discounts on

21incremental volume rather than go back to square 1.

22And note, if that's your objective to give

23people low prices, we have ways of doing that. I'm not

24preventing the discounts, just trying to make them a way

25that actually makes some sense.


1The second point is that loyalty discounts can

2actually create no cost predation. And I'm going to

3give you a quick example of this in terms of numbers.

4The reason is that what we do is we inflate the

5price of A rather than really give a discount. Imagine

6the normal monopoly price of A is 100 and you can get it

7at the normal monopoly price if you also buy B at 20.

8But if you don't buy the B, then I will raise

9the price of A to 120. Hence, the effective price of B

10is zero or certainly below cost in this case.

11Now, the key observation is that nobody actually

12pays the 120 because nobody is foolish enough to only

13buy A on an a la carte basis. Therefore, since the

14threat is credible, it doesn't have to be used and it is

15not costly.

16The difference between predation and this type

17of loyalty discount is that under predation, the firm

18actually charges below cost, and so customers benefit

19from those low prices.

20Here all that is happening is the firm is

21threatening to charge a high price if you don't go

22along. It is like the mugger who says "your money or

23your life," and when you give him your wallet, he wants

24credit for actually saving your life. Actually, I don't

25think that gets to count.


1Because there is no need for recoupment, it is

2easier to implement this. Hence, there is a greater

3danger of it. Because customers aren't necessarily

4winning along the way, there is also more reason to be


6To give you another disguised example of this,

7the following is a case where an incumbent firm had a

8market power in three goods, 1, 2 and 3, and they

9offered prices like you see in column 1.

10However, if you were to buy all four of their

11products, including their competitive fourth product,

12then you would get the discount, 16, 26, 51, so on


14If you added up those discounts, what you

15discover is that the cost of buying all of the three

16products on an a la carte basis, which essentially you

17had to do anyway because they were the only supplier of

18those three products, ended up being sufficiently high

19that you were going to save $1-1/2 million by buying the


21The end result of that was it was actually a

22negative incremental price to go and take the

23competitive product.

24Once again, that is something that is very hard

25to compete with. That leads to the following proposed


1test, which is if you have a firm which has market power

2in A and you are worried about whether or not it is

3going to extend that to another good, B, look at the

4price of the A-B bundle versus the price of A alone and

5ask how much more is the firm charging for A and ask

6could that firm itself make money selling A at that

7incremental price or B at that incremental price.

8So instead of asking whether or not the rival

9can make money selling B at that price, is the firm

10itself apparently making incremental profits or not. If

11it isn't, then what we have is a case of exclusion, and

12that exclusion can be achieved without cost.

13One of the things that is nice about this test

14is that we actually don't have to look at actual rivals

15or hypothetical rivals, we can look at the incumbent

16firm's own cost structure. The incumbent firm which

17knows its own cost structure.

18Therefore, it is well equipped to discover

19whether it is passing this test or not. It knows

20whether it is in the safe harbor or it isn't.

21There is an extra element to this test that

22David Sibley and his co-authors have emphasized, which

23is did the price of A go up or did the price of A-B, the

24bundle, go down.

25We should be more worried about the case when A


1alone goes up than when the A-B bundle goes down

2because, of course, when it is a threat, there is no

3benefit. Whereas, if the bundle has been discounted, at

4least customers are getting some value along the way.

5One point that I think the courts have really

6missed about loyalty discounts is that some of the ways

7that these rebates are paid end up being significantly

8less competitive than a straight price cut.

9So, again, think of a case where the incumbent

10has market power in A, and B is a substitute. And the

11two examples I will take you through are Scotch tape and

12generic tape or Keflin and Kefzol, two cephalosporins,

13where Keflin was the big money maker and Kefzol was the

14new product which is the competitive one.

15In the cephalosporin market, we had Lily with

16its monopoly and Keflin, Keflex, Loradine, Kaphacen and

17facing competition with SmithKline Ancef, which was the

18exact same compound as Kefzol.

19The first thing they tried doing was just

20discounting Kefzol to match the prices on Ancef. The

21problem with that was that Kefzol ended up being a

22substitute for Keflin.

23So not only did they have trouble capturing the

24market against Ancef, as prices starting getting lower,

25it started eating in on the demand to Keflin.


1Then they got wise and said okay, we will give

2you a rebate on Keflin and the other products if you buy

3enough of our goods.

4Now, note what happens here. The price of

5Kefzol ends up being high. I'm getting a million

6dollars back or some fixed amount of money back, but I

7don't end up discounting Kefzol.

8In essence, I'm bribing you to say if you buy

9all of my goods, I will give you this fixed amount of

10money. But because Kefzol keeps its price high, that

11reduces the competition between Kefzol and Keflin, and,

12hence, customers don't get that benefit.

13We also see that by its equivalent it is almost

14as if Lily says to the customer we will give you 100

15units of Kefzol for free on the condition that that's

16all you use, which of course is something again that

17rivals would have a hard time matching.

18We have the same issue in LePage's. If you are

193M, you don't want to get into a price war with LePage's

20over generic tape, because the cheaper generic tape

21gets, the more that will eat into Scotch tape prices.

22What you want to do is how can I beat LePage's

23without discounting my generic tape. Well, if I give

24them a bribe, a million dollars just to take my goods,

25even if they are high priced and you can spread out that


1million dollars over their expected sales, then you can

2say the overall deal is better for me, Staples, than it

3is for taking LePage's.

4But note the incremental cost of another roll of

5tape is high. What that means is the price to consumers

6for that tape is going to be high and there will be less

7substitution of generic for branded product.

8So in that sense, these rebates don't get passed

9on to consumers and don't threaten the incumbent


11That's an aspect of these loyalty rebates that I

12don't think has been appreciated and I think is


14Another area is that loyalty rebates make

15pricing incredibly hard to understand. If somebody

16offers 2.93, I know that is cheaper than 2.97. But if

17somebody says you get 3 percent off A and B if you buy

18B, is that a good deal or not? Well, it depends on how

19much A I'm going to buy. And sometimes I know the

20answer to that and sometimes I don't.

21Moreover, if rivals are trying to compete and

22think about how much they have to undercut to get the

23business, that means the B rival has to forecast my

24demand for A, and, generally speaking, they are not very

25well equipped to do that.


1So we have seen cases where people misforecast

2these demands, end up buying the wrong product or don't

3get discounts as large as they think.

4I have also found that actually analyzing these

5price things can often take an MBA. And it is not an

6understatement to say it costs $10,000 to actually

7figure out what price is the cheapest, and many times

8that is not worth it for the individual customer to do.

9 An issue that bothers me about loyalty discounts

10is that the price a firm charges to a customer shouldn't

11depend on who else the customer buys from. I have less

12 a problem if the price says if you buy many units,

13here's the charges. If you buy this many more units

14this year compared to last year, here's the charge.

15I think it is very funny to say to the customer,

16"oh, and if you buy 10 units from Fred, I'm going to

17charge you more money" or "if you buy 3 percent of your

18products from Fred, I'm going to charge you more money."

19The price that I charge you should ultimately

20depend only on what it is that you buy from me, not what

21it is that you buy from other people.

22Now, I realize that the effect may be the same

23through some volume discounts. But that still leaves

24many more options in an uncertain environment for a

25rival to come in than when you literally price based on


1what you are doing with your rivals.

2You will hear many what I will flat out call

3bogus justifications for bundled discounts. For

4example, it is often said that customers like bundles

5and, hence, that's a justification for doing bundling.

6Yes, that's true, but it is not a justification

7for a bundle discount. Because a customer likes it, in

8theory you could charge more for it. You don't have to

9offer it as a discount if you are providing something

10customers like better.

11We do the discount for price discrimination.

12Well, there is no room for price discrimination if A and

13B are consumed in fixed proportions.

14Moreover, the arguments for price discrimination

15generally rely on having a negative correlation between

16the two products or no correlation in valuation between

17the two products.

18For example, opera tickets and wrestling tickets

19you think of as having negative correlation. However,

20if you look at what's bundled out there, I think you

21will find that they generally have a positive

22correlation in value and, hence, don't fit the normal

23framework that we would expect price discrimination to

24fall under.

25Yes, Virginia, bundling can leverage and protect


1market power. Here is an example of how that works.

2If we have a monopolist whose demand is

3represented by 10 minus P and the cost is zero, the

4monopoly price would be 5. Profits would be 5 times 5.

5Price is 5, quantity is 5.

6I'm having the B product be competitive with a

7cost of one. So the price is one. Demand I'm making

8just to be one unit.

9Chicago School says don't sell A and B together

10at 6. I do better just to sell A alone at 5, because

11there are some people who may not want B, even at the

12competitive price.

13What I say is consider the following contract.

14If you buy my B, I will lower the price of A to 4. But

15if you don't buy my B, I will raise the price of A to 6.

16Well, if you think about the cost of that threat

17and promise, the customer is going to save at least $2

18on A by buying the B product since they are going to be

19buying at least four units of A.

20That means that it is a net savings to them of

21at least 8, which means they are willing to pay up to 9

22in order to get that discount. They will pay 9 on B to

23get that discount.

24Well, the discount doesn't cost the firm very

25much. And the reason is that discounts my price from 5


1to 4 only lowers my profits from 25 to 24. Raising my

2price from 5 to 6 also only lowers my profits from 25 to


4 So at a cost to me of only a dollar here, I can

5do something that will either reward or punish the

6customer to the tune of 8. And the reason for this is

7the monopoly is inefficient.

8So in essence, what I'm saying to the customer

9is I'm willing to be a less inefficient monopolist if

10you play ball with me and do what I'm asking on good B.

11It doesn't make sense to take out all of your monopoly

12rents on the monopoly product because that's what leads

13to dead weight losses.

14What I would like to do is some type of lump sum

15payment and incremental pricing and charge the customer

16for the right to buy my goods at a reasonable price.

17Oftentimes the way we see that happen is the way

18I charge them for being less of a monopolist is I say

19you have to buy my other goods at B at inflated prices.

20It is also the case that the bundle allows firms

21with multiple market powers to protect themselves. So

22if I have market power in A and B and charge 10 for A

23and 10 for B but only 16 for the two together, there is

24a $4 discount that any single-firm rival would have to

25meet in order to undercut me.


1Note that my average price is 8. In essence, I

2get to use that same $4 discount on multiple fronts. So

3the customer isn't benefitting $4. The customer is only

4benefitting 2 on each.

5Rivals would actually have to go 4 below. That

6is a special sauce in multigood bundling that makes the

7incumbent have an advantage over rivals. It is sort of

8why it works.

9It also explains to me why the right test should

10not be whether or not the overall bundle is above or

11below cost but whether or not the individual components

12at the appropriate incremental price is above or below


14So the Chicago School story is correct in its

15limited environment, but it misses most of the

16interesting cases that we look at when it comes to


18Even where there is one monopoly profit, that

19monopoly profit can be of different sizes. In

20particular, bundling can allow price discrimination,

21such as through metering and some of the examples you

22have seen, which, therefore, leads to greater profits to

23the monopolist but less surplus to the consumer.

24It is also the case that many of the motivations

25for bundling are dynamic, that by preventing somebody


1from getting into the B market, that may be their

2subsequent entry into the A market which is where I

3still have market power.

4It is also the case that bundling and tying

5provide potential for no cost for closure, which has the

6same effect as predatory pricing but at no cost.

7I recognize that bundles versus bundles is

8generally more competitive than individual items versus

9each other. So what I would like to be able to do is

10take the advantage of that competition without the harm.

11And the way that I do that is the following. I

12actually take the example from Johnson & Johnson who

13said, look, U.S. Surgical, you have a full line, we have

14a full line, Coke and Pepsi, you each have full lines,

15you can compete against me bundle for bundle.

16But if I don't have a full line, I will not

17count your sales in my 80 percent number or 90 percent

18number. Whatever target I make, it is only a target for

19other full-line competitors.

20 We have come our way through the deserts often

21through intuition. There are now some tests that I hope

22you will believe offer more formal approaches.

23And I believe -- maybe this is a temptation here

24-- that the theories of bundling loyalty discounts are

25now ready for prime time. So I hope you will be able to


1use them.

2Thank you.


4MR. DEGRABA: Thank you.

5Our next speaker is Tom Lambert, who is an

6associate professor at the University of Missouri

7Columbia School of Law, where he has achieved the

8university's Gold Chalk Award for excellence in graduate


10Professor Lambert's scholarship focuses on

11regulatory theory, including antitrust policy and

12business law. His 2005 Minnesota Law Review article

13provided one of the first scholarly treatments of the

14law of bundling discounts.

15Tom is a member of the eSapience Center for

16Competition Policy and is a regular contributor to Truth

17on the Market, a Weblog devoted to academic commentary

18on law, business, economics and more.



21It is an honor to be here on such a

22distinguished panel. I will talk today about bundled

23discounts entirely. I will not focus on single product

24loyalty discounts.

25A word about the scope of my remarks. I'm a


1lawyer, not an economist. I'm very concerned with

2structuring rules in a way that they can be administered

3by judges and juries and used by antitrust counselors to

4advice their clients.

5My focus is on the law, how we would structure

6the rules.

7I have a three-pronged agenda that's very

8ambitious for 20 minutes.

9Why are bundled discounts troubling, and I will

10give you the straightforward view the courts have

11adopted and most of you are familiar with this.

12Summarizing and critiquing of the leading

13evaluative approaches offers an alternative proposal

14that I think is very administrable.

15The problem with bundled discounts the courts

16have recognized is they may lead to the exclusion of an

17equally efficient but less diversified rival even if

18they are above cost.

19The classic example of this came in the Ortho

20Diagnostic case. It is I think a little bit

21unrealistic, but this is what the court wrote in its

22opinion and it illustrates the problem, I think.

23You can have two manufacturers who sell the same

24product, manufacturer A and manufacturer B. They both

25make shampoo. Manufacturer B is the more efficient


1producer. It can produce shampoo at $1.25 a bottle.

2Manufacturer A, it costs $1.50 to produce the shampoo.

3Manufacturer A, though, is a more diversified

4rival. It sells conditioner as well as shampoo.

5So by bundling its shampoo and conditioner and

6by offering an above-cost bundled discount -- and what I

7mean there is that the price, the discounted price of

8the bundle is in excess of manufacturer A's cost of

9producing the bundle -- manufacturer A can effectively

10exclude manufacturer B from the market.

11If the separate price of shampoo and conditioner

12for A is $2 and $4, so that buying them separately you

13would have to pay $6, and manufacturer A charges a

14package price of $5, that is still a dollar in excess of

15its average variable cost of four dollars. Manufacturer

16B can't compete with that.

17In order to sell its shampoo -- and any buyer

18that buys both shampoo and conditioner will have to pay

19$4 for the conditioner and will not be willing to pay

20any more than $1 for the shampoo. Manufacturer B is

21excluded despite the fact that it is the more efficient


23So the fundamental problem the courts have

24identified is that bundled discounts can lead to the

25sort of exclusion of equally efficient but less


1diversified rivals, and that's the case even if the

2discount is above cost.

3All right. I have identified six approaches in

4the case law and commentary for evaluating the legality

5of bundled discounts. I want to march through them

6quickly and explain why I think each is a little bit


8The first and the most sort of laissez-faire is

9a rule of per se legality. This is the rule that's been

10advocated most recently by Professor Hovenkamp in his

11new book, "The Antitrust Enterprise," and also the rule

12advocated by Demicci in the LePage's case.

13It basically says a bundled discount should be

14per se legal if the discounted price of the bundle

15exceeds the aggregate cost of the products within the


17The reason for this rule is not that we don't

18believe that above-cost bundled discounts can ever be

19anticompetitive. The Ortho Diagnostic example showed

20how they could lead to the exclusion of a more efficient


22Administrability concerns motivate this rule.

23The idea is that it is simply too difficult to separate

24the pro-competitive wheat from the anticompetitive chaff

25and will end up chilling pro-competitive bundled


1discounting if we don't have the sort of safe harbor,

2and so the best approach is to have a per se legality

3rule for above-cost bundled discounts, very much along

4the lines of the Brook Group rule.

5My criticism is -- well, I'm not all that

6critical. In the long run, this may be the best

7approach to take. However, I'm not willing to concede

8that at this point.

9I think the search for anticompetitive bundled

10discounts may be worth the cost, including the cost of

11deterring some pro-competitive bundled discounts.

12It is very easy to imagine instances of

13anticompetitive exclusion. Professor Nalebuff and

14Professor Sibley have modeled cases where this could

15occur. The Ortho Diagnostic example is a good example.

16I think there is a fairly easily administrable

17weeding device that can help us separate pro-competitive

18from anticompetitive bundled discounts. I will get to

19that in just a minute.

20The second approach is at the other end of the

21spectrum -- and this is an approach from the raising

22rivals costs literature. I'm thinking in particular of

23Will Tom, who will speak this afternoon, and Einer

24Elhauge, who has discussed this in testimony on hospital

25group purchasing organizations and also in his Stanford


1Law Review article defining better monopolization


3This approach says that bundled discounts are

4discounts are illegal if they unjustifiably usurp so

5much business from their rivals that their rival's costs

6are erased.

7Now, the $64,000 question here is how do you

8determine what is unjustifiable. Every discount tends

9to usurp some business from rivals. And obviously we

10don't want to ban discounts.

11The concern here is that so much business will

12be usurped from rivals that it will deny rivals

13economies of scale, make it harder for them to raise


15A couple of approaches have been advocated for

16identifying what are unjustifiable instances of raising

17rival's cost.

18Will Tom suggests in his article on the

19Antitrust Law Journal that we adopt a case-by-case test

20where the courts look to see is this an exclusionary

21usurpation of the business or a pro-competitive

22usurpation of the business.

23That is difficult because that leaves a lot open

24to the whims of juries and judges and will likely have a

25chilling effect on pro-competitive bundled discounts.


1Professor Elhauge has suggested an approach

2where a business-usurping discount is justified only if

3the discounter's business stealing, business usurpation

4occurs because the bundling has made the discounter more


6If you are stealing business because your

7bundling is making you more efficient, then that's okay.

8But if you are stealing business for any other reason,

9then that's illegal.

10I think this is a troubling approach for several

11reasons. First, it would prevent price cutting by a

12monopolist who has reached minimum efficient scale and

13can't achieve any additional distribution efficiencies

14by bundling.

15That person is not getting any efficiency

16benefits from the bundling and then would be precluded

17from cutting prices, which seems bad for consumers.

18Secondly, this approach is very difficult to

19administer. A court would have to figure out what is

20minimum efficient scale, very difficult for judges and

21juries to do.

22In addition, it has to figure out what discount,

23what amount of discount is necessary to get the

24discounter to the point of minimum efficient scale. Any

25discount beyond that would be excessive discount and


1under Professor Elhauge's test would be exclusionary.

2That is extremely difficult for judges and

3juries to administer. For that reason, this approach is

4likely to have a major chilling effect. Discounters

5discount at their own peril.

6The third approach is the approach we sort of

7see in LePage's. Everyone in this room knows it is very

8difficult to articulate a rule of law from the LePage's


10There were some key facts that were very

11important in the court's analysis there. LePage's was

12not required to prove that it couldn't match the 3M

13discount. It was not required to prove it was as

14efficient a manufacturer as 3M was.

15Instead, it just had to show that it was being

16excluded. And once it showed that, the burden shifted

17to 3M to justify its behavior.

18So if you want to take away a rule from that --

19and lots of smart antitrust counselors are trying to do

20so and advise their clients accordingly -- it would seem

21to be the following. A bundled discount is

22presumptively exclusionary if the discounter is bundling

23products not sold by rivals and is winning business from

24those rivals.

25Now, the discounter may rebut that presumption


1if it proves a business reasons justification. There is

2a suggestion in the LePage's case that that

3justification must show that the bundling saves costs

4approaching the amount of the discount, very similar to

5Professor Elhauge's suggestion in the Stanford Law


7This I believe is a very troubling rule. First

8of all, since the plaintiff need not establish its

9equivalent efficiency, this approach essentially creates

10a price umbrella for less efficient rivals.

11And there is a suggestion in LePage's that's

12exactly what happened. LePage's expert economist

13conceded that LePage's was a less efficient manufacturer

14of tape than 3M and yet LePage's won.

15Moreover, since the focus is on product line

16breadth and not whether an efficient rival is being

17excluded, this approach will tend to chill bundling,

18which has a number of pro-competitive benefits which we

19will talk about in the roundtable discussion. I assume

20that some of my co-panelists will discuss that issue.

21The third approach here -- fourth approach, I

22guess -- the approach we see in the Ortho Diagnostic

23decision, in that case, the court reasoned that a

24bundled discount is illegal if the plaintiff shows

25either that the bundle is priced below average variable


1cost, straightforward predatory pricing, or that the

2plaintiff is at least as efficient a producer of the

3competitive product but cannot match the discount

4without pricing below cost on that product.

5In other words, you have to show you are an

6equally efficient rival, and after you show you are an

7equally efficient rival, you show if you attribute the

8full amount of the discount to the competitive product,

9that will result in below-cost pricing by the

10discounter. You couldn't match that discount.

11My criticism of this rule, it is a great rule in

12theory, but this is a very difficult rule to administer.

13The plaintiff, in order to prevail, has to show

14that it is an equally efficient rival. To do that, it

15has to establish its own cost and the discounter's cost.

16In addition, there are going to be joint costs

17in here because this is a bundling case. In figuring

18out the discounter's cost on its competitive product, it

19has to figure out what percentage of the joint cost it

20should attribute to that competitive product.

21That is an incredibly difficult rule to

22administer. For that reason, I believe this rule, the

23rule of law in Ortho Diagnostic, may be underdeterrent,

24because plaintiffs are going to have a hard time winning

25these cases.


1The next approach is what I'm calling the

2original antitrust law approach. This is the approach

3that was advocated in the Areta/Hovenkamp treatise. It

4was updated this summer. I had to update my


6The original approach advocated by the treatise

7was focused on trying to fix the administrability

8problems with the Ortho Diagnostic test.

9Rather than asking if the plaintiff itself was

10an equally efficient rival, the original antitrust law

11approach said let's ask if a hypothetical equally

12efficient single-product rival would be excluded by this

13discount and without adequate business justification.

14So essentially we take the Ortho Diagnostic

15test, we lop off the part where the plaintiff has to

16show that it is actually an equally efficient rival, and

17we say if you attributed the entire amount of the

18bundled discount to the competitive product, would a

19hypothetical single product be excluded by this


21This is definitely an easier to administer test

22because plaintiffs don't have to prove the defendant's

23costs where there are joint costs. It is troubling,

24though, for a couple reasons.

25First, it prevents discount cross-subsidization.


1Consider a situation where you have a seller that sells

2products A, B and C. Its cost is $4 each. It sells

3them separately for $5 each. But it would sell the

4bundle for $13.50.

5Under the antitrust law approach, this would be

6a presumptively exclusionary discount because a single

7product seller of A that was equally efficient at a cost

8of $4 couldn't match this discount because it would have

9to charge a price of $3.50, a price below its cost.

10Now, if you think about an oligopolistic

11market -- it is not cartelized, but there is a lot of

12what looks to be tacit collusion -- if you assume the

13seller that sells A, B and C is selling in that market,

14it is great that the seller can engage in the sort of

15complicated pricing.

16Professor Nalebuff says it is very difficult to

17figure out exactly what price is being charged.

18That's a fantastic thing in an oligopolistic

19market. This sort of pricing can disrupt, this sort of

20bundling can disrupt oligopolistic pricing. In

21addition, it is a discount for customers. That would

22seem to be good in itself.

