# DEPARTMENT OF JUSTICE

> Briefs, arguments, decisions, and more.

URL: https://www.frixlaw.com/law-library/documents/agency%3Aftc%3Af55014301a8e43e9

## Record

- **Collection:** Agency decision
- **Document type:** Agency decision

## Text

1
1
2

DEPARTMENT OF JUSTICE

3

FEDERAL TRADE COMMISSION

4
5
6

MERGER WORKSHOP

7
8
9
10

WEDNESDAY, FEBRUARY 18, 2004

11

9:00 a.m.

12
13
14
15
16
17
18

FTC Conference Center

19

601 New Jersey Avenue, N.W.

20

Washington, D.C.

21
22
23

Reported by:

Rita M. Hemphill, C.V.R.

24
25
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

2
1

C O N T E N T S

2

PAGE

3

WELCOME - J. Elizabeth Callison

4

4

NON-PRICE COMPETITION/INNOVATION

4

5

Moderator:

6

Panel:

Mark W. Frankena, FTC

Ann Malester

7

Richard T. Rapp

8

Steven C. Sunshine

9
10

UNILATERAL EFFECTS

11

Moderator:

12

Panel:

51

Luke M. Froeb, FTC

Joseph Kattan

13

Gregory K. Leonard

14

Abbott "Tad" Lipsky, Jr.

15

Gregory J. Werden

16

Valerie Rabassa

17
18

COORDINATED EFFECTS

19

Moderator:

20

Panel:

Michael H. Knight, FTC

Jonathan B. Baker

21

Andrew R. Dick

22

Paul Yde

23

Deborah P. Majoras

24

Steven C. Salop

25

David T. Scheffman
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

127

3
1

C O N T E N T S (CONT'D)

2
3

PAGE
UNCOMMITTED ENTRY

4

Moderator:

5

Panel:

197

Theodore A. Gebhard, FTC

Timothy Daniel

6

A. Douglas Melamed

7

Frederick "Rick" Warren-Boulton

8

Mark D. Whitener

9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

4
1

P R O C E E D I N G S

2

WELCOME

3

MS. CALLISON:

This is the second day of the

4

Joint DOJ/FTC Workshop on Merger Enforcement.

5

contained a lot of interesting information and useful

6

panels.

7

well.

8

Yesterday

I'm sure today we will follow that up quite

I want to let you know that presentation

9

materials from our panelists' papers and slide shows that

10

they have sent in will be available on our respective web

11

sites.

12

be there shortly.

13

hearings of the workshop will be available on the web

14

sites at some future date.

15

I wouldn't look for them on Friday, but they will

Thank you.

16
17

Likewise, the transcripts of the

NON-PRICE COMPETITION/INNOVATION
MR. FRANKENA:

This morning we will be focusing

18

on the discussion of the effects of mergers on

19

innovation.

20

speak first, and then Ann Malester will go next, and then

21

Dick Rapp, and after that, we will have an open

22

discussion between the panelists and the audience, with

23

audience participation and so forth.

24
25

Our panel today is Steve Sunshine, who will

When I was asked to moderate this distinguished
attorney and economist panel on mergers and innovation,
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

5
1

my first thought was to seek input from the investor

2

community which didn't seem to be represented.

3

So I asked my friend Tom, who after all has

4

about 1,000 patents, whether he could address the

5

relationship between competition and innovation this

6

morning.

7

that he had addressed this very matter back in 1889, when

8

someone proposed that his firm, Edison General Electric

9

and its competitor, Westinghouse Electric call a truce in

10

their war to sell electric systems to cities and towns

11

around the country.

Well, Tom couldn't join us, but he reminded me

12

Tom explained that he refused to go along with

13

the proposal because, and I now quote from a contemporary

14

letter from Edison:

15

usefulness as an inventor is gone.

16

be worth a penny.

17

incentive.

18

"If we make the coalition, my
My service wouldn't

I can only invent under thought of

No competition means no innovation."
But my real difficulty with this statement,

19

though I -- after I read it I noticed that it was written

20

on April Fool's Day in 1889, so I don't know how much

21

weight to put on it, but that's a statement from an

22

important inventor.

23
24
25

Okay.

With that, could I ask Steve to get us

started?
MR. SUNSHINE:

Thank you.

I'm happy to be here

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

6
1

this morning.

2

speaker.

3

Professor Davis.

4

therefore, could not be with us.

5

extreme way to get out of doing this talk, but I did last

6

night consider that.

7

I was not supposed to be the lead-off

The lead-off speaker was the distinguished
He ate some bad fish yesterday and,
It seemed like a rather

I struggled for how to title this talk, and

8

like everything else, the inspiration came to me because

9

of something my 12-year-old son did.

10

this song by Smashmouth called "All Star," which I'm sure

11

you all have heard.

12

it for you.

13

Don't worry.

I was listening to

I'm not going to sing

There is a line in there that says, "I'm not

14

the sharpest tool in the shed."

15

between that and innovation analysis were just

16

immediately obvious, as I'm they are to everyone sitting

17

here in the room.

18

To me, the parallels

The point that I will make generally today is

19

that innovation analysis has its uses.

I wouldn't call

20

it the sharpest tool in the shed.

21

question, and hence the title, when should this tool come

22

out of the shed?

That leads to the

23

Just as an overview, I hope that these

24

propositions here are pretty non-controversial, at least

25

the first few ones are.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

7
1

Innovation is an important dimensional rivalry

2

and intervention of progress.

I don't think there's a

3

lot of dispute for this point.

I think the studies that

4

have been done over time have said that innovation has

5

been responsible for the vast majority of economic

6

progress that has been made over the last century.

7

Then when we try to actually apply it to merger

8

analysis, can we actually predict the effect of a merger

9

on innovation?

10

think it's fair to say that the effects are uncertain.

11

Of course, it's probably also fair to say that the

12

effects of market structure on goods and on the

13

relationship between market performance on goods is not

14

as clear as we would like it to be as well.

15

We will talk more about this later.

I

That leads me to this is, I think, where we get

16

a little bit more into my view, and others clearly have

17

different views, the legal and economic issues.

18

are legal issues here, too.

19

on the economics, but there are legal issues here, too,

20

that legal and economic issues require a showing -- I

21

called it "probable effect on output."

22

"output" here, I don't mean innovation output.

23

output in the goods market.

There

We have been focusing mainly

When I say
I mean

24

The Genzyme example, we will get to it, we all

25

have our views on it, but perhaps Genzyme is an example
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

8
1

where the chairman and the rest of the Commission

2

basically followed the first three bullet points, thought

3

in theory there could be an effect on innovation, but

4

found a set of facts where the merger could not have a

5

demonstrative effect on output, and hence, no case. If

6

that's what they did, count me in.

7

we have to talk about.

8

Of course, the facts,

Lastly, just in terms of what does this mean

9

practically?

10

that the practical application of innovation markets is

11

infrequent.

12

market cases.

13

industries, pharmaceuticals being the obvious example,

14

and that for reasons that I think some of my co-panelists

15

will go into.

16

My general point on this is going to be

We are not going to see a lot of innovation
They are more likely in certain

One of the many big problems with innovation

17

markets is that they are really hard to define.

18

you know an innovation market?

19

standard more than just as spotters, you know, I'll know

20

it when I see it.

21

How do

We'd like to see a

We have to account for trying to identify all

22

the conceivable sources of innovation.

23

account for how do we know what the strengths and

24

significance of the population of innovators are.

25

We have to

Also, the type of innovation that's going on in
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

9
1

the market may be relevant to our confidence in market

2

boundaries.

3

What do I mean?

Well, are there regulatory barriers that are

4

going to make innovation much more structured?

5

innovation tied to certain sets of production assets,

6

where we know that people without production assets

7

really are not effective innovators?

8

Is

For that, I would refer to the Department of

9

Justice's complaint in the General Motors/ZF

10

Friederichshafen case.

11

to an existing set of production assets, and no one could

12

innovate new truck and bus transmissions unless they had

13

those set of processes.

14

That case of innovation was tied

Then another way or type of innovation is

15

patents, but patents are really tricky.

16

patents, patents may mean there are a necessary set of

17

assets.

18

up so that the mergers are complimentary.

19

really careful.

20

When you get to

It may also mean that markets have been divided
You have to be

I won't go through the article that Professor

21

Gilbert and I wrote, but we identify certain sets of

22

factors that you need to find in order to have some

23

confidence that you have an innovation market.

24

don't have that kind of confidence, then frankly you

25

should go home at that point.

If you

If you can't define the

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

10
1

market and you can't define a few players, there's no

2

point in going on.

3

work to do.

4

If you can, of course, there's more

Defining the boundaries of the market is hard.

5

This part is even harder.

How can you actually predict

6

the competitive effects?

How can you assess the merger's

7

effect on innovation and then translate that into an

8

output effect?

9

We start off with, I think, a very obvious

10

question, structures as a means of predicting

11

performance.

12

Who is right, Schumpeter or Arrow?

Schumpeter basically says you need monopoly

13

rents, and it's possible in some industries you may need

14

monopoly rents to actually provide incentives for

15

innovation. Arrow says, on the other hand, because of

16

cannibalization concerns, expected return, that perhaps

17

monopolists do not have the same incentive to innovate at

18

the same rate as others.

19

I don't think it's fair to say the economics

20

literature thinks this is completely an unaddressed

21

question.

22

by Spence, by Stidless, that allows you to make some

23

inferences.

24
25

There has been a lot of work done by Scherer,

It's certainly not proven.

I think, also, if you go with just basic
intuition, how many around this room believe that time
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

11
1

after time, monopolists really have a high incentive to

2

innovate?

3

I think it is fair to say that just as a matter

4

of economics, we don't know for sure what the answer to

5

that question is.

6

have our own belief, but we don't know.

7

We might have an intuition.

We may

That leads us right into the second issue,

8

which is the problem of the necessity of case specific

9

evidence.

10

economics, that all of these cases have to be driven by

11

the actual facts of what's in front of the Commission,

12

what's in front of the DOJ, what's in front of the

13

plaintiff.

14

innovation competition going on between the two, and that

15

each have high incentives to innovate.

