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

1

1UNITED STATES FEDERAL TRADE COMMISSION

2and

3UNITED STATES DEPARTMENT OF JUSTICE

4

5

6

7SHERMAN ACT SECTION 2 JOINT HEARING

8UNDERSTANDING SINGLE-FIRM BEHAVIOR:

9EMPIRICAL PERSPECTIVES SESSION

10TUESDAY, SEPTEMBER 26, 2006

11

12

13

14

15HELD AT:

16UNITED STATES FEDERAL TRADE COMMISSION

17SATELLITE BUILDING, CONFERENCE ROOM C

18601 NEW JERSEY AVENUE, N.W.

19WASHINGTON, D.C.

209:00 A.M. TO 12:30 P.M.

21

22

23

24Reported and transcribed by:

25Susanne Bergling, RMR-CLR

2

1MODERATORS:

2WILLIAM A. KOVACIC

3Commissioner

4Federal Trade Commission

5and

6KENNETH HEYER

7Acting Deputy Assistant Attorney General

8for Economic Analysis

9Antitrust Division, U.S. Department of Justice

10

11PANELISTS:

12

13Jonathan B. Baker

14Luke M. Froeb

15Robert C. Marshall

16Wallace Mullin

17David Reitman

18F. Michael Scherer

19Clifford Winston

20

21

22

23

24

25

3

1C O N T E N T S

2

3Introduction

4

5Presentations:

6     Jonathan B. Baker

7     Luke M. Froeb

8     Robert C. Marshall

9     Wallace Mullin

10     David Reitman

11     F. Michael Scherer

12     Clifford Winston

13

14Moderated Discussion

15

16Conclusion

17

18

19

20

21

22

23

24

25

4

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

2- - - - -

3DR. HEYER: Okay, first, it's a pleasure to be

4here, and since you're probably less interested in what

5I have to say than what these people have to say, I am

6going to be brief before turning things over to Bill.

7I wanted primarily to thank some people, not

8only the panelists for giving us their time and soon

9sharing their insights with us, but I wanted to thank

10particular people at the Antitrust Division who have

11helped prepare this and helped prepare me.

12We have some people from the Legal Policy

13Section in the Antitrust Division, Deputy Chief Gail

14Kursh, who in an earlier life helped manage the Dentsply

15case, which you will hear more about from Dr. Reitman

16over there. One of the attorneys in her section, Joe

17Matelis, crackerjack paralegal Brandon Greenland, and

18most importantly, June Lee, one of the economists in the

19Division, who, in addition to putting up with all the

20administrative stuff, has actually contributed

21substantively.

22So, with nothing further, I am going to turn it

23over to my distinguished colleague and co-moderator,

24Bill Kovacic.

25COMMISSIONER KOVACIC: Welcome to the New Jersey

5

1Avenue Conference Facility on September 26th, the 92nd

2Anniversary of the adoption of the Federal Trade

3Commission Act. We're delighted to have you all here

4today and to focus on what I think is one important

5dimension of the assessment of what standards for

6unilateral firm behavior ought to be. Many of the

7presumptions that run throughout discussions of doctrine

8and policy involving the enforcement of competition law

9against dominant firms derive from empirical judgments

10about the state of the world. To read judicial opinions

11and see how often the opinions say "we know, it is

12believed, it is thought, the world is," and then to look

13futilely in the footnotes for what editors in journals

14would note and say "Add cite," is a striking phenomenon.

15More than that, when you take a look at the

16papers of some of the Justices of the Supreme Court,

17papers that have become available, you see how

18frequently in their deliberations they're relying upon

19hunches, judgments or assessments about the state of the

20world and the way in which business behavior has been

21used in the past, and about the significance of that

22behavior. It's impossible, in short, in looking at the

23full range of history and enforcement policy and

24judicial decision-making, to escape the significant role

25that assumptions about the state of the world play in

6

1the formulation of doctrine.

2Our aim today is to address three questions and

3to try to link empirical work that's been done or might

4be done in the future to the development of standards.

5Three questions really animate our session today.

6The first is to consider what past empirical

7work tells us about how firms become and remain

8dominant, to look back and, at least selectively, to

9take a look at what work has been done by empirical

10researchers, whether in the form of quantitative work,

11whether in the form of case studies, whether simply in

12the examination of the way in which judicial decisions

13or enforcement decisions have affected the way firms

14behave.

15Second, and more forward-looking, is to ask what

16we would like to learn if we could, what additional

17facts would we like to have if we could get them in

18principle.

19And last, based upon what we offer as an answer

20to the second question, how might we go about doing it?

21What combination of effort within public enforcement

22agencies, among think tanks, academic research centers

23or other bodies, might provide the means by which

24important empirical questions could be answered?

25Later today, as Ken has, I will acknowledge the

7

1many contributions of our professional staff that have

2made the event possible. For now, to begin, I just want

3to remind you of a couple of housekeeping details about

4the session.

5The first is to respect our speakers by turning

6off all of your communication devices. I was at a

7hearing a couple of years ago in the federal courthouse

8where the bailiff stood up and said, "If your

9Blackberries or cell phones go off, you will be

10removed." We won't remove you, but please do honor this

11convention.

12Second, those of you who want to make your way

13to the restrooms, they are through the lobby -- the

14signs are marked -- between the elevators and off to the

15right. Now and then, there are planned or unplanned

16fire drills and alarms. If one goes off, we and our

17staff will lead you out to the street, to the right,

18back through the lobby, and we will simply gather out in

19front of the building until it is possible to return.

20To begin today, we have divided our session into

21two parts. We are going to have a series of

22presentations before we take a break, and then we will

23have a larger discussion joined by two of our panelists

24who have agreed to discuss what they have heard and then

25to add comments of their own about the proceedings.

8

1To get us started is Mike Scherer. Mike is as

2renowned and significant a figure in the modern

3development of economic research and analysis at the

4Federal Trade Commission as there is. Going back to his

5time as Bureau Director in this institution and through

6his recurring assistance, research and analysis, I think

7it is fair to say that, in the illustrious collection of

8those who have served as Bureau Director of the Federal

9Trade Commission, none has been more distinguished in

10 that very hall-of-fame like collection of individuals.

11Mike is also well known for the extent to which

12not simply has he done theory, but one of the reasons we

13asked Mike to come here is Mike's particular affinity

14and interest in empirical work and the extent to which

15empirical work, as well as history and an examination of

16the past, has figured into his own scholarship.

17Mike, please, thank you.

18(Applause.)

19DR. SCHERER: Thank you for those kind words,

20Bill.

21Let me just briefly address the third of Bill's

22questions, how to learn. In many ways, I have been a

23disciple of Joseph Schumpeter, not the stuff he wrote

24about monopoly and technological progress, but what he

25wrote about how economics advances. Schumpeter argued

9

1that economic analysis was all about three things. It

2was about theory, it was about statistics, and it was

3about history. To do economic analysis right, you need

4all three, and I have tried hard to do all three of

5those things. I think in the profession now there is a

6bit of an imbalance; in particular, we do too little

7 history.

8I am not sure whether it was distributed or

9whether it is on the web or whatever, but I do have a

10background paper for the meetings entitled

11"Technological Innovation and Monopolization." It is a

12case history of seven great high-tech monopolization

13cases in the 20th Century, and the thrust of my remarks

14will be based upon that paper.

15Now, first of all, how do you monopolize? Well,

16it is pretty well known. Mergers, here we have very

17strong precedent, so I won't dwell longer. Natural

18advantages, such as economies of scale, the control of

19natural resources, network externalities and the like,

20these are fairly rare except in the traditional

21regulated industries or in those cases where you define

22the market very narrowly, as in certain pharmaceutical

23deals.

24The most interesting one is surely superior

25efficiency and especially technical innovation. These

10

1pose the hardest cases for antitrust. When a firm

2achieves a monopoly position through superior efficiency

3or innovation, one faces very difficult trade-offs. We

4should clearly, clearly be encouraging technological

5superiority, but where is the line crossed? That is the

6really tough question.

7A subset of this is patent accumulations. In at

8least two of the seven cases I analyzed, that is the key

9to how firms monopolized, specifically, General Electric

10in the lamp case and AT&T in the telephone case. We did

11not do anything about it early in the century, and

12therefore, we had a raft of problems to deal with

13beginning in the 1940s and later.

14There are some puzzles here. There is one that

15I really think the FTC or someone ought to study very

16carefully, and that's Cisco. Cisco reached its dominant

17position in the network switch business on the strength

18of about 100 acquisitions and a lot of patent

19acquisitions. Was that necessary? Would we have had

20the best market structure for the switch industry if

21antitrust had intervened against these mergers?

22I remember one time being at a cocktail party in

23Cambridge and meeting a gentleman who told -- you know

24what you do at these cocktail parties, "What do you do?

25What do I do?" He said, "Well, what I have done, I have

11

1developed a switch that is a thousand times faster than

2anything Cisco has." He ran a high-tech startup,

3needless to say. I said, "What are you going to do with

4it?" "Oh, we are going to exploit it. We are going to

5market it." The next thing I know, he is bought by

6Cisco for a couple of billion dollars.

7Now, what would have happened if this guy had

8been encouraged to develop the switch technology on his

9own? These are interesting counterfactual questions

10that ought to be explored carefully.

11I pass on very briefly to the pricing

12consequences of monopoly. It has to be brief, because

13the theory and the evidence are extraordinarily complex.

14It depends critically on entry barriers, broadly

15defined, or cost structures. In particular, if entry

16barriers are low, you have the paradox of explaining how

17a firm achieved dominance despite having low entry

18barriers.

19The United States Steel case, decided by the

20Supreme Court in 1920, bears careful examination. The

21evidence is very clear. The Bureau of Corporations did

22a superb job studying that industry. U.S. Steel had no

23cost advantage over its rivals after the Carnegie

24properties had settled into normality. So, it had no

25cost advantage. How could it preserve its dominant

12

1position? Well, the answer is it could not, and so it

2chose an umbrella pricing strategy. It set prices high

3enough to provide nice profits for everybody in the

4industry. That encouraged a flood of entry, and

5gradually, U.S. Steel's market share declined, which the

6Supreme Court saw as evidence of effective competition,

7the declining market share.

8In fact, what it was evidence of was setting

9prices monopolistically high above the entry-deterring

10level and behaving essentially sluggishly about entry,

11and as a result, we have a steel industry that inherited

12this tradition of sluggishness, of not responding to

13price signals for 50 years until it got into big trouble

14in the 1970s and 1980s.

15Well, much more important than pricing is

16technological innovation, much more important. There I

17am clearly a "Schumpeterian." The question is, are

18monopolists, are dominant firms, superior innovators?

19The theory we have on this -- and we have got a lot of

20it, and evidence, too -- the theory and evidence on this

21say there's a duality. On the one hand there are

22situations, situations mainly associated with

23slow-moving technologies, where the science base is

24changing slowly. There are situations where a

25monopolist will, in fact, be a superior innovator, where

13

1only a monopolist is able reasonably quickly to realize

2sufficient quasi-rents to cover the R&D cost. Those

3cases definitely do exist in small markets and markets

4where the science base is moving slowly.

5But there's an exception when the science and

6technology base is moving rapidly, where you have

7revolutions, the kind of revolution we have had in

8information technology in the last few decades, where

9that is happening, and/or when monopolists are reluctant

10to cannibalize the rents that they are earning on the

11products that they already have marketed. In those

12cases, firms in dominant positions are almost surely

13sluggish innovators. I say "almost surely" because

14here, too, one can find exceptions.

15The most interesting exception in recent years I

16think has been Intel. Andy Grove's book Only the

17Paranoid Survive is a really nice example. I

18participated for the FTC in the case against Intel and

19read all of Andy Grove's memoranda for several years.

20Intel was really terribly alert to new technological

21challenges and tried hard to stay abreast of them and

22not be out-competed by upstart innovators. Even so, the

23record is quite interesting. I do not have a slide

24projector, and I did not bring a slide anyway -- I

25forgot to bring it, it was the most important slide I

14

1was going to bring with me, and I forgot to put it in my

2portfolio --

3COMMISSIONER KOVACIC: We have a sketch artist

4in the back.

5DR. SCHERER: No, I will wave my arms so you can

6see. I did a graph, this was in the FTC's Intel case,

7from public data. I had a graph on which time was the

8horizontal axis, and on the vertical axis was the speed

9of microprocessors, and what one sees is two things.

10First of all, in the period when Intel had a

11monopoly, at least in 32-bit chips, where Intel had a

12monopoly, the trajectory introducting speed improvements

13was like this, quite gradual, but then AMD and then

14Cyrix caught up and got into the 32-bit technology and

15began competing with Intel, and what you see, that slope

16abruptly turns sharper. There was more rapid increase

17in the key variable of competition, the speed of the

18microprocessor, and one also found the individual new

19product points more tightly clustered, showing that more

20new products were being brought into the market as a

21result of the competition from AMD and Cyrix.

22Intel argued in the FTC's case that we are our

23own best, sharpest competitors, because we have got all

24this installed base out there, and we have to bring out

25new products constantly or people will just stick with

15

1their old microprocessors. I did a series of simulation

2analyses, and what I found was that using reasonable

3parameters, Intel would try to maintain a generation for

4five or six years in the absence of competition. When

5there was competition, however, it moved the speed of

6the introduction process to two or three years.

7Now, this blends into another aspect where you

8really have serious problems for antitrust, and that is

9the so-called fast second strategy. This is a concept

10that was introduced in the late 1960s by Lee Baldwin

11and -- I don't know his first name -- Childs, and there

12has been a good deal of theoretical development on it

13since. The basic idea is that the dominant firm holds

14back until there is a real threat -- Andy Grove's Only

15the Paranoid Survive -- and then when that threat

16appears on the horizon, the dominant firm comes onto the

17market with a new product, with all guns blazing, and

18perhaps with a whole panoply of practices to make life

19difficult for the new company. You can see them

20described in the paper I submitted for the record, but

21you clearly see this kind of conduct in Standard Oil, in

22General Electric, in AT&T, in Xerox, in IBM, and in

23Microsoft, you see at least delayed innovation, and for

24IBM and Microsoft, a powerful fast second strategy.

25How much time do I have?

16

1MR. HEYER: You have got another ten minutes or

2so.

3DR. SCHERER: Oh, okay. Then I will read Judge

4Jackson's -- I think it's the penultimate paragraph --

5MR. HEYER: Five or ten minutes.

6DR. SCHERER: -- in Judge Jackson's decision in

7Microsoft.

8"Most harmful of all is the message that

9Microsoft's actions have conveyed to every enterprise

10with the potential to innovate in the computer industry.

11Through its conduct toward Netscape, IBM, Compaq, Intel

12and others, Microsoft has demonstrated that it will use

13its prodigious market power and immense profits to harm

14any firm that insists on pursuing initiatives that could

15intensify competition against one of Microsoft's core

16products. Microsoft's past success in hurting such

17companies and stifling innovation deters investment in

18technologies and businesses that exhibit the potential

19to threaten Microsoft. The ultimate result is that some

20innovations that would truly benefit consumers never

21occur for the sole reason that they do not coincide with

22Microsoft's self-interest."

23Well, Intel pursued similar policies. Actually,

24the truth is more nuanced than what Judge Jackson said.

25What he said was basically right, but recognizing this,

17

1firms that had to compete with Microsoft or had to

2compete with Intel pursued more sophisticated

3strategies. Sometimes they simply tried to avoid areas

4of dominant firm strategic interest, and therefore, we

5may have missed significant innovations. We will never

6know what we have missed.

7But in other cases -- and I think this is the

8larger majority of cases -- what they did was made their

9appearance on the scene and then made it clear that they

10really would like to be acquired by the dominant firm at

11a very hefty price, and here we face a tough

12counterfactual question. Would technological progress

13be faster if they had seen their way clear to innovate

14independently rather than having their operations taken

15over by the dominant firm?

16Now, my own view is that open competition is

17clearly superior in inducing vigorous innovation as

18compared to situations in which one has a relatively

19secure dominant firm. The presumption of antitrust

20should be to err on the side of maintaining competition

21and especially, especially keeping both conduct

22barriers, including fast second strategies, and

23structural barriers at minimum feasible levels. This is

24hard. There is no way to evaluate such situations

25without a careful rule of reason analysis guided by

18

1appropriate economic theory. But when monopoly

2positions exist, the job can be done, and it should be

3done.

4At this, I will stop and will be happy to take

5questions. Thank you.

6(Applause.)

7MR. HEYER: I think what we are going to do is

8we are going to hold off on questions until we get into

9the post-break round table discussion. We will let each

10of the panelists go.