23A second problem with the antitrust law approach

24is there was no requirement that the foreclosed market

25be capable of monopolization, there was no requirement


1that there be entry barriers in the foreclosed market

2that the plaintiff was being excluded from.

3The revised antitrust law approach is definitely

4superior to the original. But I still think it is a

5little bit troubling.

6What Professor Hovenkamp is now saying -- which,

7by the way, seems to conflict with his book, "The

8Antitrust Enterprise" -- is that we should analogize

9bundled discounts to tying and say there is a tie-in if

10the price is below cost when the entire discount is

11attributed to the competitive product.

12Very importantly, the treatise points out there

13will not be this tie-in if there is another significant

14rival that sells all products. In the Johnson & Johnson

15versus Tyco case or U.S. Surgical case,

16Johnson & Johnson engaged in this bundling, but there

17was another significant rival that had the same bundle

18in place.

19Professor Hovenkamp would say that does not

20constitute a tie. But absent such a significant rival,

21there would be a tie-in if there was a below-cost price

22after the discount was attributed to the competitive


24The treatise then says that after you find time,

25you should apply a basic rule of reasoned approach, ask


1whether the foreclosed market is capable of

2monopolization, ask if a collaborative bundle is

3probable, ask if there are pro-competitive

4justifications for the bundling.

5This is a definite improvement on the original

6version. My criticism is why involve tying at all. It

7seems to me that the reason that we are concerned about

8tying in cases like this is that it leads to


10Why should we focus on the tie rather than

11focusing directly on the foreclosure issue?

12Here is my alternative proposal. The goals of

13the proposal is we want to condemn bundled discounts

14that could eliminate competitive rivals and result in

15price increases. We don't want to condemn other bundled

16discounts. And we want the rule to be easy to


18What I want to structure my rule to show is that

19the complaining rival has exhausted its competitive

20options. You are not a competitive rival unless you

21have done everything you can to stay in business.

22The complaining rival must have the ability to

23match the bundled discounter's efficiency. You are not

24a competitive rival if you are not as good as the



1We have to show the foreclosed market is capable

2of monopolization. We don't want to ban discounts in

3 markets that can't be monopolized because there are very

4low barriers to entry.

5Here is a proposed rule. I would have a rule

6that says that the above-cost discount, and that means

7that if you add up the cost of all the items in the

8bundle, they are exceeded by the price of the bundle.

9So the above-cost discount is per se legal

10unless the plaintiff could not match without pricing

11below cost and, number one, barriers to entry exist in,

12A, the product market in which the plaintiff doesn't

13participate, and, B, the market for the competitive

14product, a collaborative bundle is impracticable, a

15good-faith supply offer was rejected. That means that

16the foreclosed firm goes to the bundled discounter and

17says, hey, let me supply my products to you, you buy my

18product and bundle it.

19And if those are established, then the bundle is

20considered presumptively exclusionary, but the defendant

21gets a rebuttal opportunity to show that it rejected

22this good-faith supply offer because it wasn't

23attractive, either the price being offered was too high

24or the quality was insufficient.

25Let me explain how this meets all of my goals of


1protecting competitive rivals. We want to protect

2competitive rivals and only competitive rivals, and we

3want to ensure that the market that is being foreclosed

4is capable of monopolization.

5The above-cost discount is per se legal unless

6the plaintiff could not match without pricing below

7cost. That requires a complaining plaintiff to lower

8its price to the level of its marginal cost.

9That's what we expect will happen in perfect

10competition. We should demand that of a complaining


12Next, it has to show that barriers to entry

13exist in a product market in which the plaintiff doesn't

14participate. An option for a plaintiff that's

15confronting a bundled discount is to enter the other

16markets in which it doesn't participate. It needs to

17show there are some entry barriers that prevent it from

18being able to do so.

19In addition, it has to show barriers to entry

20into the market for the competitive product. That's

21required to show the market is in fact capable of


23Supercompetitive prices could be charged in that

24market without inviting so much entry that it is

25impossible to charge those prices.


1Next, the plaintiff would have to show that a

2collaborative bundle is impracticable. It cannot

3compete with the bundle by entering into agreements with

4sellers of other products to craft a competing bundle.

5These sort of cross-seller bundles are

6incredibly common. I sent my research assistant to

7Target, and he found an Olympus digital voice recorded

8bundled with batteries, Suave body wash bundled with a

9Schick razor, Colgate White-Plus teeth whitening cream

10bundled with a camera. Americans are vain.

11The prima facie case here is intended to show

12that the plaintiff has exhausted its competitive options

13and that the market being foreclosed is capable of


15Then we have a rebuttal opportunity. The

16defendant may rebut by showing that the supply offer was

17not attractive.

18The defendant has to show that when the

19plaintiff came and made the supply offer to me, I didn't

20accept it because the price it was charging me was

21higher than my cost. That shows that the plaintiff is

22in fact a less efficient rival.

23If the plaintiff can show its prima facie case

24and the defendant can't rebut, then we have an exclusion

25of a competitive rival in a market that is capable of


1foreclosure or capable of monopolization, and it would

2seem to me that liability is appropriate.

3Otherwise, I would have a rule that these sorts

4of discounts which are discounts, good to customers, are




8MR. DEGRABA: Our next speaker is David Sibley,

9who is the John Michael Stuart Centennial professor of

10economics at the University of Texas at Austin.

11Professor Sibley was previously the head of the

12economics research group at Bell Communications Research

13and served as a member of the technical staff in

14economics at Bell Labs.

15In 2003 and 2004, David served as a Deputy

16Assistant Attorney General for economic analysis in the

17Antitrust Division.

18Professor Sibley has carried out extensive

19research in the area of industrial organization,

20microeconomic theory and regulation, and his

21publications have appeared in numerous leading economics

22journals. He has consulted extensively for various

23firms and agencies, both in the United States and

24abroad, on antitrust and regulatory matters.




2The title of my talk, what have we learned since

3LePage's about bundled discounts, I guess is sort of

4inspired by the feeling of knowing not what to say at

5the Antitrust Division when the parties representing

6both sides of LePage's came to convince us either to

7support a take cert brief or not.

8There was not a whole lot the economists had to

9say. Greg Warden was on the right track when he said

10"what do the prices do?"

11It turned out you couldn't tell from the

12evidence in the record. I take LePage's as kind of a

13baseline as sort of very useful knowledge.

14What have we done since then? Well, there has

15been some progress. We understand now I think better

16the effects of bundled discounts on both foreclosure and

17customer welfare.

18I mentioned foreclosure and customer welfare

19separately here because, as we will see, it is possible

20to have a bundled discount which increases customer

21welfare and yet excludes equally efficient rivals.

22I expect that to be the case from the way

23bundled discounts can be structured. I will also talk

24about tests to determine if a bundled discount is



1This would be in the spirit of the Ortho test

2with its explication of extension by Barry Nalebuff or

3tests whether customer welfare rises or falls. The

4reference here would be what I was aware of without

5having to go to any trouble to look up more, Wrightman,

6Sibley and Roy Nalebuff.

7The tests here I guess were designed originally,

8both to try to see whether we could figure out whether

9bundled discounts are good or bad but also with a view

10toward the same type of goal that Tom Lambert had,

11administrability here.

12We wanted simple tests that didn't require you

13to calculate complicated things or use data that you are

14not likely to be able to get in practice. The result is

15we have tests that will work sometimes but not all the


17To start, I will take a very simple set-up which

18is actually I think probably the set-up behind some of

19the slides here.

20A is a monopoly market served by a firm we will

21call Firm 1. B is a competitive market. It might not

22be perfectly competitive. I think for the next five

23minutes or so I will assume it is perfectly competitive.

24But it doesn't have to be.

25Firm 1 is a seller in the B market too. I think


1for the purpose of the rest of the slide, I want to

2assume a couple things, one, that the B market is

3perfectly competitive and some B customers will buy B

4only and some will buy A only.

5A couple of preliminary observations. Starting

6from independent pricing, a bundled discount or a BD can

7raise both profits and customer welfare. That doesn't

8mean that it will actually happen by a profit-maximizing

9profit discounter, but it is capable of happening.

10We should keep that in mind. The logic is

11really as follows. Let's suppose we have a preexisting

12time that we can observe where Firm 1 is engaged in

13independent pricing and it is a monopolist in market A.

14We will assume that if it hasn't anticipated the

15onset of the regulatory rule I will be talking about,

16that the price it charged to the A market was probably a

17monopoly price.

18As Barry was saying, if the monopoly price is --

19if the price charged by firm one in the A market really

20was the profit-maximizing monopoly price, then it is

21always possible to have a slight discount on the price

22of A, which will have an insignificant effect on profits

23that Firm 1 generates in market A.

24In Barry's example, it was a $1 increase in

25profits. From a customer welfare standpoint, that is


1not an insignificant increase in customer welfare.

2That allows the firm, Firm 1, to bundle that

3slightly lower price of A with a price of B that's above

4marginal costs and still get A and B consumers to select

5the bundle in preference to buying any A at all or

6paying a bundled price for it and getting B at a

7marginal cost.

8In that situation, the A/B consumers are better

9off. They can all select the bundle that will make them

10better off.

11B-only consumers are nowhere better and no worse

12off than before. They are getting B at marginal cost

13from all the other perfect competitors out there.

14If the bundled discount in doing this has an

15out-of-bundled price no higher than the previous

16monopoly price under independent pricing, then we know

17that consumers' options within the bundle are no worse

18than before.

19In fact, we have designed the bundle to attract

20them away from independent pricing of A and marginal

21cost of B. So they are better too.

22So starting from independent pricing, which

23would be the marginal cost of B for everyone, including

24Firm 1, and the profit-maximizing monopoly price of A by

25Firm 1, we can always construct a bundled discount which


1raises customer welfare and also raises profits for Firm

2 1.

3Now, this has an interesting effect here. There

4is implicit in this the foreclosure result.

5Since the A and B customers are better off

6taking the bundle, even at a price of B that is slightly

7above marginal cost, this means that a B-only seller,

8one of those perfect competitors, can't appeal to these

9folks without charging below-market costs. Equally

10efficient providers are foreclosed and, yet, consumer

11welfare has gone up.

12Clearly I have contrived this example to make a

13point. But it is a point that I suspect in practice

14comes up often enough to make it interesting.

15It at least points out when we are talking about

16bundled discounts, we should not equate foreclosing

17equally efficient firms with lowering consumer welfare.

18In my example, consumer welfare is higher. The

19single-line producers of B would just sell to B-only


21Okay. Another point that is implicit to what

22Barry said which I should have mentioned a moment ago,

23we are going to assume here that under independent

24pricing, the Firm 1, the monopolist in the market for A,

25has not been able to extract all consumer surplus in the


1market for A.

2In principle, the firm might do this by a

3 perfect two-part tariff, for example. In practice, I

4think neither Barry nor I think this is a big deal. If

5it were, we would see lots more two-part tariffs with a

6lot fewer loyalty discounts than we do.

7If consumers' demands have some uncertainty and

8consumers know more about what their demands are, then

9you will not have a two-part tariff anyway, and you

10would find that would still be of some use.

11So far we are talking again about a monopolist

12in the market for A and everyone inside is a perfect

13competitor in B.

14Had I taken more time on this particular slide,

15I would have had a third bullet point which contrasts

16what you might expect Firm 1 to actually do with the

17possibility of raising both consumer welfare and


19In practice, you wouldn't expect the firm to be

20interested in raising consumer welfare. So profit

21maximizing in a very simple setting where B is perfectly

22competitive and products are not differentiated and the

23only thing consumers care about is price, in that

24setting profit-maximizing behavior by Firm 1 is to raise

25the out-of-bundled price of A a great deal.


1The only point of the out-of-bundled price is to

2essentially stampede consumers into buying product. In

3fact, you would give them very bad out-of-bundle

4alternatives, $10 trillion an ounce or whatever it might


6Of course, they could buy B at marginal cost

7from competitors. This puts consumers of A and B in a

8much worse position.

9So in that setting, the effect of

10profit-maximizing bundling would not be to raise

11consumer welfare. It would be to increase profits and

12lower consumer welfare.

13Let's not lose sight of the fact that if the

14out-of-bundled price of A is no higher than the

15preexisting monopoly price of A under independent

16pricing, we have the result which has the interesting

17effect, as I said a moment ago, of excluding sellers in

18B market from selling to consumers that buy A and B.

19What I will do next is to change the story and

20the market for B a little bit. What I talked about so

21far I suspect people in the audience have heard before

22from me, from what I have heard. It is on the paper on

23SSRN for a while. My coauthors and I have labored to

24extend the results and have had some progress.

25The story I will tell next, suppose that the


1market for B is not perfectly competitive. It has two

2firms, one of which is Firm 1. They produce

3differentiated products.

4So yes, there are substitutes but not perfect

5substitutes. Consumers have tastes which are some will

6prefirm firm 2's version of B and some prefer Firm 1's

7version. You have a distribution of tastes in the

8market for B.

9Some consumers want only B, but there is also a

10population of A and B consumers. If you look at those

11folks, the ones who want A, we will assume the same

12distribution of taste as regard to B. So there are some

13A/B consumers who really like Firm 1's flavor of B but

14some who really like Firm 2's.

15In this setting, the world changes a fair

16amount. Let me talk you through things before I go to

17the bullet point here.

18Firm 1 now has a much more interesting role for

19the out-of-bundled price of A than it had a moment ago

20when I assumed that the B market was perfectly

21competitive and all sellers in B produced a homogeneous


23In this case, Firm 1 realizes there are folks

24out there wanting to buy my monopoly product which

25really want to buy B from the other guy.


1The tools at my disposal if I'm Firm 1 are I

2will have out-of-bundled prices for A and B and bundled

3prices for A and B.

4I also know there are some A consumers who also

5prefer my version of B. How hard do I want to try to

6retain consumers that want to buy A but really want to

7buy firm 2's version of B?

8If I am going to keep those folks, I might have

9to really discount the price of the bundle a lot. If I

10do that, then I'm passing up profits that I could make

11on A and B consumers that like my version of B.

12So maybe I won't do it. Maybe it is better not

13to try so hard. I will simply concede A/B consumers

14that prefer firm 2's version of B to Firm 2.

15Now, I still would like to make some money off

16them. I would like them to continue to buy A from me.

17So my out-of-bundled price for A in this

18setting, although it is a high price, is no longer set

19at some infinite level that is designed solely to

20stampede people into buying the bundle. It is low

21enough so that A/B consumers that like Firm 2's version

22of B are still going to buy some A product.

23So the stand-alone price of A, the out-of-bundle

24price of A in this setting has a price discrimination

25goal as well as incentive to buy the bundle. It is a


1more complicated world.

2Now, look at the first bundle here. Compared to

3independent pricing, consumer welfare can go up or down

4assuming that Firm 2 does not exit the B market. In

5this bullet, when I say consumer welfare can go up or

6down, I mean in the aggregate. I don't necessarily mean

7every single consumer.

8Now, Firm 1 -- why would that work? Firm 1 --

9there's sort of an interesting effect here. Firm 2 has

10a tougher job with bundling than under independent

11pricing because it has to convince consumers to buy B

12from it at the expense of them having to pay a higher

13price for A.

14Under independent pricing, it didn't have this

15problem. In this set-up here, Firm 2 lowers its price

16of B because it is now competing, trying to pull people

17out of the bundle from Firm 1, which it didn't have to

18do under independent pricing.

19Firm 1's best response to that is to set an a la

20carte price for B which is lower as well so B-only

21consumers are better off in this setting here.

22If you look at the people buying the bundle, it

23is not clear whether they are individually better off or

24not. Usually some are worse off.

25In that setting, B-only consumers are always


1better off. A and B consumers may be, may not be.

2Aggregate consumer welfare can go up or down.

3There is an interesting permutation of this for

4either entry deterrents, if that's how you want to think

5about this, or driving firms to another market.

6Since Firm 2 always sets a lower price of B

7because it has to work harder to capture consumers

8because they will be tempted to buy B to get a lower

9price of A, it always charges a lower price, its cash

10flow is lower. Depending on the costs it may have, it

11may in fact exit the market.

12Look at this from another way. Imagine that

13Firm 2 has not yet entered the market but it is thinking

14about doing that and asking itself what would happen if

15I did enter the market.

16Well, the story I have gone through here depends

17on a result which is in the paper that Firm 1's best

18response to entry by Firm 2 is always to bundle.

19Firm 2, if it hasn't entered yet, knows if it

20does, Firm 1 will respond by bundling. Therefore, if

21there is some cost of entry specific to the active entry

22that Firm 2 had to incur, they may be deterred from

23entering, somewhat like the one in the tying literature,

24the paper by Mike Winston.

25But there is a difference. You recall that


1Mike's entry deterrence result depends on the equivalent

2of Firm 1 giving a precommitment to a time, meaning if

3an entry were to occur, it has to precommit to the tie.

4There is no precommitment requirement because

5bundling is what Firm 1 will want to do anyway, the best

6response. So it is possible to induce Firm 2 to exit

7even without the precommitment assumption of Winston and

8others in the tying literature.

9For a long time in this more complicated set-up,

10I didn't think we were going to get any sort of fact

11pattern that would tell us we had a safe harbor here the

12way we did in the previous story that I just told.

13My coauthor, David Wrightman, actually came up

14with one. A sufficient condition in this set-up for

15consumer welfare to be higher under bundling than under

16independent pricing, assuming Firm 2 does not exit, is

17the following.

18If the a la carte price of A or the

19out-of-bundled price of A is no higher than it would be

20under independent pricing and if Firm 2's price for B

21falls, then whatever happens with the bundle, we can

22infer consumer welfare has to have gone up, even though

23the price of B in the bundle may be a little higher.

24So if we have this fact pattern, we can conclude

25not only that overall consumer welfare is higher but in


1fact every single consumer is better off.

2A couple of remarks here. I talked about two

3kinds of safe harbor tests here. The previous result or

4model were the B markets perfectly competitive, and we

5compare the a la carte price of A under bundling to the

6monopoly price of A and we have a result.

7And in this case we do the same thing. We can

8only do that if there is a preexisting independent

9pricing regime followed by an onset of bundling.

10And in practice you may not find such a clean

11set-up. Perhaps bundling began in 1932 or something

12like that. However, in a litigation setting, the

13chances are reasonably good that you will run up against

14this set-up.

15Typically what happens is firms compete, and one

16of them will start bundling, and then there is an

17antitrust complaint. Typically there is a before and

18after if things make it to the litigation stage.

19Let me contrast this with the doability of

20Barry's test. Barry's test does not have the problem of

21needing to find a before and after situation.

22It basically lists as attributes the discounts

23to the competitive line and asks if an equally efficient

24competitor could undercut that. We could use that in

25principal using data from the firms if we didn't have


1any reason to think that was a strange point in time to


3The advantage of -- Barry has a safe harbor, and

4it's really oriented towards saying when do we exclude

5competitors. That doesn't necessarily mean consumer

6welfare is lower if in fact this test has failed.

7Okay. To sum up, then, in the right

8circumstances, at least, it seems possible that simply

9by looking at pricing patterns in order to prepare

10out-of-bundled prices so the monopoly must carry them to

11the preexisting prices or by allocating discounts to the

12competitive line, we do have some safe harbor tests at

13this point, most of which weren't around or at least

14weren't understood by us at the time of LePage's.

15Okay, thank you.


17MR. DEGRABA: Our last presenter for the morning

18is Joe Kattan.

19He has asked that I waive the reading of his bio

20and simply say he is a partner at Gibson, Dunn &

21Crutcher in Washington, D.C.

22MR. KATTAN: Thank you. I will also waive the

23use of PowerPoint, the pervasiveness of which may be

24testament to the bundling of the Microsoft Office Suite.

25I'm a lawyer, too, and I'm going to look at


1things from the perspective of the lawyer. And I want

2to start out with a fairly obvious proposition which is

3that both bundling and loyalty discounts involve price


5This is an area which U.S. law has tread very

6carefully, and for a good reason. The cost of error in

7this area, as we all know, is deterring firms from

8engaging in aggressive price cutting, which the courts

9have been loathe to do, viewing such deterrents as

10antithetical to the goals of antitrust.

11Justice Breyer back when he was in the First

12Circuit captured this idea in the Barry Wright case,

13where he said, "The consequence of a mistake here is not

14to force a firm to undergo a legitimate business

15activity that it wishes to pursue. Rather, it is to

16penalize appropriate competitive price cuts, perhaps the

17most desirable activity from an antitrust perspective

18that can take place in a concentrated industry where

19prices typically exceed costs."

20This has been the foundation of U.S. antitrust

21policy in the price arena. We have obviously seen that

22in a number of Supreme Court cases in the predatory

23pricing area, where the Supreme Court said that cutting

24prices to increase business is the very essence of



1Although this policy has its underpinnings in

2the predatory pricing area, at least from a narrow legal

3perspective, it is important to note that the Supreme

4Court has made it clear that this policy has a broader


6This point was made in the Arco versus USA

7Petroleum case which involved, as you all know, maximum

8RPM, where the court said in the context of pricing

9practices, only predatory pricing has the requisite

10anticompetitive effect.

11The reason for that it said was that low prices

12benefit consumers regardless of how those prices are

13set, and so long as they are above predatory levels,

14they do not threaten competition.

15We have adhered to this principle regardless of

16the type of antitrust claim involved.

17So at least from the legal perspective, we have

18to start from the standpoint that discounting practices,

19regardless of their form, can only violate the law when

20they result in some form of predatory pricing.

21Obviously there are economic models that attempt

22to show how various discounting practices can harm

23consumers, sometimes even when price exceeds cost.

24But the courts have stubbornly clung to this

25bright-line standard. And the reason for that is that


1courts have been loathe to sacrifice the immediate

2benefits of a price cut for the much more speculative

3possibility that some future harm to competition might

4be avoided if we curb the ability of firms to discount,

5at least in the absence of more tractable and, more

6importantly, more general economic models that can

7predict harm.

8Essentially what the courts have said, we like

9the bird in the hand, the immediate price cut, much more

10than the birds in the bush, which is the possibility

11that at some point down the line we may have a more

12competitive market and lower prices.

13This obviously embodies assumptions about the

14efficiency of progressive price cutting and about the

15cost of false positives.

16Regarding efficiencies, the courts are assuming

17that price cuts that remain above cost enhanced both

18consumer and total welfare. And with regard to the cost

19of false positives, what the courts are saying is we are

20worried very much about inhibiting price cutting that we

21view is the essence of competition.

22There are clearly several worries that the

23court's fixation with false positives has made them

24insensitive to the possibility of false negatives.

25The basic critique is that anticompetitive


1pricing conduct involving mixed bundling, involving

2loyalty, is likely to be more pervasive and more

3permanent than predatory pricing, so that the risk of

4underdeterrents is greater than in the predatory pricing

5area where this policy has its roots.

6The reason for this, and I think we have heard

7some of it today, is that while predatory pricing often

8requires a large profit sacrifice, uncertain possibility

9of recoupment, which leads to the predation approach not

10being tried very often, anticompetitive bundling or

11loyalty rebates could -- and I want to underscore

12"could" -- entail in profit sacrifice or alternatively

13enable instant recoupment.

14For that reason they are more likely to occur.

15For that reason, they are also more likely to be

16durable, which is to say it can go on for a long time.

17To me, the absence of a profit sacrifice would

18also suggest, at least in the realm of pricing and what

19we are talking about here is pricing -- not talking

20about blowing up a competitor's factory or lying to a

21standard-setting body -- is that an equally efficient

22competitor would be able to match the discounts as a

23general proposition.