I think, given the uncertainty in the

You have to believe there is important

16

Again, the GM/ZF case, I guess I have a little

17

familiarity with, was a case where the evidentiary record

18

showed years and years of the two companies actively

19

targeting each other's innovation activities, actively

20

pursuing incremental process improvements, and then with

21

the signing of their merger agreement, saying we're not

22

going to do this so much any more, but then labeling it

23

efficiency.

24
25

That was a fact specific inquiry that led to
say, you know, I think we know enough.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

Maybe Schumpeter

12
1

is right sometimes.

2

we are now in the Arrow camp with these facts.

3

Maybe Arrow is right sometimes, but

I touched on incremental versus revolutionary.

4

This is an idea of how big the innovation is.

5

ways, incremental innovations may be the ones that are a

6

little bit easier to understand how to protect.

7

be easier to define where the sources of that innovation

8

are going to come from.

9

there's much more profit at stake, so there's more

10

incentive to pursue them, and sometimes they are harder

11

to tell where they are going to come from, what the

12

source of that innovation is going to be.

13

In some

It may

These revolutionary innovations,

In the incentives, I don't think you could

14

understand the incentives to innovate until you

15

understand the nature of competition between the

16

innovations.

17

downstream level.

That's going to require looking back at the

18

Are two pharmaceutical companies going to be

19

manufacturing exactly the same chemical compound that,

20

you know, once they get to the market, they are going to

21

be in competition with one another, and there's no orphan

22

drug status.

23

will be in competition with one another.

You know, once they get to the market, they

24

That may lead you to one set of conclusions.

25

If on the other hand these products are so sufficiently
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

13
1

different they are going to be attacking different

2

markets and there will be a high expected return, maybe

3

there will be an incentive for both products to be

4

developed.

5

going to open up so many new applications and increase

6

demand that there will still be a strong incentive to

7

invest.

8

I don't know.

Maybe the second invention is

Confidence that the reduction in innovation

9

will lead to an output effect.

10

previous statements, too, but here, I think you have to

11

take into account downstream competition.

12

are downstream products that are competing against your

13

upstream products, and those are going to continue to

14

spur innovation.

15

That goes to all the

What if there

I think that what all this is saying is that

16

there are a lot of steps to go through in order to get to

17

this level of confidence.

18

the effect outweighed by innovation efficiencies?

19

Lastly, I put down here, is

Let me just throw in a point just as a matter

20

of personal opinion.

I know this is not necessarily

21

accepted by everybody.

22

whole idea of merger specific efficiencies.

23

after the fact, sitting around and saying gosh, I think

24

these guys probably could have done a contract research

25

and development joint venture.

One of my pet peeves is this
People,

To me, that is a naive

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

14
1

view.

2

I think the analysis should be here's the deal,

3

here's what life would be like absent the deal, and is

4

the world a better place or not, but to try to social

5

engineer efficiencies seems to me to be a misguided

6

approach.

7

What I've done by putting efficiencies as part

8

of the competitive effects, I haven't shifted the burden

9

of the efficiencies to the defendants.

10

still part of showing -- to show an output effect.

11

Eventually, this really belongs in the camp of the

12

plaintiff, because you can't show the effects unless the

13

efficiencies don't outweigh the other benefits.

14
15
16

Enough about economics.

I think it is

What about the law?

I'm running behind.
There have been a lot of consent decrees.

17

There has been a complaint that the Department of Justice

18

filed that did not go to trial.

19

other cases where innovation is sort of talked about but

20

not a true innovation market.

21

There have been some

Does Section 7 really allow us to have an

22

innovation market?

The few cases that exist on this, I

23

think, say that without actual sales and a line of

24

commerce, that perhaps you don't have a market, which if

25

that was true, that would doom Section 7 analysis.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

15
1

That's my view, and what Gilbert and I wrote in

2

our articles is that it would be misplaced for Section 7

3

to not recognize an innovation market theory.

4

why I say that, we would start from, I think, a pretty

5

easy proposition that says that merger analysis should be

6

designed to identify those mergers that lead to reduction

7

in welfare, to lower output.

The reason

8

The analysis that I just went through, if you

9

can go through that analysis and come to the conclusion

10

that output is reduced, you have satisfied what is the

11

basic condition.

12

market, I think, is the innovation market essentially

13

identifies the place of competitive interaction where the

14

problem arises, and then the output market actually

15

defines where the effect is felt, and it defines the line

16

of commerce and the section of the country that the law

17

is trying to protect.

18

Then the question on an innovation

In that way, I would draw a loose analogy, and

19

it's not a perfect analogy, to a vertical merger.

20

actual output market, there's no direct effect.

21

could look at input, you could look at innovation as an

22

input or an upstream product in a vertical merger, but I

23

would say if you look at the Medicorp vs. Humanas case

24

and some of the other cases cited in the stuff that we

25

wrote, there are some courts that have gone other ways,
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

In the
You

16
1

because they are thinking cookie cutter, define a market,

2

measure the market shares, prima facie case, blah, blah,

3

blah.

That's one way of doing it.

4

The second point, can plaintiff prove a

5

non-speculative effect in a reasonable amount of time? I

6

think that proposition stands for itself.

7

a defense that I would want if I was defending against an

8

individual market case.

9

importance of being able to actually have the documents,

10

the market evidence, and everything else to show that

11

this is a real problem.

12

This would be

This just goes back to the

I know I'm running short and I want to leave

13

time for some conversation.

14

this.

15

I won't spend much time on

Ron Davis, who was going to lead off, was going

16

to say burdens, where's the beef?

17

why all the fuss?

18

plaintiff.

19

Hughes say that upon the establishment of a prima facie

20

case, that the burden of production shifts over to the

21

defendant, but the burden of persuasion always remains

22

with the plaintiff.

23

I was going to say,

The burden of proof is always on the

We all know that.

We know that PNB and Baker

I think in the Genzyme case, while it was just

24

an investigation, it wasn't a case, I have to imagine

25

that the defendants satisfied a burden of coming forward
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

17
1

with evidence.

2

It seems to me that any proper application of

3

this kind of analysis would lead to the burden going back

4

to the plaintiff to persuade they were actually right.

5

If Commissioner Thompson's dissenting opinion

6

means something else, with a presumption, then he means

7

something else, which I think is not in step with the

8

case law, perhaps even for output markets, but certainly

9

not for innovation markets for the reasons that we have

10

talked about.

11

Problems of the investigator.

The first point,

12

I think, is obvious to everybody sitting in this room.

13

The investigator has a duty to evaluate the transaction

14

before it.

15

serious competition in innovation, what are you going to

16

do?

17

just let the transaction go if you think there is going

18

to be a problem.

19

In the transaction before you, if you see

You have a duty to look at it.

You don't want to

You have to dig in.

The law with the state of the economics, there

20

is no easy path here, is there?

We have just been

21

through it.

22

could be some adjustments, depending on who is involved,

23

the burden of proof is clearly going to be on the

24

plaintiff, and the plaintiff and the investigator will

25

have to show a number of difficult issues.

I think the burden of proof, while there

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

Innovation

18
1

space which will occur, the merger's likely effect on

2

innovation, the manner in which output will be reduced,

3

and again, output is output in a goods market, and when

4

the output effect will be felt.

5
6

Not easy stuff, right?

We are talking about doing this in at least two
different markets, two different spaces.

7

Again, as sort of the last resort of the

8

investigator, the importance of empirical evidence from

9

the merging parties in the markets cannot be overstated.

10

Without the evidence, the economics isn't going to get

11

you all the way there.

12

evidence.

13

You are going to need the

I won't go through this in any kind of detail.

14

I just thought through questions on how I might think

15

about doing an innovation analysis.

16

Is innovation an important dimension of

17

rivalry?

18

for price effects.

19

beer.

20

The answer is no, forget it.

Why bother?

Look

Go out with your colleagues for a

Second question, will innovation effect

21

existing product markets in a reasonable amount of time?

22

The answer to that is yes, let's not worry about

23

innovation markets.

24

Let's do a competitive effects on the goods market.

25

Gil will remember, this was the Microsoft/Intuit case.

It sounds like a goods market to me.

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

As

19
1

We toyed with bringing a case in an innovation market.

2

We said we have a real market here, why are we bothering?

3

Innovation is an important dimension of competition.

4

one of the two players is in the market, then we have a

5

potential competition case.

If

6

I had to sit back for a second and say, I think

7

a potential competition case is easier than an innovation

8

market case?

9

Well, there you have it.

Can the boundaries of the innovation market be

10

determined?

We went through that.

It's not child's

11

play.

12

Again, I put the efficiencies in with the obligations of

13

the plaintiff.

14

show there is an effect, then the plaintiff should also

15

be comfortable showing the efficiencies don't outweigh

16

the other effects.

Does the merger provide incentives us to innovate?

Again, if it's the plaintiff's job to

17

And then last, the harm to the output market,

18

without this last piece, not only do we have an economic

19

problem, but we clearly have a legal problem as well.

20
21

With that, I know it's all painstakingly clear.
Thank you very much.

22

MR. FRANKENA:

Thank you.

Ann?

23

MS. MALESTER:

Well, I must say that I'm really

24

surprised that this many people showed up this early in

25

the morning to hear us talk about innovation markets, but
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

20
1

this must mean all of you are certainly aware that we

2

have been having a debate in the antitrust world for well

3

over a decade about what the antitrust agencies should do

4

about the concept of innovation markets, and essentially,

5

should there be any enforcement actions in this area?

6

As you know, there are those who believe that

7

because the empirical economic data has not been

8

conclusive in showing a correlation between increased

9

concentration and reduced innovation, that really the

10

antitrust agencies should just throw up their hands and

11

walk away and not think about bringing enforcement

12

actions in this area.

13

There are others, and I count myself in this

14

group, who believe that preserving competition at the

15

research and development stages is really important, but

16

recognize, as Steve has pointed out, that there are some

17

real problems and a lot of difficulties in really

18

assessing how innovation markets should be defined and

19

what the antitrust agencies' role really should be there.