11Let me say a few words about Luke, eager to get

12up here. Luke has a very long title. He teaches at

13Vanderbilt. He is particularly proud of his work

14recently at the Federal Trade Commission, and I am happy

15to say I know Luke back from when he was a staff

16economist at the Antitrust Division. Despite his work

17there, he became chief economist at the Federal Trade

18Commission.

19With no further adieu, we can --

20DR. FROEB: Can we bring up the slides?

21MR. HEYER: Actually, these aren't Luke's. All

22right.

23DR. FROEB: Thank you. It's a pleasure to be

24here. Every time I go in and out of academia, I get

25more discouraged about what we are doing in academia.

19

1We work hard on problems no one cares about and publish

2results in journals that nobody reads, and so it is a

3delight to be back here working and thinking about

4important problems that people care about.

5This area is the source of the biggest policy

6disagreement between the U.S. and the rest of the world.

7The U.S. is relatively permissive towards single-firm

8conduct, while the rest of the world is not. We have

9reached agreement, by and large, on how to analyze

10price-fixing and merger cases. And while we do have

11differences about individual cases and evidence, we do

12agree on the analytical framework.

13There is no such agreement on single-firm

14conduct, and why do we have this disagreement? What do

15we really know about single-firm conduct? But more

16importantly, do we know what we don't know about

17single-firm conduct, and the message of this talk, there

18is a lot of stuff we do not know, and I think we have

19got to be really careful about policy in this area.

20Before I start, I want to thank those who have

21contributed to my thinking in this area. I thought I

22would stop taking credit for other people's work once I

23left the FTC, but apparently not for a couple more

24years.

25Okay, so why is horizontal merger analysis

20

1easier than vertical? The biggest reason is we ignore

2the long-run indirect and strategic effects of

3horizontal mergers. We focus solely on the short-run

4increases in market power, and we have relatively good

5understanding of how that occurs. Most disagreements

6focus on the magnitude of the effect and how to estimate

7it. In other words, we disagree about the evidence, but

8not on the analysis.

9 The second reason is that we have these distinct

10mechanisms through which mergers affect consumer

11welfare: unilateral effects, entry, product

12repositioning, efficiencies, and coordinated effects.

13I think we know less about coordinated effects than we

14want to, but the other mechanisms are well understood.

15To analyze cases, we gather evidence on each mechanism,

16and estimate the net effect by estimating the magnitude

17and likelihood of each individual mechanism.

18So, why is analyzing single-firm conduct harder?

19Well, we are concerned about long-run, indirect

20strategic effects. We just cannot ignore them. If we

21did, we would have a very simple analysis. And the

22second reason is that mechanisms with opposing effects

23usually appear in a single kind of behavior. Predation

24is the simplest example. In the short run, firms reduce

25price, but in the long run, we get fewer competitors.

21

1Vertical integration has the same problem. In

2the short run, we have the unilateral effect of vertical

3integration where firms eliminate the double

4marginalization. But in the long run, we might have a

5raising-rivals'-costs or reducing-rivals'-revenue

6mechanism.

7Exclusive dealing, again, has two opposing

8mechanisms. The immediate effect of exclusive dealing

9is to reduce consumer choice, but indirectly, exclusive

10dealing serves to align the incentives of the retailer

11with the goals of the manufacturer. So, balancing these

12effects is really, really difficult. They appear

13together, and we do not really have good ways of

14balancing them.

15So, for these three reasons, single-firm conduct

16is hard to analyze. There is a taxonomy that I borrowed

17from Tim Brennan that says, let's consider the simplest

18case where we have some kind of behavior that has only

19two effects, two mechanisms at work. There is a

20proximate, immediate, direct, short-run mechanism that

21we may know something about, but the effects of the

22distant mechanism are much less certain.

23There are four possible outcomes, the distant

24mechanisms and the proximate mechanisms can both be good

25or bad. Those are the relatively easy cases. Where we

22

1run into problems is when the mechanisms work in

2opposing ways, where the distant mechanism can be bad or

3good and the proximate mechanism has the opposite sign.

4When you are doing single-firm analysis,

5evidence determines which box you go in, and most of the

6kind of behavior we are concerned about goes in either

7the off-diagonal boxes. The good-bad box and the

8bad-good box, those are the ones where we run into

9problems. Most of the problem cases fall into the lower

10left box where we have a distant bad and a proximate

11good, and you can think about bundling, as an example.

12Bundling offers consumers a better price for the

13bundle. That is why they buy the bundle, and they are

14better. But in the long run, the bundle may exclude

15competitors, and that may have a negative long-run

16effect. I have already talked about vertical

17integration, but loyalty discounts and predation give

18rise to the same kinds of problems.

19So, how do we characterize the different

20regimes? The big difference between the U.S. and the

21rest of the world is that we disagree on the distant

22effects of mechanisms, i.e., what is the magnitude of

23these distant effects and how frequently do they occur?

24The Europeans are much more concerned with the

25long-run negative effects of things like bundling and

23

1predation and loyalty discounts, and so they are

2concerned with avoiding type II errors. If regulatory

3agencies are uncertain about the effects of single-firm

4behavior, they are going to make mistakes. They will

5either deter behavior which is good, type I error, or

6let bad behavior go through, type II error. And there

7is an inevitable trade-off: The only way you can reduce

8type I error is to increase type II error and vice

9versa.

10The U.S. regime is more concerned with type I

11errors. We are more concerned with deterring good

12behavior. So, we tend to regulate less aggressively.

13Europeans are more concerned with type II errors, so

14they regulate more aggressively. We cannot determine

15who has the better regime, but we can say that relative

16to the U.S., the Europeans commit more type I errors;

17and relative to the Europeans, we commit more type II

18errors.

19The "makes no business sense" standard is really

20about trying to find cases in that box so we do not

21deter any good behavior. We miss more bad behavior than

22the Europeans; but they deter more good behavior than

23we.

24So, the interesting question and the focus of

25this hearing is, how do we determine the effects? Mike

24

1correctly states that the effect question is a difficult

2counterfactual. How do we know what would have happened

3had a firm behaved differently?

4This requires comparing two states of the world,

5only one of which we observe. That is what Mike means

6about the counterfactual. We have to figure out what

7would have happened had the firm behaved differently.

8There are two ways to do it. You can construct

9a theory that describes competition, and use that theory

10to tell me what would have happened had the firm behaved

11differently.

12The other way is to use what we call natural

13experiments, and this is really a misnomer. Any

14statistician in the audience will cringe when I use the

15word "experiment," because there is nothing experimental

16about economics data. We do not get to run experiments

17with the economy, probably for good reason.

18When I talk about natural experiments, I am

19talking about comparing a market with the behavior to a

20market without the behavior, and drawing inference about

21the effect of the behavior by comparing those two

22markets. The big questions here are how well does the

23experiment mimic the effect of interest; and did we hold

24everything else constant that could have accounted for

25change. These are tough questions to answer.

25

1We would particularly want to draw inference

2about the distant, long-run, or strategic effects,

3because we know less about them, and because uncertainty

4about their effects is the source of conflict between

5policy-makers, attorneys, and economists. I hate to be

6so hackneyed, but we need more information; we need more

7research. However, do we have natural experiments that

8estimate the effects of these distant effects?

9Here is my favorite study. It is from a paper

10by Mike Vita of the FTC, and it estimates what happened

11when the appeals court overturned the must-carry

12regulations for cable TV. Local cable TV monopolists

13must carry local over-the-air broadcast channels, and in

14close areas like Baltimore/Washington, they must carry

15both the Baltimore and the D.C. stations. When the

16Court overturned those regulations, which stations did

17the cable TV monopolist drop?

18Would the Baltimore cable system drop the

19Baltimore over-the-air broadcast stations which compete

20for audience share and advertising revenue, or would

21they drop the Washington over-the-air stations where

22they do not compete and can get the same content? And

23Mike found that they dropped the channels that had the

24lower rating, and these tended to be the competitors.

25Competitors were less likely to be dropped, and Mike

26

1interprets this as evidence refuting the anticompetitive

2hypothesis. He found that in the long run a firm will

3not exclude its competitors, as long as they are

4carrying a good product. I thought it was a very clever

5kind of use of the decision to try to draw inference

6about these long-run distant effects.

7Another Whinston natural experiment is Indiana's

8ban on exclusive territories for beer distributors.

9After a state law banned exclusive territories, beer

10consumption fell by 6 percent. Here again, the author

11concludes exclusive territories were pro-competitive.

12Other experiments show that gasoline prices are

133 cents higher in states where refiners are prohibited

14from owning their own gas stations. For fast food,

15prices at company-owned stores are 3 percent lower.

16Another experiment which is pretty messy, and I have

17given this talk over in the UK, and they fight me on

18this one, on the banning of tied pubs -- so if you are a

19beer manufacturer, you can't own your own pub to

20exclusively promote your own -- you have to carry at

21least two brands of beer. Small beer manufacturers

22liked having their own pubs because they were using them

23to promote their beer, and they thought it was an

24effective way of competing against large brewers. And

25once they got rid of tied pubs, price went up and

27

1quantity went down. However, there were a lot of other

2changes that were going on at the same time, so it is a

3hard experiment to interpret. But more telling was that

4the small beer manufacturers fought the change. They

5liked being able to own their own tied pubs and to have

6exclusives with a pub so they could promote their

7brands, and sure enough, the small -- the small beer

8manufacturers were hurt by the change.

9At the same time that we were reviewing the

10literature, Francine Lafontaine, who knows more about

11franchise agreements than I, and Margaret Slade, who

12used to be at the FTC and is now in the UK, were

13reviewing the literature as well, and they used a

14different taxonomy than we did. We were trying to

15determine what can we learn about these distant effects,

16but they were looking at government-imposed changes

17versus voluntary changes, and they looked at a lot of

18the same studies that we did. Here is their conclusion:

19When manufacturers impose restraints, not only

20do they make themselves better off, but they also

21typically allow consumers to benefit from higher quality

22products and better service provisions. In contrast,

23when the Government prevents these kinds of contracts,

24the effort is typically to reduce consumer welfare as

25prices increase and service levels fall. And they

28

1conclude that the interests of manufacturers and

2consumer welfare are apt to be aligned, while

3interference in the market is accomplished at the

4expense of consumers, and, of course, manufacturers.

5I would interpret this as evidence that these

6kinds of arrangements are doing what we want them to do,

7which is the U.S.'s relatively lenient attitude toward

8single-firm behavior relative to the rest of the world.

9I do realize there is a lot that we do not know, and I

10think it is important to recognize that there is much we

11do not know.

12More importantly, how do we generalize these

13studies to cases? I am not naive enough to think that

14in a litigation context we are going to have a nice

15natural experiment that we can interpret cleanly to tell

16us what to do in a specific case. However, I am not

17sure how frequently we have been looking for experiments

18like these.

19I am much less sanguine than Professor Scherer

20that we know that much about innovation. So, you look

21at the Intel innovation, who knows what the innovation

22rates would have been had we had more people in there?

23Maybe there was room for only one firm in the market?

24It is a really tough counterfactual. I wish we knew

25more.

29

1And finally, how do we test for the effects of

2antitrust intervention? Bill Kovacic has been a real

3advocate for what he calls competition R&D. When we go

4around the world and talk to new antitrust regimes, we

5say, look, don't just adopt a regime and freeze it,

6because what if you get it wrong? Instead, build in

7some kind of feedback mechanism, and start with the kind

8of follow-up studies that are done at the FTC and DOJ.

9I think they are absolutely crucial to try to

10characterize what are we doing, and to try to figure out

11what would have happened had we done something

12differently, in hope of improving.

13So, characterizing what we do and determining

14what its effects are really tough, but there are some

15instances where we can figure out what is going on, and

16I think we have to be on the lookout for good natural

17experiments.

18I guess that is all I want to say.

19MR. HEYER: Thank you.

20(Applause.)

21MR. HEYER: Okay, our final panelist presenter

22pre-break is Professor Wally Mullin. You have got his

23bio. He is a professor at George Washington University,

24and particularly of interest to us I think here is that

25he has done a fair amount of empirical work on some of

30

1the issues we are trying to grapple with. A lot of us

2have a lot to say about theory, but he has gotten his

3hands dirty a bit, and we look forward to his remarks.

4DR. MULLIN: Thanks. I am delighted to have

5this opportunity to appear in these public hearings, and

6I thank the Department of Justice and the Federal Trade

7Commission for jointly sponsoring these hearings and, of

8course, in particular, the co-moderators today, Ken

9Heyer and Bill Kovacic.

10So, switching gears, today I want to talk about

11what lessons we can draw from the history of antitrust

12enforcement, okay? Now, these may very well be lessons

13that are kind of in the DNA of current antitrust

14enforcers, but in the interest of redundancy, I am going

15to include some of those lessons as well.

16So, the initial set of dominant firms arose out

17of the trust movement in the sort of merger to monopoly

18way. So, in saying that this should be an area of

19contemporary interest, you know, I certainly acknowledge

20that similar economic and legal conditions may never

21return; however, the historical emphasis can still

22provide a modern researcher with a relatively large

23sample of dominant firms which faced antitrust scrutiny.

24So, as an empirical economist, that is very attractive.

25 So, I am going to focus in the discussion today,

31

1in part, as reflected in my own work, on an admittedly

2non-random sample of these firms, okay, Standard Oil,

3U.S. Steel, which Mike has already talked about a little

4bit, and American Sugar Refining Corporation. So, this

5choice arises out of a variety of factors. One is sort

6of the economic importance of the firms, you know, at

7that particular time, the legal significance of the

8associated antitrust decisions, and to some extent the

9similarity and differences in their business strategies.

10In work with co-authors, I have studied two of

11these firms. I haven't published any work on Standard

12Oil, but other people here have, and obviously it's a

13well-known case in terms of monopolization law.

14So, since all three firms faced antitrust

15prosecution, we can examine not only dominant firm

16behavior, but also the effects of prosecution, and we

17can also study the effects of remedy as implemented or,

18admittedly, more speculatively, consider the effects of

19remedies that were not ordered, because in some cases no

20liability was found.

21So, let's start with Standard Oil. My remarks

22on this will be relatively brief, reflecting sort of

23comparative advantage issues. So, Standard Oil, right,

24if we want to have a poster child for different types of

25dominant firms, Standard Oil was an aggressive

32

1competitor, okay? So, while the claim that Standard Oil

2engaged in predatory pricing has been debunked by McGee,

3the company had other practices that still marked it as

4an aggressive competitor. For example, Granitz and

5Klein in 1996 published an article studying how Standard

6Oil obtained differential rebates from the railroads on

7petroleum transportation, and that is a source,

8according to Granitz and Klein, of their sort of

9supra-competitive rents, and those rebates, of course,

10advantaged it relative to other refiners.

11Of course, Standard Oil was found guilty and

12dissolution was ordered, and it was kind of alluded to

13by Mike, Bill Comanor and he have argued in a paper that

14dissolution of Standard Oil raised long-term industry

15performance, and also in that paper, this is

16counterfactual, it would have been good had U.S. Steel

17been dissolved.

18In his academic work, Bill Kovacic has argued

19that the effect of this dissolution rests in part on the

20fact that the dissolution involved formerly independent

21entities. So, one shouldn't necessarily take this as a

22dissolution child's story in which everyone lives

23happily ever after as an automatic indication that

24structural remedies in all forms and in all

25circumstances will work. You have to be sensitive to

33

1the particular facts involved, but given the fact that

2Standard Oil was organized as such that what was spun

3off were things that were in some sense formerly

4independent or had a certain amount of autonomy within

5Standard Oil in terms of decision-making, in terms of

6things like corporate culture, the enterprise was able

7to grow and prosper going forward, and so my take-away

8would be that, you know, a different remedy in another

9industry or even with a firm with a different internal

10organization and history might have unduly sacrificed

11production costs, but that is merely a speculative

12comment with a note of caution.

13So, in terms of U.S. Steel, Mike has already

14touched upon part of this. So, you know, John D.

15Rockefeller and Standard Oil is the poster child for the

16aggressive competitor. United States Steel is sort of a

17poster child for a dominant firm that may be good for

18competitors and bad for competition, which was something

19that the Supreme Court didn't realize at the time.

20So, in published work with co-author brothers,

21and it's otherwise hard to find two other Mullins, we

22have presented evidence that dissolution, which, of

23course, was never ordered, would have lowered steel

24prices in that case, in particular, and raised steel

25output. So, in particular, the pattern of

34

1contemporaneous stock market reactions to events from

2the dissolution suit, okay, basically from 1911 to 1920,

3not only judicial decisions but periods when it was

4rumored U.S. Steel might dissolve itself to basically

5avoid prosecution, and then a denial of that rumor the

6next week, some subset of the events that I mentioned

7ended up having big stock market reactions for U.S.