24We have heard about some exceptions and that to

25the extent that the rivals are excluded, they are being


1excluded on the basis of superior efficiency.

2In addition, I think we have to take into

3account the pervasiveness of bundling, the pervasiveness

4of discounts that have a retroactive feature, which is

5to say you hit a threshold and the discount applies to

6it for marginal units.

7Volume discounts are fundamentally structured

8that way. Buy a hundred units and you get 10 percent

9off is a fairly common form of doing business.

10The pervasiveness of these types of practices

11throughout the economy, the prevalence of their use by

12firms that don't have market power and have no hope of

13excluding competitors would suggest or at least caution

14that there is a good possibility that the efficiency

15explanation for these practices is the dominant one.

16Now, there are models that show that equally

17efficient competitors can be excluded even without a

18sacrifice of profits.

19But I think the issue with these models is that

20they don't necessarily show consumer welfare being

21reduced by the discount. In fact, I think some of the

22models depend on consumer welfare being enhanced. I

23guess we are really in the dark about this area of the


25Some of these models depend on consumers


1actually being better off with the bundled price

2offered. What we see is that both producer surplus and

3consumer welfare is better off.

4So what this shows is that there are

5circumstances in which bundled pricing can harm

6competitors, even efficient competitors, but they don't

7necessarily harm consumer welfare.

8Another question is whether these models at this

9point are sufficiently general to support changing the

10current legal regime.

11Professor Hovenkamp argues that they are not,

12that they cannot support a legal standard. What he says

13is that the economic modeling showing that certain

14discounts can be anticompetitive tend to be highly

15complex, often making unrealistic assumptions.

16The result is proposed legal standards that make

17impossible informational demands on the courts.

18 A more benign way of looking at this is the

19early models that have questioned the conventional

20thinking and basically challenged the Chicago view have

21worked with stylized assumptions to knock down at least

22the universality of the received wisdom and more work

23needs to be done before we know whether the results can

24be generalized enough to support a rule of law, whether

25basically what we have are some interesting footnotes


1that show that the current legal regime can lead to

2false negatives under some severe assumptions.

3To base a rule of law on the economic models, we

4have to have general models on which we can base clear,

5predictable and administrable rules. Models that show

6that anticompetitive results could happen are not good

7enough to prescribe rules of law.

8What we need are models that identify the

9particular circumstances in which aggressive pricing is

10likely to be anticompetitive and do so in a way that can

11be reliably administered within the constraints of legal


13The challenge is to have rules that capture the

14circumstances in which discounts harm competition, rules

15that do not discourage price cutting and do not serve as

16an instrument for strategic behavior by rivals to attack

17discounting by more efficient competitors, rules that

18offer sufficient guidance to business executives to

19enable them to respond and can be administered within

20the constraints of legality.

21We need rules that are general, sufficiently

22general to have application beyond a narrow range of

23stylized assumptions, do not lead to incidents of false

24positives, are capable of application by business

25executives, capable of implementation with the types of


1evidence that are available to us in the litigation

2setting rather than some idealized laboratory setting.

3The virtue of the cost-based test as a starting

4point or as an initial screen for analyzing pricing

5practices -- and that would include bundling and include

6loyalty discounts -- is that it does all of the above.

7It is highly general in distinguishing between

8discounting to reflect a seller's superior efficiency in

9price cutting that has the potential to drive out an

10equally efficient competitor.

11It avoids false positives by limiting liability

12to cases in which it is economically rational to incur a

13profit sacrifice in the hope of subsequent recoupment,

14following exclusion of a rival from a market.

15It also sets a very understandable guideline for

16business executives, what they need to understand is

17their own cost, the cost of producing the goods that

18they make. They don't need to understand what the cost

19of their rivals are. They don't need to have a more

20detailed understanding regarding the consequences of

21their business conduct on market performance.

22The test is administrable, because determining

23average variable cost, which has been the measure of

24costs used by the courts in most cases, which almost

25always is going to be a good proxy for avoidable cost,


1presents a relatively tractable problem, even though it

2is a fairly complicated one, as anyone who has been

3involved in any kind of cost analysis will tell you. It

4leads to predictable results.

5One cannot overemphasize the importance of

6generality, predictability and consistency. Unclear or

7open-ended rules can have some serious negative effects

8because in and of themselves they can deter firms from

9engaging in discounting.

10In fact, predictability is the reason why the

11predatory pricing test is a test that's grounded in a

12price-cost comparison rather than being a true profit

13sacrifice test.

14A true profit sacrifice test would condemn

15failing to maximize short-term profits. It would

16condemn failing to recover the opportunity cost

17associated with particular pricing behavior. And the

18reason we don't do that is that a rule like that would

19make pricing decisions by firms with large market shares

20basically a roll of the dice.

21So we have a clear rule that omits, I think,

22false negatives but one that is administrable and

23enables firms to base pricing decisions on an objective

24measure that is easy to follow.

25Now, what does all this mean in the context of


1the practices that we are talking about here? What

2would a plaintiff have to show in challenging a

3multiproduct discounting?

4The first thing a plaintiff has to show is it

5cannot offer the multiproduct bundle either on its own

6or in cooperation with other firms. If the plaintiff

7can match the entire bundle alone or cooperatively, the

8bundle is incapable of excluding, at least by virtue of

9being a bundle, other than on the basis of superior


11It can obviously ask whether the price of the

12entire bundle exceeds the cost of the entire bundle.

13But the bundling doesn't give the bundling firm a lever

14over its rival because we don't have the asymmetry of

15the ability of the rival to match a component of the


17To the extent that an equally efficient

18competitor cannot match an offer because consumers are

19better off with a bundle than from a la carte purchases,

20any exclusion that might occur, and we have heard that

21it might occur based on perhaps differentiated demand

22for the products included in the bundle, may show harm

23to a competitor, but they would not show harm to


25 The same principle I think would apply, as


1Professor Hovenkamp has argued, to single-product

2discounts if the price charged by the defendant is above

3 cost.

4That should be the end of the story. If the

5plaintiffs can show that it can't match the bundle

6either alone or cooperatively, then I think the Ortho

7test is probably today the best means that we have for

8identifying whether harm to competition may occur, if

9allocating the discount to the competitive product

10yields an above-cost price that is no more exclusionary

11than having the bundling firm price the competitive

12product on that stand-alone basis.

13If the bundling firm flunks that test, then we

14need to look at a couple of other things. One is has

15the price been extended to a sufficient share of the

16market to result in exclusion, because unlike the

17classic predatory pricing situation, where the predatory

18price is extended across the entire market and every

19unit is sold below cost, mixed bundling discounts may be

20extended on a selective basis.

21Critics of bundling cite this as evidence that

22the strategy is less costly than predatory pricing. By

23the same token, it also tells us the strategy is less

24likely to exclude.

25It is also tempting for plaintiffs to focus on


1what happens at the margin, at some point which is near

2the threshold that triggers the discount.

3To use an example from the Concord Boat case, in

4that case a defendant offered a graduated discount. The

5first level was 1 percent if a consumer bought 60

6percent of its requirements from Brunswick.

7Now, obviously, if you look at this from the

8standpoint of a hypothetical consumer who otherwise

9would buy 59 percent of its requirements from Brunswick,

10then you could say the last 1 percent is being given

11away for free and that is surely below cost. And that

12observation is as irrelevant as it is true.

13To compete for just 10 percent of a buyer's

14requirement, an equally efficient competitor would have

15only had to extend a 6 percent discount to match the

16Brunswick offer in that case. And obviously an equally

17efficient competitor can match the offer to compete for

18the entire amount of the business, in which case its

19discount doesn't need to be any larger than the

20Brunswick discount.

21It has to match the discount dollar for dollar,

22which is a 1 percent discount.

23I think it is also important to take the

24duration of arrangements into account. Supporters of an

25interventionist approach assume that bundling loyalty


1discounts can go on forever because of the absence of a

2profit sacrifice.

3In a litigation setting, this is an empirical

4question. We know in the exclusive dealing arena, the

5law presumes that arrangements of a year or less are

6presumptively legal.

7Where we are talking about an arrangement that

8does not require exclusivity but simply offers an

9economic incentive to buy more from the seller, the same

10presumption should be applicable here.

11Finally, I would say the plaintiff has to show

12real rather than conjectured harm. This means exclusion

13of the plaintiff that results in harm to consumers. We

14have seen in too many cases -- Ortho being one of them,

15but there are many others -- the spectacle of a

16plaintiff that is actually doing well in the

17marketplace, claiming that a rival's pricing practices

18are making it hard for it to compete, that it would have

19done better.

20"I would have done better" shouldn't be the

21basis for a monopolization. The basis for a

22monopolization case ought to be exclusion.

23Again, there are models that attempt to rebut

24the Chicago thesis. Maybe down the road they would call

25for a reevaluation of the law, but I think at this point


1they are too ambiguous in terms of impact on consumer

2welfare and too limited in their assumptions to support

3a change in the legal regime.

4Thank you.


6MR. DEGRABA: We are at the break portion of our

7morning festivities. It looks like we should get back

8here in about 10 minutes.

9We will reconvene. It is 11:00 now. So 11:10.


11MR. DEGRABA: Welcome back. We will continue

12for about 50 more minutes.

13This is the sort of the moderated discussion

14part of our day, where we will present a number of

15propositions up here on the slides.

16These propositions are not meant to represent

17necessarily the views of the Commission. They are

18simply statements made to generate some discussion.

19But before we move to the propositions, I want

20to go through each of the presenters and ask if any of

21the presenters want to respond to anything any of the

22other presenters said.

23So we will start in the same order that we did.

24So we will ask Barry. Is there anything you

25want to say in response to something someone else said?


1PROFESSOR NALEBUFF: I guess I was a little

2disappointed in the lack of response in terms of the

3presentation we had from Mr. Kattan in the sense of I

4think it is manifestly the case that bundled discounts

5do not mean lower prices, and that's actually a

6distinguishing feature of them.

7The example I gave you, historical prices that

8had always been charged for these goods that you saw

9sort of 1, 2 and 3, suddenly the firm says if you don't

10buy all of my goods, now this competitor, I will raise

11them substantially.

12In fact, the evidence would suggest that I'm

13going to raise them above what would be the monopoly


15It is really a case of if you don't give me your

16wallet, I'm going to shoot you. To say that is now good

17for the consumers because they have the absence of being

18shot and calling that a discount strikes me as just a

19perversion of what's really going on here.

20The fact is that when you do predatory pricing,

21you actually buy the stuff cheaply. But when you see a

22bundle, quote, "discount," it is not required that the

23customer actually was ever paying that high price. It

24is only that they weren't being threatened to be charged

25that high price.


1That is a fundamental distinction which has to

2be recognized, and it is one of the reasons why we want

3to treat these things differently.

4It is also, of course, central to David Sibley's

5perspective of when we see the price being raised for

6the single product by the absence of the bundle, that

7really should set off alarm bells.

8In terms of Tom's presentation, just one thing

9to observe is there are times when the purpose of the

10exclusion is not to raise the price in the B market.

11And in particular, there are cases I have seen

12where firms would never be able to raise the price of B,

13would always be competitive, but nonetheless, equally or

14more efficient B firms were excluded because that was

15going to be the platform for them to come into the A

16market and, therefore, attack monopoly, because it would

17be establishing sales force, establishing manufacturing

18in that territory.

19So one has to look not just at the potential of

20the B market to be monopolized but what is the purpose

21of the exclusion here.

22MR. DEGRABA: I will give Joe a chance to

23respond to the statement that bundled discounts don't

24always mean lower prices.

25Joe, do you want to respond to that at all?


1 MR. KATTAN: It is certainly an empirical

2proposition that can be tested, whether bundled prices

3or unbundled prices represent an increase in the price

4vis-a-vis the price levels that prevail in the absence

5of the bundle.

6I have not seen evidence suggesting that that's

7the pervasive way in which we encounter bundled


9Now, is it theoretically possible to set up a

10construct that says here is what a seller can do, it can

11jack up the prices on an unbundled basis and offer a

12discount that simply takes you back to where you would

13have been in the absence of the bundle.

14Yes, that is obviously arithmetically plausible.

15But the question is how common is that empirically and

16whether that is something that in the context of

17litigation also lends itself to the kind of proof that

18we have in the litigation setting, particularly when you

19have changes in quality, performance, product attributes

20that may take place over the same period of time.

21MR. DEGRABA: Thank you.

22And Tom Lambert, do you want to respond to the

23exclusion doesn't always mean an increase in the price

24in the B market?

25PROFESSOR LAMBERT: Sure. May I bundle my



2MR. DEGRABA: Is somebody controlling the volume

3on this thing? Let's continue.

4PROFESSOR LAMBERT: I will bundle my response to

5Barry, along with my comments.

6Joe began by quoting the Barry Wright case, and

7I believe that that is absolutely spot on in this

8context. So I will quote a bit from that.

9Then Judge Breyer writes, "Unlike economics

10laws, an administrative system, the effects of which

11depend on the content of rules and precedents only as

12they are applied by judges and juries in courts and by

13lawyers advising their clients."

14And here is the key part. "Rules that seek to

15embody every economic complexity and qualification may

16well, through the vagaries of administration, prove

17counterproductive, undercutting the very economic ends

18they seek to serve."

19So my response to Barry's two points, one is the

20point that he makes in response to Joe, that you could

21have these phony discounts where you have jacked up the

22price and then said "hey, we are giving you a discount,"

23and he seems to think those are very pervasive, it seems

24to me that most of the bundled discounts I see are not

25like that at all.


1McDonald's, go to McDonald's.

2PROFESSOR NALEBUFF: They are not the ones the

3antitrust cases are about.

4PROFESSOR LAMBERT: Correct, but they are the

5ones that are affected if we adopt a rule of law that is

6designed to deal with .0001 percent of bundling cases.

7That to me is troubling and it's exactly what Judge

8Breyer is talking about in Barry White.

9Sometimes we have to sacrifice the last 1

10percent to protect the 99 percent.

11It also seems to me that we could deal with

12those cases by -- the legal rule I propose presumes the

13legality of above-cost bundled discounts.

14It seems to me that it would be possible -- and

15I think Joe mentioned this in his talk -- to identify a

16phony discount. If you see a price jack and then a

17discount, well, that's not really a discount.

18So we could withhold the presumption of legality

19for that type of bundled discount.

20And then I guess I would make the same point

21with respect to Barry's argument that sometimes bundling

22is done not to charge supercompetitive prices in the B

23market, foreclose competitors and get monopoly power in

24that market but instead to monopolize the A market.

25That very well may be the case, and this is


1incredibly casual empiricism, but my guess is that

2doesn't happen enough to justify writing a law, a rule

3that is so complex that it chills the procompetitive

4bundled discounting.

5I am very much motivated by this idea that we

6have to make these theories workable as laws,

7understandable, administrable by juries and judges and

8useful to counselors who are advising their clients.

9MR. DEGRABA: Thank you.

10David, do you have any comments on anybody's


12PROFESSOR SIBLEY: Yes. This is more in the

13nature of something I meant to say but forgot rather

14than a comment.

15One thing that motivated some of the original

16work that Greenlee and Wrightman and I did on bundled

17discounts was try to figure out as nonlawyers what

18branch of antitrust law was it appropriate to use in

19analyzing these.

20We concluded predatory pricing wasn't the right

21branch. For a while we thought that tying was. The

22reason for that is if the B market is perfectly

23competitive and the A market is a monopoly, the only

24function of the bundle discounts is to set an

25out-of-bundled price of the monopoly product so high


1that consumers are faced with a choice of buying at high

2bundled prices or buying the competitive product and

3buying none of the monopoly product at all.

4That is really equivalent to a tie.

5The version of the model which involved two

6firms offering differentiated products tells us that no,

7there is lots more going on with bundled discounts than

8simply being equivalent to a tie.

9The out-of-bundled price on my third slide

10wasn't in place simply to force people to face the

11option of buying the B market at a competitive price but

12no A or buy the bundle at outrageous prices.

13It actually played a price discrimination role

14and people buy at that price. That suggests to me that

15if there is a general legal theory of bundled discounts

16to be had, it is not predatory pricing and it is not

17always going to be the same as tying either.

18It is going to be something else, and I don't

19know what it is.

20I would like to comment on something that both

21Joe and Tom said about sort of requirements that the

22plaintiff would have to meet in a bundled discount case

23in order to prevail.

24One of them was that the plaintiff would need to

25show that it is impossible, either alone or with someone


1else, to assemble a bundle that would match the bundle

2being offered by the defendant in such a case. I think

3that rule has a lot of appeal to it.

4There is one big caveat. Of the many things we

5have learned about bundled pricing from Barry, one of

6them is that the more symmetrical firms are, the more

7you would expect head-to-head competition to be very


9If in fact a firm were able to match the items

10in another firm's bundle, the result of that might

11simply be low-profit dog-eat-dog competition bundle

12versus bundle.

13Knowing that to be the likely result, the firm

14might not incur the cost of assembling that bundle in

15the first place.

16So it could well be that you could ask a

17plaintiff why haven't you assembled a counterpart to the

18defendant's bundle, and it seems perfectly possible, and

19the plaintiff would say yes, it is perfectly possible,

20it is just it would be unprofitable for me to do it

21because all I'm doing then is bundling myself into a

22price war and it is not worth the cost of doing it.

23So I have a choice, either get into a price war

24which isn't going to make any money or stay where I am

25now and be excluded.


1Although I think it is an appealing rule, you

2could probably anticipate a come-back like that.

3I do want to quibble some with the notion that

4the existing models are simply too complicated or too

5data intensive to say anything useful. Before I do, let

6me make it clear I'm sure there are fact situations in

7which that charge would be true.

8But the test that Barry has advocated which is

9sort of the refinement of the Ortho test, I don't think

10it is any more complicated to implement that test to do

11a lot of things that we do in antitrust cases.

12I don't view that as having any incremental

13complication or data requirements over and above things

14we engage in any way.

15Secondly, if you have a situation that is

16analogous to what Greenlee and Wrightman and I looked

17at, where there is a before and an after independent

18pricing of model discount, really a test for consumer

19welfare would simply be to say are the out-of-bundle

20options that consumers face after the bundle discounts

21are put into effect better or worse or the same as prior

22to the bundling.

23If the out-of-bundled price for A is no higher

24than the previous price of A and if prices in the B

25market are either the same are have gone down, then we


1can infer that whether people take the bundle or not,

2they can't be worse off and they are probably better off

3than they were before.

4That is not a complicated test. It does involve

5some work, sure. It is a lot less complicated than a

6number of things I have had to do in antitrust cases.

7MR. DEGRABA: Thank you, Dave.

8Joe, do you have any comments on anything that

9was said here?

10MR. KATTAN: Certainly.

11I am wondering whether the second model that

12David talked about is one that has antitrust


14It certainly shows that a firm that's facing

15differentiated demand in the second product market can

16raise its profits by bundling not necessarily for the

17purpose of excluding the second firm but simply in order

18to extract more surplus.

19But that usually does not fit within the

20paradigm of antitrust cases. It is simply charging a

21higher price of its consumers.

22PROFESSOR SIBLEY: I would agree that if there

23is no foreclosure effect, I probably wouldn't worry

24about it.

25Some of the -- the model does suggest there can


1very well be, that the institution of bundling by Firm 1

2puts the other firm in a situation where it has to

3charge lower prices than it has before, and this can

4make it viable.

5MR. DEGRABA: Thank you.

6We now move on to the propositions. Can I have

7slide 2 up there, please.

8We have assembled a set of propositions that we

9will read and ask anyone who would like to comment on

10them to comment. Some of them have actually been

11covered at least in some part in some of the talks, but

12we will go through these anyway.

13The first proposition is single-product

14discounts should be per se lawful if the overall price

15for all units exceeds cost.

16Is there anyone that disagrees with that


18PROFESSOR NALEBUFF: And, again, if it turns out

19that somebody could replace another firm 100 percent,

20sure, no problem.

21But the Concord Boat case is a great one where

22in fact there is evidence that they had a monopoly for

23some share of the market based on installed base.

24So it wasn't realistic. Maybe you could get 20

25percent or maybe 50 percent of the market, but you


1weren't going to get 100.

2The chart that I showed you was also from other

3cases like that. Do we think that rivals should always

4have to do 100 percent replacement? That's a pretty

5strong test for a firm that's not yet proven itself.

6It may well be that incumbent buyers want to

7only take 10 percent chance, 10 percent of their supply

8before they decide to go whole hog in this.

9So you have to ask do these discounts that go

10back to volume 1 really provide an opportunity for

11somebody to come in at a reasonable scale or not. If

12the answer is no, then I think we have a problem.

13 MR. DEGRABA: Anyone else?

14PROFESSOR SIBLEY: Let me comment on what Barry


16Barry, if the proposition were rephrased in the

17following way, would you have a problem with it?

18Single-product discounts are lawful if sellers, each

19seller in the market can serve 100 percent of each

20buyer's needs. In other words, each seller can bid for

21all of each consumer's business. Then there shouldn't

22be a problem, should there?

23PROFESSOR NALEBUFF: They may have the capacity

24to do it. It turns out it may well be the buyers are

25unwilling to have a sole source, or they may be


1unwilling -- the incumbent supplier might have a

2monopoly over 40 percent, in which case the fact that I

3could supply 100 percent and undercut them isn't really


5MR. DEGRABA: David, would your proposition be a

6problem if there was a significant amount of product

7differentiation amongst the competitors so that a

8particular consumer may not be willing to switch 100

9percent out of supplier A into supplier B?

10PROFESSOR SIBLEY: Yes, I might have a problem

11then because the buyer then is faced with competition

12really only for all his business and the buyer might not

13want that. Knowing that, one firm might end up with a

14lot of that buyer's business, charging a higher price.

15Really I guess it is only if everything inside

16is as homogeneous as you would like and sellers are all

17perfectly positioned to serve each buyer and there is

18none of the stuff Barry was talking about, then it

19should be okay without further ado.

20MR. MEYER: Is there some increment of the

21volume where it would be relevant to you that the price

22is above cost?

23In other words, if there is an installed base,

24if you leave that out of it and ask about the

25contestable units, would it be relevant if as to those


1units the price is above cost overall?

2PROFESSOR NALEBUFF: You have to ask is there

3some normal, sensible way that a firm could come into


5If you can say the picture I first showed you, a

6firm could only come in for below 5 percent of the

7market in some sensible way, in some area between 30

8percent and 50 percent.

9That didn't strike me as a normal thing. The

10other point is what's wrong with having significant

11discounts on incremental units rather than going back to

1201? You could achieve very similar objectives. We are

13not stopping firms from cutting prices.

14So I think in general, firms aren't going to

15offer negative prices under that scheme, but they may

16offer low prices over longer ranges. That strikes me as

17pro consumer. Since I have another way of achieving

18price discounting that doesn't have that exclusionary

19effect, why not use it?

20MR. KATTAN: I think the same logic would apply

21to simple volume discounts, say if you buy 100 units, I

22will give you 5 percent off. That has exactly the same

23effect as something that may be more individually


25If you look at Concord Boat, you can see that


1the rivals there didn't have to compete for very much

2business in order to -- unless you assume that 6 percent

3discount would have forced them to be below cost. They

4wouldn't have to compete for very much business to be

5able to recover their cost and compete against the


7I think the other issue is how do you determine

8which is the inframarginal business that you are going

9to say this is not contestable, it is sacrosanct, it

10belongs to the monopolist and they are really competing

11only the following units which are incremental units to

12which we would allocate the costs. I think that gets to

13be incredibly complicated because in most cases it is

14not going to be clear what is inframarginal and what's


16PROFESSOR NALEBUFF: The solution is to ask for

17an equally efficient firm, the monopolist itself, what

18units could it compete for. You can see what range it

19 is.