20

First, let me just give you two reasons why I

21

think that it's vital for the agencies to preserve

22

competition and be vigilant in protecting competition at

23

the R&D stages.

24
25

First of all, because in the long run, a merger
that reduces the pace of innovation can be far more
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

21
1

harmful to consumers than a merger that results in a

2

price increase of quality decrease.

3

Second, because in the real world, where

4

companies compete every day, it's very clear that

5

competition is the primary incentive that spurs increased

6

level of innovation.

7

In preparing for this workshop, I re-read the

8

testimony that the FTC received during the 1995 hearings

9

on global competition.

10

the economists could only agree that it was really

11

inconclusive and they weren't sure what the evidence

12

showed, the business officials who testified during those

13

hearings were, and I'm quoting the words of the staff's

14

report, "Unanimous and emphatic in their view that

15

competition is the primary incentive for innovation."

16

I spent well over a decade at the FTC as an

What struck me is that although

17

assistant director in the merger division.

18

hundreds of documents, and talked to dozens of business

19

officials, who told us that they supported that

20

conclusion.

21

I have seen

When you look at the reality of the way

22

businesses make their decisions, I think it's clear that

23

typically businesses invest more resources, work harder

24

and work more quickly in their research and development

25

efforts when they are faced with the possibility that
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

22
1

they will be beaten to the market by another firm that's

2

working to develop a competing product.

3

We saw evidence in our investigations that

4

companies often spend a great deal of time tracking the

5

research and development projects of other companies that

6

are working in their general area.

7

information about the schedule and progress of competing

8

R&D programs, and they make investment and priority

9

decisions based on the level of competition they believe

10

they face.

Companies seek

11

They also confirmed that being the first to

12

introduce a new or improved product can be critically

13

important to many industries.

14

that gives you increased market acceptance, lock in of

15

customers, there are many reasons why in some industries

16

reaching the market first is very important for future

17

success.

A first mover advantage

18

Having said all that, I really do agree with

19

Steve when he explained that it is much more difficult

20

for the plaintiffs to predict under what circumstances a

21

merger is likely to reduce the pace of innovation, and we

22

know that it's not that easy to even predict when a

23

merger will result in a price increase.

24
25

In many cases, it is difficult to identify all
the potential sources of innovation, or to identify
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

23
1

whether the merged firm will have both the ability and

2

incentive to reduce the pace of innovation after the

3

merger, and finally, even harder to decide what type of

4

efficiencies, if any, the merged company will be able to

5

achieve in combining their R&D programs.

6

I think it's for that reason when you look at

7

the past 10 years in enforcement at both the Justice

8

Department and the FTC, you will see first a relatively

9

small number of cases where the agencies alleged only a

10

reduction in innovation, not tied to an existing current

11

goods market, and second of all, you see those actions

12

that there are generally in the pharmaceutical and

13

defense industries, where first of all, it is quite easy

14

to identify generally the companies that are

15

participating in research and development, and where the

16

barriers to new companies beginning development are

17

extremely high.

18

Let me turn for a moment to the Genzyme

19

investigation, which is the most recent pronouncement by

20

the Commission, and which brought forth three separate

21

statements by the Chairman and Commissioners Thompson and

22

Harbour.

23

That decision could be the subject of an entire

24

panel, so I'm just going to talk about two points.

25

First, really the question about whether in innovation
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

24
1

markets, there should ever be a presumption of

2

anti-competitive effects at any level of competition, and

3

second, my own views of what Genzyme might tell us for

4

future Commission enforcement decisions.

5

On the first point, I was frankly puzzled by

6

the seeming importance that all the statements placed on

7

whether or not there should be a presumption of

8

anti-competitive effects.

9

feeling that in existing goods markets, where

10

presumptions are raised at much lower concentration

11

levels than was the case in Genzyme, but the antitrust

12

agencies bring an enforcement action every time a

13

presumption of legality is raised.

14

nothing could be further from reality.

Somehow, that left me with the

We all know that

15

I think that the data that the Justice

16

Department and the FTC released two weeks ago on

17

horizontal mergers and the concentration levels and which

18

ones resulted in enforcement versus closing, make it very

19

clear that even in existing goods markets, there are

20

highly concentrated markets where merger may reduce the

21

number of competitors from three to two and even two to

22

one, where in some cases, the Commission did not bring an

23

action or the Justice Department did not bring an action.

24

What does the presumption mean?

25

that the merger deserves a really close scrutiny, and
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

It only means

25
1

that the agency in question will spend as much time as it

2

has to develop an intensive fact based analysis.

3

I don't think there is anyone at the Commission

4

who believes that Genzyme did not deserve that kind of

5

close scrutiny.

6

Commissioners spend a great deal of time and effort in

7

analyzing the large investigatory record, and in

8

assessing what the facts were and what the analysis

9

should be, and that is really what I think the

10

presumption means, that it's a case that warrants that

11

kind of scrutiny.

12

Both the agency staff and all of the

Finally, let me just talk for a minute about

13

what the impact of Genzyme is and what it tells us about

14

the Commission and its likely enforcement in the

15

innovation markets.

16

First, I think clearly the Commission's

17

decision indicates that the current Commission is likely

18

to approach any innovation market analysis with a lot of

19

caution, but there are four reasons why I think you

20

shouldn't read too much into what the Commission may do

21

in other cases based on the Genzyme decision.

22

First of all, I think it's important to note

23

that not one of the three statements that were issued was

24

signed onto by anyone other than the author of that

25

statement.

Two of the Commissioners, Commissioners
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

26
1

Swindle and Leary, although they voted to close the

2

investigation, did not join in any of the statements that

3

were the public statements that were issued.

4

Second, in the Genzyme case, both companies,

5

Genzyme and Novazyme, were actually in the late clinical

6

stages of drug development, which is earlier than the

7

other innovation market cases where the Commission has

8

investigated and taken action before, and that means that

9

the likelihood of both products making it to market is

10

less than in the other cases the Commission has looked at

11

in the pharmaceutical area.

12

Third, the drugs in question were covered by

13

the Orphan Drug Act, which grants a seven year market

14

exclusivity to the first drug that makes it to market.

15

think that fact also makes it less likely than in a

16

traditional pharmaceutical market that both drugs would

17

in fact compete in a future goods market, and finally,

18

the Genzyme acquisition was not reportable under the

19

Hart-Scott-Rodino Reporting Act, and the transaction had

20

closed well before the Commission was in a position to

21

assess its legality.

22

Trying to design a remedy in an innovation

I

23

market in a consummated merger where the core assets are

24

scientific personnel, know how, access to academic

25

researchers, and many other human factors, really
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

27
1

presents quite a daunting task, and the vast majority of

2

cases that the agencies face, which isn't the typical

3

Hart-Scott-Rodino pre-merger review, I don't think raise

4

the same level of difficulty in designing and

5

implementing a remedy.

6

In sum, I think Genzyme was a very important

7

case because it gave the Commission an opportunity to

8

look at a really complete investigatory record, and to

9

take the time, which often doesn't happen with Hart-

10

Scott, to really debate the issues.

11

Obviously, different people at the agency came

12

to different conclusions on what the facts were and what

13

the conclusions should be.

14

exercise at the Commission.

I think it was an important

15

I also think the specifics of that case and of

16

the market in question and the fact that there wasn't a

17

majority opinion on what the appropriate analysis should

18

be, at least should leave everyone with some caution in

19

predicting what the Commission will do in future cases.

20

MR. FRANKENA:

Thank you, Ann.

Dick?

21

MR. RAPP: I'm in the situation that many FTC

22

speakers are in.

I have to issue a disclaimer before I

23

begin, because there are many of my colleagues at NERA

24

that disagree with my extremist positions.

25

over there, the man with his fingers in his ears, has the
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

Tim Daniel

28
1

look of participatory mortification on his face, is one

2

of those.

3

not shared by everybody at my firm.

4

I want you to know that my point of view is

I do find myself once again as the extremist in

5

the group, wishing to repeal entirely the use of

6

innovation market analysis.

7

For those of you who do not know me, I want to

8

assure you that it is not that my position is what it is,

9

not because I'm an anti-interventionist generally, that

10

the good government shouldn't put its aura in, et cetera,

11

that is a poor characterization of my views about almost

12

everything else.

13

I am an opponent of the use of innovation

14

market analysis frankly for an arcane epistemological

15

reason, and that is there is an absence of underlying

16

theory, and in a nutshell, what I mean is that antitrust

17

works well when it is based upon economics.

18

a science.

19

explanation, a hierarchy of general laws and law like

20

statements.

Economics is

What that means is that it uses scientific

21

The horizontal merger guidelines are a

22

wonderful example of economic science for policy

23

purposes.

24

demand elasticity, which is right in the mainstream of

25

the laws of economic science, going all the way up to the

The SSNIP test and other devices invoke the

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

29
1

law of demand.

2

Cournot's welfare triangle, Stigler's theory of

3

oligopoly, all underpinnings of the merger guidelines are

4

inherent, and the guidelines have a strong theoretical

5

basis.

6

facts as well.

7

concerned.

8

market analysis is basically reasoning by analogy.

9

policy premise, as I see it, is the proposition whether

10

the social gain from stopping an R&D merger exceeds the

11

social gain from letting it go through is predictable,

12

that you can tell which way that inequality will run.

13

I think the only basis for that that I can

By the way, we observe consistency with empirical
That is secondary as far as I'm

I care most about the fact that innovation
The

14

perceive is an analogy to the relationship between

15

competition, quantity and prices in goods markets, and

16

the analogy is false.

17

article, as many of you know, in 1995, that captures my

18

views about that subject.

19

I won't dwell on that.

I wrote an

Looking now in the five minutes I have, because

20

I will not impinge on the discussion period, where you

21

get to drag out of me in conversation, let me make just

22

three quick points, two of them relating to Genzyme and

23

then an attempt to be a little constructive instead of

24

being a naysayer.

25

accuse me of bothering an ugly baby.

Steven here is sick of having to
Let me see if I

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

30
1

can't do something better than that.

2

Whose baby is that?