8Steel, indicating that there was news sent to the

9securities markets in those particular events, and in

10those weeks, the stocks of customers, in particular, of

11U.S. Steel, particularly the railroads, reacted in a way

12that suggested that the stock market believed that

13dissolution would have lowered steel prices.

14So, interestingly -- and this is a bit in

15contrast to maybe what Mike Scherer was talking about --

16one of the things I also find of interest, and this is

17part of the tension of monopolization law, is that there

18are parts, going back to things that might have

19potentially been sources of market power, that

20contemporary scholarship would suggest maybe were, in

21fact, efficiency-enhancing. So, in particular, U.S.

22Steel was losing market share over time, and you might

23think, well, wait a minute, is there some sort of scarce

24factor upstream from steel production that they could

25use and acquire in order to foreclose entry, you know,

35

1or at least put a limit on that, right?

2So, historically they were vertically integrated

3into iron ore properties, as the Carnegie properties had

4been, and during the period where they were undergoing

5antitrust scrutiny at the start of the 20th Century,

6they added to that a significant amount by long-term

7leasing the iron ore properties of the Great Northern

8Railway and James J. Hill. So, that is why they are

9referred to as the Hill properties. And that was viewed

10as anticompetitive by contemporary antitrust authorities

11for some reason, as I will sort of talk about in the

12next slide, but that is not only criticized by the

13standing Congressional Committees -- the Federal Trade

14Commission wasn't around at the time -- but the Bureau

15of Corporation's report criticized it, and, in fact,

16U.S. Steel ends up cancelling the lease in 1911 in part

17to try to forestall prosecution because this was that

18big of deal to the Department of Justice at the time.

19Okay, so what might be some of the lessons we

20take from there? So, as before, of course, the law

21should protect competition, not competitors. You know,

22it strikes me -- as I said, I recognize that this would

23be known by the contemporary court, but it is a good

24case to assign students, because you have them read the

25case, and, of course, the Supreme Court is praising U.S.

36

1Steel because its competitors had such nice things to

2say about it at trial, and the contrast with Standard

3Oil is pretty stark. U.S. Steel's anticompetitive

4effect is not only due to single-firm conduct in a

5narrow sense, but U.S. Steel's actions in organizing the

6Gary dinners, which it later abandoned, clearly had a

7collusive intent, and they were also bad for

8competition, although good for competitors.

9So, another tension of monopolization law is

10that even a firm with market power may have

11efficiency-enhancing innovations, right? So, the easy

12case would be in which, you know, if you wanted to do

13some variation of the diagram, the easy case would be,

14oh, there are firms that have market power and there are

15firms that have cost reductions, and they are completely

16disjoint. I say empirically, that is not the case. In

17fact, in terms of work that we have done, U.S. Steel was

18a firm with both elements.

19So, in a paper with one of my brother

20co-authors, okay, we didn't have a falling out over the

21difference in these papers, orthogonal to that issue,

22the paper with Joe Mullin examines the Hill ore lease,

23and says that, on balance, that it seems to be best

24explained as being efficiency-enhancing rather than as

25vertical foreclosure.

37

1There are several reasons for this. So, if you

2sort of back up, the underlying problem of developing an

3iron ore mine is a problem of relationship-specific

4investment, something that was studied later by

5transaction cost economics, both for kind of developing

6the mine or the investment in the mine, which, of

7course, is not mobile once it is sunk, and also

8development of transportation to get the ore or some

9variation of the ore to market, and that transportation,

10given where those mines were, was over the Great

11Northern Railway, which otherwise would have owned the

12mining rights.

13So, the specific contractual terms that were in

14the lease, which caused the Bureau of Corporations to

15scratch its head circa 1906, has been studied by people

16like Crocker and Masten. So, one example of this is

17they had a take-or-pay provision which was quite large,

18so U.S. Steel was basically committed to making these

19large payments, and, in fact, during the initial period

20of the execution of the lease before it fell under

21antitrust scrutiny, they were, in fact, investing --

22they were basically scaling up to exploit that property

23at a very high level.

24And it's striking, also, in the sense that you

25might imagine some notion of vertical foreclosure or

38

1barrier to entry would be, oh, well, they are going to

2acquire this iron ore. They have other iron ores. They

3don't need to exploit it to produce right now. They are

4just going to sit on it and prevent anyone else from

5gaining entry to it, but, in fact, they invested heavily

6in trying to exploit the iron ore.

7It is possible, of course, it had an

8anticompetitive effect, so it is not so much a -- you

9know, a complete nesting of the hypotheses, but rather,

10sort of saying, our judgment, my judgment, the bulk of

11the evidence would be that that particular aspect of

12their innovation was something that was

13efficiency-enhancing.

14And, of course, the challenge for contemporary

15antitrust enforcers is what sort of humility should they

16exercise when faced with some sort of business practice

17that they don't automatically have an obvious efficiency

18explanation for? Now, obviously the staff and other

19people are going to be aware of transaction cost work,

20et cetera, right, but presumably, we will figure out 20

21years from now other reasons why some firms might have

22some sort of purpose. That doesn't necessarily mean

23that the behavior is necessarily benign, but that's the

24situation that requires the people to look at it.

25So, finally, love of my life, American Sugar

39

1Refining. So, David Genesove and I have written a

2series of paper on this. This is one of those things

3that you don't necessarily know what you're getting into

4when you start. So, in a paper that recently appeared

5in the Rand Journal, they profitably engaged in

6predatory pricing, and that was one of their business

7practices.

8Now, these joint hearings have already included

9a rich discussion of predatory pricing in an earlier

10session, so I won't recapitulate that now. We might get

11into some element of that in the discussion. David and

12I noted in the paper that compelling evidence of

13predation is rare. That is reflected not only in the

14academic consensus, but obviously also in the case law,

15but we think the evidence that we present in the paper

16in this case is compelling.

17So, in terms of a couple of things to point out,

18American Sugar engaged in predation. They didn't prey

19on all entrants. Every single entry episode didn't

20trigger predation or didn't trigger immediate predation;

21however, the nature of the market was such that after

22they preyed, they acquired the entrants and other fringe

23firms at lower buy-out prices. So, in a sense, if they

24were making the dynamic calculation, they were sort of

25saying, well, here's some small firm, it's entering, you

40

1know, no big deal. As more firms enter, they are sort

2of like, okay, well, now it's time to prey and buy

3people out and raise up our market share.

4In terms of trying to rationalize the

5observations under different theories of business

6behavior, that manipulation of rivals' beliefs played a

7very big role as in some of the reputation models. So,

8once again, it is not as if they sent out a clarion call

9saying that, oh, they were going to prey and then they

10were going to buy people out, so, in fact -- precisely

11because there were multiple firms they were basically

12preying on simultaneously, there are cases in which they

13basically made an arrangement with one of the firms to

14say, okay, well, fine, we are going to buy you out, here

15are these terms, but let's keep this secret, and so --

16and then continue the war, and then buy out the other

17firms.

18So, in some sense, part of the aspect of kind of

19buying out firms and engaging in predation is that the

20process is sort of the reverse of what we are calling

21the free-rider problem when you form a trust, right? If

22you form a trust, you are going to restrict output, and

23so people will want to stay outside of it and just take

24advantage of the output lowering entity.

25Conversely, if there's predation going on, and

41

1you know there will be a buy-out and the predatory

2pricing is going to end, of course, people also want to

3free-ride on that. So, the manipulation of rivals'

4beliefs is I think part and parcel of being able to be

5successful.

6So, there was a monopolization suit, and it

7stretched on over a period of time, that eventually

8resulted in a consent decree. But there are some other

9sort of, you know, maybe, you know, happy lessons here

10that antitrust serves as a deterrent on a variety of

11levels. Part of the rationale of the antitrust law is

12to be punitive, but obviously you also want to think,

13well, gee, you hope other firms get the message and we

14don't have to go prosecute them, or this firm in the

15future, once bitten, twice shy, and so will behave

16 better, and have some sort of implicit consent decree.

17So, there are two examples of this, and one

18deals with American Sugar and one deals with other

19firms. So, during its monopolization case, American

20Sugar underwent sort of partial "voluntary" dissolution,

21so this was before the consent decree, because of the

22government victories in the American Tobacco and

23Standard Oil cases.

24So, focusing on American Tobacco or Standard Oil

25as cases, those basically had a spillover effect on the

42

1behavior of another firm, in this case American Sugar,

2and presumably other firms. The difficulty of the

3non-random sample is, of course, it may be that the

4whole universe of firms behaved differently, which is a

5reason why people should do more work on it.

6Later on, there is also an impact on American

7Sugar itself. David Genesove and I also studied not a

8single-firm conduct, but in terms of collusive conduct,

9we studied The Sugar Institute of the twenties and

10thirties, of which American Sugar was the largest and

11most important member, but no longer as large as in 1911

12or 1914.

13So, this is noted in our AER paper, even though

14it wasn't the focus of that paper, which was that the

15legal representatives of American Sugar at these

16basically collusive meetings within the industry were

17very sensitive to things like discussion of price. That

18was a part of the battle, in a sense, within The Sugar

19Institute, one person complaining to his boss, oh, gee,

20we are never allowed to do anything that's going to have

21any real effect, and so that may just be the wise

22counsel of American Sugar at the time, but one has to

23think that the fact that they had had this antitrust

24 prosecution was something that empowered people within

25the firm to say, okay, compliance is important. It is

43

1certainly something you think that going forward would

2be an important part of antitrust enforcement.

3So, all I have for now.

4 (Applause.)

5COMMISSIONER KOVACIC: Thanks, Wally.

6I would now like to invite Jon Baker to present

7his comments. Jon, as you know, like Mike and Luke, is

8part of the galaxy of superb economists who have headed

9the Bureau of Economics at the FTC. In addition to

10Jon's affiliation with the Commission, in many ways he's

11been what I consider to be hitting for the scholarly

12cycle. Not only has he done excellent quantitative

13work, both at the Commission in matters such as Staples,

14but also, in his own published work, he has contributed

15wonderfully to theory. In studying the deliberations

16that took place over the Verizon-Twombly matter, I many

17times went back and referred to Jon's paper on two

18Sherman Act dilemmas from the early 1990s. And quite

19apropos for this panel as well, Jon, like so many of our

20presenters, has a good aptitude for history, reflected

21not only in his survey paper in the JEP on competition

22enforcement, but also in his recent paper in the

23Antitrust Law Journal on the development of widely

24accepted norms and standards, and his political

25bargaining paper. We are delighted to have Jon here

44

1today.

2DR. BAKER: Thank you. Thank you, Bill. That

3was a very nice introduction. It is not what I would

4expect from a case book co-author, but I appreciate it

5anyway.

6COMMISSIONER KOVACIC: I should have added, he

7is the co-author of the most astonishing and --

8MR. HEYER: Copies on sale in the lobby.

9COMMISSIONER KOVACIC: During the break, there

10will be the signing process --

11DR. BAKER: And I am always delighted to be back

12to see all my former FTC and Justice Department

13colleagues. I worked with Ken and Luke back in the old

14days at the Antitrust Division.

15Well, so let me -- I have a -- sort of several

16comments on what we have heard this morning. They are a

17little bit disjointed, and I will just get into them and

18see how far we get.

19The first is on the question of what can we

20learn from the old monopolization cases. On the one

21hand, there are very few of them. They are often high

22profile, but there aren't many, and a lot of them were

23reviewed when antitrust standards were very different

24than they are today and when ideas about remedies were

25different than they are today. I don't think we would

45

1remedy the Standard Oil monopoly were that to have

2appeared today anything like the way it was remedied

3then. We would have tried to get the parts that were

4broken up to engage in head-to-head competition from the

5beginning.

6So, there's something funny about this exercise.

7The -- you wouldn't -- it's a little like saying, well,

8what can we learn about merger analysis from studying

9Pabst and Von's, you know, some poster children of

10 merger cases that are no longer thought to be good

11precedents, although they are technically controlling

12Supreme Court precedents, as an aside.

13Well, what we learn from Mike Scherer and Wally

14Mullin, I think, is something that perhaps we have

15always known, which is the value of careful

16case-specific analysis. This is what the judicial

17system at its best makes possible.

18Now, that's not to say that the courts have

19always undertaken this -- the adversarial system has

20always forced the same level of analysis that later

21scholars have been able to bring to these cases. I

22mean, it took 50 years, but the Mullin Brothers finally

23got to the bottom of the U.S. Steel case. One would

24like that to have happened, in the case itself. But on

25the other hand, it shows you the power of case-specific

46

1analysis to hear Mike and Wally go through what they

2have learned about these cases.

3That's not to say that their conclusions are

4undisputable, but the kind of analysis they do, they can

5focus in on the issues, and it really does support the

6kind of work that we do in the enforcement agencies and

7the courts.

8Now, let me move on to say something about the

9issues Luke raised. It struck me, one interesting point

10is the short-term focus, Luke says, of our antitrust

11thinking. He didn't quite put it this way, but I mean I

12guess I'm a little -- I read it in the light of also

13thinking about a paper that John Lopatka and Bill Page

14wrote where they argued that antitrust enforcement

15courts are more congenial to -- or the decisions, I

16suppose you would say, the decisions are more driven by

17the short-term benefits and costs than the long-term

18ones.

19If you take that perspective and think about

20Luke's charts, it seems to me that one message is we

21shouldn't just give a free pass to all those kind of

22practices in the lower left box of Luke's taxonomy:

23Price predation, bundling, vertical integration and

24loyalty discounts. These are things where I think Luke

25says the proximate effect is good and the distant effect

47

1is bad.

2Now, I suppose that my characterization of the

3implication of those boxes is a little different from

4Luke's, but in order to go beyond the picture Luke drew

5to an enforcement regime that gives a free pass -- well,

6free pass is a little strong -- but that makes it tough

7to bring cases in the lower left-hand box, you have to

8take another step in the logic. You have to argue, as

9some people do, things like the Government can't do a

10good job analyzing these practices, separating out the

11two kinds of effects, and remedying it, and you have to

12conclude that the costs of one type of error are greater

13than the other. There's a whole additional apparatus

14that we have to apply before we can reach the conclusion

15that antitrust should be hands off on all these

16practices.

17In thinking about Luke's taxonomy a little more,

18I started thinking about most favored customer clause

19cases or most favored nation clause cases. The Justice

20Department for a while had an enforcement program

21involving dominant firms that instituted these kinds of

22practices. It was a dominant health insurer that had a

23most favored customer clause in its contracts with

24healthcare providers, and I'm thinking of -- was it

25Delta Dental, there's a bunch of Delta Dental cases, and

48

1I think there's some other ones.

2So, the idea was the provider, the doctor or the

3dentist or whatever it was, wouldn't lower rates to

4rival health insurers without also lowering it to the

5dominant provider, let's call it Blue Cross, and so that

6makes it impractical for the rivals or the entrants to

7make procompetitive deals; that is, rivals to Blue

8Cross. Insurers want to come in and say if you give me

9lower rates, I'll funnel more business to you, the

10provider, and we will both do better, and then this

11creates competition for Blue Cross.

12Of course, these most favored customer clause

13provisions can also result in collusion by making

14discounting more costly, but we are in the dominant firm

15context here, so we will put that aside.

16The interesting thing about these most favored

17customer clauses as a practice is that there are

18efficiency justifications that are often offered, but in

19a health care setting, they are not very plausible. The

20best efficiency justifications are either preventing

21opportunism when futures markets are unavailable, which

22sometimes happens in long-term contracting where you see

23these kinds of provisions, or perhaps signaling low

24prices where buyer search is costly, and these are the

25kind of -- here, we're thinking there about retail

49

1businesses selling to customers.

2Perhaps Luke will say to me I just moved these

3provisions in the health care context from his lower

4left box to his upper left box, where the efficiency

5justification isn't very good, and so there isn't a

6problem, but I think if you accept what I have gotten to

7so far, that these provisions can be troublesome for

8dominant firms to contract using them in many of these

9health care contexts, you have to ask, well, when we

10move outside the health care context, perhaps to one

11where the efficiency justification is potentially more

12plausible, don't we have to analyze? Don't we have to

13think about whether the bad guy story and the good guy

14story -- which is more powerful as between the two? So,

15my take from Luke's taxonomy is we ought to think hard

16about practices in the lower left-hand box and analyze

17them as best we can.

18On natural experiments, Luke, I think you missed

19an opportunity when you were talking about experiments.