20If it turns out the discount is small enough, as

21it may have been in Concord Boat, so that a large range

22of entry is possible, I'm not worried about it. It is

23an empirical question.

24You can say 1 percent doesn't work, 5 percent

25does, up to 30 percent does, but you can't go beyond


1that. Okay. So show what the range is based on what

2these discounts create.

3PROFESSOR SIBLEY: Let me point out, really

4supportive of Barry's feeling, that this really isn't

5all that undoable. There has been a fairly recent -- I

6know that antitrust or patent was used, but in a case I

7was involved in in which we had discounts like this.

8You can calculate which units have negative

9prices associated with them and what level of entry you

10would need to achieve if you were a new entrant and

11wanted to cover costs. The prices do look bizarre, just

12like Barry's picture, but it is really not that hard.

13PROFESSOR LAMBERT: Can you do it ex ante if you

14are the business planning to give a loyalty discount?

15PROFESSOR SIBLEY: Ex ante, the picture is

16actually somewhat easier. Right? It is one thing to

17think about an incumbent being there and there are a lot

18of reasons why incumbents are hard to unseat. Perhaps

19you are dealing with a loyalty discount scheme, which

20makes it tougher.

21Ex ante, we could all be competing with loyalty

22discount schemes.

23PROFESSOR NALEBUFF: If you can't do it ex ante,

24because it is so damn hard. If you as a seller can't

25figure out what it is, the buyer will probably have


1trouble, the rival would have trouble.

2I would hope that a firm who is setting a price

3could actually figure out what its profits are at

4different levels.

5MR. KATTAN: I think the reason the firm

6offering the discount can't do it ex ante is it doesn't

7know the scale at which the entrant is able to enter,

8which is something known to the entrant, the firm

9offering the discount.

10I think part of what we need to do here is to

11make sure that the test that we apply is one that is

12based on information which is available to the firm

13that's offering the discount and doesn't depend on

14things that are outside of ability to know or control.

15MR. DEGRABA: Thank you. We will move on to the

16next slide, slide number 3.

17We have heard about this a little bit. The

18LePage's decision's vagueness is likely to chill pricing

19behavior that enhances consumer welfare.

20I will ask sort of a two-part question, one to

21the lawyers and one to the economists.

22The one I want to ask the lawyers is so what

23counsel are you giving to your clients, if at all, if

24you have run into this problem or what other counsel

25 have you heard other attorneys giving to their clients?


1For the economists, have there been any other

2empirical studies or any other data in the market that

3might suggest there's a problem? Anyone?


5I think the answer -- the proposition is

6correct. And if you want empirical evidence, it is not

7very rigorous, but go to Google and enter "client alert

8LePage's," and you will end up with pages of client

9alerts from law firms saying "warning, this practice is

10potentially troubling, be very, very careful," blah,

11blah, blah.

12It is likely, I believe, that that means there

13is a chilling effect.

14In terms of counsel to clients, I would say give

15bundled discounts at your own risk, be very, very

16careful before you do it, and you might want to think

17about whether your rivals could compete with those

18discounts, even if your discounted price is above your


20Another piece of advice that I would give is

21something that Barry mentioned, I think -- maybe not --

22and this is based on the Johnson & Johnson versus

23Applied Medical case.

24PROFESSOR NALEBUFF: I mentioned it.

25PROFESSOR LAMBERT: I knew you did but I wasn't


1sure if it was in private conversation.

2In the Johnson & Johnson versus Applied Medical

3case, Johnson & Johnson was giving bundled discounts

4primarily to compete against an equally diversified

5rival, U.S. Surgical, a small, less diversified rival,

6Applied Medical, and several others who couldn't compete

7because they didn't sell the full product line.

8And after receiving complaints, Johnson &

9Johnson responded by carving out the purchases of those

10smaller rivals. And nonetheless, Johnson & Johnson got

11sued by Applied Medical even after this act of


13The judge granted summary judgment in favor of

14Johnson & Johnson on the basis of those carve-out

15purchases. So this does seem to be a way that a company

16can protect itself.

17MR. KATTAN: I certainly agree that the

18vagueness of LePage's is problematic.

19I think one of the things that is not clear from

20LePage's -- and when I have talked to people who have

21been associated with the case, I have gotten different

22answers on this -- is whether 3M simply showed that the

23price of the total bundle exceeded cost or whether it

24actually passed the Ortho test.

25People who have studied the record tell me that


1they don't think that 3M passed the Ortho test. That

2still creates a problem for us because the burden of

3proof normally would be allocated to the plaintiff to

4show that the Ortho test wasn't met.

5But to the extent that this is a case where the

6discounts allocated to the competitive products resulted

7in a below-cost price, it may be less exceptional than

8we think it is and we just need to wait and see how the

9law develops in this area.

10MR. DEGRABA: Joe, could you articulate what the

11Ortho test is for anybody in the audience who doesn't


13MR. KATTAN: It says that we will allocate the

14discount in a multiproduct bundle discount to the

15competitive product.

16So the example would have been the shampoo and

17conditioner example that I think Tom used in his


19In this case, the question was whether

20allocating the discount to the generic transparent tape

21would have resulted in an above-cost or below-cost

22price. I gather that the record is silent on that


24So it may be simply -- what the case may come

25down to, then, is really who bears the burden of proof


1in showing whether the Ortho test has been satisfied as

2opposed to some of the broader readings that have been

3given to the case as basically setting a formless and

4vacuous test for exclusion and a Section 29 test.

5MR. DEGRABA: Thank you.

6You want to say something, David?

7PROFESSOR SIBLEY: I don't know whether LePage's

8has had a chilling effect or not because to answer that

9question, I would have to know all the firms that have

10thought about doing loyalty or bundled discounts but

11have chosen not to.

12I will simply observe that there are antitrust

13cases going on now or recently concluded which, if you

14believe the plaintiffs in those cases, involve firms

15that have been engaging in bundled discounts in time

16periods subsequent to LePage's.

17It may have chilled such activities in some

18senses, but it certainly hasn't stopped them.

19MR. DEGRABA: We will move on to the next slide,


21PROFESSOR NALEBUFF: One more comment.

22One should also take the perspective that the

23vast majority of bundled discounts that we see out

24there, whether it be the Happy Meal at McDonald's and

25the like, wouldn't come close to the test we are


1describing here in terms of leading to incremental

2products being below cost.

3Those in my view are just red herrings in terms

4of thinking about the type of bundled prices that you

5would see. They are not affected by LePage's and they

6are just not relevant for any discussions we have here.

7MR. DEGRABA: Okay. Our next proposition is a

8bundled rebate or discount can exclude an equally

9efficient single-product competitor, even if the

10postdiscount price of the bundle as a whole is above


12We have talked about that at some length here.

13There is actually kind of two off-shoot questions I want

14to talk about.

15The first is what happens if we instead of

16looking at existing competitors in the market, what if

17we were to also consider entry deterrents. How would

18entry deterrents be considered in this proposition?

19PROFESSOR NALEBUFF: I have written on that. In

20the paper in the quarterly Journal of Economics, the

21challenge is that same rebate gets to be used in

22multiple dimensions, and, therefore, it makes it less

23profitable for somebody to come in.

24It is also rational that a firm would want to do

25that rebate and do that bundling in the face of



2It is also the case that it limits the potential

3market of a rival to consumers who like that entrant's

4product and don't like the A product and therefore can

5shrink the potential market available to an entrant.

6Bundling is one of the most effective tools to

7prevent entry that we know, I think.

8PROFESSOR SIBLEY: Let me follow along with what

9Barry is saying.

10In the last line of my talk, when I had the two

11firms offering differentiated product, it is not only

12the case that it is possible to exclude an equally

13efficient B competitor in that set-up where the overall

14price of the bundle exceeds the cost of the bundle.

15In fact, the individual prices would all be

16higher than costs as well. In no sense are you pricing

17below cost, and, yet, you can exclude an equally

18efficient competitor.

19MR. KATTAN: I think the question is whether in

20these models that show that an equally efficient

21competitor can be excluded, consumers are better off or

22worse off.

23At least as I read the exclusionary bundling

24paper by Professor Nalebuff, in one of the examples he

25gave with the A and B products with one of his


1propositions today actually showed that both consumer

2welfare and producer surplus go up.

3So total welfare goes up, consumer welfare goes

4up, and yet an equally efficient competitor gets


6And the question is do we want an antitrust

7policy that says that we are going to punish firms for

8conduct that actually raises consumer welfare.

9PROFESSOR SIBLEY: I think we are not to

10punish firms for conduct that raises consumer welfare.

11The sort of policies at least the economists at

12the table have been talking about are not policies which

13are finely designed enough so they attempt to root out

14consumer welfare reducing activities, consumer welfare

15increasing activities, but simply to construct safe

16harbors. That is, if the following is true, then

17consumers are not harmed.

18It does not mean that if the following is not

19true, they are benefitted necessarily or are harmed.

20But at least the safe harbor test we have been

21talking about I think are on sound ground there.

22By the way, you asked earlier about examples of

23what Barry was talking about in terms of bundled

24discounts just involving sort of fictitious discounts.

25Barry went through a pharmaceutical example.


1Who are the firms in that one, the Keflin and the rest

2 of it?

3PROFESSOR NALEBUFF: That was SmithKline and


5PROFESSOR SIBLEY: That's a case where the

6discount was in some sense fictitious.

7What the defendant did there was to raise the

8out of bundled price 3 percent and give them a 3 percent

9discount on the bundle. That was pretty close to what

10you have heard us talking about.

11MR. KATTAN: My recollection is that that case

12could be addressed by application of the Ortho test.

13PROFESSOR SIBLEY: It was addressed by the

14application of the Ortho test. The point is simply it

15isn't just a figment of economists' imagination that

16these things could happen. They did in that case.

17MR. KATTAN: What you are citing is a 25-year

18old case. If that's the only example we can come up

19with in 25 years, I'm not persuaded that it is


21The question is, do we need a test that is more

22stringent than the Ortho test, or is the Ortho test

23adequate to address the kind of concerns that have been

24articulated through these models?

25PROFESSOR SIBLEY: It depends on what your


1concerns are. The Ortho test is a test designed to see

2whether a single-line firm can undercut a bundle.

3You can say if you want to call that a test of

4anticompetitiveness, that's what it does. It gives a

5sort of safe harbor.

6The sorts of things that Greenlee Wrightman and

7I talk about were not tests for that, does consumer

8surplus go up or down, when can we be assured it only

9goes up. There are circumstances under which it is an

10easy test to do.

11PROFESSOR NALEBUFF: The one place where Joe and

12I do agree is what I proposed is really a modification

13of the Ortho test.

14There are some parts of the test that are

15missing. For example, it turns out the right time to

16apply the test is not ex post. It is ex ante. It is

17when the consumer is trying to decide who to buy from.

18Therefore, you have to use the anticipated

19volumes, not the ex post volumes, which can often be a

20challenge here. You also have to use the incremental

21cost as opposed to thinking about what I'm selling, the

22bundles or just selling things individually.

23Subject to correcting for what expectation

24should be and how you measure costs, actually I think it

25is the way to go.


1MR. DEGRABA: Let me skip ahead here to slide

2number 7 because it is a related question.

3The proposition here is loyalty discounts,

4either single product or bundles, should never be

5condemned without applying some kind of price-cost test.

6Do you agree or disagree? Or kind of agree?

7PROFESSOR SIBLEY: It kind of depends. If what

8Barry and I call the B market is perfectly competitive

9and the demand for A and B are independent and all that

10sort of thing, then in that case you can say whether

11consumer welfare has gone up or down or stayed the same.

12Simply by comparing the out of bundled price to

13the prebundled price, to the independent pricing level

14of the monopoly good, you don't need to know anything

15about costs.

16MR. DEGRABA: Outside of the nice, clean test on

17prices, is there any other conditions under which you

18would condemn a bundled discount without a price-cost

19test? Is this essentially a price-cost issue?

20PROFESSOR NALEBUFF: I have this general matter

21and issue where my price depends on what it is that you

22buy from other people as a general statement, as opposed

23to the price I charge you depends on what you buy from


25So that to me -- it is of the form I will charge


1you one price if you buy from David and another price if

2you buy from Joe.

3I think that is problematic, as opposed to my

4pricing depends on what you buy from me.

5PROFESSOR SIBLEY: If I was going to use that as

6a test, if I'm a bad guy and I want to charge you more

7if you buy from Fred as opposed to me, can't I always

8mimic an anonymous-looking thing just by appropriate

9choice of quantity discount with grade points which

10happen to exclude Fred?

11PROFESSOR NALEBUFF: You can try and do that.

12It is much more difficult to do it. I didn't claim

13excluding that is going to be perfect. When you do it

14directly, it is problematic and I shouldn't -- we should

15know how.

16MR. MEYER: If we grant you an exception for the

17moment for discounts specifically or rebates

18specifically keyed to purchases from identified

19competitors, leave that off the table, is there some

20kind of price-cost test safe harbor that you would

21acknowledge is appropriate here?

22PROFESSOR NALEBUFF: That first one is basically

23a statement of my price to you depends on the market

24share I get. My market share test is ultimately a test

25that you don't buy from somebody else.


1Those things are very common. They are not

2exceptional out there.

3MR. MEYER: You wouldn't limit your exception to

4specifically identified purchasers? You would say if

5there is anything that is keyed to how much the consumer

6is buying?

7PROFESSOR NALEBUFF: Not volume. Percent.

8Ultimately --

9MR. MEYER: If you have an estimate of the

10customers' total needs, don't you also have an estimate

11of their share?

12PROFESSOR NALEBUFF: I have an estimate. The

13price will depend on what they buy, an absolute amount,

14not punishing them for buying something from another


16MR. KATTAN: If you buy 800 units from me, you

17get a 5 percent discount, that's okay, even if I say to

18him if you buy 600 units from me, you get a 5 percent


20PROFESSOR NALEBUFF: I'm much happier with that

21than saying i will take away your discount if you buy

22anything from David.

23MR. MEYER: Defining this exception to mean

24market share discounts, where do you end up after that?

25PROFESSOR NALEBUFF: I think if you pass the


1modified Ortho test, if you would like, so that the

2incremental price, based on expected volumes and such,

3is above the incremental average variable cost, you are

4fine. And you are using your own cost in doing that

5test. Because you could offer a competing B product by

6itself without any difficulty.

7MR. MEYER: Is there congruence, David, between

8that statement and the situation you were describing of

9the conditions where an increase in the out of bundled

10price for A goes up or doesn't go up?

11PROFESSOR SIBLEY: I guess in some sense. I

12would want to think about that more. Simply saying buy

13from Fred, pay a lot for A.

14PROFESSOR NALEBUFF: I thought the question was

15something else. I thought we all agreed on that safe

16harbor, by the way, in terms of if the incremental price

17for B compared in the A/B bundle story is sufficiently

18high, then it actually isn't below the actual variable

19cost to B, for the firm selling it, I think we all

20believe you are in no danger.

21The question is what about if that test isn't


23PROFESSOR SIBLEY: If it isn't passed, then it

24is hard to tell. It is not a simple test. At least I

25personally don't have anything ready for primetime on



2MR. DEGRABA: Let's go back to slide number 5.

3It says a loyalty discount that allows a competitor to

4operate profitably at some scale can never be harmful to


6 Basically what we want to know here is is the

7sort of antitrust objections to loyalty discounts

8strictly one of driving competitors out of the market or

9can there be serious harm to consumers simply by

10shrinking, if you will, some competitors' output.

11PROFESSOR LAMBERT: I would say, just as a

12factual matter, sure, there can be harm to competitors

13and to consumers by shrinking the rivals' output through

14a discount.

15The problem is, beating a dead horse here, we

16have to come up with a way to write a rule that

17implements that notion, and that requires us to know

18something about minimum efficient scale, which is almost

19impossible to know.

20So while I would concede that it is possible to

21harm rivals and harm consumers by reducing the rival's

22scale by usurping so much business from them with your

23loyalty discount, nonetheless we should have this

24Hovenkamp legality rule if the discounted price is above



1It could be met by equally efficient rivals.

2The discounting practice might actually affect rivals'

3efficiency by diminishing their scale.

4But I can't think as a lawyer of a way to design

5a rule that doesn't have a chilling effect if we are

6having to focus on what is minimum efficient scale and

7what amount of a discount is permissible before you

8usurp so much business that you prevent someone from

9achieving minimum efficient scale. I think that is too

10hard to administer.

11MR. MEYER: What if you instead define the

12defense, which is if the plaintiff is continuing to

13operate profitably in the market for B, even if it is at

14much lower volume than it had or market share than it

15had, then the plaintiff's claim fails?

16PROFESSOR LAMBERT: I would certainly have that

17defense. I would say that if a plaintiff can match the

18discount --

19MR. MEYER: He may not have been able to match

20the discount for all the customers to which it was

21operating but still operating in the market for B is my


23PROFESSOR LAMBERT: Yes, I would give that


25MR. KATTAN: That is exactly what happened in


1the Ortho case.

2PROFESSOR SIBLEY: I think people who know the

3facts of LePage's better than I may tell me I'm all

4wrong here.

5As I recall, LePage's didn't claim it was going

6out of business. It just had a lower market share and

7it wasn't making as much money as it was before.

8If this rule were applied to LePage's, I guess

9it would have been over in favor of 3M. Let me

10speculate as well. I don't know if this is true, and I

11haven't thought about it before this second.

12Even if we accept that a rival can only compete

13profitably for a subset of consumers, maybe based on

14some peculiar behavior scale of economies, something

15like that, nonetheless, if the other firm, the one that

16is not the rival in this case is pricing some other set

17of consumers very high, it may be possible for the

18rival, even though it can't serve the entire set of

19consumers, to sort of skip around between subsets that

20it does in fact serve and keep prices down that way.

21PROFESSOR NALEBUFF: It seems to me that this

22can still be a problem. And actually we saw a recent

23case against Briggs and Stratton here in the lawn mower

24industry, where some of the rivals were making some

25money and others were actually losing so much that they


1were exiting the business.

2The view was that if the type of loyalty

3payments had been different, those rivals would have

4been at 20 percent of the market, they would have been

5at 50 percent and the competition would have been much

6more vigorous in that industry, that the customers would

7have been able to have a whole collection of different

8companies to buy from, that there would have been a lot

9more innovation going on here.

10So, if you are able to keep your rivals at 10

11and 15 percent, they may choose not to invest in this

12business, not to try and expand it. And I think there

13can be tremendous harm in the long run here.

14MR. DEGRABA: Anyone else?

15Okay. Thank you.

16I have time for one more before we break for

17lunch. We will move to slide 8, which reads "In a

18loyalty discount case, intent is relevant to proving


20Do you agree or disagree? That comes from

21LePage's, by the way.

22MR. KATTAN: The question is intent to do what?

23Every firm intends to take business away from its

24rivals. When I discount, I'm hoping that by offering

25the discount, I'm going to get more business for myself


1and that my rivals are going to get less business.

2That intent certainly shouldn't have any bearing

3on the outcome of the case. You can assume that it did

4in every case.

5MR. MEYER: What if the intent were the converse

6or the flip side of that, which is intent to achieve

7some business justification, if you will -- I'm not

8interested in what those might be -- evidence that there

9wasn't a desire to exclude rivals or that that wasn't

10the dominant driving factor in the business's behavior?

11MR. KATTAN: I think that presents a more

12complicated question. I think if you have a test that

13focuses on objective factors, did I price below or above

14cost, if I priced below cost, did that exclude

15competitors, that you probably would not need to go

16through things like that.

17PROFESSOR NALEBUFF: I think this actually gets

18to some of what Tom was asking about, which is is this

19market ultimately monopolizeable, and I would extend

20that to is there something else that you could achieve,

21maybe not monopolize B but A.

22It is harder to understand why firms would be

23engaged in this type of exclusion if there was no

24ultimate benefit for them. I also share the view that

25trying to either look for evidence of intent one way or


1the other is sufficiently manipulable or hideable that

2I'm worried about playing that game.

3You would have the advantage the first time it

4is being done in that people aren't aware of it. So you

5can have a lot of bad evidence.

6And, of course, people say things that they

7don't really mean in ways when they get into court that

8can often not sound as good as sometimes they really did

9mean it to.

10MR. DEGRABA: Okay. Given that it is 12:00, I

11will thank the panelists for all of their insight.


13MR. DEGRABA: We will reconvene at 1:30 after a

14tasty lunch.

15(Whereupon, at 12:00 p.m., the hearing was

16recessed, to be reconvened at 1:30 p.m. this same day.)











1AFTERNOON SESSION          (1:30 p.m.)

2MR. MEYER: Let's get started.

3Welcome to the second of today's sessions on

4loyalty discounts. My name is David Meyer. I'm the

5Deputy Assistant Attorney General at the Antitrust

6Division. I will be monitoring this afternoon's session

7with the help of Patrick DeGraba, who is at the Bureau

8of Economics at the FTC.

9The Department and the FTC are sponsoring

10jointly this series of public hearings on single-firm

11contracting to help advance the development of the law

12concerning the treatment of unilateral conduct under the

13antitrust laws.

14Transcripts and other materials from prior

15sessions are available on the DOJ and FTC Web sites, and

16in due course, hopefully soon, the transcripts of

17presentations from today's sessions will also be posted.

18Our next hearing will be December 6th -- that's

19next Wednesday -- addressing misleading and deceptive


21Today's session concerns the law and economics

22of loyalty discounts.

23Bundled discounts or rebates involving two or

24more products have been a hot topic in antitrust forums

25for some time, particularly since the LePage's decision


1of several years ago.

2In addition to bundled discounts, today's

3panelists are also addressing single-product loyalty

4discounts, sometimes referred to as first-unit

5discounts, by which a seller provides a discount on all

6units sold once certain targets are reached, not just

7the discount on the incremental units sold above or

8beyond the set targets.

9Our morning panel offered many interesting

10comments and observations about loyalty discounts of

11both sorts, and we look forward to learning more from

12this afternoon's panelists.

13This afternoon's speakers are, starting with Tim

14Muris, a George Mason University Foundation professor of

15law. He is of counsel at O'Melveny & Myers and, as

16perhaps all of you know, a former chairman of the FTC.

17He also has the distinction of having headed

18both the FTC's Bureau of Competition and the FTC's

19Bureau of Consumer Protection.

20PROFESSOR MURIS: Not at the same time.

21MR. MEYER: That may be debatable.

22Our second panelist is Daniel Crane, who is an

23associate professor at law at the Yeshiva University

24Benjamin N. Cardozo School of Law.

25Our third panelist will be Janusz Ordover, who


1is professor of economics at NYU and a former Deputy

2Attorney General in the Antitrust Division.

3And, finally, Will Tom, who is a partner at

4Morgan, Lewis & Bockius and a former deputy director of

5the FTC's Bureau of Competition.

6More detailed bios are available out front. So

7I will not bore you with all of the accomplishments of

8all of these esteemed panelists.

9The organization of the panel will be as

10follows. Each of the four panelists will deliver a

11presentation, approximately 15 to 20 minutes. We will

12then take a short break.

13When we return, we will start with each panelist

14having an opportunity to take a few minutes to respond

15or comment on the presentations made by the other

16panelists, at which point after hopefully only 10 or 12

17minutes, we will turn to a moderated discussion among

18the panelists and with the panelists.

19Unfortunately, I will not be able to take

20comments or questions from the audience. We plan to end

21around 4:00, but if the discussion is lively and

22entertaining, we don't have any necessary hard and fast

23end time.