I want to point out -- I'm going to go through

3

this whole slide -- it's really the first point that I

4

wish to remark to you about, and that is that I am an

5

admirer of Chairman Muris' statement in Genzyme.

6

remarkable to me for the fact that it changes the basic

7

policy rule.

8

with peculiar circumstances of the case, but one should

9

not lose sight of the pun on the word "competition."

10

What we are talking about now is a race to

It is

It has to do, as Ann described very well,

11

monopoly.

It has nothing to do with the kind of

12

competition that is output expanding, price reducing

13

competition, the likes of which we talk about when we

14

talk about horizontal mergers.

15

to do with this.

Market power has nothing

16

The underlying characteristic of -- let me say

17

the most basic finding in the majority statement is even

18

though they are nowhere near the end of the race, that

19

the pace of the two parties is predictable, the issue is

20

not who is going to come in first.

21

the second will come in at all or come in sooner rather

22

than later.

23

It's whether or not

The fact that there is no possibility for delay

24

of the first one getting to the deadline as a result of

25

the merger means the merger has no negative impact.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

31
1

That's my quick two second reading of the majority

2

viewpoint without all of the reasoning that follows.

3

All I wish to say is that as an analysis of

4

incentives, it makes a great deal of sense to me.

5

quite novel.

6

The lawyers in the group can consider how it would play

7

at Section 7 or FTC Section 5 enforcement, but I think

8

it's unusual.

9

It's

That doesn't bother me as an economist.

My real focus is on Commissioner Thompson's

10

dissent, because I have to say in the debate between the

11

extremists and those who favor the innovation market

12

approach, as far as I'm concerned, I can't be both a

13

player and a referee, but I have to say as far as I'm

14

concerned, my reading of Commission Thompson's dissent

15

led me to say game over, I win.

16

This is a litany of what can go wrong if merger

17

analysis is applied using its very premise, that is to

18

say reason by analogy to the goods market.

19

You talk about a merger, a monopoly, and speak

20

about market power, even though there are no goods around

21

and market power is not the issue.

22

I know about presumption of anti-competitive

23

effect without either theoretical or empirical basis.

24

There is this -- now we are getting back to the lesser

25

things that I won't dwell on so we can get to the
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

32
1

conversation part -- this struck me as extraordinary.

2

There is a difference to be expected in the

3

resource allocation of Genzyme, arising from the

4

difference between trying to get the first approval with

5

orphan drug exclusivity, to save a lot of children's

6

lives and make a ton of money, and trying to get first

7

approval with orphan drug exclusivity for a longer period

8

of time for the sake of first mover advantage.

9

I can go along with all of those who say

10

competition spurs innovation.

It makes a certain amount

11

of sense.

You see how the notion can be, I would say,

12

misused.

It represents to me a believer in competition,

13

certainly an oddly crabbed view of human nature.

14

There is a further litany of errors, errors in

15

economics and logic, but I believe here, in Commissioner

16

Thompson's dissent, we can talk about them if you like,

17

my worry is that some day President Kerry might appoint a

18

Commission with this frame of mind, and that is the

19

reason why a line of by and large sensible -- I won't say

20

unobjectionable -- to me, understandable decisions both

21

by the Pitofsky Commission and the Muris Commission, are

22

not good evidence that we are okay with the innovation

23

market concept.

24
25

Let me see if I can do what Chairman Pitofsky
did to me at the 1996 hearings, to try and find some
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

33
1
2

compromise in my extremist position.
At the very end of that panel, he put the

3

following question to me.

He said, look, just imagine

4

that Boeing is building the next airframe.

5

jet engine makers.

6

cam systems and start designing the new jet engine that

7

would go along with that.

There are two

They are about to sit down with their

8

Do you propose that the Federal Trade

9

Commission, if they came to us and said we want to merge,

10

we want to do a joint venture, we want to turn those two

11

research efforts into one, that we ought to just go away?

12

My answer to him was no, sir.

What I wish I

13

had said is no, sir, and here is why.

No, sir, because

14

you can do a goods market analysis, even though the first

15

good, the design and character of either good has not

16

been invented.

17

research effort and whether or not they will pour it on

18

to a greater degree on R&D spending or effort, if they

19

merge, because you can't predict that.

You have the means to analyze not

20

What you can do is you can say, look, this is a

21

case where there are going to be two goods competing with

22

one another versus one, if the merger or the joint

23

venture takes place.

24

guidelines reasoning to deal with that, and that makes it

25

okay with me.

We can use horizontal merger

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

34
1

My attempt at a constructive end to my

2

extremist talk here is to say that to the extent that we

3

are -- let me put it in terms that Mark used when he

4

wrote us a series of possible questions to discuss.

5

I think the distinction between innovation

6

markets that pay attention to R&D effort, about which we

7

know very little and predict very little, the distinction

8

between that and future goods markets, which to me means

9

taking only those cases where we can actually foresee

10

what the goods are and what the number of players, the

11

number of pricing voices will be, and make some attempt

12

at doing horizontal merger guidelines analysis.

13

That distinction between innovation markets on

14

the one hand and future goods markets is very great, and

15

as far as I'm concerned, one that should strive and be

16

well, and the other should quietly go away.

17

Thank you.

18

MR. FRANKENA:

Thank you.

Just following up on

19

that, I'd like to ask Ann a question.

20

in 1995, the antitrust agencies had completely listened

21

to what Dick said, and had said that henceforth, in going

22

forward, we won't be defining innovation markets.

23

think innovation is important and we think competition

24

may affect the rate of innovation, but we're not going to

25

delineate innovation markets, so we don't want our staff
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

Let's assume back

We

35
1

to be doing that.

2

My question for Ann is what difference would

3

that have made in the types of investigations, things

4

like that?

5

MS. MALESTER:

I think there are maybe two

6

examples that I can use to show you why I think Dick's

7

approach presents enormous dangers to competition, that

8

would allow mergers to proceed, that I think would have

9

really, really serious anti-competitive impact.

10

The first might be similar to the case that the

11

Justice Department brought, and I think it was during the

12

time Steve was there, Lockheed/Northrop Grumman.

13

It was really a step further back from the

14

Boeing example that Dick and Chairman Pitofsky talked

15

about, where the Justice Department raised the concern

16

that there were only three companies left in this country

17

that really had the capabilities of designing military

18

aircraft, and if Lockheed were permitted to buy Northrop

19

Grumman, that would be reduced to two.

20

There were no specific plans for a fighter

21

aircraft, but as we all know, in military procurements,

22

there are eventualities that arise that very quickly turn

23

the Defense Department into needing something new to meet

24

a new military challenge.

25

Even though there wasn't a future goods, a
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

36
1

specific program or product on the drawing board, both

2

the Justice Department and the Defense Department felt

3

very strongly that one of the reasons to block that

4

merger was to preserve the capabilities of designing and

5

producing a particular weapons platform.

6

I think that is a step sort of removed from

7

Dick's position of we have to have a specific good that

8

we are looking at, even if the first one hasn't been

9

produced yet.

10

proceed would have very serious anti-competitive effects

11

for our economy, and in that case, for our national

12

security.

I think it's a case that if allowed to

13

The second point I want to make is moving even

14

further back away from the future goods markets, and this

15

may be something that is specific to the pharmaceutical

16

industry, and that's an industry that I did a lot of work

17

in, but may be applicable to others.

I'm not sure.

18

That is apart from looking at what the impact

19

of eliminating the merger is on the future goods market,

20

is it going to mean that there will only be one company

21

selling a product at some point.

22

There is actual competition going on during the

23

research and development phase when two companies know

24

that they are both trying to develop the drug, and want

25

to get approval first.

Those companies will, for

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

37
1

example, do clinical trials that involve many more

2

patients, to make sure they get a broader range of

3

patients with different kinds of seriousness of disease.

4

The kinds of clinical trials they do, the

5

amount of effort and outreach to patients they do, there

6

are benefits right now in the present of having two

7

companies competing to develop a drug, completely

8

separate from the actual benefits, potential benefits,

9

that we are not sure will be realized because we don't

10

know yet if both companies will successfully develop the

11

drug.

12

I think that may not be completely responsive

13

to your question, Mark, but I think that gives an idea of

14

the different kind of scenarios that gives me concern, if

15

we take Dick's position to say absent a very identifiable

16

good that companies are actually producing or are about

17

to produce, the agency should simply step away.

18

MR. SUNSHINE:

Let me just jump in here, too.

19

I was actually tremendously comforted by Dick's last

20

example, but if we go back to the engine example, and

21

engines take years and years to develop, but if it's that

22

situation where Dick sees a potential enforcement action,

23

what we are talking about is the area of competitive

24

interaction between the two companies is in research and

25

development.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

38
1

Yes.

There will be a goods market, but that

2

goods market is a long way off.

3

we normally consider to be within the reach of Section 7

4

of the Clayton Act.

5

Usually outside of what

Innovation market is an immediate step to

6

identify where the activity of the two companies come

7

together and where there might be an actual reduction.

8

In the case that Dick put forward, there will

9

be less innovation competition in developing this new

10

engine some years off, which will translate itself into

11

output.

12

If you want to say, no, no, that's not an

13

innovation market, that's a future goods market, fine.

14

Now we are talking semantics.

15

MR. RAPP:

Are we talking semantics then if we

16

rule out the application of the innovation market concept

17

to the Genzyme case?

18

we could make about how many goods would otherwise go to

19

market, what the future market would look like.

20

are no such goods.

There, there is no prediction that

There

21

I see that as a sharp distinction.

In other

22

words, my primary criteria is that the analysis not deal

23

with issues like the degree to which R&D spending will be

24

cut back or reduced.

25

some future, some prediction about a future goods market.

It has to do with an analysis of

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

39
1

That's the distinction I want to draw.

2

MR. SUNSHINE:

How do you get there?

How do

3

you get to the output effect on the goods market without

4

understanding what's going on at the research level?

5

MR. RAPP:

It maybe that what I'm talking about

6

comes very, very close to the concept of potential

7

competition, where we are already very close to home.

8

You can do that even if the design of the goods

9

hasn't occurred, if you can predict the future.

10

a closed conversation.