20I have a new motto for the FTC, and this really would be

21your motto, not mine, "We fool around with the economy

22every day." Natural experiments are fine in

23principle -- that was just a joke -- natural experiments

24are fine in principle, and I basically am sympathetic to

25what Luke was trying to do with them.

50

1Tim Bresnahan and I have a recent paper where we

2talk about something similar. We say that a key

3challenge for antitrust analysis and empirical

4industrial organization economics going forward, which

5is not recognized in antitrust to the same extent that

6it's recognized in economics, is to exploit similarities

7among related industries that focus an inquiry involving

8the industry and the firms under study. We have some

9examples different from Luke's, but I think the spirit

10of the exercise is similar. An important question, even

11assuming it's a good natural experiment, is what

12generalization you can make from it.

13 Tim and I think that the right generalization is

14the level of the industry. In other words, I would look

15at some of the examples that Luke has about -- oh, I

16don't know, gasoline divorcement or something like that,

17but not -- and perhaps that would create a presumption

18about gasoline retailing, but I wouldn't connect the

19dots and generalize to all vertical restraints. All of

20Luke's examples, for example, in his representative

21studies are about manufacturer- distributor

22relationships in consumer products. They do not tell us

23much about most favored customer clauses, for example,

24in health insurer contracts with providers.

25Finally -- I am not sure how much time I have

51

1left. Do I have time left? Okay.

2MR. HEYER: Is it good?

3DR. BAKER: It's not as good as what's happened

4already, Ken. I don't get better.

5No, I think I'll just stop right there, and I

6will -- it's not that good, Ken.

7MR. HEYER: Save it for the discussion, all

8right.

9DR. BAKER: We will save it for the discussion.

10Thank you.

11(Applause.)

12MR. HEYER: The final person we are going to

13hear from before the break is Cliff Winston, who you'll

14see is a long-time economist at The Brookings

15Institution and has done just an incredible amount of

16empirical work, largely having to do with regulated

17industries but not exclusively, and partly because he's

18really taken on some tough challenges empirically, he

19seems like a perfect person to invite to talk here, and

20let's just hear from Cliff.

21DR. WINSTON: Thanks a lot for inviting me to

22this conference.

23Let me, since I'm a little bit on the fringe in

24this enterprise, sort of tell you my context and how I

25was thinking about this and eventually how I synthesized

52

1what we have heard.

2When Jim Taronji called me about this, my sort

3of immediate perception was you were planning a series

4of conferences that were basically assessing the

5antitrust activity at the federal level of DOJ and FTC,

6and I naturally thought this, and it turns out that -- I

7had just finished a book called Government Failure verse

8Market Failure that looks at all areas where the

9government intervenes in trying to correct market

10failures, including but certainly not limited to market

11power, but information problems, externalities, public

12good, public production and the like, and figured, well,

13this is right along the lines of what I have just

14written up, and so I can sort of look at what you're

15doing from this perspective.

16But I also pointed out that I was going to be

17away a couple of weeks before the conference and

18literally just got back late the night before, so it

19would be good if I got the presentations beforehand.

20Otherwise, you know, I would have to be on the fly, but

21I thought there obviously might be difficulties in

22getting things to me, and I was checking my web when I

23was in Europe, but late last night, I realized a couple

24had come in, but unfortunately one was in WordPerfect,

25and Brookings doesn't use WordPerfect. I assume Mike

53

1does this as a protest against Microsoft. I like

2WordPerfect better. So, I didn't have them, but I did

3have a fall-back position.

4What I was going to do was sort of outline a

5template, in general, about how I would assess the

6performance of a federal agency and what recommendations

7that I might make in terms of improving performance, set

8that up, say, okay, and I'll just plug in everything I

9hear in these areas.

10So, let me outline the template and then just

11make a few comments on what we've heard. So, the first

12thing in general that I would ask and think about for

13any federal agency is, is there compelling evidence of a

14problem to begin with? That is, you know, are there

15some stylized facts, summary measures of welfare, you

16know, that something is going on, you know, information

17problems are costing consumers hundreds of millions of

18dollars a year, monopoly is causing similar kinds of

19costs?

20Okay, the first thing, just get a big picture

21overview, when I do these things with transportation, it

22is very easy, because I can just point to graphs of the

23lake, there is a problem, congestion going on, airline

24delay, going up, there's a problem, not too much

25controversy about that.

54

1The second question one would ask, you know,

2what is the scholarly evidence -- when I mean the

3scholarly evidence, I mean quantitative, welfare type

4calculations, and certainly counterfactuals isolating

5the effects of other factors, on first market failure,

6what do we know about how markets are performing or not

7performing, since they may be the source of the problem,

8and government failure, that is, how are governments

9doing in all of this, and third, government success.

10So, you know, here are the things you want to look at

11from the bottom up, the little pieces of evidence that

12we look at to assess the agency.

13Then the third thing, since this really is a

14scholarly enterprise, when I ask the big picture

15question, where is the field going? You know, since

16we're getting a lot of the intellectual infrastructure

17from the scholars who work in the area, how does the

18field look at this problem? What kind of research are

19they doing? Where are they likely to help in the

20future, if at all? Are there incentives the agency

21could give to researchers to sort of get them focused on

22problems that they are interested in, so on and so

23forth?

24And then finally, you know, given one, two and

25three, where do we go from here? How do we put all this

55

1together and say, okay, here is how I think you can

2improve your performance and your interventions, or here

3is what I think we, you know, we need to know before we

4can give confident recommendations.

5So, let me go through these now with an eye

6toward what has been said and what has not been said

7about them. Okay, first, the big picture question, I

8didn't really hear exactly what I was looking for there,

9but there's a reason. It's really hard. They are

10trying -- and I think it is one of the big problems --

11maybe the biggest problem with industrial organization,

12is unlike other fields in economics, there isn't this

13stylized fact that you're constantly facing that reminds

14you of what's going on out there.

15It's not like in labor economics where you hear

16about what the unemployment rate is, okay, or the

17percent of people below the poverty line. You hear

18these numbers, you know, these are the kinds of things

19that researchers get to work on in dealing with this.

20It is not like trade where we hear what's going on with

21the dollar, the trade balance. Recently, it just came

22out about we now have sort of have negative net capital

23funds, I assure you now a lot of paper is going to come

24out about this, trying to explain it to us, what is

25going on, so on and so forth. You can think of a whole

56

1bunch of things, but when you talk about IO, yes, your

2instincts are well, we want some measure of economic

3welfare, but that's not presented by the Commerce

4Department. It's hard to construct that kind of thing.

5Now, that said, there was an effort to do that.

6In the sixties, there was a lot of effort to think of

7things in terms of concentration ratios, and that was

8sort of our stylized fact, and there was even a

9Commission, the Neal Commission, you know, that met and

10made recommendations about, you know, deconcentration of

11industries that exceeded a 70 percent level of

12concentration, and that may not be something that people

13take seriously today, but there was a time when that was

14sort of an orientation towards thinking about IO and

15even antitrust policy, okay?

16But there really isn't that, which is a bit of a

17concern, because you never sort of know, well, are you

18working on a problem that's really important? And the

19only one who talked about that was Luke in terms of

20motivating -- while we care about this, and he said this

21in terms of, you know, apparent disagreement or I would

22say just different approaches toward antitrust policy

23between the U.S. and the EU, and I just simply say,

24well, does that signify different concerns with the same

25problem?

57

1To the extent the U.S. is less aggressive and

2more permissible and allows certain things to go on,

3does it basically feel that competition is pretty

4intense, and maybe this is just signifying we really

5don't have that much of a problem, whereas in Europe,

6they might feel that there is more, but this is

7certainly something to think about.

8Okay, secondly, the scholarly evidence on the

9various issues, you know, first, looking at market

10failure -- and I agree completely with Mike, it's an

11excellent point, a point that is not made enough, that

12too much of economists' orientation on market failure is

13static inefficiencies, so price distortions and the

14like, where so much of the big gains from policy

15improvements are the dynamic ones, because that's the

16counterfactual that you don't see.

17So, if you look at what we've learned about

18deregulation in terms of what regulation we're doing,

19the big ticket effects were suppressing innovation,

20right? So, there you get, you know, more than first

21order effects. You get really big effects, you know,

22shifts of cost curves as you completely change what

23you're doing, shifts of demand curves where you provide

24new products, okay? So, to the extent that a dominant

25firm is working like a constrained regulatory policy,

58

1you know, the effects can be big.

2Now, that said, you know, measuring these things

3are very difficult, and, you know, it's not clear to me

4that we really have hard evidence on this kind of thing.

5I think the anecdotes are informative, but it would be

6nice if there was a really strong body of literature on

7the dynamic effects of delayed innovation, so on and so

8forth.

9I would also add, though, just for balance, more

10emphasis on the self-correcting nature of markets. All

11the time you are listening to these firms, they are all

12dinosaurs, right? Look what's happened to them all.

13Mike mentioned U.S. Steel. Look what happened to them,

14right? And it was foreign competition, the mini-mills,

15right? I mean, look at the auto companies, you know,

16look at Ford, GM, and it's amazing. You know, go on

17down the line. Now, this does take time, but I think,

18you know, it's important to keep in mind the

19self-correcting nature of markets in all of this.

20Along with that, then, is the parallel of

21government failure. Now, there are parallels of all the

22policies we're talking about. Antitrust is not made in

23a vacuum. Everything that you're talking about

24intersects a lot of major policies. Trade protection,

25for example, right? You know, more often than not we

59

1hear about, well, we need more competition in the

2airline industry. Yes, let's allow cabin -- oh, no, we

3are not going to do that. So, here you have trade

4policy effectively working against what antitrust policy

5is trying to do.

6We talk about technology policy with no mention

7of what happened in the early 1980s with the change in

8the patent law, right? Patents are going up now,

9lawsuits are going up now, you know, talk about, you

10know, impact on innovation and technical changes, look

11what's done in technology policy. That's not antitrust

12policy, but it's the crazy patent system that we've got

13now with, you know, the change in the '82 Act.

14Regulatory policy, Luke's point was fair enough

15about cable behavior, but again, it's a regulatory

16policy that's facilitating that, you know, the whole

17communications regulatory policy is screwed up. Again,

18this is not antitrust's, you know, cross to bear, but to

19some extent, it is. So, where you have a policy that is

20constantly at cross-purposes with other areas of what

21the Government is trying to do, it is going to make it

22very difficult for you to figure out to do, but I might

23add, the first best thing to do would be to have a

24technology policy, regulatory policy and trade policy

25that makes some sense, okay?

60

1Government successes, you know, I think the key

2thing on the government successes is almost more of the

3learning rather than the status assessments. You know,

4Standard Oil was interpreted as a success, and let me

5just suggest that there is some controversy about that,

6Bob Crandall and I head our exploration on antitrust

7policy, and you know, our look at what the

8counterfactual evidence was that, you know, there was

9very little that we could see from changes in prices, if

10one wanted to use that as a measure of welfare, and it

11is certainly not a reasonable starting point for what

12Standard Oil did.

13I think the more attractive thing that I would

14point to about antitrust is the learning just how one

15thinks about problems in terms of anticompetitive --

16what was initially thought of as sort of knee-jerk

17anticompetitive reaction as to whether these things were

18really efficiency-enhancing types of behavior and also

19just the nature of dynamics, how things are changed, and

20I think that's where antitrust policy has gone and is

21certainly a lot better.

22Now, the big thing about all of this and my

23concern about this whole area is the effectiveness of

24this evidence accumulated, because that's what you

25really want. In certain areas, just to go to a

61

1completely different area, you know, one's seen study

2after study about congestion policy in this country,

3every one of them, huge welfare losses, the Government

4ought to have efficient pricing, and no one is really

5disagreeing with that. There are obviously variations

6from here to there, but the evidence really builds

7beautifully, and you can just sort of drop it on

8somebody's lap and say, okay, look, deal with this, and

9it's easy to do that.

10Here, it is quite hard. I mean, yes, there are

11fragments of evidence, cases here and there, and as I

12said, what Crandall and I attempted to do was actually

13get a base case for a starting point of saying that

14this -- and if you disagree with that, fair enough, but

15at least build on that, reshape it, and then start

16adding more, and frankly, the disappointment has been,

17at least in the reaction to that paper, is, you know, I

18could -- is predictable either pro or critical antitrust

19people reacting to it, but in terms of actually new

20evidence being added to the enterprise, that just

21doesn't seem to be what idle people care about these

22days, which leads to my third concern, where is the

23empirical IO field going? And there was very little

24mention of that here, and with good reason.

25I mean, it is not clear where it is going in

62

1relationship to your interest in what is going on here.

2I mean, my sense, as I would say more of an observer

3than a participant, that empirical IO is sort of trying

4to get "uber" dynamic model of industry behavior, you

5know, that's what we're looking for, for the -- what's

6the word -- the Holy Grail, I guess that's because I saw

7The Da Vinci Code on the plane. That's what we are

8trying to do, and to the extent there's empirical work,

9it's pretty much demonstration papers, right?

10I mean, a lot of them are really pretty trivial,

11you know, you can get data on it -- and I won't go into

12examples, but you know what I'm talking about, and you

13know, who cares? And they don't care. They just want

14to show, yeah, I can get something estimated with some

15generalized method of moments estimator and add some

16structural stuff and something is going to get there,

17and yeah, I'll talk about an industry, about some hotel

18off a Nebraska highway, no one cares, but you know, the

19results actually made sense.

20The question is, where is this research going?

21Now, I don't want to rule this out, because this is a

22big ticket item. If people can succeed -- and this is I

23think really the positive spin on it -- in really

24building, you know, a structural dynamic model of an

25 evolution -- structural dynamic model of the evolution

63

1of industry, to hell with these case studies. You have

2got your tool, right? You just use this, run through

3any policy scenario, and you could figure out, you know,

4where things are going, what you ought to be doing, and

5that is your guidance.

6Well, you know, we've tried that with Keynesian

7models (ph), we have tried that with rational

8expectations, we have tried that with real business

9cycles, you know, in a sense it's a parallel to macro

10that we are really going to figure out in a big picture

11way analytically how markets behave, industries behave,

12and that will be your guidance for policy.

13So, you know, that's where it's going. It's not

14intersecting I think small case studies will build up,

15it is not doing thing in terms of big picture facts,

16even motivating what's going on, what people view to

17within industry seems to be more the availability of

18data and possible consistency with the analytics they

19want to pursue.

20All right, so, you know, where does that leave

21us? Well, you know, there are three ways to go, and to

22some extent you can pursue them simultaneously, you

23know, you can think about first looking more what the IO

24field is doing, the general model, that kind of work, or

25I would say more constructively try to focus that kind

64

1of work on the types of problems that you are interested

2in.

3The case evidence, I guess, you know, my concern

4there is just whether it's accumulating, is it likely to

5accumulate, because otherwise it won't be all that

6helpful. You will continue to just have patches of

7evidence that just don't seem to bind together to tell

8you anything in general.

9My interest is really going back to the first

10one, which was abandoned, and probably for good reason,

11is getting broad summary measures -- welfare measures of

12industries, conservation measures is obviously one, and

13work on quantifying the welfare loss from monopoly --

14and that line of research obviously had its problems --

15but there was a start of work I remember by Bobby

16Willig, Dansby and Willig on trying to come up with

17industry performance measures that I thought was

18promising, but I think it went out very quickly as

19people turned over to conduct, and so that work never

20went anywhere.

21But I think that it might be useful to think, at

22least in some way, along those lines for this agency.

23There are broad ways of gauging industry performance,

24you know, is there really something systematically wrong

25with what is going on with U.S. industry? Are we seeing

65

1anything that is now, you know, sort of really

2threatening a $13 trillion economy, or, okay, there are

3some bad guys, we know that, every once in a while

4certain things are going to go on, but the truth is

5markets are self-correcting, the world is getting more

6competitive all the time, you know, what do we have to

7do?

8I would not say at this point we're ready to say

9where to go. I would just sort of step back and reflect

10on various approaches and see what makes the most sense.

11(Applause.)

12MR. HEYER: Okay, we are about to take our

13break. We are going to be joined afterwards, there will

14be some remarks and discussion involving two of the

15 other panelists, Dave Reitman and Bob Marshall. I would

16encourage people to think during the break about maybe

17picking up a little bit on what Cliff ended with some

18and other comments that were made about, say, the issue

19of empirical anecdotes and what can be generalized from

20them or not, should we be focusing more on case-by-case

21analyses, or is there some kind of broader policy

22guidance we can learn from the empirical work?

23Anyway, let's take our break, and we will come

24back -- what, 15 minutes?

25COMMISSIONER KOVACIC: About 15 minutes.

66

1MR. HEYER: Fifteen minutes, all right.

2(A brief recess was taken.)