24The doors will be locked. So don't worry about

25that. Before we start, I need to cover a few


1housekeeping matters.

2First, as a courtesy to everyone and given the

3way the electronic system works here, I would ask you

4all to turn off your cell phones and Blackberries or at

5least turn them off of transmit so they don't cause a

6problem with interference.

7Second, as you may know, restrooms are all the

8way across the hall past the elevators where you came in

9this morning.

10Third, and this is a required safety

11announcement here at the FTC, if the building's alarms

12go off, move calmly and quickly but act in the manner in

13which you are instructed to. You will be exiting

14through the main entrance if necessary. Presumably

15there will be a lot of FTC folks who know what they are

16doing. Just follow them.

17With that, I would like to introduce and welcome

18Tim Muris.

19PROFESSOR MURIS: Thank you very much for the

20very kind introduction.

21In the long time since I left law school -- and

22I think I look younger than my actual age -- I have had

23a lot of jobs and six of them in the federal government.

24With apologies to my many friends at the Antitrust

25Division, four of them were at the FTC Commission, which


1I guess makes me an FTC guy. But I have had a deep

2fondness and respect for both agencies.

3I'm going to talk today a lot about some

4experimental economic work. Let me put it in an overall


6I do want to disclose that I was retained by the

7United States Telecom Association, the views in the

8paper, and I will express views that are my own as well.

9And this slide presents a framework that we all

10know, I believe, which is the basic economic framework

11about not just economics but about what a legal system

12needs to do.

13A legal system needs to be efficient, needs to

14minimize some of the error costs and indirect costs, and

15I believe we all know a lot about both of those. We all

16know about type I, type II, and the direct costs makes

17livings for lots of us.

18The history of Section 2 is one that ought to

19give us -- which I have written and many people have

20written on -- one that ought to give us great pause. It

21has largely been a history of mistakes, not exclusively.

22As someone who launched the most aggressive use

23of Section 2 of any enforcement head since the '70s, I

24hope the pattern and history of mistakes doesn't



1There obviously have been some good cases, I

2believe, along the way. But as Hovenkamp says in the

3second bullet, the scope and meaning of exclusionary

4behavior remains, indeed, very poorly defined.

5There's a few key cases in the bundling world

6and in the broader world. Brooke Group clearly wanted

7to minimize the type I error, recognized the high type I

8costs, rejected the theoretical possibility of harm as a

9sufficient basis for liability and focused on market


11Probably the most important thing about Brooke

12Group is Brooke Group is about having a bright-line test

13that is administrable for judges, juries and parties.

14I think my good friend Greg Warden phrased it

15best, that it is a recognition that we don't want to

16contemplate making mistakes in this area. I don't know

17if I quoted Greg exactly, but I think I paraphrased in

18the spirit.

19Concord Boat is another important decision. It

20doesn't address bundling but single-product market share

21discounts in a manner that is consistent with the varied

22cost approach of Brooke Group.

23The discounts were above cost. They are

24ordinary business practices often used in competitive

25markets. They are not unlawful exclusive dealing.


1The bottom line, it was not a Section 2


3Then we come to LePage's. And whatever one

4thinks of potential problems with bundled discounts, I

5think it is hard to find supporters of the standardless

6LePage's opinion.

7It is an opinion that did not exercise caution,

8a very poorly articulated theory of harm and an

9incomplete record.

10If you believe, which is usually a good thing to

11do, to take the opinion at face value, the jury could

12find a dominant firm liable under Section 2 based on the

13possibility that bundled rebates, regardless of their

14effect on consumer welfare, could exclude an equally

15efficient competitor.

16The point is that when you apply the

17standard-free approach of LePage's to Section 2

18liability, you are going to likely have high error costs

19from false positives.

20We know, and particularly because we live in a

21world where bundles are everywhere, bundles can reduce

22transaction costs in both the purchasing and selling

23side of the market, they can serve as an alternative

24traditional advertising, and they can be used, and a

25very important part of the literature, they can be used


1by companies to give retailers strong incentives to

2promote and sell their products and services, which is

3an efficient and important vertical control function.

4As an example, one can think of bundling. The

5consumers of telecom products and services demand

6bundles. And we live in a world increasingly where the

7competition of the future is between the traditional

8so-called wireline companies and the cable companies

9selling consumers bundled products, video, data, voice,

10and now the cable companies are even offering wireless.

11That's just one of -- we could go forever on

12examples of bundling.

13 Now, the economic literature on exclusionary

14bundling indeed shows that bundling can exclude

15competitors. And from that it is possible that

16anticompetitive harm could exist. They certainly don't

17show that such harm is likely.

18These models contain many restrictive

19assumptions. They don't consider efficiencies from

20bundling, and they have not been tested for robustness

21or empirical application to the real world.

22They simply don't show whether the potential for

23anticompetitive harm outweighs the benefits from


25Now, there have been lots of suggested theories,


1and one of the ones that would be a vast improvement

2over LePage's but I still think has problems is the idea

3of excluding a hypothetically equally efficient


5This in itself focuses by itself -- I believe

6Professor Hovenkamp has supported this but now in a much

7narrower version than he originally did.

8By itself this focuses on harm to competitors,

9not competition, and the bundled discount would exclude

10-- using this test, would exclude a bundled discount

11that could help consumers.

12My basic problem with the test can be summarized

13in one simple sentence, which is all else equal, how can

14a firm that offers you less of what you want be equally

15efficient with a firm that offers you more?

16And I think the government in its 3M brief had a

17sentence alluding to the fact that, indeed, the whole

18concept of equally efficient might be a difficult

19proposition here.

20This is an example. I won't go through the

21arithmetic because I do want to get to the experimental

22economics. This is an example that Professor Hovenkamp

23uses, and it's an example clearly where bundling

24increases consumer welfare.

25Yet, you can see that the alleged equally


1efficient competitor is excluded.

2Let me move to what I in my ways believe is a

3significant contribution to moving the ball forward on

4the issue of bundled discounts, and that is work that

5was done by Vernon Smith and his colleagues at the

6Interdisciplinary Center for Economic Sciences at George

7Mason, where I teach, so-called ICES.

8Vernon is one of the fathers of experimental

9economics, and for that work he received the 2002 Nobel

10Prize in economics.

11Experimental economics uses laboratory subjects

12to test the validity of economic theories.

13One of the many good things about experimental

14economics is in numerous settings, experimental

15economics has been shown to be consistent with the way

16the real world works.

17One of the most interesting things --

18experimental economics is sometimes criticized because

19they use college students. I have watched the

20economists compete with the students. And the college

21students have nothing in mind but making money in these

22experiments, and the economists are often trying to

23think of some theory. And the college students

24invariably kill them.

25And I participated in these experiments myself.


1My colleague at George Mason, Bruce Kobiashi, can

2remember going down many a time for beer money, to pick

3up a little money. But you had to be good at the


5What ICES did was they conducted an experimental

6evaluation of various theories of anticompetitive

7bundling using a baseline case and then variations on

8the case.

9The variations included changes in the

10correlation of reservation values. I don't know if

11Professor Nalebuff and Sidley talked about that this

12morning, but the relationship of the reservation values

13is extremely important in this literature, the existence

14of efficiencies from bundling and the introduction of a

15fringe competitor to the monopolist.

16What you have is an A market with a monopolist.

17100 percent of the literature assumes 100 percent that

18also sells in the B market. And in the experiments, the

19B market was served by up to three sellers of only B.

20The baseline experiment tested cases in which

21bundling by the monopolist was first prohibited and then

22permitted. There are lots of details about this that

23are available in the paper.

24The baseline results showed that despite no

25efficiencies and despite a setting under which you would


1think that bundling could be problematic, consumer

2welfare still increased, although it wasn't

3statistically significant.

4And then when you added efficiencies, that is,

5transaction costs, savings or the assumption of perfect

6correlation and demand, the welfare-increasing effects

7of bundling rose.

8Here are again the details that are in the

9paper, and the slides are available, and obviously we

10have given you a set of the slides.

11What was measured was consumer surplus, total

12surplus and the number of competitors. And then the

13variations were the transaction cost savings, the

14negative correlation.

15 The Stigler paper on black booking, which was

16one of the first papers here, had as opposed to perfect

17positive correlation, perfect negative correlation,

18which was a situation that allowed price


20Under each of these, sometimes it was

21statistically significant and sometimes it wasn't, but

22there were not welfare losses. And only when bundling

23was efficient did statistically significant exclusion

24occur, which is interesting.

25Now, what happened is after these -- I consulted


1throughout this experimental process. We were hoping

2that what we would find is under these very simple

3conditions here, you have anticompetitive effects of

4bundling which these models had predicted, and then we

5would tweak them some and see what happened.

6Since we couldn't find anticompetitive bundling

7effects, what we decided to do was modify the demand

8conditions to make exclusionary bundling more


10The reservation value for the B good was raised

11to greatly exceed the reservation value for the A good

12for a lot of consumers, which meant there was more to

13get out there.

14Under the modified demand conditions, the

15bundled discounts can exclude competitors in the B

16market. And, indeed, welfare did fall, but it was very

17little and it wasn't statistically significant.

18Look at the conditions that had to occur. There

19were extreme assumptions regarding the demand in the B

20market. There was perfect positive correlation between

21the A and B market demand.

22There was no fringe seller in the market, and

23there were entry and exit frictions in the sense that if

24you entered, you had to stay for some periods, and if

25you exited, you had to stay out for some periods.


1And here are the results. You can see there was

2a big drop in the number of competitors. And there was

3actually a drop in surplus, both consumer and total, but

4very small and not statistically significant.

5Now, then the results of the experiments were

6changed to add -- one of the nice things about

7experiments, you can hold a lot equal and make a lot of

8variations, to add a fringe seller, with the fringe

9seller having a small fraction of the capacity of the


11With the fringe seller, the total surplus

12increased, and also they tested the effects of removing

13those entry and exit frictions that I talked about.

14Those alone reversed the negative welfare results that

15were shown in table 2.

16And here in table 3 are the results, showing the

17various effects of what I just described.

18Let me briefly say what a lot of this means, and

19I do hope to do this under my 20 minutes.

20We have the various tests. The hypothetically

21equally efficient competitor test, for reasons I stated

22before, it is overinclusive, and it would condemn

23bundled discounts that increased welfare.

24The de facto tying test requires knowledge of

25the hypothetical monopoly price in the absence of


1bundling, which is a price generally not easily

2ascertained, to say the least.

3If deviations from perfect competition in the B

4market or other alternatives are considered, this

5further complicates the test.

6I don't know what Professor Sidley talked about

7this morning, but in the second paper he did, he showed

8that his results of anticompetitiveness depended on

9perfect competition in the B market.

10And, of course, perfect competition in the B

11market is not -- perfect competition anywhere is not

12something we find much of in the real world.

13Based on the experiments, the conditions under

14which this de facto tying will emerge are very limited.

15And, of course, we already have rules about tying.

16I think that the most appropriate test would be

17a modified Brooke Group test, which would be based on

18the bundled price exceeding the bundled cost. It would

19minimize the cost of false positives, and it would be


21And in that sense, in the absence of evidence

22that the cost of false negatives from anticompetitive

23exclusionary bundling is large, I submit that we should

24use the modified Brooke test.

25So just a few points in conclusion.


1I believe and I think that Barry Wright's case,

2as Professor Crane and others have stated, Professor

3Lambert, who was here this morning -- clearly the

4federal courts have ruled on single-product pricing


6But that is not true for bundled discounts.

7LePage's is standard free and it has spread beyond the

8Third Circuit. Read Pease Health, a Ninth Circuit

9District Court case which is discussed in the paper.

10Given the history of Section 2 and given that we

11are dealing with price application of Section 2 to any

12exclusionary conduct, particularly this one, we should

13follow a cautious approach that is consistent with that

14applied to the single-product pricing in order to

15minimize the sum of error and direct costs.

16And the courts and those subject to potential

17liability would benefit from guidance that reduces the

18uncertainty that LePage's has created and stems the

19general application of the Third Circuit's flawed


21I'm hoping that these hearings, among many other

22good things they will do, will accomplish that result.

23Thank you very much.


25MR. MEYER: Our next speaker is Daniel Crane,


1who, as I said, is an associate professor of law at the

2Cordozo School of Law. He is also counsel at Paul,

3Weiss, Rifkind, Wharton & Garrison.

4PROFESSOR CRANE: It is my pleasure to be here

5today. I'm going to speak primarily about bundled

6discounts, the LePage's issue, and I'm afraid I will

7repeat some of what was said this morning, but hopefully

8to good effect.

9I should disclose I am been involved and

10continue to be involved in some cases as a lawyer in

11which bundled discounts are at issue, though, of course,

12the views I present today are my own.

13I will not use any slides. We can just listen,

14I hope.

15At the outset, let me stress that the rule that

16I'm going to defend as a safe harbor, the Ortho test, or

17perhaps the Ortho test with some modification, is not a

18rule that is perfect.

19It is not a rule that will achieve a state of

20affairs where all discounts that are harmful to

21consumers will be said to be illegal and all that are

22pro-competitive will be lawful.

23But what I think is important as a lawyer, of

24course, now as an economist, is to articulate

25practicable, workable rules for the courts. This is


1particularly important for private litigation.

2At the end of my remarks, I will talk about how

3I would welcome a perhaps different approach in

4injunctive cases brought by the government than in

5private litigation. Because of some features of private

6litigation, I am quite confident that the LePage's test

7is doing far more harm than good.

8Let me sort of set forth four background

9conditions against which I will defend the Ortho test as

10a safe harbor in litigation.

11The first background condition which has been

12addressed already today is that bundled discounting is

13pervasive and has many pro-competitive or competitively

14neutral reasons.

15When a practice is widespread and usually

16competitively neutral or pro-competitive, courts should

17be particularly concerned about condemning instances of

18that conduct without very strong proof of

19anticompetitive consequences because of the dangers of

20false positives.

21Just to give an example about how widespread and

22pervasive bundled discounting is, when I was working on

23my article for the Emory Law Journal about bundled

24discounting, I looked at my e-mail, and lo and behold, I

25got an e-mail from the ABA antitrust section advertising


1"Market Power Handbook" and "Econometrics," two separate

2 books, for a package price which is a 12 percent bundled

3discount off of the retail price of the two


5I ran into Dan Rubinfeld a short while later,

6who is the editor of one of the books, and I promptly

7served on him a complaint for monopolizing the economics

8and antitrust literature market.

9Of course, bundled discounting happens in all

10sorts of places where there can be no suspicion of

11anticompetitive conduct.

12That doesn't mean that cases like that will be

13litigated, but what it shows you is when you have a

14pervasive practice, there are often likely to be good

15explanations for it, which should make us particularly

16reluctant to condemn instances where there might be

17anticompetitive consequences without very strong reasons

18to do so.

19A second background condition is that bundled

20discounts in commercial contexts are often driven by

21buyers rather than sellers. Significantly, many of the

22recent bundled discount cases to be litigated did not

23involve sales to end consumers but to retailers or

24manufacturers acquiring components or other large

25oligopsony or monopsony buyers.


1Large diversified buyers often leverage their

2buying power across multiple product lines to exact a

3bundled discount from the manufacturer. And, of course,

4these buyers have a strong incentive not to demand

5discounts when doing so would weaken competition in the

6markets that supply them.

7So my own view is that bundled discounting,

8often driven by buyers in these cases, the incentives of

9the buyers themselves are much better at controlling

10competition than litigation is.

11This morning Barry Nalebuff said that oftentimes

12in bundled discounting cases, one of the problems with

13bundled discounts is that they sort of conceal what the

14real price is, it is hard to know what the real price

15is, it can cost up to $10,000 for a buyer to really

16compare apples to apples unbundled versus bundled.

17When we are talking about buyers like Wal-Mart

18or GPOs and the medical devices cases, AMD, Intel,

19Broadcomm, Qualcomm, Information Resources versus A.C.

20Nielsen, many of these litigated cases involve very,

21very large sophisticated buyers.

22So even if it is true that sometimes comparing

23apples to apples is difficult, if you look at the cases

24litigated today, it seems to me that is simply not an



1A third background condition is that I believe

2that courts need rules and not merely open-ended

3standards, particularly in sensitive areas of price

4competition and Section 2.

5I have a forthcoming article in the Washington

6and Lee Law Review about the rules and standards debate

7as applied to antitrust law where I argue in favor of

8bright-line rules to immunize defendants from lawsuits

9in cases involving particularly private litigation and

10unilateral practices.

11Committing economic policy to juries in cases

12like LePage's is a really miserable way to run a legal


14A fourth and related background condition is

15that a growing literature -- including my co-panelist,

16Janusz Ordover, has written in this area -- shows that

17firms can strategically misuse antitrust law to prevent

18pro-competitive behavior by their rivals.

19As I will discuss in a few minutes, I believe

20that many of the recent bundled discounting cases

21involve frustrated competitors seeking to deny

22commercial advantage to a more diversified rival, not

23firms that are in any real danger of being excluded from

24the market.

25Well, with these background conditions as


1considerations, let me say a word about why it is not

2sufficient to analogize bundled discounting to tying or

3exclusive dealing, as some cases like LePage's have


5In a tying case the consumer is required to take

6the tied product if he wants the tying product as well.

7In a bundled discount case, the consumer always has the

8choice to buy simply the tying product.

9Of course, it is possible that the discount is

10so large that it would be economically irrational for

11the consumer who wants both products to buy just the

12tying product and then purchase the tied product

13separately from the plaintiff or a smaller, less

14diversified firm.

15But that would only be the case if the plaintiff

16was unable to offer a discount that would make the

17consumer indifferent on whether it accepted the

18defendant's package discount or bought the two items a

19la carte. Of course, that's a question that tying

20analysis lacks the tools to answer.

21Similarly, an exclusive dealing analysis focuses

22on whether the defendant's contractual practices

23foreclose a substantial share of the relevant market to


25But foreclosure is, again, an empty concept


1unless it means that rivals cannot compete for the


3To put it another way, if a rival would be able

4to profitably match the defendant's bundled discount,

5what we have is ordinary price competition and not


7The problem with using exclusive dealing

8analysis to assess bundled discounts is that exclusive

9dealing analysis begins with the assumption that

10whatever contracts are covered by the relevant contracts

11are foreclosed to rivals, a fact which is not even in

12evidence yet in bundled discounting cases.

13So either a tying analogy or a bundled discount

14analysis, without first looking at sort of economics of

15the discount, is really putting the cart before the


17Let me now turn to the rule for bundled

18discounts that I will defend as a safe harbor. Janusz

19tells me he created this on the back of a napkin. So we

20will leave it to him to give you a history of this.

21The rule traces back at least to the Ortho

22decision. It was adopted in some litigated cases,

23including Information Resources versus Dun & Bradstreet,

24a Southern District of New York case, and I believe is

25reflected in Professor Hovenkamp's 2006 supplement to


1the antitrust treatise.

2A plaintiff challenging a seller's bundled

3discounts as predatory must show as a minimum

4requirement that the bundled discounts resulted in at

5least one product in the package being sold at less than

6cost after reallocation of the discounts on other

7products in the package to the predatory product.

8Another way of saying this is that the bundled

9discount is not unlawful unless the effective price of

10the product in the competitive market is below cost,

11taking into account the discounts on noncompetitive

12products that the consumer would forego by buying the

13two products individually instead of in the package.

14In my Emory Law Journal article, I propose a

15number of additional showings the plaintiff would have

16to make. I won't discuss those here, in the interest of


18I want to defend the sort of core concept

19underlying this analysis which is an analogy of two

20predatory pricing, although with some modifications for

21bundled discounts, which is what I think the LePage's

22court says should not be done, we should not analogize

23to predatory pricing. I argue we should, although it

24may take some qualification of the rules.

25The basic argument is based on the Brooke Group


1standard itself. Under Brooke Group, a single-product

2discount is per se lawful unless it results in pricing

3below an appropriate measure of cost.

4I will put aside what the appropriate measure of

5cost should be and assume we know what it is in a

6particular circuit. Let's call it X.

7What that means is that a defendant would have

8an unqualified right to offer a discount on the good Y

9so long as the price continued to exceed X.

10Now, suppose the defendant offers a bundled

11discount on goods Y and Z and that the plaintiff sells

12only Y. The plaintiff cannot offer a discount on Z.

13But by reducing the price of Y, it can make up for the

14discounts that consumers would forego by continuing to

15buy the goods unbundled.

16So long as the effective price for the

17single-product firm is not below X, it is no more

18disadvantaged than if the defendant had offered the same

19above-cost prices through a single-product discount on


21Since a single-product discount resulting in a

22price above X would be per se lawful, a multiproduct

23discount resulting in effective price above X should

24also be lawful per se.

25Let me sort of answer some arguments against


1this logic that I have seen discussed in economics

2literature, actually moreover in litigation briefs,

3which are probably not a good source of economic

4reasoning, but nonetheless, let me discuss a few things.

5One sort of very common argument is that where a

6single-product predation is expensive and risky because

7it involves sustaining losses for an indefinite period

8of time to drive out or discipline rivals, the

9multiproduct predator sustains no losses from the

10bundled discount because it can cross-subsidize the

11discount in the competitive market with discounts off

12the prices in the monopoly product.

13Of course, the diversified firm that uses a

14multiproduct discount to exclude rivals is also

15sacrificing profits with the hopes of long-term


17A discount of one dollar off the monopoly

18product for the purpose of subsidizing the campaign of

19exclusion in the competitive product is economically

20identical to a single-product firm taking a dollar out

21of a bank and subsidizing single-product predation.

22Unless there is rate regulation over the

23products and the bundled discount is somehow being used

24to fool rate regulators, this cross-subsidization story

25doesn't really hold up.


1This morning Professor Nalebuff argued that

2loyalty discounts can create noncost predation by

3threatening to inflate the monopoly price. In his

4model, the monopolist says I'm going to jack up the

5monopoly price of the monopoly product unless you take

6my bundled discount.

7Of course, as Professor Nalebuff recognized,

8this only works if the threat to jack up the monopoly

9price is credible.

10Of course, the reason that the defendant has not

11charged a price higher than the current price, the

12current monopoly price is that any further price

13increase would by definition be unprofitable because

14there would be substitution to other products.

15So the buyer has its own very credible threat,

16which is if you jack up the price even further, I will

17substitute to other products. By definition, the

18profit-maximizing price being charged already is one

19which will become less profitable to the seller if he

20jacks up his price even further.

21Although I'm not saying it could not happen, it

22is certainly the case that the threat I will raise my

23price where the defendant is already charging the

24profit-maximizing price gives rise to another threat by

25the buyer, which is in that case I will substitute to


1something else.

2A second argument that is sometimes made is that

3bundled discounts, unlike single-product below-cost

4pricing, can go on indefinitely. If that is true, it is

5because there is no sacrifice in profits and no need of

6future recoupment.

7That suggests to me that the reason that the

8diversified firm is able to offer the bundled discount

9indefinitely is that there are legitimate business

10reasons for doing so that do not depend on the exclusion

11of competitors.

12Of course, competitors can be excluded from the

13market by any number of strategies, but those should

14typically not be of concern under the antitrust laws if

15they reflect reasons that have legitimate business

16justifications. That is to say, business justifications

17that make the practice profitable, even assuming

18continued competition.

19A third and final criticism I will just touch on

20briefly is sort of really a criticism of the equally

21efficient competitor hypothesis in the Ortho case.

22Under the Ortho formulation, the plaintiff would

23have to show that it is as efficient a producer as the

24defendant in at least the competitive product.