11

MR. SUNSHINE:

This is

But just on the potential

12

competition, you would then relax the two year time

13

frame.

14

MR. RAPP:

15

MR. SUNSHINE:

16
17
18
19
20
21

Sure.

It means nothing to me.

Now, I think we are just in

different legal labels for looking for the same effect.
MR. FRANKENA:

All right.

You will need to get

a microphone.
MR. DANIEL:

I'm Tim Daniel, the aforementioned

NERA colleague of Dick.
I would be curious -- I liked the jet engine

22

example a lot.

Suppose one flows into that going into

23

the production process that Boeing and everyone else

24

knows that there is some probability that this engine

25

simply will not be a success, that the project they have
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

40
1

on the table is for an airplane that requires advances in

2

science and advances in material science or something

3

where there is some probability that engine won't work.

4

So, you don't know with certainty there will be such an

5

airplane in need of such an engine.

6

I would be curious if the panelists generally

7

could just speak to how that analysis might change if

8

that were thrown in.

9

MR. SUNSHINE:

I think it's an essential part

10

of the analysis.

11

on what exactly are the incentives of the parties, and

12

what effect is the combination going to have, factors in

13

terms of are there efficiencies that come from it, the

14

necessary investments, because of the size of the project

15

and the risk of the project, whether it requires one firm

16

instead of two.

17

Again, the idea has got to be a focus

All those considerations go into the question

18

of whether this makes sense or not.

19

we have talked about before, applying straight horizontal

20

merger guideline standards to this innovation question is

21

misplaced.

22

For all the reasons

If this project is so risky that two firms need

23

to do it together or it won't get done, I think that

24

answers the question.

25

Let me stop there.

I'm sure Dick doesn't feel

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

41
1

differently about it.

2

different view.

3

I don't know, Ann, if you have a

MS. MALESTER:

It seems to me it is not only

4

the question of is the standard that Dick puts forward

5

requiring that we know that a particular product be

6

successful, that they will in fact design an engine,

7

which I think really is a pretty extreme position, but I

8

would take the added step of saying if there are only a

9

very small number, in this case, two companies, that have

10

the technology and know how and expertise to design

11

aircraft engines, then we should be concerned about

12

allowing that merger for future aircraft engine

13

developments that aren't even yet on the drawing board,

14

and that was really why I brought up Lockheed/Northrop.

15

Stealth technology.

That was another reason

16

why the Defense Department and Justice Department felt it

17

was critical to keep a number of companies that knew how

18

to --

19

MR. RAPP:

The quick mental analysis that you

20

did to reach that conclusion doesn't really have much to

21

do with an analysis of what is going to happen to R&D

22

efforts.

23

if it is decades down the road, how many goods are going

24

to be available at the end.

25

it's processible in economics, as opposed to assertions

It has to do with being able to forecast, even

I'm okay with that because

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

42
1

about whether more R&D or less R&D effort will arise and

2

whether that's a better or a weaker thing.

3

somewhere in there, but I won't bother illustrating it.

4

A decade or so ago, we had 27 drug companies,

This slides

5

big drug companies.

6

Is anybody prepared to say that major losses to R&D

7

effort have arisen because of the combination of those?

8

I don't think we are in a position to say that the

9

connection between concentration and innovation --

10

These days, we have seven of them.

MS. MALESTER:

I think that's a very, very

11

different proposition than what we have really been

12

talking about, which is competition or the lack of

13

competition, not whether or not there are 27 or 10, I

14

don't think in any current goods market that would

15

concern us.

16

I do think -- I don't think anyone on this

17

panel has said that we can look at the dollars spent on

18

R&D and say it's more or it's less and that means we are

19

going to get a better product, we are not.

20

Those are really not the issues on the table.

21

The question is whether having competition in designing a

22

new weapons system, for example, is going to provide the

23

purchaser with a better product in the end, a more

24

innovative, a more radical breakthrough.

25

I think there are quite a number of historical
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

43
1

assessments you can look at that say yes, they do.

2

MR. HOVEN:

I'm John Hoven, Justice Department.

3

I'd like to make the point that each of you

4

were just talking about tools in the shed, and there are

5

many others, and yours are relevant for some purposes and

6

not for others.

7

a good illustration of that.

8

I think the Lockheed/Northrop merger is

The presumption of the innovation markets

9

analysis is that innovation is a very predictable

10

process, just like building bricks.

11

and you get a product.

You spend the R&D

12

A good example is the expectation that the kind

13

of innovation we are looking at is innovation that has a

14

known identifiable product at the end.

15

extent, that was true, and to some extent, that was not

16

at all true in Lockheed/Northrop, that innovation in many

17

industries is a process of ideas coming together in

18

unexpected ways and you want to preserve an industry

19

structure that allows that to continue to happen, and in

20

particular, in some cases, an innovation structure that

21

generates ideas, products that nobody can think of yet.

22

I think to some

I think the general process of innovation

23

analysis ought to be one in which one inquires how does

24

innovation take place in this particular industry, and

25

maybe it's the kind of approaches you are discussing, but
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

44
1

there are a lot of others that should be examined as

2

well.

3

MR. SUNSHINE:

I think certainly in the

4

framework that I put forth, and I think in the framework

5

that actually Muris puts forward in the statements, is

6

the recognition of what you said is exactly right.

7

The question is should we examine the

8

innovation market and under what circumstances can we

9

actually accomplish innovation effects.

10

minority of cases.

11

predict it, then you should go home.

12

It's a small

If you can't have the confidence to

Once you say that, then if you're talking about

13

innovative activity, there are three categories of what

14

can happen.

15

the future product.

16

in existing product, or it can be innovations that won't

17

be used at all.

18

It can be innovations that will be used in
It can be innovations that are used

That last category I don't think antitrust

19

enforcement should care about.

For the first two

20

categories, now we are talking about legal theories, is

21

it a competitive effect of a goods market, is it a

22

potential competition case.

23

to that is.

24

have to have developed your legal theory and put it

25

forward.

I don't care what the answer

When you get into a court, you are going to

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

45
1

To me, that's the connection.

Find the

2

innovative effect, put it with confidence under the given

3

set of facts, if you can.

4

in the business.

5

there's a case.

6

If you can't, you shouldn't be

Identify an output effect, and then

MR. RAPP:

May I just add a sentence with that

7

in mind?

8

statement in Genzyme, but I think, and this is the

9

troubling part from my standpoint, that it is also

10

consistent with the dissent, in that Commission Thompson

11

may have used the same reasoning and had sufficient

12

confidence in his ability and that of his advisors and

13

staff to see through these issues.

14

example of what can happen when innovation market

15

analysis goes wrong.

16

It is perfectly consistent with the majority

MR. SUNSHINE:

I think it's an

I agree with your criticism of

17

Commission Thompson's dissent, and I think it is wrong.

18

I think in that dissent, he does make a straight analogy

19

to a goods market, for the reasons we have talked about,

20

that is improper.

21

Commission, I think it would be misguided.

22

what Chairman Muris says.

23

If that were the policy of the

MR. FRANKENA:

That's not

I have a question for Dick.

24

Based on Chairman Muris' statement, and if you just

25

accept the facts, are you comfortable with that word of
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

46
1

"analysis," of trying to reason out the effects on

2

incentives and so forth?

3

MR. RAPP:

Yes.

I'm comfortable with

4

"analysis" as long as it reaches the conclusion that is

5

functionally the equivalent of not making innovation

6

market policy.

7

One of the reasons that I admire what the

8

Chairman's statement does is that it reasons sensibly by

9

analyzing the incentives to an outcome that says let's

10

leave this alone, and what I find hard to imagine, and

11

maybe it's the shortage of imagination on my part, I have

12

a great deal of respect for our former chairman who would

13

advocate this policy, of Steven and Rich Gilbert, who are

14

the parents, but I cannot imagine a well reasoned

15

economic statement -- a statement that is well reasoned

16

in economics that says and now let's intervene.

17

MR. FRANKENA:

Taking a broader view here, one

18

of our panelists wrote in the mid-1990s antitrust laws

19

and merger enforcement in particular have not focused

20

sufficiently on the consequences of market power for the

21

pace of industrial innovation.

22

I'm just wondering broadly whether you think

23

the agencies are doing sort of the right amount, too

24

much, and so forth.

25

MR. SUNSHINE:

Those words sound strikingly

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

47
1

familiar.

2

I don't know where they came from.
Again, those words were written in 1994 and

3

1995.

4

there has been a whole set of investigations by both

5

agencies where there has been real focus on innovation.

6

There are cases where I would certainly agree

7

with Dick that innovation analysis may have been carried

8

to the extreme or inappropriately applied or applied

9

without basis.

10

innovation is under enforced, but as I said today and I'm

11

glad we stuck with the analogy of tools in the shed, it

12

is a tool in the shed.

13

the shed.

14

circumstances.

15

sure.

16

I think that during the time period after that,

I would not take the position today that

It is not the sharpest tool in

It's one that should be used only in the right
I don't feel it's under used today for

MR. RAPP:

I'm going to not answer the question

17

directly but raise an issue that arose out of the global

18

marketplace report, the 1996 report, and it was

19

specifically quoted in Chairman Muris' statement, which

20

puzzles me enormously, and I wonder whether I am the only

21

one.

22

That is this business of saying we recognize

23

the potential infirmities with this approach, so we are

24

going to be very cautious and conservative and apply it

25

in circumstances where only small number mergers are
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

48
1

involved.

2

Is that a fair statement, Ann?

3

MS. MALESTER:

4

Do you mean small number of

competitors?

5

MR. RAPP:

Small number of competitors.

6

that is not conservative.

7

one that thinks that?

8

avoiding waste, in that if you don't carry innovation

9

market analysis into every seven into six merger, you are

10

receiving resources in some sense, but in those settings,

11

in the large number settings, you are going to have --

12

the likelihood is that the merger is going to be allowed

13

because somebody else is going to win, and if there's a

14

false positive that arises in your innovation market

15

analysis, in your merged number setting, it's not going

16

to be that consequential because somebody else is going

17

to get to the finish line.

18

That's radical.