3MR. HEYER: Okay, so let's resume.

4The way we thought we would do it is Dave

5Reitman and Bob Marshall are going to give short

6presentations before we get into what hopefully will

7begin with a round table discussion where maybe some of

8the panelists and the discussants will comment on what

9went on this morning and respond to one another,

10elaborate on one another's comments, and then if we run

11out of things to talk about, Bill and I will have a lot

12of important questions as well.

13So, we will begin with Dave Reitman. Usually

14when people introduce others they say, "It's a pleasure

15to introduce so and so," even if they don't know them

16from a bar of soap. Dave is a pleasure for me to

17introduce because I know him very well, and he is

18relatively soft-spoken but incredibly talented

19economist, and he has one other thing that makes him a

20particularly valuable addition to this panel, I think,

21is that unlike most of us who have done a lot of maybe

22talking and thinking about some of the issues that are

23raised by the topic, Dave has worked in the trenches on

24them.

25He was the Government's expert witness in U.S.

67

1v. Dentsply and did an extraordinary amount of both

2theoretical and empirical work on that case in the

3course of testifying, and he also did a great deal of

4empirical work in support of our experts in the American

5Airlines case, which, sadly, never actually got to

6trial, but I'd be interested in Dave's comments both

7general and specific on these issues.

8Dave?

9DR. REITMAN: Thanks, Ken.

10As Ken suggested, I just want to give a few

11comments today as an antitrust practitioner about the

12value of empirical tools, empirical work, in presenting

13an antitrust case. It's really become clear listening

14to the panel this morning that in doing a case, often we

15are really talking about exceptions, that even if you're

16convinced that exclusive dealing 90 percent of the time

17or 99 percent of the time is beneficial, leads to lower

18prices and some of the things Luke had in his slides,

19still we're looking for the exceptions at the time when

20it's used as a deterrent device or an exclusionary

21device, and so the question is, what kinds of tools can

22you bring to bear when you are looking at a specific

23firm in a specific industry and a specific practice?

24Again, as Ken said, my background, my tenure at

25the DOJ, I was involved in two extremely lengthy

68

1litigated Section 2 cases, and both of them involved a

2fair bit of empirical work. American Airlines, I really

3think there was a tremendous amount of empirical support

4for a variety of elements of the case, and then

5Dentsply, the Government ended up commissioning a survey

6to try to measure some of the effects that were going on

7in that market.

8Now, if you look just at those two cases, you

9have to say that neither of those was a great

10testimonial as to the value of empirical work actually

11going forward and presenting the case. In American, as

12I said, there was all this empirical evidence brought to

13bear, and yet the case never made it past the summary

14judgment phase. In Dentsply, the survey was presented

15and the analysis based on it was presented at the

16District Court level. The District Court Judge threw

17out the survey as being unreliable and decided against

18the Government. Then the case was appealed to the Third

19Circuit, which without the benefit of the empirical

20evidence, was nevertheless able to reverse the decision

21and decide in favor of the Government.

22So, you might look at that and say, it doesn't

23seem like the empirical evidence contributed much.

24There are other cases along those lines that you could

25point to in recent years where you would say it's not

69

1clear that you really need to have the empirical pieces

2in there. So, just to give one more example, if you

3look at the LePage's case, where a lot of the

4commentators looking at that have said, it really would

5be nice if we had more evidence here, more data, so we

6could decide between these competing theories on whether

7this is procompetitive or anticompetitive. The

8Solicitor General on the cert petition before the

9Supreme Court really echoed the same things, we really

10would just like more information, and yet the plaintiff

11was able to present that case and win it without having

12done the kinds of empirical things that the commentators

13would have liked.

14So, I'd like to just spend a few minutes looking

15at the American case and the Dentsply case and talk

16about what really is the value of going through and

17doing the empirical exercise, and it may be just by the

18magic of self-selection that in this room we're kind of

19preaching to the choir, but nevertheless...

20Let's start with the American Airlines case.

21The airline industry is one where companies involved

22collect a lot of data themselves and the Government

23collects a lot of data. So, there's a tremendous amount

24of data that's been a mainstay of the empirical IO

25literature, and so it's only natural that a

70

1monopolization case involving the airline industry would

2have a lot of empirical work in it.

3The Government's main expert in this case, Steve

4Berry, is a preeminent empirical IO economist, and he

5brought, as I said, empirical evidence on virtually

6every point made, and a lot of that is not in the public

7record, as there was no trial, but just to give a sense

8of the scope of the empirical effort, you may recall

9that what turned out to be the Government's main test

10for predation when the case went up for appeal was what

11was called Test 4, which suggests that there are at

12least three and maybe a lot of other tests that

13economists turn to to try to find the right way to take

14the data and to sort it out and to say this is the right

15way to classify what is predatory and what is not.

16So, what, again, is the value of having that

17empirical test for predation? And to answer that, let

18me just go back a little bit farther in time. Not long

19after I started at the Justice Department, Joel Klein

20came aboard as Deputy Assistant Attorney General, and he

21was making the rounds to the different sections to

22introduce himself, and when he came to EAG, one thing I

23remember from his presentation was he quoted from "The

24Four Quartets" by T.S. Eliot, and he quoted, "We shall

25not cease from exploration, and the end of all of our

71

1exploring will be to arrive where we started and know

2the place for the first time."

3I actually have no idea at this point what

4Joel's point was for quoting that, but it does seem to

5apply nicely to the American case. The theory of what

6happened, the basic story never changed from the very

7beginning, before the complaint was filed, which was

8American added a bunch of flights and routes where it

9competed against low cost carriers and drove them out of

10the market, but the understanding of the way that

11mechanism worked, really why it worked and what it was,

12really only evolved by really years of wrestling with

13the data and trying to get a handle on what was going

14on, and so the end, when we looked at sort of the final

15presentations and the appellate memos, we said that the

16Justice Department really seemed to know what they were

17talking about and what they thought had happened, which

18was that American Airlines was able to, by adding

19flights, was able to take demand away from its competing

20low-cost carriers in a way that it simply couldn't do by

21lowering prices or by removing fare restrictions, but

22the cost of that was to reduce load factors and push

23American up to that increasing part of the marginal cost

24curve to the point where the incremental cost of adding

25these additional flights was above both the average cost

72

1of serving the route as a whole and also the incremental

2revenues received from the passengers.

3So, there's a test that, you know, when you

4arrive back at the place you started, you understand it,

5and I certainly don't want this panel to start to brew

6up a fight about whether that was a right theory or

7whether there really was harm there. The only point is

8that we really didn't understand what we were saying,

9what we had, until that process of wrestling with the

10data, really getting into it and being able to say, this

11is the test, which at least for this company in this

12industry in these markets is able to distinguish what

13looks like predatory behavior from all the other routes

14they had, which, you know, generated essentially no

15false positives.

16So, anyway, whether that's a legal analysis is

17for the courts to decide, but that was the value of the

18test there.

19If we could turn to the Dentsply case, which is

20sort of toward the other extreme in terms of the amount

21of data available, this is a market where exclusive

22dealing had been used for at least 15 years. Following

23the kinds of things Luke was saying earlier, we looked

24around for what we could use as a natural experiment,

25and one thing that may be a potential was to compare the

73

1policy in this country with other countries, but that

2was ruled out fairly early on by the Court. So, we were

3left with not a whole lot of empirical evidence to go

4on.

5To fill in the gap, what the Government

6commissioned was a survey of dental labs, which are the

7consumers of the dental teeth that were subject to

8exclusive dealing, and among other things, the survey

9asked respondents how they would choose among brands of

10teeth given various prices and distribution

11combinations, and so from those responses, you can then

12map out demand, service, and estimate or quantify what

13the anticompetitive effects were from the exclusive

14dealing policy both in terms of pricing and in terms of

15market shares, and that quantification was important.

16Dentsply has been characterized by some as, you

17know, as an easy case, or as in Luke's slide this

18morning, it's one where the aggressive behavior was bad

19in the proximate term and bad in the distant term,

20right? But the only reason we're able to say it was bad

21all around is because the District Court ruled that the

22procompetitive explanation and justification that

23Dentsply put forward was pretextual.

24If you look at the case before the decision,

25before the trial, before even the decision to bring the

74

1case, it's not at all implausible that exclusive dealing

2would have some advantages in aligning the incentives of

3Dentsply with its dealers and that that would generate

4some benefits. You may recall the particular mechanism

5that Dentsply eventually put forward seemed to be

6inconsistent with the facts, and so given how long

7exclusive dealing had been in the market, it was tough

8to be able to say how much competition would benefit by

9removing the restrictions on dealers, or to say that the

10benefit from eliminating competition or eliminating the

11restriction would be larger than these amorphous

12benefits from aligned incentives without some sort of

13systematic study of customer preferences.

14As it turned out in the case, of course, the

15weighing -- it turned out -- it proved to be easy,

16because we could sort of rule out procompetitive

17benefits, but more generally, looking forward, there's

18almost always going to be this kind of possible

19trade-offs between the procompetitive and

20anticompetitive story, and some quantification is vital

21in determining that effect.

22So, that leads to a third benefit of empirical

23analysis in looking at these kinds of monopolist

24practices, which is just in terms of lending conviction

25about understanding what really happened or what we

75

1think is happening in that particular market. We could

2talk about this both in the context of American and

3Dentsply, but I am going to stick to Dentsply, because

4as Ken said, I was a testifying expert in this case, and

5I suppose as a testifier, there is not a huge difference

6between saying what could have been happening in a

7market and what did happen. In both cases, the

8disparate evidence you gather from different sources and

9try to piece it together in unified whole, which gives

10you the best plausible explanation of what was going on

11in the market, but at least for me, it made a great deal

12of difference in crossing over from could have happened

13to it did happen to be able to actually see that effect

14quantified in the survey data.

15That is to say, my conviction that Dentsply's

16dealer criterion had actually harmed competition was

17crystallized just by being able to see it in the numbers

18after analyzing the consumer preferences that came out

19of the survey that had been commissioned, and it

20crystallized it in a way that I wouldn't have been able

21to achieve just by looking at documents and depositions

22and all the other evidence, even though all of that

23other stuff was consistent with the same conclusion.

24Now, of course, the lessons we drew from the

25survey were not uncontested and will never be

76

1uncontested in this manner of case, and the level of

2conviction didn't seem to make much difference to the

3District Court, since they concluded that the survey

4itself was unreliable, but I do have to believe that the

5whole testimony was made stronger by having conviction

6about key parts of it that were reinforced by the survey

7and that empirical evidence contributed a great deal to

8that sense of conviction.

9So, that's really all I wanted to say as sort of

10a little ode to the value of empirical research in these

11cases. Hopefully, not a eulogy, I don't think it's a

12eulogy, but there's value in knowing what you have,

13value in having confidence in that, and then just being

14able to quantify how much difference it makes in

15competition, and those things are not always going to

16carry the day, like they didn't in these two cases, but

17they are nevertheless important to preserve for future

18cases.

19 Thanks.

20(Applause.)

21COMMISSIONER KOVACIC: Thank you, David.

22Our last presenter before we turn to a

23discussion is Bob Marshall, who heads the economics

24department at Penn State and co-directs ITS Center For

25the Study of Auctions, Procurements and Competition

77

1Policy. Bob's on leave this year. He's serving during

2that time as a partner at Bates White.

3Our interest in asking Bob to come today, again,

4is related to a major strain of his own research. He

5frequently has married both empirical work and theory, a

6great deal of it dealing with auctions, procurement and

7collusion. Bob's going to tell us a bit about lessons

8that might be derived from that body of work for

9dominant firm behavior.

10Bob.

11DR. MARSHALL: Thank you, Bill. If you got too

12flowery, I knew that means you would be late with some

13of the things you owe me as a co-author, so it's good to

14hear that it didn't get out of hand. I am going to give

15a brief overview and then I will get into some of the

16slides.

17So, I do a lot of thinking about cartels and

18cartel behavior, so I understand Section 2 is not about

19cartels, but a cartel is like, I would argue in many

20cases, a single dominant firm, and cartels often go

21beyond just the suppression of interfirm rivalry in

22their actions. In fact, I am going to show you a number

23 of things where they go into behaviors that we would

24think about as Section 2 violations. So, what we are

25going to try to do here is tell a compelling story that

78

1we can get some window into understanding Section 2

2through the behavior of cartels, and hopefully there's

3some additional tractability in terms of empirical

4analysis that comes from that. So, that's the gist.

5So, there's some fundamental difficulties of

6Section 2 analysis. So, benchmarks are real important

7in terms of doing analyses particularly of cartel

8behavior. We like to think we have got a period of

9time, for example, when firms are acting in a

10noncollusive manner, and then we can look at this other

11time period of alleged conduct to see what's going on.

12With ongoing dominant firm behavior, that's often not

13there, and that creates some difficulties with doing

14Section 2 type analyses.

15Then there's an issue of what is legal and what

16is not for a dominant firm, and that usually doesn't

17arise in the analysis of cartels. When a cartel

18suppresses interfirm rivalry and then it goes off and

19predates and then it goes off and engages in exclusive

20dealing, no one calls us to say, "Well, I wonder if that

21predation was really predation or if the exclusive

22dealing was really exclusive dealing of an

23anticompetitive nature." The fundamental premise that

24cartels function under when they get together to

25suppress interfirm rivalry is to suppress competition.

79

1So, when they engage in these behaviors, it's somewhat

2doubtful to think that they're thinking about some

3social good that is not about suppressing competition.

4So, I have already explained that we can think

5of a cartel as being something like a single dominant

6firm, and they can be highly heterogenous. Some are

7struggling to maintain internal cohesion and stability.

8Defections might be occurring; finding a mechanism that

9works may be difficult. For others, those things might

10be easy to attain and settle in very quickly. The

11central goal is the elevation of prices and profits, but

12then we see these other behaviors that start to merge,

13and I will go through examples, predation, blocking of

14entry, exclusive dealing, bundling, tying. Again, part

15of cartel behavior.

16So, there's some interesting empirical questions

17that are immediately posing themselves here. Why do

18some cartels engage in these Section 2 like violations

19but others don't? And what's the advantage of looking

20at this through the lens of cartels? Well, there is a

21rich discovery record typically in place for some

22cartels because they got busted, and because a lot of

23them got busted, it means that we're able to look at

24starting dates, ending dates, and we're able to say, Oh,

25okay, so this is when the behavior began; this is when

80

1it ended. This is when the antirivalry behavior began;

2this is when it ended. This is when the monopolization

3behavior began; this is when it ended.

4Now, you may say, well, perhaps those things are

5coincident and difficult to separate, the antirivalry

6behavior and the Section 2 behavior. A lot of times

7what we will see as we look through some of these cases

8that I'll pose here is that the anti-rivalry behavior is

9the first thing that happens. You have got to get that

10set up first when there's running of a cartel. It's

11then later, as the cartel reaches some maturity, that it

12starts to investigate other sources of profit, and

13that's where we get to the Section 2 violations.

14I do this when I teach my "Economics and

15Collusion" course at Penn State. These are Porter's

16Five Forces. Now, in business school, this is basically

17Business School 101, so let me explain why I put this

18diagram up and what it is. These are the five forces of

19competition that affect a firm's profits. So, this is

20from Michael Porter's competitive strategy book.

21In the middle of this diagram is interfirm

22rivalry. For some reason I have been told not to refer

23to that as the green zone, but in the green is the

24interfirm rivalry, okay? So, this is whatever it may

25 be, differentiated product/price competition, whatever

81

1this may be that's limiting profitability among the

2competitors in the industry.

3Now, what are these other four forces on the

4perimeter? Well, at the top we have threat of new

5entry; on the right, bargaining power of buyers; down

6below, whether the goods produced by the firms in the

7industry have substitutes or compliments; and on the

8left, the bargaining power of suppliers. So, if we have

9a lot of substitutability, we have a lot of entry

10possibility, et cetera, well, profits are going to get

11hurt by that, and if we don't have those things, profits

12will be helped.

13So, I would argue the following: Cartels at

14their initiation work on the green zone, they are

15limiting interfirm rivalry. That's the Section 1

16violation. Once they get that nailed down, they then

17often venture out into the blue zone. So, blue is

18Section 2; green is Section 1. That's the way I view

19that diagram.

20So, I want to talk about some examples here, and

21this is all based, by the way, on a co-authored paper

22with my co-author Randy Heeb and Leslie Marks (ph),

23who's at Duke University, and Randy is at the Bates

24White office here. So, what are the examples of

25monopolization behavior from recent cartel cases? So, I

82

1am going to give you five cases, four listed here and I

2will read another one, and that's not a recent one. I

3had to go back to Stocking and Watkins and pick up

4another example from there.