25This has been criticized on the grounds that


1what if the plaintiff is a new entrant, it is trying to

2achieve economies of scale, it is the defendant's very

3practices that prevent it from reaching economies of

4scale, so how can a defendant say that he should have

5been equally efficient in order to sue.

6For present purposes, I don't want to take a

7position on this, although I do think the equally

8efficient competitor hypothesis is probably correct. I

9would be happy with a test that simply required some

10price-revenue comparison, because that would at least,

11different than LePage's, force the focus back on to

12whether the plaintiff really had options or simply

13whether the defendant's discounts were exclusionary in

14the market.

15I think even that would be a substantial

16improvement on sort of the open-ended standardless

17approach of the LePage's case.

18Let me conclude my remarks, then, by arguing for

19the need for bright-line rules in unilateral

20exclusionary practices cases, particularly in private

21actions for damages.

22As I noted at the outset, it is not difficult

23for law professors and lawyers and economists to create

24sort of armchair assumptions about markets that show

25exclusionary practices in various forms.


1Nonetheless, adoption of bright-line rules is

2necessary to prevent strategic misuse of antitrust law

3by rivals.

4Many of the cases brought in recent years are

5not cases brought by very small firms that are on the

6margins of the market. These are cases brought by very

7dominant, large firms with very substantial market

8shares, with a good bit of profitability and a recent

9history of success in the market.

10One has to ask if the exclusion story doesn't

11seem to be strong on its face, what's really going on


13The answer may very well be, although I can't

14prove what any individual plaintiff intends, that there

15is simply an effort being made to prevent more

16diversified firms from using their diversification as a

17competitive tool.

18Now, it is one thing to say that we will commit

19these issues to juries, as LePage's did, but of course

20jury trial is extremely rare.

21The statistics from the U.S. courts

22administrative offices show that there are approximately

23nine civil antitrust cases a year out of about 860 that

24are terminated. So less than 1 percent of all private

25antitrust cases will end up before a jury.


1LePage's is the exceedingly rare case. What

2happens is that all the cases either are dismissed on

3summary judgment or a motion to dismiss or they have to

4settle, they have to settle because defendants simply

5cannot take the risk of going to trial.

6So what happens, then, is unless the courts are

7given good, solid, antitrust rules that can serve as

8screening devices on a motion for summary judgment or a

9motion to dismiss, the only question becomes how big is

10the price tag, the settlement that the defendant has to

11pay to avoid the trial.

12I think this is sort of a culture that LePage's

13has encouraged. Courts often interpret the LePage's

14standard as really a commitment of these issues to

15juries, although we haven't had lots of jury trials. In

16some cases it is because defendants have had to pay to

17get out of them.

18Let me conclude by saying though I think rules

19are necessary as screening devices for private

20litigation, I'm actually more sympathetic to

21experimentation by the government with different

22theories of exclusionary conduct.

23I think one of the unfortunate things that has

24happened in unilateral cases is that these same rules

25that have been designed to protect against abusive


1litigation by private plaintiffs have been applied

2wholesale to government cases, and I think a good

3example of this is the U.S. versus American Airlines

4predatory pricing case, where the court simply used sort

5of off-the-rack rules that were really not sort of

6designed to prevent against abusive private litigation

7but didn't take into account the differences that occur

8when a government sues.

9So while most of my comments have been very

10skeptical about bundled discounting cases and supportive

11of strong rules to weed out these cases early on in the

12litigation, except the most meritorious cases, I do

13think the government, whether FTC or DOJ, as plaintiff

14should be given more latitude.

15Thank you.


17MR. MEYER: Our next speaker is Janusz Ordover,

18who needs no introduction. He is professor of economics

19at NYU, as I mentioned, former deputy Assistant Attorney

20General of the Antitrust Division and a frequent

21participant in antitrust debates of all sorts.

22PROFESSOR ORDOVER: Thank you very much, David,

23for the kind words. As I say, I always need an

24introduction just to keep my name in front of the

25public, like Coca-Cola and Pepsi or Marlboros, maybe.


1My topic today is loyalty rebates. I frequently

2do something which is inappropriate which is create way

3too many slides. I have not deviated from my strategy

4here either.

5What I do have is way too many slides. Plus, I

6have also asked the organizers to post two of my papers

7which deal with the issue of loyalty rebates and which

8are in order to show the economists' schizophrenia are

9quite adversarial to each other.

10I mean that one tries to demonstrate

11circumstances in which loyalty rebates, back to first

12unit type discounts, are potentially anticompetitive,

13and we actually demonstrate how they can be so. And the

14other paper in which the same kind of loyalty rebates

15turn out to be powerfully procompetitive.

16I have spanned the universe of possible

17outcomes. The big challenge is to try to figure out how

18to marry these two approaches. It is at this marriage

19level that the huge challenges to economic modeling are

20likely to come about. I will come back to these papers,

21of course, in a very short minute.

22There is no point to running you through the

23usual introductions, as you have been here this morning,

24many of you.

25We also already talked about what the loyalty


1rebates entail.

2The main point that I wanted to make, other than

3the fact that we don't know yet enough about their

4economic effects in the wide variety of settings, the

5point that I really think is worthwhile keeping in mind

6is we both need more empirics, more technical research,

7and also I think a lot of bright-line rules, because I

8believe strongly that absent bright-line rules, we are

9going to create mischief on both sides.

10Remember, there are two types of errors. We

11always forget that.

12The point that is I think worthwhile is the

13interesting aspect of these loyalty rebates that really

14comes to play as the driver behind the variety of

15outcomes from models of these settings. And really the

16loyalty rebates create complex links in the product

17space between the supplier and the consumer.

18These links could be across volume, across time

19or across products. Because of these links, because of

20these externalities that come about as a result of the

21loyalty rebates or bundled rebates, different

22manifestations, the typical analyses that we have are

23difficult to carry out.

24Normally we do not think very easily in terms of

25mathematical modeling or empirics in which there are


1these kind of versions of almost network-like effects.

2In this case, the network-like effects are much

3more concentrated as between the pair of transactors and

4spilling into the outside world in which the third

5parties are being affected one way or the other by the

6internal contractual arrangement between the seller and

7the buyer.

8It is quite true, as I have seen in litigation

9myself, that often it is the actual buyer that is

10requiring or asking for or demanding the creation of

11those kinds of links.

12Of course, I missed here Professor Einer

13Elhauge, who would have talked about these kinds of

14links extensively in the context of GPO purchasing

15practices, which are being litigated as we speak.

16I also have to fess up that I have an interest

17in the outcome of these litigations.

18In any case, it is the nature of these links

19that creates complexity for economic modeling of the

20sort that I think is illustrated in some kind of

21examples that have been put forth.

22This is the Hovenkamp example that Tim Muris

23already took apart. So I don't want to waste my time on

24that because this example actually proves nothing. It

25proves nothing because it is not embedded in any known


1economic model of anything.

2I second Professor Muris's point that learning

3from stripped down examples is a dangerous thing to do,

4that we really have to rely on the complete and deep

5understanding of the circumstances in which the practice

6takes place and understand fully and well all the

7economic forces that act upon the practice, the

8transacting parties as well as on the third parties and,

9in particular, on consumers ultimately, ultimately


11A simple example that people have often used

12showing that it excludes an equally efficient competitor

13is, okay, so what, is there any problem with exclusion

14of this equally efficient competitor, assuming -- again,

15I agree here with the previous speaker -- that what it

16means to be an equally efficient competitor is subject

17to debate.

18Indeed, some of you may be as old as I am.

19Although -- do I look younger or older than you? We

20will debate that later.

21What I'm trying to say is that when the issue of

22that kind of an equally efficient competitor came out

23way back in the Turner treatise days, in the context of

24multiproduct firms predating against single-product

25firms, Professor Areta said -- I think it in a letter to


1Will Dommel commenting on the Ordover paper efforts --

2saying there is absolutely no reason to give a

3multiproduct firm a leg up in competition against the

4single-product firms because there is no reason why

5should we take into account these deep potential links

6on the demand side or on the supply side or the cost

7side and the cross-elastic side in terms of lowering the

8benchmark price against which the rival ought to


10Now, that was Professor Areta probably now 20 --

11maybe 18, 19 years ago. I think our thinking has deeply


13We do understand a lot of complex relationships

14in terms of the efficiencies that are involved, whether

15it is on the cost side, gains from a multiproduct

16production, whether it is on the savings side, the

17bundling effects from offering a wide variety of

18products in order to minimize efficiency, inefficiency

19of transacting and so on, coupled again with the Barry

20Nalebuff point, which is now very fashionable at MIT and

21at Harvard, when economists talk about so-called

22shrouding. You know what the hell it means, is it some

23kind of religious ceremony? No.

24It involves marketing practices precisely of the

25sort that make it very hard for a consumer to figure out


1what the price is. Is shrouding good or is it bad? We

2don't know. It depends on the model and the facts.

3The facts are the driver of our analysis when

4properly slotted into some well-understood economic


6So what I want to do is to give you a quick run

7through the two papers with Greg, who threatened to be

8here but I don't see him, thank God. Otherwise, he

9would take me to task for misrepresenting our research.

10The research in fact can be misrepresented or

11represented in a variety of ways. It goes back to

12something that happened to me and Steve Salaw in

13connection with our paper on vertical issues where Steve

14viewed that as a theorem, proof that vertical

15relationships could be anticompetitive. I viewed it as

16a proof that circumstances under which vertical

17relationships could be anticompetitive is actually

18difficult to implement.

19We had the same paper, and the two authors

20agreed to stay neutral on the subject. The same paper

21can been seen from a variety of perspectives.

22What it is that Greg Schaefer and I have tried

23to do, and we are hard at work at probably a few more

24versions of these kind of analyses, is to construct

25economic scenarios which I think are plausible as


1opposed to two-by-two examples that do not reside in any

2well known market setting -- other than Ortho -- in

3which these types of loyalty rebates which is what I

4want to talk about or back to first-unit discounts do

5emerge as equilibrium offers.

6Remember that much of the problems we had with

7examples is it is never tested whether or not what is

8happening in the example is an equilibrium or not. If

9it is, what is the gain that underlies the example.

10Greg and I have specified two sets of cases in

11which these kind of loyalty rebates as equilibrium

12offers. One is the one in which -- the first one is the

13one in which exclusionary loyalty rebate does come


15This is a model, a very stripped down model.

16Let me take you quickly through it. The papers are

17posted. Probably incomprehensible for the lawyers in

18the audience, but maybe not.

19The setting is straightforward. It is stripped

20down. We have two competitors, one of which is, quote,

21unquote, "dominant" in the following simple sense, that

22is, it is capable of producing output for the whole

23market. Whereas, the other, the rival, the smaller

24competitor, the entrant can only produce one unit of



1From a social welfare standpoint, ideally we

2would like one unit to be sold by each. And the reason

3for that is there a heterogeneity of preferences. The

4consumers who would like to buy in the marketplace, they

5like the incumbent's product and some other people like

6the challenger's product.

7In an equilibrium, we would like to see people

8being optimally served, which takes me back to the

9question of what do I mean by an equally efficient


11In the model that we have constructed, each firm

12has the same marginal cost of production, but their

13products are not equal. So they are equally efficient

14on the cost side, but they have heterogeneous offerings,

15which is an environment where economics is not entirely

16clear, what do we mean by an equally efficient


18When I am selling A and you are selling B, and

19they are not perfect substitutes for each other, it is a

20bit of a challenge to give a crisp and clear definition.

21In that model, there are some assumptions that

22actually have to be made in order to create a

23circumstance whereby an equilibrium exclusionary offer

24arises, i.e., an offer that denies consumers the ability

25to purchase the product they would like to get.


1Again, this goes back to my misspent youth with

2Shaffer and Salaw. You can argue amongst yourselves --

3I will be happy to chip in later on -- whether the

4conditions that we have specified are necessary, are

5they sufficient and more or less are they realistic,

6because the usefulness of the model stems, at least in

7my view, from modeling circumstances that are not so off

8the wall as to give no guidance to anything. But the

9conditions we have specified I think are of interest.

10For example, in the paper we have the assumption

11that the incumbent can supply all of the market but the

12entrant can supply only at most one unit. Whether it is

13one unit versus two or two versus five is not

14necessarily an issue. But it is the foundation of the

15differential that exists.

16The second assumption that I think is important

17is that the model has two periods involved. Remember I

18told you about the links that are being created through

19these exclusionary, potentially exclusionary offers.

20Here the link is intertemporal. That is,

21through time. And it is that fact in the model that

22actually is another of the key drivers.

23In period two, the buyer becomes locked in to

24the seller or the sellers from whom it purchased in

25period one.


1What does that mean? In a normal economic

2model, it would mean people would beat their brains out

3to get the second period profits and would give them all

4up in the first period.

5Remember the Supreme Court profound economic

6analysis in ITS v. Kodak on that subject.

7So, no restriction on feasible sale of

8contracting. However, the entrant faces a financing

9constraint, a cap on how much it can borrow against its

10potential period two lock-in gains in period one.

11Now the question comes in whether you are the

12believer in the old fashioned finance literature or more

13inclined to the new old fashioned financial literature

14in which the financing constraints are in fact a fact of

15life for a variety of reasons. And I refer you to

16Turro's new brilliant finance textbook.

17The second aspect of this whole thing is that

18the entrant cannot commit to its second period price in

19period one. How realistic is it? I don't know. It

20depends on the setting.

21So there are two key assumptions or three that

22limit the capacity of the entrant. The inability to pay

23for all of the first period battle, either with borrowed

24money or with the second period money, whether these are

25realistic, that depends on the particular circumstance.


1And I believe that there are settings like that in which

2these conditions are likely to be satisfied.

3In such an equilibrium, such as we do have

4equilibria in which the entrant gets no sales, the

5incumbent makes the sales despite the fact that he is

6going after or she is going after or it is going after

7the marginal unit which is less valued to consumers

8being supplied by the incumbent firm versus the


10Now, again, this goes against the grain of the

11Chicago -- perfectly on time -- against the Chicago view

12of life, which is why would anybody pay to gain sales

13against somebody who can offer those same sales more


15And the answer is well, there are these

16intertemporal links. These kind of relationships do

17change the analytics. Moreover, and here is why these

18constraints that we have talked about are key.

19Moreover, how much you have to pay in order to

20steal or to grab or to sell that second unit is clearly

21tied to how much the rival, the entrant can pay to keep

22it for itself.

23If the price is low because the rival can only

24offer the buyer a penny but you would be happy to offer

25two pennies to something that would lead you to three


1cents gain tomorrow, you are going to do that.

2The Chicago view is again somewhat too

3simplistic in terms of the underlying economics. I'm

4not saying as a matter of empirics it is flawed, but as

5a matter of underlying economics, we all know we are not

6Chicagoans anymore. I believe that's the right place to

7be, out of Chicago, leaving Chicago yet again.

8It is a slide. To switch direction completely,

9Shaffer and I with his graduate students have come up

10with another model in which in fact the efficiency of

11the first-unit discount rebate comes out in a very

12stripped down equilibrium as well.

13In that model, there is no competition. There

14is only a supplier that has a monopoly, and he is facing

15two states of the world of which one is the high demand

16and the other one is the low demand.

17As you all know, obviously, from your

18microeconomics textbooks, in such a world the seller

19would like to create incentive to sell as much as

20possible in the high-demand state.

21But the buyer may not want to reveal whether it

22is a high-demand state or not. You have this asymmetry

23of information.

24If there is an asymmetry of information, you

25have to implement some kind of sophisticated pricing,


1which we see everywhere. It is that sophisticated

2pricing that in fact is the explanation why we see these

3kind of schedules in real life.

4Now, in our model, we have a typical

5self-selection equilibrium that comes about. The

6benefit from the loyalty rebate is clearly going to

7accrue to the seller, not to the buyer, because the

8loyalty rebate gives greater power to price


10We don't know whether price discrimination is a

11good thing or a bad thing as a general economic

12proposition. In our model, we would say, look, if the

13driver behind the loyalty rebate is to incentivize the

14downstream, these buyers or a buyer could be either in

15the high state or the low state, and that should be

16enough to stop somebody trying to condemn the particular

17loyalty rebate as being potentially anticompetitive.

18So you can see that is the theorem right here.

19If you compared this diagram relative to the prior

20diagrams, you can see where the difference comes from.

21As I said, the basic insight of that paper is

22that loyalty rebates permit more efficient price

23discrimination than simple two-part tariffs because of

24the nondifferentiability of the outlay schedule of the

25self-selection point chosen by the high-demand buyer.


1Price discrimination is not always welfare

2enhancing, but we don't believe there should be public

3policy prohibitions for reasons to discourage the use of

4loyalty rebates for such purposes.

5Here I will stop, given that you have my slides.

6Here are the references for those of you who are

7interested. Greg will be happy to send you our papers

8if you ask for them.

9Thank you very much. And I hope I was not way

10too confusing.


12MR. MEYER: Thank you very much.

13Our final panelist is Will Tom, a partner at

14Morgan, Lewis & Bockius here in Washington.

15As I mentioned earlier, Will has also been the

16deputy director at the FTC's Bureau of Competition.

17MR. TOM: Thank you very much, David.

18I will make up for Janusz's too many slides by

19having none at all.

20I am also going to free ride on all the previous

21panelists, both this morning's and this afternoon's, by

22assuming that you have heard all their presentations,

23you are now up to speed on all that they have said.

24So I will not rehash any of the previous

25discussions or points that were made, which may lead


1some of you to wonder whether there is anything more

2left to be said after all of the education that you have


4I do think that one of the things that comes out

5pretty clearly in hearing the lawyers and the economists

6and listening for some of the differences between what

7they are doing here is that the lawyers are looking for

8rules that you can apply in real litigation situations

9and a state of imperfect information.

10We have had a lot of talk about the precise

11contours of those rules and what models can guide us in

12formulating what those rules are.

13At least for us simple-minded lawyers, the

14attraction of the incremental revenue versus incremental

15cost or Ortho standard or whatever you want to call it

16is that it is simple enough for us to understand, and it

17can actually provide some guidance. It provides

18guidance on which most lawyers for a fairly wide

19spectrum of so-called Chicago School or post-Chicago

20School adherents can agree on.

21But it obviously provides that guidance only

22when the incremental costs and the incremental revenues

23are known.

24And it seems to me that the interesting problems

25in actually deciding the cases is how the case should be


1decided in the large number of cases where it is not

2known or where that is the very subject of the

3litigation, with the two sides arguing for different

4factual inferences.

5One of the things that struck me in hearing the

6lawyers talk, particularly this morning, is that there

7wasn't a lot of mention of the legal framework and the

8legal doctrines by which these kinds of rules were

9introduced in the first place, particularly in the

10predatory pricing scenario.

11The question for the factfinder in these rule-of

12-reason kinds of cases is simply in a Section 1 kind of

13case, a vertical case where you have a contract and

14therefore you can bring it under Section 1, does the

15anticompetitive harm exceed the procompetitive benefit.

16In the Section 2 case, it is, "was the defendant

17able to apply or maintain monopoly power as a result of

18the conduct or did it dangerously threaten to do so?"

19And the way the rules and the economics come

20into play is in helping the court decide what kinds of

21inferences are permissible from the evidence, or in the

22words of a famous case from way back in the '60s or

23'70s -- I guess I'm showing my age -- if a frog be found

24in the party punch bowl, one can infer the presence of a

25mischievous guest, but not the presence of spontaneous



2That is the role of economics or that has been

3the role of economics. That's how it has guided us in

4the question of what inferences are possible, what

5inferences are reasonable, from the facts that are


7So suppose, to take a hypothetical or a

8paraphrase of a hypothetical that was used within the

9Supreme Court in the last couple days, suppose you had

10board of directors' minutes that said we are adopting

11this practice even though it will be costly, even though

12it is not going to earn us any profits because it will

13cut off our rivals' air supply and ensure we will not

14have serious competition for a generation.

15In the absence of proof by the plaintiff that

16the Ortho test is failed, can defendant get summary

17judgment, or does plaintiff get to a jury, having

18presented that evidence?

19Well, I guess those of us who still remember the

20law school side of the house -- and I realize that all

21of us antitrust lawyers have slowly gravitated over the

22years to being economists that simply haven't studied

23enough to get a degree have to ask, what is the legal

24framework, what is the legal system, how does the law

25control what the role of the district judge is or what


1the role of the jury is, what the role of the Court of

2Appeals is?

3You would think that that case goes to the jury,

4at least unless defendant can prove that this could not

5possibly have caused the acquisition or maintenance of

6monopoly power.

7Now, it may be that people write documents all

8the time, as someone earlier said, that they don't mean

9or they are just deluded, and there may be a defendant

10who can prove that. A lot of the real questions in

11these areas devolve into questions of burden of proof.

12When you get to the question of when is the

13legal system confident enough to take those kinds of

14questions away from the factfinder and to impose rules

15that say this case cannot go to the jury, we will decide

16it as a matter of law that such an outcome is right, you

17are really looking for the kind of confidence that we

18have in the predatory pricing area.

19I think Tim started out his presentation with a

20little bit of a refresher course on decision theory,

21which I think is very apt, that we are all trying to

22minimize the administrative costs plus the costs of

23error, and having sensible administrable rules to do

24that is a very valuable thing to do.

25But at the end of the day, the question is in


1order to have a basis for applying that kind of rule, do

2we have the kind of confidence that the cost of the

3false positives in this kind of setting is going to so

4swamp the cost of the false negatives that we should

5simply say no, this kind of inference is not


7And I think, to borrow again from things that

8I'm sure Tim Muris and others have said, one of the

9virtues of the market is that it tends to be

10self-correcting; whereas, misguided government

11intervention tends not to be self-correcting, but,

12rather, is persistent for a long time.

13I think we should be cautious in this area as

14well in applying per se rules that essentially cut off

15the debate and end up not being self-correcting,

16because, of course, if these instances of loyalty

17rebates are per se lawful, unless plaintiff meets the

18burden of proving something that is very difficult for

19plaintiffs to prove, then those cases will never be

20brought and you will not have the opportunity for the

21development and refinement of those legal rules.

22So I think I am much more comfortable with

23presumptions and with rules of thumb that can be

24overcome in the particular case. And in this

25connection, I am somewhat taken by Dan Crane's


1suggestion that the legal standards might be different

2purely in injunctive cases from the treble damage


4I do think that to a large extent in antitrust

5laws, our view of the substantive legal rules are shaped

6by the institutional setting in which those rules are

7developed, and properly so, because that tells you what

8the cost of the false positives are, at least to some


10In a setting where you don't have treble

11damages, where the relief is purely prospective, you can

12perhaps afford to experiment a little bit more or to be

13somewhat more precise in the way you apply complex legal

14rules or complex economic theories.

15Here I am not going to please the Department of

16Justice representatives or any of my former colleagues

17at the Department of Justice. Because of the different

18institutional settings that apply to those two agencies,

19it is much easier for the Federal Trade Commission to do

20that sort of thing than for the Department of Justice.

21I have long found the portion of the Areeda-

22Turner treatise that talks about applying a lower

23substantive standard in Department of Justice injunctive

24proceedings somewhat problematic because they are

25applying the same statute that is applied in private


1cases, and to some extent the court doesn't have the

2freedom to write different rules for the two different

3sides, unlike Section 5, which is entirely different.

4Indeed, if you go back and look at the

5legislative history of the Federal Trade Commission Act,

6it seems to be one of the very purposes for which the

7Commission is created is to explore some of the cutting

8edges, if you will, of the law and allow this expert

9body to define prospective rules of the game in a way

10that doesn't punish companies for past conduct that they

11did not have reason to believe was unlawful.