To me,

Am I the only

It's conservative in the sense of

My definition of non-conservative radical

19

enforcement is to take this controversial concept to look

20

only to those mergers where it will be consequential

21

because you are intervening in a merger that is going to

22

determine the output of innovation, if you will, and

23

where if you get it wrong, there has to be harm as a

24

result.

25

do mean by "conservative," but we shouldn't lose sight of

It's a semantics issue on the one hand, what you

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

49
1

the fact that we are only doing it into two to one

2

mergers, so it must be benign.

3

opposite is true.

4

MS. MALESTER:

At least in my view, the

I think really the reading of

5

what that global report meant and what the Chairman, I

6

think, alluded to, is simply the consensus that an

7

unilateral theory is by far the most likely if we are

8

going to apply an innovation market at all, and generally

9

speaking, you are looking at a very small number of

10

companies that have the specialized assets, so to speak,

11

to be innovating in the market before you are even

12

starting to think about unilateral theories.

13

was the impact of that.

14

MR. RAPP:

15

MS. MALESTER:

That really

I don't disagree.
In terms of your point of yes,

16

it's radical because where it makes a difference, we will

17

just turn it around and say it makes a difference, and

18

from my point of view, because I think protecting

19

competition where it makes a difference is important, I

20

think it is the right place to put our efforts.

21

MR. RAPP:

If that statement were always

22

prefaced by the prior statement, that the reason to be

23

conservative is that we are dealing in a realm of

24

uncertainty when we can't be all that confident that the

25

decisions that we make will be the correct ones, if it
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

50
1

weren't for that, I could go along easily.

2

combination of those two things together strikes me as

3

problematic, my only point.

4

MR. SUNSHINE:

The

Let me ask you one question tied

5

to Genzyme.

Suppose that the drug in question did not

6

qualify for orphan drug status.

7

outcome of the case?

8

you who are not pharmaceutical, it's not a winner take

9

all situation.

Does that change the

Orphan drug status, for those of

10

MS. MALESTER:

11

that go into coming to a decision.

12

orphan drug situation adds a complication, but in and of

13

itself, shouldn't bar there being a case altogether, and

14

it's one factor you look at in a very large number of

15

factors in assessing whether or not the Commission should

16

take action.

17

MR. FRANKENA:

I think there are a lot of facts
In my own view, the

Just one final question.

Is

18

there anything that you think the antitrust agencies

19

could or should try to clarify about their approach to

20

analyzing the effect of mergers on innovation?

21

anything left they could actually do?

22

MR. RAPP:

Is there

I think conversations like this are

23

illuminating.

I can't think of anything other than that.

24

You know, seeing Greg Werden and Luke over there sitting

25

side by side leads me to ask a question.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

If it's not

51
1

fair, forgive me.

2

Are there or have there been interactions

3

between the agencies about the differences in the

4

application of it?

5

the comparison.

6

drugs gets a lot more cases where entry barriers are high

7

and things like that.

8

or not there is interaction between the agencies on

9

enforcement of innovation markets.

10
11

I won't ask for specifics.

Obviously, the FTC, with its focus on

MR. FROEB:

I'm curious to know about whether

Not on specific cases, but

certainly on general policy matters, I'd say yes.

12

MR. WERDEN:

13

MR. SUNSHINE:

14

None that I participated in.
I thought Greg and Luke were

working on the 2004 innovation market merger guidelines.

15
16

That's

MR. FRANKENA:

Thank you very much.

I'd like

to thank both our panelists and the audience.

17

(Applause.)

18

UNILATERAL EFFECTS

19

MR. FROEB:

20

unilateral effects analysis.

21

today.

22

Welcome to the session on
We have five speakers

We have Greg Leonard.

He's here.

I know he's

23

here.

We will start with Valerie Rabassa from the

24

European Commission, followed by Joe Kattan, Greg Werden,

25

and Tad Lipsky, and then Greg Leonard, if he gets here in
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

52
1
2

time.

He will bring up the rear.

There he is.

I want to thank the panelists for agreeing to

3

come on relatively short notice.

I also want to thank

4

the staff who has put together this conference, and also

5

worked on the enforcement data, getting the enforcement

6

data together, and releasing that as part of our desire

7

to be more transparent to the external bar and the

8

external community.

9

I also want to acknowledge the efforts of my

10

predecessor, Dave Scheffman, who began much of the work

11

that is now just coming to fruition.

12

By 1999, they had displaced the structure

13

conduct/performance paradigm in industrial organization

14

economics, thinking on pricing and output coordination in

15

oligopolies had evolved considerably from the view that

16

it made coordination almost inevitable.

17

While economists never entirely rejected

18

coordinated effects theories for mergers, they did reject

19

exclusive reliance on them, and they had more plausible

20

theories for many cases.

21

that unilateral effects analysis appeared prominently in

22

the 1992 horizontal merger guidelines, which were jointly

23

promulgated by the FTC and the Department of Justice.

Thus, it was not surprising

24

The unilateral effects analysis satisfied the

25

attorneys' demand for simple intuition that they could
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

53
1

understand and explain to a judge, and at the same time,

2

it satisfied the economists' demand for more rigorous

3

analysis.

4

structural theoretic models to make quantitative

5

predictions of unilateral competitive effects.

6

It wasn't long before economists began using

In a price setting model, price goes up.

In an

7

option model, firms don't have to bid as aggressively to

8

win.

9

bargaining power.

10

we learned that mergers aren't usually profitable with

11

equality setting models.

12

In a bargaining model, the merged firm gains
In the equality setting model, well,

The controversy surrounding unilateral effects

13

analysis has focused on the application of structural

14

models to individuals cases, their main virtue is they

15

force assumptions to be made explicit and they provide a

16

clear mapping for the facts of the case to the effects of

17

the merger.

18

The shortcoming is that the models are

19

necessarily unrealistic and abstract away from the

20

important features of the industry. As such, the results

21

may be quite misleading.

22

Unfortunately, we have little evidence on

23

whether these models can accurately predict the effects

24

of real mergers.

25

about what is the best way to analyze unilateral effects

Instead, we are left with controversy

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

54
1

theories.

2

What follows is we will hear from Valerie

3

Rabassa, Joseph Kattan, Greg Werden, Tad Lipsky, and Greg

4

Leonard on this and more general topics.

5

with Valerie.

Let's start off

6

MS. RABASSA:

Thank you, Mr. Chairman.

7

Let me talk about the Lagardere case.

It's a

8

European case which took place last year in France.

9

was analyzed by the European Commission.

10

the leading 2003 cases of the media industry.

11

Let me talk about the transaction.

It

It is one of

It was very

12

simple.

13

the first and second player of the book industry.

14

book industry in Europe is characterized by a very high

15

level of vertical integration.

16

industry.

17

publishing right, and who sell the rights to the

18

publisher, and distribute the books to the retailers who

19

distribute or sell the books to the final consumer.

20

given the particularity of these cases was an econometric

21

study carried out for the Commission by Professor Marc

22

Ivaldi, and it studied basically the effects of trash and

23

trees in the downstream segment.

24
25

The transaction concerned a transaction between
The

Different animals in this

In the upstream market, you have the

And

It was very interesting because it was the
first time that we incorporated an econometric study into
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

55
1

a final decision.

In this study, they will in effect

2

measure, as you know, the entire concentration, the fix

3

in price, and then we have the impact to the final

4

consumers who produce and choose a profit of the firm.

5

So this was the launch with VUP, if Lagardere

6

had decided to increase price obviously unilaterally,

7

some of the final consumers would turn to the other

8

competing publisher, who may reach VUP.

9

thought of the concentration with VUP has shared a

10

subsidiary of Lagardere to accept part of this

11

competitive pressure, and so can recover part of these

12

customer.

So I've always

13

In this case, we use a very strong model, the

14

nested logic model, which is quite adapted to the book

15

industry, which is characterized by differentiated

16

products.

17

family of the discrete choice model, which was very

18

interesting in this case because obviously you cannot buy

19

too seminal a book.

20

The logic model, as you know, is from the

The consumer in this case made a discrete

21

choice, a model, a different set of economists and then

22

choose a book on the concerned list.

23

consistent with the Bertrand Model of competition and the

24

estimation was three stage least-squared.

25

Simulation is

So the reasons why it's quite interesting is
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

56
1

because we find that the end price of the books were

2

significantly increased in the downstream segment.

3

Obviously, we were looking at the consumer surplus and we

4

find that consumer surplus will also fall significantly,

5

which was equivalent to a very negligible part of the

6

turn-over of industry in the field I'm generally tied to.

7

What was wrong obviously there is that the

8

price increase was linked with the market size, and we

9

were able to reconstitute the market size in this case.

10

Again, we decided to receive -- the results

11

were quite robust to incorporate it into the final

12

decision, and we have used a boot strap method to

13

construct confidence intervals that are quite often used

14

in the econometrics field and we find that there is only

15

five percent variability that the price rise to the

16

concentration that could be including significant a

17

significant interval of plus or minus one percent, with

18

the mean value of the price change.

19

Results were quite robust because of the very

20

high number of observation to at least 10,000.

21

different statistical traits were quite significant, and

22

the main parameters, I mean the marginal ATAF of a given

23

book and the intra-brand correlation were quite stable.

24

Altogether I feel that's reasonable to quite robust as

25

for when we decide to incorporate this econometric study
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

The

57
1

into the final decision.

2

Thank you very much.

3

MR. FROEB:

Joe, you're next.

4

MR. KATTAN:

I want to talk about, I think,

5

differences between the ways that lawyers tend to look at

6

unilateral effects and the way economists do, and

7

particularly with reference to the new models, the merger

8

simulations that are being done increasingly in

9

unilateral effects cases.

10

I want to begin by just talking a little bit

11

about the 1992 guidelines and what they say, and to

12

contrast that a little bit with the new simulation models

13

and pose the question whether we are asking the same

14

questions, that is do the guidelines give guidance or do

15

these simulation models answer the same questions that

16

the guidelines are asking.

17

I think it's pretty clear that the 1992

18

guidelines really changed the way that people look at

19

unilateral effects before 1992.