5But let's start with citric acid. So, this is

6vitamins in training is a way you could view citric

7acid. The guy who ran citric acid was promoted to run

8the vitamins cartel. So, this is an important cartel in

9the history of Section 1 violations. And, of course,

10what they're trying to do, these firms, is suppress

11interfirm rivalry. This is a section from the European

12Commission decision regarding what part of the action,

13part of the conduct of the citric acid cartel. So, they

14were very bothered by entry by Chinese manufacturers,

15particularly into the European community, so those

16customers who were buying from the Chinese were

17targeted, and there were specific predation against the

18Chinese targeted at those customers. They were going to

19undercut those customers, and this list of customers was

20referred to as the Serbian list, and then there was

21frequent discussions that went on about how that

22predation activity was progressing.

23Now, when you read stuff like this in European

24Commission decisions, it becomes very clear very quickly

25it's not just about the suppression of rivalry amongst

83

1themselves. Once they have got that nailed down, as

2members of the cartel, they start to reach out into

3other mechanisms that they could use to increase

4 profitability.

5Carbon brushes, this is also a story about

6predation, and I'll just go to the next slide quickly

7and show you a particular example on German

8reunification. There was an East German company, EKL,

9and there was a pesky little noncartel firm, and so two

10strategies were agreed. None of the members of the

11cartel would supply any graphite to EKL, that's the

12basic raw material in making a carbon brush, the block,

13carbon block, and EKL would be denied any market share

14by systematically undercutting it with all customers, so

15that it would not be able to sell anywhere. EKL was

16taken over by one of the cartel members in 1997. Again,

17targeted predation at a noncartel firm.

18Now, keep in mind, again, this is a cartel that

19 begins and ends. This predation begins in '92, well

20 predating the beginning of the cartel behavior. So, we

21have got the antirivalry behavior, that gets

22established, that gets set in place, then the

23monopolization behavior begins, okay?

24Then there is also things like standardization.

25The cartel implements a ban on advertising, not to

84

1advertise or participate in sales exhibitions.

2In vitamins, agreed-upon elimination of

3competitors, and in this case, we're buying out

4competitors, Coors, that's the folks who make beer, and

5we're -- the two major cartel members here, Roche and

6BASF, are racking up the purchase price in proportion to

7their market shares.

8The European Commission goes on to talk about

9the use of the bundling of the basic vitamins into

10premixes as another mechanism by which the cartel

11predated against downstream blenders, so you have to

12look -- you have to understand a little bit of what

13happens here.

14Hogs and chickens and cattle get fed a premix of

15vitamins, and there were groups in the marketplace who

16would actually mix the vitamins together and sell the

17premixes to be added to the feed, and so to eliminate

18those pesky competitors in the downstream market, strong

19actions were taken by Roche and BASF to drive them out.

20The European Commission notes in particular, if

21you go to the second bullet here, it says, "In

22addition," referring to Roche and BASF, "they enjoyed

23greater flexibility to structure prices, promotions and

24discounts and had a much greater potential for tying."

25Again, we are not talking about just the suppression of

85

1interfirm rivalry here. We are well into Section 2

2violations now.

3Sorbates, we're talking here about -- this is

4another European Commission decision -- the blocking of

5entry to the marketplace. And then I went back and just

6pulled something from Stocking and Watkins regarding

7General Electric and the incandescent electric lamp

8cartel. Together with other lamp manufacturers, it made

9exclusive contracts with the manufacturers of

10lamp-making machinery and in bulbs and tubing, binding

11them to sell goods exclusively to General Electric and

12the companies associated with it or to sell to competing

13companies only at discriminatory prices. So, this is

14part of the action of the cartel.

15So, let me just as an aside say, standing issues

16about cartels are confusion to me at this point.

17Noncartel firms don't have standing because they are

18always the beneficiaries of cartel behavior. That seems

19a bit odd to me just an aside here given the fact that

20these Section 2 violations are existing, well documented

21in the record, with regard to the noncartel firms, but

22that's just an aside.

23I would just like to say that I think that this

24is a rich avenue for potential empirical investigation,

25again, because we have got clear benchmarks in place.

86

1We can also get a clear look at the discovery record

2associated with cartel behavior and start to see when

3these kind of behaviors, the Section 2 violations, are

4implemented by the cartels, look across industries,

5cartels in different industries, and see who was doing

6these kind of activities, which industries are not

7engaged in those kind of activities.

8I'm hopeful that this illuminates as a potential

9or at least gets investigated as a potential some of

10these ambiguities that have existed in the past with

11just looking at single dominant firms as being the

12source of data and empirical inference.

13 COMMISSIONER KOVACIC: Thank you, Bob.

14(Applause.)

15COMMISSIONER KOVACIC: Before we have the more

16open-ended discussion among all the panelists, I'd like

17to give our first four presenters an opportunity simply

18to comment on what took place or to add additional

19thoughts that came to mind. Could I simply go through

20the order again, go with Mike, Luke, Jon and Cliff?

21Mike?

22DR. SCHERER: Well, lots of things I found

23stimulating, so I'll have to be very, very selective.

24I think the thing that struck me most was

25Cliff's distinction between the European Union and the

87

1United States. There are two points I'd like to make

2there. One is a puzzlement; one I think I understand.

3It's been said by several of the panelists that

4the European Union has been more aggressive in some

5sense towards dominant firms. They have tended to

6pursue an abuse of dominance standard, whereas our

7approach has been mainly structural combined with some

8elements of conduct.

9On the other hand, the Europeans have been

10severely limited because when they tried to go against

11abuse, as in, for example, the Hoffmann-La Roche Valium

12case and the Volkswagen case, they ran into big troubles

13ascertaining what an abusively high price was or an

14abusively high level of profits was, and in this sense,

15 they are going back to the caveats that Judge Taft

16expressed in the Addyston Pipe case more than a century

17ago, but I think there's something else going on.

18I think the ghost of Friedrich Hayek haunts the

19Europeans in the sense that Hayek argues that you simply

20cannot tell what an abusive price is. The European

21community ran into this squarely in Microsoft. They

22were unwilling -- at least initially, they realized in

23the end they had to -- but they were unwilling initially

24to state the fees that Microsoft could command for

25licenses to its intraoperability information. And even

88

1more seriously, when they required the provision by

2Microsoft of an unbundled version of Windows without the

3media player, they allowed Microsoft to sell both

4products at an identical price. The obvious thing to do

5would have been to set a price differential, but they

6refrained and have continued to refrain from doing this,

7and therefore, virtually no one has taken the unbundled

8version when you could get a more complete version.

9The Europeans have a serious problem. When you

10look at our past compulsory licensing cases, you see we

11were much more willing to intervene and said, "Here's

12the reasonable royalty that you can command."

13Now, the other thing about the Europeans is

14this: Beginning with a conference at Fontainebleau in

151965 and then the book by Jean-Jacques Servan-Schreiber

16and then another conference in Germany in 1976, and God

17knows what else, the Europeans have adopted the policy

18of encouraging large dominant national champion

19enterprises with the express purpose of competing with

20the United States technologically. In most respects,

21 they have failed.

22In most areas of modern technology, they have

23lagged the United States, and partly I think because we,

24on the one hand, following the sage advice of Chairman

25Mao, have encouraged 100 flowers to bloom. The

89

1Europeans have tried to cultivate their national

2champions, and they just didn't have the diversity

3required to achieve technological innovations. The big

4exception was in a couple of high-scale economy

5industries. One is the provision of nuclear power

6plants, and the other is the provision of aircraft,

7although they are having trouble there now, too, but for

8a while, Airbus was doing very, very well.

9I think there really are important lessons to be

10learned here, and they need to be studied much more

11carefully than they have been thus far.

12A point that Luke made, and I think Bill Kovacic

13made it, too, and it is very, very important, that we

14should be doing follow-up studies on areas in which we

15have intervened. We did this, among others, in Xerox.

16The FTC specifically commissioned a study by Tim

17Bresnahan of the results of the Xerox case, which found

18that it had been quite beneficial. Xerox did its own

19study by David Kearns in a book entitled Prophets in the

20Dark. It found that the entry of Japanese competition,

21which was facilitated by the FTC intervention, had a

22remarkably salutary effect on prices, reliability and

23technical change in the copying machine industry.

24Let me end with one footnote on the marginal

25paper, vitamins. I happened to be a consultant for

90

1Eisai in the vitamin E case. One should not look into

2these things without taking into account international

3trade rules and how they shape the framework within

4which international agreements appear. Specifically, in

5the case of Eisai, Eisai was a newcomer to the vitamin E

6market. They began entering the U.S. and European

7markets, and the chairman of Eisai was called into a

8meeting by the head of Hoffmann-La Roche's vitamins

9operation and was told, I quote exactly, "If you yellow

10bastards don't join our cartel, we will drive you out of

11both the U.S. market and the European market with

12antidumping suits."

13What happened after then is very complex, but

14there remains in my mind at least a puzzle. I couldn't

15find any change in Eisai's pricing behavior after they

16allegedly joined the cartel. The one thing observable

17that changed is that they began shipping more of their

18output to China and they began dumping their excess

19output in China. Why, I don't know, whether it was

20because China was growing rapidly or that was a cartel

21facilitating device, I do not know. There are

22interesting stories here to be explored.

23Thank you.

24COMMISSIONER KOVACIC: Thank you, Mike.

25Luke?

91

1DR. FROEB: Thanks. I just want to say a couple

2of things.

3First of all, to talk about Jon Baker comments

4about how to balance the good proximate effects against

5the bad distant ones, and there's two ways to do that,

6you know, empirically or use some kind of model,

7theoretical model that helps you do that, and if you

8kind of contrast the way we balance horizontal, you

9know, efficiencies against unilateral effects, we have

10well-developed models that allow us to make the

11trade-off. I just don't know of any well-developed

12models that would allow us to make those kinds of

13trade-offs, and furthermore, if we held our prosecutions

14of these Section 2 cases to the same levels or same

15standards that we did our merger cases, I mean, I think

16it would be very difficult to bring good cases in those

17instances.

18I want to talk a little bit about what Cliff

19Winston said about where is the empirical literature

20going. In economics, young IO economists demonstrate

21their technological expertise by building structural

22models and, you know, trying to estimate them, and they

23ignore, you know, trying to figure out, well, what's the

24effect of things like Wal-Mart entry, you know, what is

25Wal-Mart doing or what -- doing follow-up studies,

92

1because they seem so pedestrian, yeah, anybody can do

2that, you know, you just have to gather the evidence

3and, you know, control for competing factors, and so

4there's a natural bias in the economics literature

5favoring, you know, structural technical modeling, even

6when it's not appropriate, and we see that a lot. I

7think that is one reason for the dearth of good

8empirical evidence in industrial organization, because

9we have this fetish almost with structural modeling.

10I want to agree with what Cliff Winston said

11also about the real problem is, you know, empirically,

12you know, antitrust cleaning up trade, regulatory or

13lousy patent policy. I mean, when you look at the

14recent acts at the FTC bringing a lot of cases that

15wouldn't exist but for the people abusing the patent

16system, or I remember when I was back at the DOJ, we

17challenged a merger between Westinghouse and GE in

18electrical generators because Toshiba was out of the

19market because they had been selling machine equipment

20to the Russians to make submarines, so the Commerce

21Department said, "Hey, you can't bid on electrical

22generators in the United States," and that, you know,

23would have made the merger okay, but, you know, we

24blocked the merger because they were out of the market.

25I want to note that Dave Reitman's Dentsply

93

1case, he was able to estimate the proximate effect. He

2wasn't able to estimate the distant effect, which wasn't

3an issue in the trial because the judge said, "Hey,

4there's no possible, you know, beneficial effect of

5these exclusionary practices," but he was able to

6estimate the proximate effect, not the distant one, and

7I think the real challenge empirically is on these

8distant effects, these indirect strategic effects.

9I think that's all I want to say, and -- well, I

10guess I would say to Bob, when you see these vertical

11restraints in these cartels, I mean, suppose I form a

12cartel upstream and I buy some downstream or put the

13downstream guys out of business or refuse to deal with

14them, I mean, there are certainly procompetitive

15justifications for that given that you have a cartel.

16COMMISSIONER KOVACIC: Wally?

17DR. MULLIN: I would like to pick up on this

18interplay between economic research, whether done at the

19university or a think tank, and antitrust practice. So,

20I've neither done any antitrust cases nor have I

21estimated a discrete choice demand system. However, I

22guess you can imagine talking about developing clinical

23facts, which a judge or even an antitrust enforcement

24agency might think are too bound up in the particular

25circumstances to really be admissible.

94

1I mean, if you said, okay, here are three or

2four or five not tools but three or four or five, you

3know, examples favors saying there's predation, is that

4going to mean that the American case doesn't survive a

5summary judgment? I don't know. I would be doubtful.

6The argument I guess in favor of some sort of

7methodology, right, is that, yeah, if the tool works,

8then you can use it in lots of arenas. Operating very

9quickly, so it's not a Section 2 example, but my sense

10is that a lot of mergers involve firms that produce

11differentiated products. The state of the art circa

121975 on estimating those models was not great. Berry

13Levinsohn Pakes (BLP) offered a big methodological

14improvement. Previously, the profession knew there were

15problems with the standard approach. We just kicked it

16under the rug and BLP took on a very difficult problem.

17So, from their papers you can say, okay, well, I don't

18just know something more about the automobile industry,

19I can use this in other settings.

20I guess the question that I have heard others

21raise in other contexts in terms of the way the

22industrial organization field has gone in certain

23universities is whether -- maybe we did need to make

24progress on the demand side and now have a better sense

25of how to estimate demand, but we're industrial

95

1organization economists. We study also the supply side,

2at least at this point in the development of the

3literature, it cannot yet say okay, here are some tools

4in terms of supply that would allow you to make these

5sort of counterfactual predictions. For example, if

6this particular exclusive dealing isn't available, this

7is how the market will change and this is how firms will

8operate differently, which is a real cost of pursuing

9 models on motels in Nebraska or something like that.

10 COMMISSIONER KOVACIC: Thanks, Wally.

11Jon?

12DR. BAKER: Thanks, Bill, a couple quick things.

13First of all, I need to be a law professor for a

14moment. When Bob Marshall talked about Section 1 and

15Section 2, what he really is saying is a distinction

16between conduct that's collusive and exclusionary.

17Probably you would attack all of that conduct in the

18context of the cartel cases that Bob was referring to.

19The exclusionary conduct, you would probably attack it

20under Section 1 of the Sherman Act, not Section 2. But

21when we're talking about monopolization under Sherman

22 Act Section 2, typically the conduct is exclusionary,

23and so that's why Bob thinks it's instructive to look at

24 the exclusionary conduct for the cartels.

25I actually think there's a close connection

96

1between exclusionary conduct and collusive conduct,

2because you can think of exclusionary conduct as

3creating an involuntary cartel or a coerced cartel.

4Think about it this way: The dominant firm would like

5to collude with a fringe rival, a prospective entrant or

6whatever, but the rival doesn't go along, so the

7dominant firm has to force the fringe rival or

8prospective entrant to compete less aggressively, cut

9back on output, not expand, whatever it would require,

10and it does that with a panoply of exclusionary

11techniques, raising rivals' costs, reducing their access

12to the market or whatever, and the result is that

13industry output falls below the competitive level, not

14by voluntary agreement among the firms the way a cartel

15would, but essentially by coercing the maverick. It's

16an involuntary cartel; that is how I like to think of

17it. So, they are closely connected.

18My other comment on the conversation we have had

19here, have had today, is about the problems of assessing

20the "but-for" world. That was brought up I think by

21several people here, Bob and Wally and Mike I think all

22alluded to it, and probably everyone else did, too. To

23make this concrete, I started to think about the Intel

24case that the FTC brought in 1998, which was when I was

25bureau director. It was settled in 1999 I think after I

97

1had left, and it's the case that Mike was referring to

2where he was going to be the witness for the Federal

3Trade Commission.

4The basic idea was that Intel refused to deal

5with certain customers, cutting off their access to

6technical information about upcoming new microprocessor

7products that the customers needed if they were going to

8be able to design complimentary products like personal

9computers, and they did all this as a way of coercing

10the licensees -- or, I'm sorry -- yes, getting the

11rivals to license their microprocessor technology to

12Intel. That was the story that the Commission told, and

13the rivals included Digital Equipment Corporation or

14DEC, Intergraph and Compaq.