12I think given how much there is to talk about

13among the panelists and how late it is in the day, I

14think I will stop there and leave as much time as

15possible for any discussion.

16Thank you.


18MR. MEYER: Thanks very much. I think we will

19take about 10 minutes as our break. It looks like it is

20about quarter to three.

21If we could all be back here in five minutes to

22the hour, that would be great.


24MR. MEYER: We are ready.

25We will start, as I said, with an opportunity


1for each of the panelists to comment on or reply to or

2question the others about their remarks.

3Just to shake things up a bit, I'm going to

4suggest that we alter the order and start with Dan.

5PROFESSOR CRANE: Sure. I guess I would like to

6respond to one thing that Will said, which was the

7hypothetical memo to the board of directors about the

8reasons for a discount and how it could be exclusionary

9of rivals.

10The problem I would have with a legal standard

11that focused on the intent of the defendant is that

12usually it will not be a memo to the board of directors

13but an e-mail to some third-tier manager that has some

14inflammatory war metaphors for its metaphor about

15crushing a competitor. And it won't be just one, it

16will be three or four or five or six of these strung

17together from millions of documents. You will always

18find these in someone's files.

19Although the memo to the board of directors

20might be better evidence, in private litigation, if we

21even raise intent as a consideration, it is those third

22and fourth-tier manager e-mails that will become the

23evidence that get to the people in the jury, even though

24those e-mails really tell us very little about the true

25efficiency consequences of the bundled discount program.


1I agree with Richard Posner that intent evidence

2is evidence of anticompetitive conduct only two people

3who sort of are foolishly taken by sort of aggressive

4language. And juries certainly can be influenced by


6I think in the Brooke Group case, the post-jury

7or post-trial interview showed the jurors had no

8understanding about oligopoly, the average variable cost

9test, but they were highly influenced by Brown and

10Williamson's war documents. To me, that is not a good


12MR. MEYER: Thanks very much.

13Janusz, any thoughts?

14PROFESSOR ORDOVER: I think that we are all

15pretty much in agreement on a lot of aspects of how to

16approach these kind of business practices.

17My only question would be actually to Tim Muris,

18whether or not we really have that much faith in

19experimental economics to create the edifice of a big

20chunk of antitrust laws, what it is that

21well-incentivized graduates, undergraduates or even

22faculty of the law school can do in these games.

23I historically have been rather skeptical of

24experimental economics. In this case, I think my

25skepticism is probably heightened by virtue of the fact


1that the kind of environments in which litigation

2actually takes place, the market settings in which the

3actual litigations take place are very hard, I think, to

4reproduce in the pure experimental setting.

5I'm not saying there is no insight to be gained.

6I'm trying to figure out whether or not this is enough

7to say that we should allow X or that we should disallow

8Z. I would say it is not.

9It may be an interesting angle to look at

10matters through the prism of these experiments. But I

11would hate to have someone go to court and say that

12Professor Vernon Smith, how much I admire his work over

13the years, has shown that the experimental setting with

14three firms, a bunch of graduates, X, Y, and Z cannot

15happen, therefore the case should be dismissed.

16I don't know whether you would go there. But I

17would say that one shouldn't even try to go there.

18That's my strongest reaction.

19As to the Ortho test, of course I find it rather

20attractive. The problem in that setting again, the test

21was somewhat limited as to the broad application,

22because it did involve again a very specific set of


24There was only one buyer, Red Cross, which

25needed a whole panoply and did specify a whole panoply


1of red blood tests that it needed, and it commanded the

2two offerors to give them bundled and unbundled pricing.

3Where the issue arose, where I really fell down

4flat on my face, was because we had no cost data for

5anybody to be able to apply any of these scratchings on

6the napkin that I have generated as a foundation for

7this whole analysis.

8But again -- so now the question does arise

9whether what the court did there and how they looked at

10the allocation of margins and so on would be directly

11translatable into other circumstances.

12So from an intellectual standpoint, the source

13of that test is of course the so-called compensatory

14pricing test that Bobby and I have invented since 1980.

15 I'm sort of asking myself those questions

16because I see the possibility for the application. But

17I also understand the limited setting in which the test

18actually had its traction may not have the kind of

19traction that we would need in other contexts.

20MR. MEYER: Thanks, Janusz.

21When DOJ develops and opens its museum on

22loyalty discounts, it will ask you to donate that


24PROFESSOR ORDOVER: I think it is part of the

25record. Mr. Weinstein, whoever was the lawyer for the


1other side, actually he attached it to my deposition.

2It should be someplace. I think I have it.

3MR. MEYER: We will leave no stone unturned.

4Since you have now posed two questions for Tim,

5one that you just asked him about, experimental

6economics, and the earlier one about age and looks, Tim,

7you can respond now.

8PROFESSOR MURIS: We will leave the second one

9to a market test.

10Let me make four points. You will probably hear

11me either way. The first is just a point I repeatedly

12make to the world, which is that when people say

13Chicago, they are talking Posner and Bork, who don't

14even -- Posner and Bork are the most extremely

15differentiated on mergers. But Posner and Bork had

16certain views that were not the views of a so-called

17Chicago economist.

18I don't consider myself a Chicago economist. I

19like what is called the new institutional economics.

20As a matter of fact, what is called Chicago

21economics before 1960 invented and dismissed as

22empirically irrelevant raising rivals' costs, variable

23proportions as an explanation for why tying is

24anticompetitive and why, RPM could be anticompetitive in

25certain circumstances, all this by 1960.


1Posner and Bork came along, particularly in the

2vertical practices, with what I think was a restrictive

3and extreme view.

4On experimental economics, there are some

5economists who are concerned about experimental

6economics. I think there is enormous validity to

7experimental economics in the sense that basic

8theoretical propositions of economics are verified in

9the lab.

10The beauty of the experiments is that one can

11take Janusz's paper, which I obviously haven't studied

12and deals with a different problem than was modeled in

13the bundling, you could take that paper and you could

14run it in the lab and run various differences and see

15what happened.

16In this world, for better or worse, experimental

17economics is the one-eyed man in the kingdom of the


19We are dealing with almost complete ignorance

20about the empirical effects of bundling. We are taking

21a ubiquitous practice in nonmarket power settings and

22saying in market power settings there are problems here

23based on an extreme set of assumptions.

24My third comment is about bright-line rules,

25which Will was talking about. Even -- I guess this will


1turn out to be a point that Janusz may want to talk

2about as well.

3Even in Brooke Group, which appears to be a very

4bright-line rule, courts are pushing it. Janusz

5testified for Northwest in the Spirit -- is that --

6PROFESSOR ORDOVER: Yes, I did successfully the

7first time around.

8PROFESSOR MURIS: Right, right.

9The point is that the Sixth Circuit, Brooke

10Group or no Brooke Group, was pushing the envelope


12And I think what we ought to do is look at the

13world as we know it, and the world as we know it is a

14world in which LePage's has caused lots of damage. We

15have highly theoretical evidence of problems without

16real world evidence.

17Of course, the experiments were designed to push

18and test and find exclusionary bundling and didn't. But

19someone can go run modifications if they want, which

20leads me to my fourth point, which is the Ortho test.

21The Ortho test, as they say, Parker, is

22obviously much better than the standard LePage's world.

23The problem is that it is so easy for -- one of many

24problems with Ortho besides the fact that it would

25condemn efficient practices is that it is so easy to


1turn the safe harbor into the test.

2I assume Professor Crane would reject that.

3Professor Hovenkamp rejects that. But it would be very

4easy to cross over that line.

5When you are shifting presumptions makes a

6difference. One of the interesting things on the

7Twombley argument Monday was Justice Stephens through

8the course of the argument, it appeared -- who knows

9exactly, obviously -- it occurred to him that if he

10allowed the complaint, from his questioning, if you

11allowed the complaint, there was going to be some fact

12that was going to survive a motion for summary judgment.

13As a practical matter, maybe you did want to

14scream at the complaint level. I wrote an amicus brief,

15along with some other people at O'Melveny, supporting

16the petitioners in Twombley. So I obviously have a dog

17in that hunt.

18Recognizing these real world practical

19considerations, as Will and others have said, is

20absolutely essential.

21 MR. MEYER: Thanks.

22Will, your final opportunity.

23MR. TOM: Let me just respond to Dan Crane's

24last remark by repeating something that Joe Kattan said

25this morning, and that is "intent to do what?"


1I think there is a danger in a lot of these

2areas in using broad classifications to stand for a

3whole bunch of disparate things, and that applies to

4evidence as well as it does to some of the economic

5issues we have been discussing here today.

6I think most courts nowadays confronted only

7with the intent evidence that says "let's crush our

8competitors" would say that that is insufficient

9evidence to go to the factfinder.

10Whether you should then sweep into that every

11other piece of evidence that you find in an internal

12company document I'm highly dubious about.

13MR. MEYER: Okay. With those comments, I think

14we will turn now to the propositions.

15In these hearings, we have been using

16propositions merely as a starting point for discussion

17and not necessarily as a set of propositions that

18reflect the agencies' views either for enforcement or


20If we go to slide 3, we will start with this. I

21think perhaps we might have something like agreement,

22but I will ask.

23The proposition is the LePage's decision's

24vagueness is likely to chill pricing behavior that

25enhances consumer welfare. Agree or disagree?



2MR. MEYER: We all agree?

3PROFESSOR CRANE: I do agree. And from a

4client-counseling perspective, I have been on a number

5of calls in cases where someone is not a defendant but

6simply trying to figure out what they can do and what

7they can't do.

8Without being too specific, for attorney-client

9privilege reasons, if you have any moderate degree of

10risk aversion, you can guess what the answer is.

11It is oftentimes the case that you probably

12wouldn't get sued, but you don't want to be the person

13who gives the advice that we could bring in the smart

14economist and convince the court to dismiss the case on

15summary judgment.

16You tell them you don't want to invite

17litigation at all and it is always better to try to

18unbundle a discount than to face the prospect of


20MR. MEYER: I certainly understand the need to

21mask the specific facts. But are the situations that

22you are describing ones where, at least in the mind of

23the company involved, there is a clear pro-competitive

24motivation or rationale for wanting to structure a

25discount program and they are asking can we do this


1without fear of litigation or is it where they have the

2structure and they are being asked is this going to pose


4PROFESSOR CRANE: It really varies. Even what a

5pro-competitive justification is I'm not always clear


7There are certainly cases where clients are

8asking. Sometimes these bundled discounts are customer

9driven, and large diversified buyers are putting

10pressure on sellers to give them a concession for

11buying, and that is simply responding to pressure from

12the client.

13Sometimes there is a question simply about using

14as a competitive advantage, not to necessarily exclude a

15rival, but because you think you can increase your

16market share through a discount that takes advantage of

17your diversification.

18I think certainly I tell the client if the

19discount is one that looks like it is going to really

20harm the competitor to the point of extinction,

21obviously you shouldn't do it. Even far short of that,

22clients often think about this as a competitive


24PROFESSOR ORDOVER: Would the answer differ in

25the following two settings?


1One, there are bundled discounts but we have

2something called mixed bundling. In other words, you

3offer a bundle, there is a good price, like the one that

4ABA offered for two volumes of writings, but there is

5also a stand-alone price.

6Because of the not total disattractiveness of

7the stand-alone prices, people can avail themselves of

8buying one of the volumes and then buying a substitute

9product somewhere else. But there are recognizable

10efficiencies from bundling.

11Does one get protected under any of these

12LePage's standards or their progeny from the challenge

13if you do indeed offer mixed bundling and you also

14demonstrate that people are buying at the stand-alone


16PROFESSOR MURIS: That's an important point. I

17think the hypothesis of the attack with LePage's is you

18don't have a de facto time. It is calling it mixed


20MR. MEYER: Is the problem with LePage's from

21the perspective of its vagueness and potential to chill

22behavior, which I think we all agree to, is the problem

23the lack of a safe harbor, the lack of a concrete cast

24or the focus on the impact on rivals or something else

25or all of the above?


1PROFESSOR CRANE: I think it is the lack of a

2concrete test. Even though we had the Ortho standard,

3it would be sometimes hard in a client-counseling

4situation to anticipate how that would come out in


6I think most bundled discounts would clearly

7meet the Ortho safe harbor. And it is not even a

8question of most cases.

9What I think that would do is change the culture

10of this issue in the courts, where you could tell a

11client that only in really sort of egregious cases of

12bundled discounting that has a clearly exclusionary

13effect on single-product rivals will a court condemn it.

14That will certainly change your willingness to

15say go ahead and do it.

16MR. MEYER: Let's turn to proposition 6.

17This problem situation is as follows: Because

18lower prices immediately benefit consumers, we should be

19extremely careful not to adopt legal rules that can

20result in false positives, that is, condemn legitimate

21price cutting.

22Do we agree or disagree with that proposition?

23MR. TOM: I think the disagreement here will be

24 more on whether this proposition is one that is relevant

25to the loyalty discount kind of setting rather than


1agreement or disagreement with the proposition itself.

2I think we all agree that in general we like

3lower prices to consumers as long as it is not an

4exercise of monopoly power. You will not get a lot of

5disagreement on that.

6MR. MEYER: Is your question, Will, whether in

7certain situations the loyalty rebates that are being

8offered to particular customers actually result in the

9overall prices paid by them being higher rather than

10lower in the short term, or is this a long-term versus

11short-term problem you are identifying?

12MR. TOM: Even in the short term, there are

13issues of what would the stand-alone prices have been,

14absent allowing it.

15MR. MEYER: Have we seen any cases where the

16prices in the short term were higher?

17MR. TOM: There was one mentioned this morning.

18PROFESSOR CRANE: The SmithKline case this


20The discussion this morning was that in

21SmithKline, the offer was a 3 percent increase

22accompanied by a bundled discount to buying the package,

23which would suggest there was the possibility that even

24in the short run, the defendant was not sacrificing

25profits immediately by taking market share from


1single-product rivals.

2MR. MEYER: Fine. Going back to the beginning

3here and with Will's amendment, if the bundled discount

4or loyalty discount results in lower prices in the short

5term, we all agree that care should be taken to avoid

6chilling such conduct?

7PROFESSOR MURIS: I agree. Let me add, I wasn't

8here this morning, but I assume that Professor Nalebuff

9was probably the most aggressive on behalf of his

10various rules.

11If you look at Tim Brennan's comment on his

12paper, it shows that in the equilibria, consumers are

13better off in the short run virtually all the time.

14That's the nature of excluding, what it means.

15So the theory is really a long-run theory. It is not a

16theory in the model. But that is really the theory.

17And that's I think a very strong reason to agree with

18the proposition that we need to be very careful.

19PROFESSOR ORDOVER: I think to emphasize what

20Tim said, I agree 100 percent. And that is in order to

21close the model of these adverse effects, you really

22have to have the second stage or the third stage and

23when something bad actually does happen from a price

24discount, unless you can show that, you don't have a leg

25to stand on in the rest of the case.


1It could be a complicated set of issues to be

2addressed, these intertemporal linkages, the R&D

3incentives. But if the marketplace is not of the sort

4that it is susceptible to exclusionary conduct over a

5long haul, then I think we should really be very

6protective of price cutting.

7I think where the problem comes in is much of

8the literature on loyalty rebates, as summarized in much

9of Professor Elhauge's writings, actually show this

10concept -- sort of like the rug carpet dealership or the

11vitamin store where you always get 20 percent off. They

12don't say what the benchmark over which you are

13discounting is.

14There is that issue. The equilibria in many of

15these games, the discount is off of what appears to be a

16super-monopoly price, and then really it boils down to

17another point Tim made very importantly earlier today,

18which is to say is that a credible threat for the

19incumbent firm to say if you don't buy it from me, I

20will charge you monopoly price plus 15 percent on top of


22Again, that is a complicated analytical issue,

23whether or not this is a credible threat or not. It

24really much depends on how you view this monopolist

25power to guide the transactions.


1After all, you can say the same thing to a

2monopolist who says, "look, I'm charging you $10 for the

3widget." You say "hell with you, I'm not paying $10."

4He says, "okay, okay, I will charge you 9." Then the

5whole thing begins to unravel.

6Every monopolist issue is that of credibility.

7I think as Carlos pointed out, when the monopolist

8cannot stick credibly to his threat, the monopolist

9competing against himself will drag the price down to

10his marginal cost.

11We have the same question here. How credible is

12the super-monopoly price as a way to enforce an

13equilibrium in which everybody is paying close to

14monopoly price, which is what the outcome is in the

15naked exclusion model.

16That's the story of that basic model which

17Elhauge finds very attractive.

18MR. TOM: For a clarification point, aren't most

19of these models that are based on a super monopoly price

20for the monopolized good and a discounted price for the

21competitive good ones in which commitment is not

22necessary because the purchaser of the bundle does not

23face a price increase? That is, the excess of the

24monopoly price on good A is no greater than the discount

25on the competitive price.


1PROFESSOR ORDOVER: You end up in equilibrium

2with something close to the monopoly price. In the

3naked exclusion model, you end up with an equilibrium

4where everybody is getting a penny off the dollar price.

5That is again supported by what some people may

6consider not credible threats of how the firm will

7behave out of equilibrium.

8That is the same problem in all of these models

9potentially, actually, other than the Ordover-Shaffer

10model in which the equilibrium is supported by credible


12I'm talking about game theory stuff. I don't

13know whether it makes any difference to anybody here.

14If you are trying to be serious about it, you try to

15model it seriously. It is very difficult because it

16does require this credibility.

17MR. TOM: The credibility issue is that you will

18still give the discount on the below marginal cost on

19the competitive product even if he doesn't buy the

20monopoly product?

21PROFESSOR ORDOVER: Right. Or if somebody

22refused to transact with you, that you will not revise

23the market off.

24MR. MEYER: Didn't you mean the other way

25around? If you don't buy the competitive product, the


1monopolist will still charge just the monopoly price,

2not the super-monopoly price.

3PROFESSOR MURIS: Right. A couple points.

4On FTC.gov, you can find in terms of naked

5exclusion the originator of the concept, Michael Winston

6-- unfortunately, this was a workshop we had that turned

7out to be on September 11, 2001, which was a pretty

8crazy day.

9Anyway, the economists, they all stayed and

10talked. And he said he didn't have a clue whether this

11has any empirical significance or not, which I think is

12an honest position.

13In terms of the super-monopoly price, the

14de facto time is the special case here. The reason the

15Nalebuff thing is so important if it had empirical

16significance is it is above-cost exclusion with mixed


18One of the interesting results of the

19experiments that I did talk about is mixed bundling

20still occurs a lot, virtually under every setting.

21Mixed bundling again being where they are selling the

22stand-alone as well as the bundle offering and selling


24And a further point of interest of the

25experiments and in terms of -- remember, when you talk


1about exclusionary behavior, we are all agreeing the

2point is on welfare, not on excluding competitors.

3When ICES tweaked their model to try to really

4push and show that bundling decreased welfare, they did

5show big-time exclusion. But they showed very small

6reductions in welfare, not statistically significant,

7even under very extreme assumptions.

8MR. MEYER: Let's go to proposition number 5, if

9we could.

10PROFESSOR MURIS: I think he is trying to

11confuse us.

12PROFESSOR ORDOVER: Like Lenin, two steps

13forward, one step back.

14MR. MEYER: I think the comments that Janusz and

15Tim made may be a good segue to this proposition, and

16that is a loyalty discount that allows a competitor to

17operate profitably at some scale can never be harmful to


19Anyone want to take that one on?

20PROFESSOR ORDOVER: I think that to use such

21things as "never," even in the proposition --

22MR. MEYER: How about taking it on as usually


24PROFESSOR ORDOVER: I think we just don't know.

25I think if the competitor can operate profitably, then


1there has to be a showing to condemn the practice that

2somehow that scale which it can operate is so

3sufficiently constricted as to render basically the

4competitor, the rival marginally profitable, much less

5constraining of the market outcome than in a less

6constricted equilibrium.

7So the question is that of the benchmark,

8really. The consumers benefit from having competition.

9If the scale is sufficiently large, the

10competitor cannot be profitable and exert competitive

11pressure. Then I would say that's good enough. If the

12competitor is completely marginalized, it is one of the

13few competitors that can exert any kind of competitive

14pressure, I believe that possibly could be a

15circumstance that may require some remedial


17MR. MEYER: Others?

18PROFESSOR MURIS: Let me preface this by saying

19all of my comments reflect this basic framework of the

20efficient legal system.

21George Stigler once wrote a piece where he just

22numbered the comments, the first one, of course, being a

23Chicagoist, this is just a coast theorem.

24Everything I'm saying is in the context of

25efficient legal rules. Here the models that people are


1positing are models of complete exclusion. If we have

2quote partial exclusion, Janusz is absolutely right; I'm

3sure you can conjure up a situation where that is bad.

4As a practical matter we ought to be cautious if

5the exclusion is partial in terms of false positives.

6PROFESSOR ORDOVER: I agree. The European

7Union's white paper, pink paper, whichever color they

8use on anticompetitive conduct has some complicated rule

9dealing with something called the suction test and the

10loyalty rebates.

11I tried to figure out what it means empirically,

12how to apply it. It strikes me as a rather difficult


14If the competitor can operate profitably, the

15burden shifts drastically against the complaining rivals

16to show that something else could have happened but for

17this conduct that truly would benefit welfare.

18I would apply a very strict test to what it is

19that can be shown or should be shown. It was a minimum

20showing in such a circumstance from the competitor.

21MR. TOM: In fact, you are looking for the

22rival's marginal cost to be raised in such a way that

23the perpetrator can raise prices.

24MR. MEYER: In the spirit of jumping around, I

25think we will go to slide 8.


1We have already heard a little bit of debate

2about this in the earlier dialogue, but to state this

3proposition. In a loyalty discount case "intent is

4relevant to proving monopolization." Quoting from


6I will start this, conscious of the prior

7comments, by first asking whether if you have intent --

8maybe we can all agree on this. If the evidence is

9simply that the defendant intended to cause harm to his

10rivals, to drive its rivals out of business, to raise

11their costs, to steal sales from them, is that ever

12enough to get to a jury?

13PROFESSOR CRANE: Just to repeat what I said

14before, part of the problem is I don't know what a

15corporation's intent is.

16A corporation is a fictional person. Will's

17suggestion that we can't simply lump all intent in the

18same category maybe is right in theory.

19When you get to actual litigation, if the legal

20standard is framed as an intent-oriented standard or one

21where intent is a relevant proposition, how do we

22separate out the different kinds of intent and at what

23stage in the litigation?

24Is this a role for the court in summary judgment

25to sort of talk about different kinds of intent and sort


1of sort them out as a screening device?

2I think that would get rather difficult to do.

3Again, it would also create a predictability problem.

4In most cases, the objective economic evidence

5will be available for something like the Ortho test, and

6it really should not be necessary to go to intent.

7MR. MEYER: Any other reactions to the first


9MR. TOM: To your specific question, certainly I

10think everyone would agree on 1 and 3. I'm not sure

11that everyone would agree that a demonstration that your

12plan was to raise your rival's cost would necessarily

13get a free pass.

14PROFESSOR ORDOVER: Competition is about killing

15your rival, really, or diminishing its capability as far

16as you can do that.

17The real question is is it done in a way that is

18conducive to consumer welfare or done in a way that

19harms it for horizons we are comfortable to deal with,

20whether you can sort it out efficiently without running

21into these other problems or without having the current

22rival abusing the system.

23I think one should not view intent as really a

24bunch of nasty e-mails. One should look to intent as a

25manifestation of business practice that has a likelihood


1of harming competition.