20

basically thought of in terms of monopoly or dominance,

21

and the insight of the guidelines was that unilateral

22

effects can arise outside the monopoly context.

23

Unilateral effects were

The merger of Daimler and Chrysler does not

24

exactly have the same effect as the merger of Daimler and

25

BMW, and if you were just looking at HHIs and market
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

58
1
2

shares, you may get a misleading answer.
Some people thought we were returning to

3

submarkets, which was a term that had been discredited at

4

least within the Beltway.

5

litigating a case on behalf of the agencies, I would rely

6

on the submarket nomenclature simply because it is

7

something that the courts are familiar with and are

8

comfortable with, regardless of whether that nomenclature

9

really makes a lot of sense in terms of the way we do the

10

unilateral effects analysis.

11

Certainly, if I were

I think it is certainly the case that inclusion

12

of unilateral effects was the most important change in

13

merger law since the 1982 guidelines, and had a very

14

profound effect on merger enforcement, and it wasn't that

15

very long ago that one could hear people talking about

16

whether unilateral effects was in fact the only valid

17

theory for looking at mergers.

18

swung back a little bit.

I think the pendulum has

19

I think it's fair to say that what we saw in

20

the guidelines is a synthesis of good economic theory,

21

but also practical judgment about the qualities of the

22

tools we have available to us in merger investigations,

23

and how refined a judgment we can make about the effects.

24

To repeat very briefly what the guidelines say,

25

not to state the obvious, but to give a benchmark for
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

59
1

what we are looking at, the guidelines focus on spacial

2

separation between products and product space.

3

is on localized competition where certain sellers compete

4

more closely with one another than with other sellers who

5

are in the same product market.

6

The focus

According to the guidelines, that requires that

7

there be a significant set of shares of consumers who

8

regard the product of the merging parties as their first

9

and second choices, and certainly the way that lawyers go

10

about applying the guidelines is by asking very often

11

about whether we are dealing with first and second

12

choices.

13

when trying to look at a case for the first time.

14

That's one of the first screens one applies

We have a 35 percent screen for the merged

15

entities' market share, the debate whether that is a safe

16

harbor or something else, but in fact, we have seen cases

17

where the market shares were lower than that that have

18

been brought.

19

I think these reflect a pragmatic tradeoff

20

between quality of the analytical tools that are

21

available in merger investigations and the theory.

22

certainly possible to argue, in fact, it has been argued

23

by Greg Werden and others, that you can have an

24

anti-competitive effect in a merger where the merging

25

parties are not the first and second choices of
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

It is

60
1

customers, and you can certainly draw the theoretical

2

picture that would be supported and have numbers to

3

support it.

4

The question certainly in 1992, and not

5

accounting for all developments and the analytical tools

6

that have taken place since then, was what are the tools

7

that we have and how refined a judgment can we make, and

8

the judgment was, at least at that point, we are going to

9

look at next best substitutes.

10

Now, the way that the lawyers tend to approach

11

the issue of unilateral effects is, I think, familiar to

12

most people here.

13

You define a market, and you do that because not only the

14

guidelines say that, but Section 7 case law says that.

15

More broadly, I think doing an antitrust case

You look at markets, first of all.

16

without defining what the market is is a recipe for

17

disaster.

18

There was a recent case involving Rambus

19

Company that is in the news, where Infinion brought both

20

antitrust and fraud claims against Rambus.

21

claims were thrown out because Infinion forgot to define

22

a geographic market for memory chip technology, something

23

that probably would have taken two paragraphs in the

24

expert reports.

25

The antitrust

I think economists often take the position that
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

61
1

defining markets is superfluous if they can get to the

2

answer more directly, and that may or may not be right,

3

but certainly from the viewpoint of the courts, if you

4

don't define the market, you may find yourself out in the

5

street, given a case like Infinion vs. Rambus, where the

6

definition of the geographic market was clearly a red

7

herring.

8

Once you define the market, I think the lawyers

9

focus on the issue of next best substitutes.

Why?

10

Because that's what the guidelines say, and the analysis

11

is driven by interviews and documents.

12

is to ask for way too many documents so that you can have

13

boxes and boxes of documents stacked up in the hallways.

14

I think there is a critique that one hears from

A lawyer's fetish

15

the economists that lawyers use models but don't

16

articulate them with sufficient particularity.

17

Certainly, the legal method does not require articulation

18

of the model the way the economic method does, and when

19

you look at the economic model, it tells you what the

20

assumptions are and if

21

the limitations are.

22

the models are.

23

assumptions are not made, what
Lawyers don't tell you exactly what

The model effectively if we look for

24

significant market share.

We look for next best

25

substitutes, and that embodies certain assumptions on the
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

62
1

effects of the internalization of loss sales.

That is

2

sales that would have been lost to the merger partner

3

that now is internalized as a result of the merger, the

4

effects of that on the merged firm's incentives.

5

It's certainly true that the economic models

6

spell out their assumptions with greater particularity.

7

Lawyers have an advantage.

8

courts are lawyers.

9

thrown out on Daubert grounds, you know, the economic

10

models, which may be equally valid, are subjected to a

11

different kind of scrutiny, at least once you get it to

12

the court.

The decision makers and the

You cannot have a legal argument

It's a different issue at the agency.

13

The economists, I think it is fair to say, are

14

less bound by the guidelines because the methodology of

15

looking at what the rules are and applying them, that is

16

a legal methodology, and in fact, I've heard countless

17

times from various economists in various contexts,

18

mergers and otherwise, that the whole process of defining

19

markets and driving the analysis that way is an

20

artificiality, and if we have the tools that allows us to

21

get to the answer more directly, you know, why bother.

22

There is certainly an institutional bias

23

against the approach that the lawyers use.

I think the

24

economists have a desire to try to get closer to "the

25

real answer."

The tools that the lawyers use are fairly
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

63
1

crude in many cases, making predictions about the future,

2

relying on assumptions, some of which are stated, some of

3

which are unstated, and have various degrees of support,

4

and obviously, if through the application of the data

5

methods, data analysis, you can get closer to the real

6

answer, why not try to do that.

7

It is certainly the case that the economic

8

approach has the advantage that the models have well

9

articulated specifications, which the legal approach does

10

not.

11

One of the things that we see now is that

12

economists have an insatiable appetite for data.

13

lawyers want documents.

14

a belief that virtually everything can be solved through

15

data exercises.

Economists want data.

The
There is

16

We have these models, and we can talk about --

17

Tim made the comment to the effect that the analysis not

18

helped by relying on the jargon, but I think there are

19

questions about whether the constraints of Bertrand

20

models are realistic in the context of differentiated

21

markets because they allow one point for differentiation,

22

which is price, and in fact, at least the way the issue

23

is posed in the guidelines, we are looking at a host of

24

variables that affect the positioning of different

25

competitors in product space.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

64
1

The insinuations are with us, and I think they

2

are going to be with us for a long time, whether or not

3

the lawyers like it, we are going to have to deal with

4

it.

5

Modeling simulations can be mutual if we are

6

all comfortable that they are attuned to the market

7

realities, that they are based on defensible assumptions,

8

and that's obviously where the debate has been.

9

One criticism has been that a small difference

10

in the assumptions can make very big differences in the

11

results.

12

economists is well, we have different ways of doing

13

things, we should probably do it more than one way and

14

impose some kind of reality check on your analysis and

15

also try to use the method that's the most conservative

16

from the standpoint of the party that you are.

The answer that I think you will hear from the

17

I'll pose the question whether there is a

18

conservative assumption from the standpoint of the

19

plaintiff, as all these models seem to be designed to

20

show a price increase.

21

Part of the issue that I have is price

22

increases are assumed in these models based on structural

23

assumptions, even though we are trying to run away from

24

structural analysis.

25

The question is what are we trying to show
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

65
1

here?

Are we trying to show what the magnitude of the

2

price increase is.

3

that's not what we are trying to show, but I think that's

4

where it comes out.

I think everybody will tell you no,

5

When you get somebody telling you that I can

6

predict with 95 percent degree of confidence that you

7

have a price increase of 7.4 percent, you know, some if

8

it even has a decimal point, it sounds so precise that

9

you can debate that.

10

says, well, if you accept all my assumptions and that's

11

also before I take into account issues like repositioning

12

of product entry and efficiencies.

13

Of course, there's a footnote that

There is a risk, I think, of false empiricism

14

where you have these results which are so precisely given

15

with high degrees of confidence and decimal points, you

16

know, somebody is telling you this is the price increase

17

they are going to get.

18

I think this is fairly paradoxical, but the

19

economists' approach is more likely to identify problems

20

with a merger than the more traditional way the lawyers

21

do, simply because the models are designed to predict a

22

price increase, and because these models will predict a

23

price increase, even in relatively unconcentrated

24

markets, I think they tell us something, which is either

25

the models have a problem or the guidelines are wrong
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

66
1

because the models will predict a price increase in

2

situations where the merger would not even get a second

3

look under the guidelines, things that are below 1500.

4

The economists' answer is well, we are not

5

really trying to predict the price increase, we have to

6

take into effect entry repositioning and efficiencies,

7

but in fact, if you accept this methodology, then what

8

you have done is really shifted the burden of proof to

9

the merging parties, because you have said on the basis

10

of some data analysis that you have a price increase

11

unless these other factors come into play, so now please

12

prove to me that this price increase will not happen.

13

In fact, to really say to the parties in most

14

cases prove repositioning, because anybody that has done

15

a lot of mergers will know that most mergers, the

16

efficiencies that we see are not efficiencies that affect

17

the incremental costs in a meaningful way.

18

efficiencies that I've seen tend to be overhead

19

synergies.

20

real.

21

the end of the day, they are not the type of efficiencies

22

you can expect to translate into a beneficial price

23

effect.

24

very hard to show entry, particularly when you are

25

talking about differentiated product markets, where you

Most of the

They are very important and they are very

They drive merger economic efficiencies, but at

Entry is kind of a defense of last resort.

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

It's

67
1

have large investments integrated with differentiation,

2

that is promotion, advertising, and also distribution

3

arrangement.

4

The parties are now in a position of being told

5

your merger is anti-competitive, you have to prove it's

6

not going to have a price effect, and really your resort

7

is repositioning.