15So, Intel was trying to get leverage in

16unrelated commercial disputes involving the scope of

17competing intellectual property rights. The theory of

18the case was that what Intel did to cut off these

19customers from the technical information diminished the

20incentives of those three Intel customers, as well as

21all sorts of other firms that are similarly situated,

22whether they are Intel customers or they are otherwise

23dependent on Intel, to develop new innovations relating

24to microprocessor technology.

25Just to give Intel's side of the story, they

98

1defended by saying that the conduct alleged in the

2complaint didn't diminish the incentive of any firm to

3develop new innovations of any kind. So, that was the

4dispute.

5The case was settled with an agreement that

6prohibited Intel from -- I wrote it down here --

7impeding, altering, suspending, withdrawing, withholding

8or refusing to provide access by any microprocessor

9customer to -- oh, dear, I don't know what I wrote down

10here -- some sort of information for reasons related to

11intellectual property dispute with such customer -- et

12cetera -- or basing any supply decisions for

13general-purpose microprocessors upon the existence of an

14intellectual property dispute.

15So, the question is, all right, this case

16against a big firm, it was technically a Section 5 case,

17but it was basically a monopolization case, how do you

18tell whether the consent made any difference? That's

19the question I am trying to set up. The theory would

20have to be that this consent encouraged rivals to

21innovate in ways to take on Intel, and before they

22didn't have the incentive to do that, and maybe that

23makes sense.

24I think that the kind of markets you're talking

25about are winner-takes-most generally, and it's hard to

99

1believe that Intel wouldn't keep innovating in those

2markets even if you did something to make it easier for

3the rivals to innovate, too.

4But how do you prove or disprove that theory?

5We know that AMD, a key rival, has been successful in

6the last couple of years, but that doesn't settle the

7issue. What we have to do is somehow construct a "but-

8for" world and figure out how AMD would have done there.

9We don't know whether AMD's success has anything to do

10with this consent or not just from what I've recited as

11the facts.

12I guess what I am driven to, I'm not sure what

13we would do. I think the best we could practically do

14is probably use Section 6(b) of the Federal Trade

15Commission Act to review the R&D plans and the marketing

16plans of Intel and AMD and the other firms before and

17after the case, assuming all the documents are still

18available, and depose key executives and see if Intel

19and its rivals changed their strategies -- we could

20probably find that out -- changed how they thought about

21innovation, the kind of innovation they went after, what

22they would do with them and the like.

23The point of this exercise is that it shows how

24hard it is to construct the "but-for" world in any

25actual case in order to either figure out the violation

100

1in the first place, which was the point of some of my

2colleagues here, or to evaluate how well we did in

3bringing the case and remedying it.

4I don't view this as a reason not to bring

5cases, by the way, but I know that some people do.

6That's my comment.

7Go ahead, Cliff.

8DR. WINSTON: Just two brief things, and let me

9sort of shape them more toward ultimately, what advice

10do we give Bill and Ken? Presumably at the end, they

11will say, what should we do to make sense of all of

12this?

13You know, my comment on -- really about the

14method -- the IO methodology is just more of a caution

15about the difficulty of just focusing on, you know, can

16we pull studies together and amass, you know, a core of

17useful knowledge that way, and my caution was really

18historical.

19If we turn the pages back to the sixties, the

20 leading empirical enterprise of the day was basically

21concentration and profit progression. I mean, there are

22scores of those, and along with that was the policy

23issue of, you know, should we have a deconcentration

24policy in America as the focus for antitrust? And, you

25know, these studies evolved certainly from, you know,

101

1noneconometric approaches, contingency tables and the

2like, to more sophisticated econometric approaches, but

3ultimately the enterprise basically collapsed, obviously

4concerns of heterogeneity and concerns that, in the end,

5the concentrated industry is the good one, this is a

6good thing we should be having, and there's just none of

7that around at all, and no one even sort of looks at

8that for much guidance.

9Dick Schmalensee I remember in The Handbook of

10IO tried to summarize that and offered, you know, 20

11stylized facts that sort of stretches what you get out

12of it, and I'm concerned that, you know, in the sense

13the empirical IO we have got today may go in the same

14way for a somewhat different reason, but ultimately,

15there is a somewhat destructive nature of the

16enterprise. It's extremely competitive, and it's

17extremely easy to raise the stakes at every -- you'd be

18surprised.

19I mean, you know, at this point I would say BLP

20has done a brilliant job of market share capturing,

21nothing short of brilliant, among the best I have ever

22seen of intellectual importers, and people think

23naturally of, well, they have a nice demand system and

24so on and so forth, but I think you will see, as certain

25other papers come out, there are real cracks in even

102

1what they've got, you know, for every model for which

2you want to try to capture heterogeneity, you can point

3out why there are problems in the way they are doing it,

4and so almost every study can be replied with that as

5the methodology pushes harder and harder and harder and

6excludes more and more people and almost makes it

7virtually impossible to understand for a lot of people

8in practice.

9I'm just wondering where all of this ultimately

10is going to go and thinking, well, we can use this

11still, you know, the simplest thing is in courts, but we

12can't, because obviously the other side is going to come

13back and use more technical things and just smash what

14you do, and so I am concerned about ultimately where all

15this stuff is going to converge in a constructive way.

16You know, that said, then, you know, what then

17would I say to emphasize? And I think this has been

18touched on, but maybe not enough, and that is the

19deterrence aspects of antitrust policy. I mean,

20sometimes, you know, I am interpreted or at least my

21paper with Crandall was interpreted saying we ought to

22abolish antitrust intervention, and that's ridiculous,

23we never said it, and I certainly don't believe it, but

24the importance really of antitrust is in deterrence,

25and, of course, that's your success story, but it's also

103

1the most important and difficult thing to quantify.

2So, the challenge, I would suggest, at this

3point, where you could get help but certainly it's a

4challenge at this point, is trying to find the areas

5where there is evidence that we are clearly deterring

6other areas, but what for going after Microsoft, who

7would have known, all right, regardless of what people

8think on that case, you know, other things that may be

9done, and that may ultimately be the strength that a lot

10of people think of antitrust and certainly the thing

11that also needs to be emphasized and systematized, but

12at this point, obviously, that's eluded our ability to

13do that kind of thing.

14MR. HEYER: Well, I want to give at least -- if

15Dave and Bob want to say a couple of words. Otherwise,

16we can throw out some very insightful, stimulating

17questions.

18DR. REITMAN: Well, we could end up looping

19quite a bit here if we go round and round, but --

20DR. MARSHALL: Fire away.

21DR. REITMAN: Yeah.

22MR. HEYER: Well, you guys can respond first

23maybe.

24I had one question I alluded to at the end of

25the morning session that I wondered if everyone could

104

1comment on, sort of a general question about the value

2of individual anecdotes and studies, a number of which

3have already been discussed, as compared with or maybe

4related to what Cliff had referred to as the Holy Grail

5and what I know Luke, some of his work has suggested is

6broad policy guidance.

7I mean, to what extent do folks think we are

8able to learn enough from individual studies to base

9policy and priors on versus doing what, say, serious

10case-by-case analyses in determining the effects on an

11"as it comes in the door" kind of basis?

12Anyone? Professor Scherer? Luke?

13DR. FROEB: I think that the broad aggregate

14studies suffer from, you know, aggregation bias, and

15it's very difficult to draw inference from the large

16down to the small. I think it's much easier to go from

17the small to the large. And the studies that we've been

18doing at the FTC have shown that, say, for example, when

19you're using census data and industry-level studies,

20you're missing a whole lot that's going on at the

21individual level, and I think you ultimately learn a lot

22more by going as narrow and as case-specific as

23possible.

24COMMISSIONER KOVACIC: Mike?

25DR. SCHERER: I somewhat disagree. What's the

105

1value of anecdotes? As Zui Griliches used to say, "The

2plural of anecdote is data." The humor of that escaped

3you.

4DR. WINSTON: Wasn't it Stigler who said it?

5DR. SCHERER: Maybe he learned it from Stigler,

6I don't know.

7In any event, you have got to do all this stuff.

8You have got to do case studies. You have got to do

9data. You have got to integrate all the case studies.

10All of these things need to be done in order to get

11something like generalized knowledge.

12Well, I guess that's all I'll say on that.

13MR. HEYER: Jon?

14DR. BAKER: Well, my reaction to this and to

15some of the other comments here is that I think the

16economics literature has been a little bit -- I have a

17different perspective, shall I say, on the development

18of empirical IO, which is that one of the big movements

19has been away from cross-industry studies, which have

20all sorts of problems that people here have described,

21to individual industry studies, where you can learn

22about -- which effectively control for lots of the

23differences across the industries. There's been a lot

24of learning about individual industries.

25I'm just thinking of all the studies in Tim

106

1Bresnahan's IO Handbook chapter, Peter Reiss and Frank

2Wolak have a recent chapter that surveys a bunch of

3studies, too, and there is just a wealth of knowledge

4that -- the unit of observation in empirical IO has

5shifted from the economy as a whole, across all

6industries, to individual industries, and we've learned

7a lot. Even when those structure-conduct-performance

8studies are still done, they are all done largely on

9related industries, as with the Leonard Weiss book I'm

10thinking of from a while back.

11You can use what you learn about individual

12industries too, as I was saying before, to create

13presumptions about related industries that you can argue

14about what you know about retailing from retailing

15industries and how it works. I'm thinking of Dean

16Schmalensee's testimony in Microsoft. He was talking

17about how software markets have certain kinds of

18competition generally and that that observation probably

19applies to operating systems. Then the Government comes

20back and says, well, maybe that's an exception. The

21presumption frames the analysis appropriately.

22So, there's a lot you can do with individual

23industry studies to learn about related industries that

24I think we're undervaluing here.

25COMMISSIONER KOVACIC: David?

107

1DR. REITMAN: I just want to add that I think

2you have to recognize that Section 2 cases are just

3distinct from other kinds of antitrust cases in how

4unique the behaviors are from case to case. So, it's

5hard to generalize from, for example, our merger

6analysis, which has benefited greatly from being able to

7go back and forth between cases and theory and getting a

8body of theory, which can then identify the cases and

9the time.

10There is so much individuality to any particular

11set of bundled discounts, where a particular mechanism

12that a firm predates, it's hard to see that even

13generalizing from case studies or whatever is going to

14add a whole lot to the analysis of a particular case,

15even if it's necessary to some extent for the law. As

16far as the analysis goes of what's going on in a

17particular industry, I'm not sure how you can use that

18very well.

19COMMISSIONER KOVACIC: David, if I could follow

20up on that, as you reflect on your experience with the

21two cases you discussed, and if you were looking ahead

22to try to extract more general observations from those,

23is there something about an investigative methodology or

24an analytical approach that you might derive from those

25experiences?

108

1Suppose you were thinking at the time you left

2the Division about how to leave behind or to make more

3concrete know-how that you had extracted from your

4experience analyzing the cases and as a potential

5testifying expert. Are there specific lessons that you

6would have derived from those that you think would have

7informed the analysis that you would use in future

8cases?

9DR. REITMAN: Well, the clear one I think is

10from the Dentsply case, that the survey that we did

11there seems to be fairly rare, at least on this side of

12the Atlantic, although if you go across to England and

13Europe, it seems like it's fairly routine as part of a

14gathering of consumer information to do it

15systematically through a survey, and the survey really

16is just that, it's -- instead of interviewing a bunch of

17customers, it's a way of systematically getting a

18representative sample and asking the same sorts of

19questions in a way which could be quantitatively

20analyzed, and so I think that technique was helpful in

21Dentsply. It could be helpful in a lot of

22monopolization cases.

23COMMISSIONER KOVACIC: Do you have an impression

24about the arena in which, in many ways, so much of the

25information we're talking about ultimately has to be

109

1applied? Was the decision of the trial court simply to

2reject the empirical study that had been done? Is that

3just an outlier that we're going to encounter when we

4bring cases? Or is there something to be learned there

5about how to present evidence in a way that ensures that

6it doesn't simply die at the doorstep of a preliminary

7motion but makes its way into the resolution of the

8case?

9MR. HEYER: Objection, calls for a legal

10conclusion.

11DR. REITMAN: There are certainly things to be

12learned there about how to actually conduct the survey

13in order to be able to get through the hurdles of

14reliability that the Court needs and rightly should

15require. I don't think the analysis in the Court, at

16least in Dentsply, really went beyond that, and so I'm

17not sure what further lessons, but I do think you can

18get over that hurdle. There may be additional hurdles

19in terms of different sides looking at the same evidence

20and, you know, making different conclusions from it and

21the Court trying to figure out what to do with it and

22such that we will have to wrestle with later, but the

23first hurdle in terms of getting things admissible I

24think you can overcome.

25DR. WINSTON: I would just -- one thing, and you

110

1can probably enlighten me on it, the whole discussion is

2sort of taking place in a political vacuum, you know,

3it's like antitrust policy proceeds, you know, that we

4do the analysis right, find out what's going on and

5bring the case. I mean, obviously all this proceeds

6with a lot of political constraints and, you know,

7within your department, you know, how you want to frame

8the case, the kind of people you want to bring in, the

9cases you want to go after.

10I mean, I think all the things that Mike was

11saying I agree with completely, that you want to draw on

12as much evidence as possible, different sources,

13different people, but all of this is constrained by just

14political forces within and outside your agencies, and,

15you know, how you grapple with that ultimately may be as

16important as any of the analytical things that you

17solve.

18MR. HEYER: Do you want to take this one?

19COMMISSIONER KOVACIC: What forces would those

20be?

21One reason that the FTC's anniversaries are

22interesting to me is that my own appointment is tied to

23the 26th of September. As the sands go through the

24glass, I have five years before the appointment comes to

25an end. So, one question for me, given that I have

111

1perhaps a bit more influence in how decisions get made,

2is how the agency should invest its resources. One

3possibility that Mike referred to before, and it's

4implicit in the comments that all of you have made, is

5that one way to begin to use empirical methods to assess

6the appropriate course in future policy making is to

7examine past decisions to enforce or not to enforce.

8As Mike said before, my first assignment at the

9FTC in 1979 was to work with a young Assistant

10Professor, Tim Bresnahan, in the formulation of the

11Xerox study. I think in principle that any institution

12ought to go back and look at completed matters, and for

13purposes of some public discussion and revelation,

14should make the results of that process available.

15That's clearly a sensitive matter and I suppose

16political in this sense: How do you develop a norm or a

17standard that encourages ex post review in a way that

18does not raise suspicions that you're picking topics for

19study or examination simply to show up your predecessors

20or in some way to reinforce a predilection or set of

21preferences that you brought to the process?

22I think we could agree generally that there are

23tremendous methodological challenges in doing such

24studies well. I don't put those aside as being

25insignificant by any means. There would be a difficulty

112

1in implementation.

2My own preference would be that you would try to

3develop an internal norm that puts money in the budget

4every year to do that kind of work -- that is, that some

5of it be done every year, that there be an expectation

6such that outside observers would ask every year. "What

7matters are you going to look at this year? Which

8projects are you going to launch this year?"

9Second -- you can't model this in a formal way,

10this is simply a matter of leadership and choice --

11incumbent leadership would be willing to pick matters

12that could be sensitive to them. For myself, if I were

13to pick mergers, I would be quite happy to see in the

14relatively near future (that is, during my time here),

15an examination of the cruise lines decision. I was

16general counsel here when that transaction took place.

17The FTC and three other jurisdictions studied the cruise

18lines merger. I'd like to see if we got the answer

19right. I'd also be interested in taking other matters

20where we intervened and failed, Arch Coal being one.

21I'd also like to take up the possibility that Jon

22mentioned, that is at least with respect to the case

23study component of matters, that there always be an FTC

246(b) matter in progress; that is, that it always be part

25of the research agenda, perhaps with the possibility,

113

1again, of using it to examine somewhat more

2microscopically matters in which the agency intervened

3and did not intervene.

4To do that in a way that creates confidence that

5it is being done in a technically acceptable and

6even-handed manner requires a great deal of political

7skill and judgment. One needs to make sure that the

8evaluation process is perceived internally and

9externally as being a neutral, truth-seeking exercise

10rather than in some sense as a political exercise.

11That's one thing an agency can commit itself to do.

12The further question would be, what's the right

13forum? Should something be done intramurally? Should

14these be partnerships with academic institutions, or

15think tanks, such as the AEI-Brookings Joint Center on

16Regulation? Should it be done with specific centers of

17research within the university community? What are, in

18the language of international relationships, the

19modalities for doing this kind of work? How it should

20be conducted is another issue. To do it well and in a

21way that would be regarded as a neutral, truth-seeking

22exercise, as opposed to simply an effort to vindicate

23one's own judgments or to discredit the judgments of

24one's predecessors is politically a very delicate

25matter. It would also be a politically delicate matter

114

1to take one other matter we have mentioned here that's

2of keen interest to me -- to look at the question of how

3the antidumping system serves as the punishment

4mechanism for cartel coordination. To even begin to put

5a toe in the water in that kind of research work would

6require a great deal of care to see how warm the water

7was and to decide in what part of the pool you are going

8to step in first. As a general matter, I can't help but

9think that it's impossible to look at the question of

10cartel coordination at home and abroad without

11accounting for that.