2From my perspective, when I teach my kids about

3competition, they say what do you do when you run

4experiments at NYU, yes, you try to vanquish your rival,

5but try to do it in such a way that is conducive to

6welfare. Let them figure out what that means. That's

7the true story.

8MR. TOM: Is that the jury instruction?

9MR. MEYER: What if instead of being evidence of

101 and 3, as the shorthand we will use, and I think what

11that means, if I'm recalling my own comment, is evidence

12of a desire to kill the rival or eliminate the rival or

13steal sales from the rival, instead of that you had

14documents or testimony that constituted a very detailed

15analysis of the reasons why the business wanted to

16engage in this practice of structuring the discounts in

17the way they were structured that had appeared on its

18face to have nothing to do with hurting the rival or

19excluding the rival.

20Would that be probative in some way?

21MR. TOM: It seems like it would be probative of

22an efficiency justification or lack of competitive

23effect, if I'm understanding your question right.

24PROFESSOR ORDOVER: It could be probative of the

25fact that the practice makes sense, that the competitive


1practice irrespective potentially of how it affects the

2competitive marketplace, if that's your question.

3PROFESSOR MURIS: Under Brooke Group, since I'm

4arguing for modified Brooke Group, you have this

5price-cost safe harbor. If you fail that, you need to

6show the entity competitive effect. There are obviously

7places where the law makes intent relevant, including

8Norr Pennington, for example, especially in the

9so-called pattern case because of the nature of the

10First Amendment protection.

11But here I think you need -- this is one of your

12other propositions, if I'm jumping the gun. I think you

13do need price-cost benchmarks to start with.

14PROFESSOR CRANE: Intent is certainly relevant

15in an attempt-to-monopolize case, because intent is a

16specific intent crime and the Supreme Court has made

17clear that intent is relevant.

18But intent in a case like Spectrum Sports would

19only come in in addition to a showing of exclusionary


21As to that element, a legal defense might

22concern LePage's in that it seems to make intent part of

23that element of the offence, which is anticompetitive or

24exclusionary conduct, which seems to suggest that even

25if you have weak evidence, sort of economic evidence of


1exclusionary conduct, that intent can make up for the

2weakness in that showing, which I think should not be


4MR. MEYER: What are your thoughts on this

5question? Can good intent save you even if it turns out

6that you were wrong?

7For example, there are detailed analyses that

8all the prices are going to be above cost, no matter how

9you measure them incrementally or in the aggregate, and

10it turns out there was a math error. How does that case

11come out?

12PROFESSOR CRANE: That's the historical accident

13standard, where you monopolize completely by mistake. I

14will use that on my antitrust exam. It is a

15hypothetical case.

16I don't think the defendant should have a

17defense that we had benign intent. But, of course,

18pro-competitive justifications as the explanation for

19the conduct could be like the defendant's intent. I

20think, of course, that's always permissible.

21PROFESSOR ORDOVER: Especially in certain areas

22of business conduct, for example, R&D, research, it may

23turn out it is more costly than you planned or more

24successful than you thought it was going to be.

25When you have business activities with


1themselves, random outcomes hard to predict, it is key

2that one should not hang somebody for a circumstance

3that is one of the possible many outcomes, most of which

4or at least ex ante believe that you are going to be

5acting in a pro competitive manner.

6If you embark on an R&D program which may cause

7some problems for your competitors but it turns out you

8are now going to be spending 10 percent more, somebody

9said if you knew you were going to spend 10 percent

10more, now you are killing us.

11It is the sort of ex ante nature of the

12calculation that is the right way to look at it.

13MR. MEYER: Janusz, you promised one step back.

14We will go back to slide 7.


16MR. MEYER: Loyalty discounts, either single

17product or bundled, should never be condemned without

18applying some kind of price-cost test.

19Tim, I think you said you agree with that.

20PROFESSOR MURIS: Sure, for the reasons of

21administrability and an efficient operation of the legal


23PROFESSOR CRANE: I would add in addition to

24what Tim said also just for the purpose of disciplining



1The problem with sort of open-ended standards

2that don't contain sort of concrete legal rules is that

3the district courts tend to interpret these as

4invitations to punt issues downstream to juries, and

5that then leads to forced settlement because people are

6risk averse and don't want to go to trial.

7Part of this is not simply from business

8planning purposes. It is also to give a more

9disciplined structure to motions to dismiss, and for

10summary judgment that allows very serious screening of

11cases so that only the very most meritorious cases ever

12make it to a jury.

13MR. TOM: I'm not sure if this one is right or

14not. The reason I say that is that what you are

15essentially saying is the application, the passing or

16failing of a price-cost test is part of plaintiff's

17burden of proof and that without meeting that burden,

18the plaintiff should fail.

19Maybe that's right. Maybe we know enough about

20these price-cost tests and we know enough about the

21ability to prove this that it should be part of

22plaintiff's burden.

23On the other hand, if you take my board of

24directors hypothetical, if you will, maybe one should

25say, well, the ultimate question under the law as it has


1been handed down to us is is this conduct on that

2pro-competitive or anticompetitive.

3Plaintiff has come forward with some evidence.

4If defendant is able to rebut it by application of the

5price-cost tests, then we will accept that as a trump.

6But it is not part of plaintiff's prima facie case.

7I don't know which one is right. Maybe the

8economists on the panel or others can give us all some

9empirical basis for knowing which is more likely to lead

10to better results.

11MR. MEYER: It sounds like your alternative

12approach as you have described it would mean there is no

13safe harbor that a business can rely upon but, rather,

14that cases would go to summary judgment, past summary

15judgment to the jury if there is any evidence from which

16a jury could reasonably find --

17MR. TOM: No, I don't think that's quite right,

18because, of course, if it is a trump, if the price cost

19is a trump, it is defendant's trump, of course the state

20of the record on summary judgment may be such that there

21is no question of material fact in dispute as to that

22trump. Then it doesn't go to the jury.

23It is really -- I think that the difference is

24not whether these cases automatically go to the jury.

25The difference is who has to come up with this evidence


1and which way do these cases get decided under a state

2of uncertainty as to the price-cost test.

3Or another way of putting it is is our knowledge

4of the price-cost test so superior to any other

5knowledge that we can bring to bear on the ultimate

6question of competitive effect that we should make it

7part of the prima facie case.

8MR. MEYER: Let me flip the question around a

9little bit and ask let's say it were an affirmative

10defense so that at summary judgment the defendant could

11prevail if it demonstrated there was no dispute that

12prices were above cost.

13Is it a different answer in that case or you

14still want to allow more of what I will call an

15open-ended inquiry into that.

16MR. TOM: If defendant can show that, that's

17pretty convincing.

18MR. MEYER: Janusz, any thoughts?

19PROFESSOR ORDOVER: As an economist, I'm very

20fond of tests that are clear-cut and also try to compare

21some sort of price to some sort of cost.

22But I think a price versus cost test is a very

23ambiguous standard because we already have heard today

24that there could be average price, average cost, there

25could be marginal price vis-a-vis opportunity cost,


1which is what the Ortho cost was, where the marginal

2price was the incremental revenue under tests that Ortho

3could have provided against Abbott. And the question

4was which of the allocation of the costs ought to be

5brought into the particular calculation.

6So there is nothing wrong with price versus cost

7tests. The question is is it the right test in each and

8every case that involves possibly anticompetitive


10As an economist, I really don't know. If I were

11to be advising petitioners, I would say let's try to

12come up with as clear rules as we can. We ought to be

13comfortable with understanding the meaning of the price

14and the meaning of the cost in the test, comfortable

15with advocating the correct price and the correct cost.

16MR. MEYER: Is there a clear rule you would be

17comfortable with, Janusz, as to a particular price and a

18particular cost as a safe harbor for these kind of

19loyalty discounts?

20PROFESSOR ORDOVER: In the Ortho test, I thought

21the rule, which was already my prior work and which is

22consistent with much of the telecommunications

23regulatory practice, where it came from, the efficient

24component pricing rule, which is the progeny for all of

25this, I thought was a good rule. And I would like to


1see that be applied if possible.

2But there could be circumstances in which one

3can try to argue that it is not the right one, that a

4better calculation would be to look at the average cost

5versus average price of some sort.

6I think that in, for example, U.S. versus

7American Airlines, I thought that the relevant test

8would be applied to profitability of the route, because

9the contestable object there was the route or a large

10portion of the route, as opposed to marginal flight,

11which was not what the gain was all about.

12I don't have a hard and fast rule, and I would

13like to be able to argue for some degree of flexibility,

14in part because different circumstances may call for a

15different version of this thing called the price-cost


17MR. MEYER: Is there any rule that you would say

18a monopolist or a firm --

19PROFESSOR ORDOVER: Let me come back next year.

20MR. MEYER: -- could take comfort in as a safe


22Is there some minimum least common denominator

23in your various approaches so that you would be

24comfortable with a rule that said these situations will

25never be the source of Section 2 liability?


1PROFESSOR ORDOVER: I don't think that I'm that

2smart or that obnoxious to have such a vision.

3Again, as I said, I said something along these

4lines in the Ortho test. I thought that was not a bad


6I also think in different settings, much broader

7increments of output ought to be the test. I think it

8has to be looked at in the specifics of a particular

9case as much as possible.

10MR. MEYER: Tim?

11PROFESSOR MURIS: I think then-Judge Breyer said

12we just have to remind ourselves, in Barry Wright,

13"unlike economics, law is an administrative system, the

14effects of which depend on the content of rules and

15precedents only as they are applied by judges and juries

16in courts and by lawyers advising their clients" in this

17pricing area.

18We get ourselves away from price-cost benchmarks

19and we are lost, I think.

20That's what the world was like when predatory

21pricing cases were brought at the drop of a hat.

22I sat in a Commission conference room in 1975

23and 1976 when the coffee case was debated, and it was a

24case that everyone in this room would now regard as nuts

25but was seriously being pursued as any of the above.


1And what happened there was when Proctor &

2Gamble was relieved from a Commission order that

3prevented it from expanding that was the condition of

4purchasing Folger's, it immediately marched into the

5east into General Foods' territory.

6General Foods, panic faced with the Proctor &

7Gamble of the 1960s, 1970s, created horrendous

8documents, and the FTC wanted to bring the case and they

9voted repeatedly not to bring it, and the staff kept

10bringing it back until they voted to bring it.

11Again, it was a manifestation of lots of things,

12but there were lots of predatory pricing litigation,

13lots of uncertainty. And it was ended by Brooke Group,

14and I think purposely so.

15Even now we still have fighting at the edges.

16It was ended by a test that will have some mistakes, but

17I think it is essential.

18PROFESSOR ORDOVER: I want to comment on this.

19I think as you so beautifully said, the thing that makes

20me be less sure than I generally try to be of myself is

21that the source of that price quote status was a notion

22that in a perfectly competitive environment, the firm

23would not go below marginal cost. That's what it was.

24That was the foundation.

25The problem that we have is in many of these


1cases that we are dealing with, all of them do not take

2place in perfectly competitive environments.

3Because they don't take place in perfectly

4competitive environments, the question then becomes what

5lessons can we learn out of a perfectly competitive

6model that would illuminate the competitive effects of

7these interactions in markets which are by definition or

8by experience two or three standard deviations from the

9perfectly competitive ones.

10I really want to make sure that we don't get

11ourselves entangled in this price-cost test as being an

12economic foundation of anything. But I am perfectly

13happy to view those as being the right things to look at

14given the administrability and the clear-cut statement

15that one can make to the firm that is trying to compete

16hard in the marketplace.

17PROFESSOR MURIS: Perfect competition exists

18nowhere. I would agree with that.

19My favorite example is the hot dog vendors out

20there on the street. When they rise their price, they

21don't lose all of their sales. That means they have a

22downward-sloping demand curve, period.

23That is because of transaction costs and various

24things, not because of market power. Even Areta and

25Turner in the article admitted that firms for lots of


1reasons would price below something that looked like

2marginal cost for their average variable cost proxy.

3I obviously accept that and understand, as you

4said, that administrability is a key part of the


6MR. MEYER: I will jump to proposition 2, which

7is an extension from the prior proposition. Maybe we

8will make progress if you will move backwards.

9This is a quote from Herb Hovenkamp's recent


11"Single-product discounts should be per se

12lawful if the overall price for all units exceeds cost."

13Janusz, why don't we start with you.

14PROFESSOR ORDOVER: I was paying attention to

15something else.

16MR. MEYER: Talking about various price-cost

17tests and situations where you thought a broader

18calculation of price cost -- is this one?

19PROFESSOR ORDOVER: I think, again, it much

20depends on the circumstances. I think that if the

21average price is above cost, then again there is a

22burden-shifting exercise saying, well, there are these

23discontinuities or jumps in the loyalty schedule and

24they have potentially serious competitive effects.

25Is there a reason why we should not look at the


1average price versus average cost as being the right

2indication of how competition will play itself out?

3It could be that in some particular settings,

4comparing the two averages may just be inadequate in

5trying to really sort out all the potential competitive

6effects. But I would try to do that through the burden

7kind of shifting as opposed to per se blanket rule.

8In particular, in Ortho it was quite clear that

9Abbott was going to get an average return on all of its

10five tests that were way above its total average cost

11across these five tests. There was still a potential

12competitive issue.

13MR. MEYER: This statement is limited by its

14terms to a single-product situation.

15PROFESSOR ORDOVER: Wait a second. If you

16believe in the competitive equilibrium model, every good

17is a single different thing.

18We shouldn't get all hung up on this just

19because I called something -- there is something to be

20said about the uniformity of widgets versus not.

21But what about airline flights? Is flight 05

22the same product as flight 07? How should we look at


24We have to try to think a little bit more

25broadly as opposed to saying this is a bundled rebate


1and therefore two different products as opposed to the

2single product and, therefore, the 17th widget is the

3same thing as the 15th widget. That is all true.

4In particular circumstances, the 17th widget

5gets a certain kind of weight in how the equilibrium

6outcome looks that you have to try to pay attention to.

7In my opinion, we should not hide behind just

8the differences in the product names but in the economic

9circumstance that is driving it.

10PROFESSOR CRANE: To the extent this is directed

11at the Concord Boat situation as opposed to the LePage's

12situation, it is correct. One can imagine circumstances

13where single product loyalty discounts or volume

14discounts, market share discounts could have

15anticompetitive consequences.

16The same sort of discipline that one needs in

17litigation for bundled discounts also applies in cases,

18in fact, applies arguably even more in cases involving a

19discount on a single product.

20I think this actually is a law today in Brooke

21Group quite clearly and it is appropriately law.

22MR. TOM: Let me just take exception to that.

23I actually find what you just said a little bit

24surprising in light of the fact that you were defending

25an incremental revenue, incremental cost test in the


1multiproduct situation. And I think Janusz is

2completely right, that it can be very difficult to

3distinguish single product from multiproduct situations

4as a theoretical matter.

5So in terms of appropriate safe harbors, the

6ones that we have mostly had on the table today have

7been either incremental revenue, incremental cost or

8average revenue, average cost, which I think Professor

9Muris was advocating.

10I think this proposition can only be justified

11on the administrability and cost of false positives and

12false negatives kind of argument because there are

13certainly plenty of possibility proofs that show that

14you can have anticompetitive effects in this situation

15even with overall price exceeding overall cost.

16So the piece that I'm not hearing -- and maybe

17Hovenkamp lays it out in this article, and I haven't had

18the opportunity to read it -- is how do we know, what do

19we know about the prevalence of false positives or the

20prevalence of false negatives and the cost of false

21positives and the cost of false negatives?

22Is this situation such that you would advocate

23an average cost, average revenue rule rather than an

24incremental cost and incremental revenue rule?

25MR. MEYER: For the benefit of all of us, the


1average revenue/average cost rule is what you are

2saying --

3MR. TOM: What I read this proposition to say,


5MR. MEYER: Divide total units by total dollars.

6PROFESSOR MURIS: Let Dan respond.

7PROFESSOR CRANE: I wasn't trying to defend any

8particular measure of cost, whether it is variable cost

9or average total cost.

10I was simply suggesting that you should use a

11cost-based test in all cases involving single-product


13Again, even in a classic predatory pricing case,

14what the appropriate measure of cost should be is

15something that there is a lot of debate over.

16Without defending any particular cost test,

17though, I think that the proposition is correct, that is

18to say, whatever the appropriate measure of cost is, if

19that cost is recouped on the overall sale to a client,

20then the discount that created the overall sale should

21be legal.

22PROFESSOR MURIS: Perhaps Professor Hovenkamp

23had some idea of long run here. It doesn't matter. If

24you are going to apply these tests, in the short run

25real world, you will have to separate out the variable


1costs, I would think.

2You can have differences in different industries

3and how you define costs. And the airline case raises

4lots of complex problems, I agree. But I don't think I

5read this to say that we are talking averages total cost

6versus average total revenue.

7MR. TOM: Sorry for being less than clear. I

8wasn't really addressing what kind of cost is

9appropriate in the Brooke Group kind of situation.

10What I was addressing was the kinds of tests

11that have been applied in the writings on Concord Boat.

12Do you look at the incremental sales that were

13induced by the loyalty program and look at the revenues

14from those incremental sales and compare it to the

15incremental cost or do you apply a Brooke Group test

16that says you take all of the sales, all of the revenues

17and compare it to all of the costs for all of the sales.

18That's all I was saying. Frankly, I don't know

19which is the right test. I think if finding out what

20the facts were cost free and error free, then I would

21think this is clearly the wrong test.

22PROFESSOR ORDOVER: If you take the Concord Boat

23stylized example in which the challenger can go

24profitably after a particular dealership in the view of

25the loyalty schedule that applies to the dealership,


1obviously there is nothing to debate anymore, right?

2If indeed it is profitable to serve that by

3virtue of the fact that what the incumbent is charging

4is sufficiently above cost, whatever the right measure

5is, then you would think the effective or efficient

6challenger should be able to squeak under it somehow and

7capture the sale, which is why these price-cost tests

8make some economic sense.

9But, again, the issue is what it is that can be

10challenged and how much of an obstacle it is if you are

11required to challenge just the margin.

12MR. MEYER: If you all have a few more minutes,

13I would like to ask one further question, taking us out

14of the realm of safe harbors.

15Assume that whatever safe harbor is out there is

16not applicable, and we are now asking the question

17should the court condemn a particular loyalty discount


19What sorts of business justifications or

20efficiencies should the defendant be entitled to bring

21forward to escape liability?

22And, for example, perhaps it is obvious that if

23there is a particular efficiency associated with

24incenting a bundle, that that ought to be clearly

25cognizable, but what about simply the lower prices that


1are being paid by consumers in the short run or gains in

2share that the firm realizes by making its bundle more

3attractive to those consumers. Reactions?

4MR. TOM: I'm not sure how we quite leapfrogged

5from the safe harbor to the efficiency justification.

6It seems to me we have skipped the anticompetitive step

7in between.

8You can fail the price-cost test, but you would

9still want some sensible explanation of how this gives

10the defendant power over price, how prices go up as a

11result. And if price doesn't go up or indeed goes down,

12then I think you never get to those efficiencies.

13 MR. MEYER: Assume a plaintiff is coming forward

14and arguing that you are going to be excluding your only

15competitor by pricing this way, that you won't have any

16competition because a competitor cannot match the

17bundled price or the program. Assume that.

18MR. TOM: Then you may get into a debate I don't

19like to get into about consumer welfare versus total

20welfare, which is a little too theological for my taste

21or at least for my knowledge.

22So I will leave that to more expert folks.

23MR. MEYER: What justifications can a firm offer

24for successfully excluding its rival using some kind of

25pricing program like this?


1PROFESSOR CRANE: I think obviously there are

2plenty of pro-competitive reasons, like it costs less to

3sell the bundle. Those are obvious ones.

4The real question would come up if the plaintiff

5met whatever its prima facie case was and then the

6defendant was put to the burden of responding through

7some sort of explanation for why they offer the discount

8package. And things like price discrimination would

9come up.

10To the extent that mixed bundling is explicable

11because it is device for price discrimination, how

12should that cut? Price discrimination could be good for

13output. It can increase output. It can reduce output.

14Very hard to show sort of which way that cuts.

15So to me, any explanation that the defendant

16could offer that's accepted as the true explanation that

17is not an exclusionary explanation should be legitimate.

18MR. MEYER: That sounds like a no economic sense


20PROFESSOR ORDOVER: It is a good one.

21PROFESSOR CRANE: It is a pretty good one. We

22have some support on the panel for it.

23I think the sacrifice test or no economic sense

24test is difficult as a starting point. When it comes to

25defenses, it makes some sense, I think.


1MR. MEYER: Other reactions?

2MR. TOM: Let me just pose a question on that

3last one.

4 I said I didn't want to get into the total

5welfare versus consumer welfare. But I want to know if

6that's the question you are posing.

7You are hypothesizing that the result of this

8conduct is that prices to consumers go up. That is,

9whatever the efficiency justification, it doesn't lower

10the monopolist cost sufficiently that the price actually

11goes down. Am I correct in understanding that?

12MR. MEYER: That's a good question. I wasn't

13 being nearly so theological.

14MR. TOM: Go ahead.

15PROFESSOR MURIS: There are lots of them, and I

16prefaced them at the beginning and in the paper,

17specific efficiency justifications one can think of to

18stick with bundling.

19We need to step back and realize we are in a

20world where bundling is everywhere in very competitive

21markets. That in itself is an enormous empirical

22proposition of the efficiency benefits of bundling.

23PROFESSOR ORDOVER: I think that is

24undisputable. In fact, it is the case with many of

25these kinds of loyalty rebates as well.


1You go to Starbucks. You used to get your 10th

2cup of coffee free if you bought nine. Then you have a

3big discontinuity.

4MR. MEYER: No one else can sell you that 10th

5cup, right?

6PROFESSOR ORDOVER: It makes you drink the 10th

7cup and get jittery.

8There are some good reasons for stimulating

9demand, especially when you have a world in which the

10marginal cost is really low and you want to drive

11demand. It is a very powerful driver.

12When you have asymmetric information between the

13buyer and the seller or in many of these environments

14that people talk about, the GPO is insisting on

15discounts that are not volume driven but share driven in

16part on this theory that the differently situated

17hospitals are to be equally treated. And just because

18you are a small hospital, you can only buy 10 units of

19X, and if you are the big one, you can buy 100 units.

20You should not be somehow disadvantaged because of that

21because you are under the umbrella of the GPO.

22Some people say that is silly or what. There

23are -- if you go back to the case that was not quite

24fully litigated, Virgin British Airways case which

25pitted Schmazi against Bernheim, two pillars of


1antitrust and higher economics.

2There was a lot of discussion as to the

3usefulness of these various mechanisms as drivers of

4volume of sales at the travel agency level, which is

5where much of the action was.

6Rewinding the Areta paper by ten years, we will

7learn a lot of what the economics was at that time.

8Schmazi had a large number of defenses that he

9put forth why share-driven contracts were in fact

10efficient or optimal in some cases, and Bernheim took a

11somewhat different legal, working for Virgin.

12It is a case which we have not cited here, but

13it has a lot of levity in economics.

14MR. MEYER: We could go on forever here, but we


16 I want to thank everyone on the panel for an

17excellent discussion. Thank you all for attending.

18The next session will be next week, December

196th, I think, on misrepresentation and deceptive

20 practices.

21Thank you all for coming.

22(Whereupon, at 4:08 p.m., the hearing was





1C E R T I F I C A T I O N O F R E P O R T E R



4DATE: NOVEMBER 29, 2006


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: 12/18/2006






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