12DR. FROEB: Based on the kind of studies we did,

13you can't learn something from every follow-up study,

14and I think it's really important to be opportunistic,

15and I think Mike made a study of the Appellate Court

16decision overturning the must carry laws provided a

17really nice natural experiment where we could learn

18something, and being opportunistic on something like

19that, it takes a lot of judgment about are we going to

20be able to learn anything from this? We've talked about

21the difficulties of counterfactuals, and I think you

22have got to be very careful about that.

23MR. HEYER: Let me raise another question for

24folks to talk about that was touched on earlier,

25particularly Professor Scherer got into it when talking

115

1about innovation and dominant firms.

2In trying empirically to get at some of this

3stuff, the effects of remedies, the performance of

4dominant firms, I was wondering if there's anything we

5can usefully do empirically having to do with more

6long-run issues, incentive issues for firms to become

7dominant or for firms to be acquired by dominant firms,

8perhaps? I think Professor Scherer had suggested

9that -- seemed to suggest, at least, and maybe I'm

10reading it wrong -- that maybe the harms from

11constraining some of the larger firms, at least in the

12innovation arena, might not be too great, might be worth

13it, you could get short-run benefits, long-run maybe as

14well, but we can't tell.

15I'm wondering if we know anything about long-run

16effects, whether anything empirically can be done in

17that area.

18DR. WINSTON: Well, there, whatever you do, you

19are going to have to interface the patent system just in

20general with technology policy in this country. In

21other words, you know, what you first want to start with

22is, you know, just positive economics, you know, how is

23it -- we understand innovation, which is obviously very

24important and a very difficult thing to do, and layered

25on top of that is going to be, you know, technology

116

1policy, and just it does have an influence on that.

2So, you know, whatever you are going ahead with,

3you just want to caution yourself that your answers are

4going to be shaped to a large extent by the

5institutional environment that exists in this country.

6DR. SCHERER: It should be done. It is really

7hard. Obviously the longer time frame you deal with,

8the more historical artifacts you have to factor in. I

9think the way you get around that is to look at a broad

10array of cases and try to see how did it work in one

11case and not work in another case.

12A really interesting one to study, I do not

13think it has been studied, is the United Shoe Machinery

14case. United was dominant in inventing and developing

15shoe machinery, but Judge Wyzanski found them guilty of

16monopolization around about 1955 or so. I happened to

17interview them in a quite unrelated context in 1958, and

18they said this was a case where we really had the wrong

19policy. Wyzanski said I'm not going to break them up

20now, and there were good reasons for not breaking them

21up, but I am going to leave the Sword of Damocles

22hanging over their heads. We will come back five years

23from now and see whether they ought to be divested.

24And so here's USM sitting there with this

25possible divestment if they don't get their market

117

1shares down in the future. So what did I find in 1958?

2They were saying, we're not putting our R&D into shoe

3machinery. We're putting it into diversification

4activities. And what then happened -- and again, it's a

5big fast-forward -- what happened eventually was that

6they became noncompetitive in the shoe machinery

7business. Italian firms, maybe they would have done so

8anyway, Italian firms became the leading suppliers of

9shoe machinery in the world, and United Shoe Machinery

10gradually just declined to nothingness.

11We ought to be studying cases where we clearly

12failed as well as cases where we think we might have

13succeeded.

14DR. MULLIN: And this doesn't give a specific

15methodology, but some insight might actually come from

16the kind of, you know, cross-industry comparison or at

17least looking at the experience of other industries,

18even ones in which we don't think there's some problems

19with competition. So, for example, you know, Scott

20Stern and Josh Gans have a series of papers about

21basically licensing in biotech, as they say, licensing

22the gale of creative destruction. Before you look at

23the data, you might think, oh, they are these small

24people, they are going to come up with something that's

25going to leapfrog Lilly or something like that, a Lilly

118

1product, but in actuality, what they will end up doing

2is end up being acquired through some sort of licensing.

3Effectively their competitive advantage is innovation

4and not dealing with regulatory hurdles, et cetera, and

5it makes more sense for it to be joined with incumbent

6pharmaceuticals.

7Now, once again, you might imagine that a

8different world where Lilly would shrink because it's

9been leapfrogged by competitors, but by the same token,

10you know, presumably the current system leads to

11innovation at the biotech level because they basically

12know they have got this opt-out in terms of an external

13capital market. They know if they get a hit, they are

14going to be acquired and they don't have to go through

15the whole costs of taking the drug to market themselves.

16DR. SCHERER: Absolutely right. My daughter is

17research director of a small biotech startup, and she

18knows she can't -- if they go into Phase II testing that

19her firm can't do it. So, they expect either to license

20out or be acquired.

21COMMISSIONER KOVACIC: To what extent is the set

22of institutional arrangements by which agencies actually

23bring and prosecute cases something that has to be

24examined as well? I think that many of you, if not all

25of you, have been involved in litigation episodes,

119

1either inside the agencies or outside the agencies. I

2was struck at David's comment about how in the course of

3American Airlines the basic intuition that led to the

4decision to prosecute remained the same over time, but

5perhaps the understanding of why it was a good case may

6have changed in significant respects over time.

7I suppose in any one instance, in deciding to

8prosecute any one case, the agency not only makes

9decisions in general terms about whether there's a

10sustainable theory, but has to make decisions about

11whether to gather information, what information to

12present, what is ultimately going to be persuasive to a

13reviewing tribunal.

14One element of the equation that we have to

15consider not simply the functionings of specific firms,

16industries, and economy as a whole, but the means by

17which agencies themselves formulate and present cases

18basically the mechanism by which theories and ideas are

19ultimately transmitted into specific cases and how those

20cases are pursued.

21DR. WINSTON: I mean --

22COMMISSIONER KOVACIC: There are larger

23questions of institutional capability.

24DR. WINSTON: And/or institutional constraints.

25I mean, there has been some political economy literature

120

1about the role of Congress or, you know, funding sources

2and how they affect what the agency does. There was

3a -- I can't remember, but a while ago, wasn't there a

4study on -- saying how FTC cases were influenced by

5Congressional funding in terms of, you know, you weren't

6going after cases or areas where somebody was high up on

7a committee in Congress because that could affect your

8funding? That kind of stuff has been around for a

9number of years.

10COMMISSIONER KOVACIC: Yes.

11DR. WINSTON: I haven't seen recent work on

12that, but, you know, there's that kind of political

13economy reality in terms of your dealings with Congress

14and the President, of course.

15COMMISSIONER KOVACIC: But I'm saying that, even

16in the instances where you've decided to go ahead, one

17key variable is the skill, the shrewdness, with which

18the institution actually pursues a given matter.

19DR. SCHERER: Let me say, my greatest failure.

20Because I had a long connection with Detroit, when I was

21director of the Bureau of Economics in the seventies, I

22put very high priority on beginning an investigation of

23the automobile industry. It was clear they were headed

24for trouble. Who was it? I think it was Cliff who

25talked about how -- yeah, Cliff talked about the

121

1dynamics that got GM and Ford into their present pickle.

2Well, it was clear already in the seventies that

3they were heading for trouble, and the objective of that

4investigation was not primarily to bring an antitrust

5case; it was to illuminate to the public and to the

6Congress what was going on, and the whole thing failed.

7If we had succeeded, I think we might have avoided some

8very serious mistakes. The industry might have learned

9some things, the public would have learned some things,

10the Congress would have learned some things.

11I didn't see that case going into litigation. I

12saw it as performing the FTC's historical role of

13telling the public what the hell's going on in American

14industry.

15COMMISSIONER KOVACIC: Jon?

16DR. BAKER: I was going to add that in the paper

17I alluded to before with Tim Bresnahan, we talk about

18two ideas for increasing the institutional capacity of

19the traditional system to use economic learning, one of

20which is to think about limited rules for neutral

21experts, and another is for the enforcement agencies,

22particularly the economists, to identify and codify

23relevant generalizations about industries from the

24empirical economic literature and make that available to

25courts.

122

1You all do try to do something sort of like that

2in Schering, essentially in that whole line of cases

3where the FTC is effectively relying on the idea that

4generic drugs, when they enter, the price goes down for

5the brandeds, and you're thinking "what can we learn

6from that about the importance of generic entry to

7create a presumption about why practices that might

8discourage generic entry would be a problem?" Well,

9taking generalizations like that and writing reports and

10having that available for courts is a way to increase

11everyone's institutional capacity.

12DR. SCHERER: The fact is that the FTC's report

13on generic drug entry and patent extension strategies by

14branded drug firms was superb.

15COMMISSIONER KOVACIC: I guess the humbling

16thing for me is Schering. The investment in the

17generic drug study was a major decision of Bob

18Pitofsky's in 2000 to start the project, handing the

19baton to Tim Muris, who made a major decision to

20continue to devote resources and make it a high

21priority. I think the study was enormously illuminating

22and an excellent example of how 6(b), which we have

23talked about before, ought to be part of the

24Commission's portfolio.

25I am not asking everyone to accept the wisdom of

123

1the Schering case on the merits (though I think you

2should), but you had decades worth of FTC activity in

3this area, you had the FTC's investment in the empirical

4study in question, and you had related work that the

5Commission had done. All of this was presented to the

6Court of Appeals, and the FTC received exactly the

7amount of deference that a wayward child would receive

8from a parent, which was none at all. The decision of

9the administrative law judge was accorded great

10deference.

11On the other hand, the decision of the

12Commission, with this affiliated research, received

13none. What is humbling when one walks into difficult

14areas of analysis of this type, internally we have to

15ask, I think, are we bringing to bear the assembled

16knowledge in an effective way for a reviewing tribunal?

17You don't get something very far saying, well, that was

18an error by the Court; there's another erroneous court.

19Yet another court has failed to get it right. They

20ultimately are the gatekeepers we have to work with.

21But in this instance, that was unsuccessful in a fairly

22traumatic way.

23MR. HEYER: One process point that I think might

24be worth considering, although I'm not quite sure how to

25get this in front of whoever makes the determination, in

124

1talking to some international folks, they have a process

2in some jurisdictions where they actually have the

3testifying economists, maybe even the consulting

4economists, the Court essentially has them discuss,

5debate, reach consensus with one another on things that

6they can agree on and things that they still disagree

7on, and to some extent it helps cut through a lot of the

8confusion that any layperson or court is going to face,

9and, you know, there are going to be some remaining

10differences, but that seems like it might be an

11efficient thing to do, perhaps within the Division or

12the FTC and perhaps within courts as well, to have that

13sort of process.

14DR. BAKER: Let me make a comment. I want to

15advertise something else now, which I was the --

16MR. HEYER: It's not another article, is it?

17DR. BAKER: No, no Tim Bresnahan on this one.

18I was co-chair of a task force of the Antitrust

19Section of the American Bar Association on which Luke

20participated last year the Economic Evidence Task Force.

21We did a long analysis of various options like these and

22laid out some pros and cons. We didn't reach a

23consensus as a task force on it, but I think you would

24find it very interesting and instructive, and I believe

25if it is not now it will soon be available on the

125

1Antitrust Section web site for everyone to take a look

2 at.

3DR. SCHERER: Actually, I had an experience, I

4was hired as an expert by Judge Will in Chicago on the

5 glass bottles case. Part of my task was to do what you

6suggested. Individually I met with the experts from

7each side, posed questions that essentially went to

8their differences, and tried to see what areas of

9agreement could be found and what new research or what

10new analyses could be found that might illuminate the

11differences. We got pretty close to getting a rational

12settlement of the case, except that one economist on the

13final day of testimony strayed from the chosen --

14MR. HEYER: The script?

15 DR. SCHERER: -- chosen path, and then so turned

16off the jury, the jury so disbelieved him, that although

17he was right on the merits, they disbelieved him and

18rendered a verdict that was totally nonsensical.

19COMMISSIONER KOVACIC: I know we are close to

20the end of our time for today. I had a couple of

21closing remarks for the session, but I wanted to give

22our panelists another minute or so, if you have other

23thoughts you would like to bring up.

24DR. MARSHALL: Well, I just had one comment

25about the implied -- well, the suggestion that you had

126

1implied, Bill, regarding the funding of research

2programs coming out of either the FTC or the DOJ. I'm

3not savvy about the political nature of all of that. I

4am generally quite happy with what I see coming out of

5the academic literature since I am not one to look down

6at the shoulders I am standing on and speak pejoratively

7about where I'm resting, but I think that if the DOJ and

8FTC were to somehow jointly put forward data that was of

9remarkable quality, you can move research programs that

10way.

11The academics will latch into rich sets of

12quantifiable information and coordinate on that if it is

13good enough. If they see that there is lots of economic

14content in there that they could never get their hands

15on otherwise, you will move research programs that way,

16and that doesn't require creating some kind of, you

17know, NSF-like program within the FTC/DOJ.

18COMMISSIONER KOVACIC: Other closing thoughts?

19(No response.)

20COMMISSIONER KOVACIC: Ken?

21MR. HEYER: No, I just wanted to thank everyone

22again. I learned a good deal, and I know it's not an

23easy matter to come to something like this, and on

24behalf of the others as well, I wanted to thank

25everyone.

127

1COMMISSIONER KOVACIC: As Ken did earlier in

2thanking June, Joe and the team at the Department, I

3want to thank the folks at the FTC who put this session

4together. Those of you who have ever organized

5anything, even a discussion around a lunch table, know

6that this doesn't happen automatically. This takes an

7incredible amount of work by the organizers. Jim

8Taronji, Pat Schultheiss, Doug Hilleboe, Elizabeth

9Argeris, and David Balan at the Commission were the

10folks who along with June and Joe, Ken, put this session

11together.

12I also want to thank the speakers again. In

13some ways, to ask what we've learned, what we would like

14to know, and how we go about learning what we like to

15know are impossibly difficult questions to address in a

16short period of time. To do this, we could only ask

17people whose skills were equal to doing the impossibly

18difficult. That's why this group is here. I want to

19thank them for taking their very precious time to share

20their ideas with us today.

21I'm grateful for everyone's willingness to have

22this session today. I think that it is truly the

23marriage of theory and practice that is so important to

24formulating good policy. I think that the empirical

25dimension, both the broader scale inquiries using the

128

1taxonomy that Cliff laid out for us, from the broader

2economy-wide perspective down to the industry-wide

3level, to the firm-wide level, down to cases, is a mix

4that's very important to what we have to do.

5Perceptions of the past deeply influence current

6views about what policy should be. In many ways, they

7set the presumptions about what policy is today, not

8just at home but also abroad. There are interesting

9opportunities to embed within agencies, and I speak of

10my own institution, a norm that makes this a routine and

11significant part of our agenda, every bit as important

12as bringing the cases; doing the research on which cases

13rest, looking at past enforcement events or

14nonenforcement events as a way of considering the way

15ahead, collaborations with researchers on the outside,

16maybe the idea, on a limited basis, of regularly

17convening a workshop at which promising empirical work

18or promising paths of work are done, something that can

19be done inexpensively in an illuminating way, and the

20possibilities that we haven't talked a great deal about,

21though we have touched upon some, for cross-border

22comparisons.

23It's also striking to see the number of academic

24centers like Bob's, like the joint project that Cliff is

25so deeply involved in, that have counterparts in Europe

129

1where, week-in and week-out, at different centers,

2interesting research along these lines are being done,

3so that what work was done might have a truly

4cross-border dimension to it.

5I'm fond of the title that Earl Weaver chose for

6his autobiography: It's What You Learn After You Know

7It All That Really Counts, and that's why continuing

8attention to doing good empirical work strikes me as a

9day well spent.

10Thank you all.

11(Applause.)

12 (Whereupon, at 12:33 p.m., the hearing was

13concluded.)

14

15

16

17

18

19

20

21

22

23

24

25

130

1C E R T I F I C A T I O N O F R E P O R T E R

2DOCKET/FILE NUMBER: P062106

3CASE TITLE: SECTION 2 HEARING

4DATE: SEPTEMBER 26, 2006

5

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

10belief.

11

12DATED: 10/9/06

13

14

15

16SUSANNE BERGLING, RMR-CLR

17

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

19

20I HEREBY CERTIFY that I proofread the transcript

21for accuracy in spelling, hyphenation, punctuation and

22format.

23

24

25DIANE QUADE