8

common.

9

implicitly the burden of proof to the merging parties by

10

effectively assuming a price increase and really saying

11

the only issue is how big a price increase.

12

Repositioning is pervasive.

It's

At the same time, we are in effect shifting

Another question to ask is what happens to the

13

next best substitutes analysis.

We talked at the

14

beginning about what the guidelines say.

15

say, well, we look at whether the products of the merging

16

parties are first or second choice of the large share of

17

the customers, and what we wind up with are legit models,

18

and really instead of being differentiated by product

19

attributes or by promotional issues, products are

20

differentiated by their sales level.

21

differentiation that the analysis assumes because that is

22

really what we are looking at.

The guidelines

That is the form of

23

Of course, we are assuming identical cross

24

elasticities of all products with respect to a given

25

product, which given the assumption that some products
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

68
1

are closer substitutes to each other than others, it

2

seems to at least raise some issues.

3

The more complex models, they impose great

4

cost.

They have an insatiable appetite for data.

There

5

are issues with using retail level data as a proxy for

6

wholesale competition.

7

Dave Scheffman and others have put together a

8

very detailed analysis of that, and it deals both with

9

implementation issues, what do you do when a price

10

increase takes place in mid-week, but the data you have

11

collects prices on a weekly basis, as well as conceptual

12

issues, that most wholesale prices now really are two

13

part tiered and relatively fixed and a variable element,

14

and obviously consumer prices are an one part tier.

15

The issue that I guess I will end with is

16

whether these models follow the guidelines.

17

source of anti-competitive unilateral effects that the

18

guidelines are talking about has to do with the spacial

19

separation of product, are we really obscuring that issue

20

by looking at models that are defining differentiation

21

effectively based on the levels of sales, are we also

22

shifting the burden of proof to the merging parties

23

implicitly by doing that.

24
25

If the

I was actually heartened to see that one of the
recent presentations Luke did showed sensitivity to the
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

69
1

cost of doing these types of analyses.

2

anybody who is doing a merger now is kind of forced to do

3

that as a defensive measure because the economists are

4

going to come to you and say, well, this is what DOJ is

5

going to be doing, and we need to do this because

6

otherwise, we are going to be unprepared for their case.

7

It's another burden of merger enforcement on top of the

8

endless boxes of documents.

9

The fact is

My feeling is that the current models are going

10

to have problems in court, but at the same time, most

11

cases are decided in front of the agencies, and parties

12

will be foolhardy to ignore these models because they

13

have become part of that decision making process, and

14

what needs to be done, which is very difficult because we

15

don't have the investigative tools to look at consummated

16

mergers and see what happens and calibrate the

17

assumptions in these models to actual market results, is

18

to try to refine them and make them better.

19

MR. FROEB:

Thank you, Joe.

Valerie, just

20

because Joe raised this question, did you look at the

21

retail sector to see what kind of contracts they had with

22

publishers?

23

tiers?

24
25

Did they use mark up pricing or two part

MS. RABASSA:

In the Lagardere case, we had a

big problem to gauge that, because frankly, the sale of
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

70
1

books to retailers and wholesale was the sale of books to

2

the final consumer.

3

right data.

4

problem of getting data.

We had many problems getting the

We can talk about this problem later, the

5

MR. FROEB:

Thank you.

Greg?

6

MR. WERDEN:

Before I get to my prepared

7

remarks, I want to respond to a couple of things Joe

8

said.

The first one was a serious misquotation of my

9

views.

He paraphrased the guidelines, which I will

10

actually quote to you, and then said I disagreed.

11

don't.

12

I

The sentence I will quote says "Substantial

13

unilateral price elevation in a market for differentiated

14

products requires it to be a significant share of sales

15

in the market accounted for by consumers who regard the

16

products of the merging firms as their first and second

17

choices."

18

I totally agree that is a very nice way of

19

putting the point, notice it said significant share, not

20

a large share,

21

share, and you don't need a large share.

22

wants to then translate this into saying there is some

23

global ranking of the products, one, two, three, four,

24

and the merging products have to be closest in this

25

global ranking.

I think Joe wants to switch it to large

That is wrong.
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

Second, he

71
1

The crucial thing to know is that typically in

2

differentiated products, different consumers rank the

3

product differently.

4

product and some people will buy another one.

5

That's why some people buy one

What the guidelines say, which I totally agree

6

with, is are you using a significant number that is

7

viewed as the first and second choices.

8

Secondly, he suggested that modeling, the kind

9

that economists do, merger simulation and other things,

10

is more likely to find mergers problematic than whatever

11

it is that lawyers do instead.

12

statement because that's just totally contrary to my

13

experience.

14

I find this a remarkable

One of the reasons Luke and I got into this 12

15

years ago was the lawyers had tremendously exaggerated

16

notions of likely anti-competitive effects of mergers,

17

and the only way we could think of to bring them down to

18

earth is with economic models, and it has worked.

19

In a lot of cases in which the models have

20

shown that it was implausible there would be significant

21

price effect.

22

enough, even with very small efficiencies, that the net

23

price effect would be positive.

24

didn't bring because of the modeling.

25

It was implausible they would be large

These are cases we

It's hard to say, because we did do the
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

72
1

modeling, but when most of the modeling said there was no

2

significant effect, we didn't bring the case, in every

3

single one of those cases that I know of.

4

I think modeling has probably kept the agencies

5

from bringing some cases, although it might have caused

6

them to bring some other ones they should have brought,

7

and in both cases, that was for the better.

8

Starting with my prepared remarks, the concept

9

of merger simulation.

If the same well specified

10

oligopoly model reasonably describes the outcome of the

11

competitive process, both before and after the merger, a

12

lot of conditions, then that model can be used to

13

generate a quantitative prediction of the merger's

14

unilateral competitive effects after it is first

15

calibrated to match the prices, shares, et cetera, that

16

would prevail but for the merger.

17

All it says is if we run a model in our

18

standard tool kit that fits the industry, and we think it

19

is also going to fit the industry, except for the merger,

20

second assumption, and if it predicts what we care about,

21

like prices and maybe more, then we know how to use that

22

model.

23

and shares and elasticities or whatever we observe for

24

the industry, and to generate what the post-merger

25

equilibrium looks like, taking into account the

We know how to calibrate it to match the prices

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

73
1

internalization of the competition between the merging

2

firms.

3
4

We can put numbers on what the guidelines talk
about in some cases.

5

From a plaintiff's perspective, and since I've

6

worked for the Department of Justice, that's the

7

perspective I usually take, I think that's very helpful

8

in several respects.

9

to tell the judge that there is a significant number of

10

customers who view the merging products as first and

11

second best alternatives, and it's quite another thing to

12

say the likely implication is they will raise price 10

13

percent.

In the first place, it's one thing

14

I think you are much more likely to win the

15

case if you can make the latter statement than if you

16

can't, and if you have something to back it up, of

17

course.

18

Plus, there is a way to put things in

19

perspective, like cost reductions.

20

off if you do a quantitative analysis.

21

otherwise.

22

You can trade them
You can't

That's really my second point here in the

23

slide.

There are certainly cases in which mergers may

24

have significant efficiencies or in which we think even

25

if we don't have specific evidence of efficiencies, it's
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

74
1

very plausible the merger will have some efficiencies.

2

It's very easy to take those into account in merge

3

simulation and very easy in some cases to show that

4

plausibly the net effect is a price decrease, not a price

5

increase.

6

We don't bring those cases.
Merger simulation can be an useful way to focus

7

an investigation or try by identifying which factors or

8

assumptions really matter.

9

modeling exercise that one assumption gives you one

10

result, another assumption gives you a very different

11

result, then it becomes a very crucial fact which is

12

which of those assumptions is the right one or the better

13

one.

If you can show in the

That can be a very useful process.

14

You can also find some things really don't

15

matter much, so you shouldn't put your resources on

16

figuring those things out.

17

Limitations.

The fundamental limitation is no

18

economic model is going to capture every nuance of

19

competition in the real world, but you don't have to.

20

All you have to do is capture enough to be able to

21

usefully predict.

22

Second.

Price increase predictions can't be

23

any better than rough estimates.

There is a whole lot

24

going on that isn't in the model.

All kinds of things

25

are going to happen in the world that you can't possibly
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

75
1

anticipate.

2

effect of the merger, but that's a whole lot better than

3

no idea of the likely effect of the merger.

4

You are only given some idea of the likely

Third.

Merger simulation is basically designed

5

to predict price effects, possibly other short term

6

effects of the merger.

7

long run evolution of the industry, but the good news is

8

Section 7 doesn't care that much about the long run

9

evolution of the industry, which in any event, we have no

10

tools in law or economics to really predict very well in

11

most cases, so what merger simulation gives us is

12

something that the law is looking for, or what it can

13

give us.

It's never going to predict the

14

The basic theme of my talk is that before you

15

use a merger simulation at trial, and for that matter,

16

before you use it for any other purpose for which you

17

give significant weight to the actual predictions, it

18

should have to pass an admissibility screen from the

19

rules of evidence.

20

Daubert, Joiner, Kumer Tire, and many antitrust cases of

21

less prominence.

22

Rule 702 interpreted by cases like

What I take away from Rule 702 and the case law

23

applying it as it applies to economic testimony in

24

antitrust cases is that testimony is admissible if the

25

witness is an expert in the relevant field of economics,
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

76
1

if the testimony employs said methods from the relevant

2

field of economics, and if the testimony reliably applies

3

those methods to the facts of the case.

4

I'll elaborate these a little, mostly the

5

third.

I think the required expert knowledge is a fairly

6

low hurdle.

7

Heisenberg, a famous physicist, who defines an expert who

8

is someone who knows the worse mistakes that can be made

9

in his subject and who manages to avoid them.

10

kind of expertise we need in any area, including merger

11

simulation.

12

make, and you have to know better than to make those.

I like to quote a lot from Werner

That's the

There are some really bad mistakes you can

13

What merger simulation does is use absolutely

14

standard, well accepted tools from economics, so in that

15

sense, theoretical sense, merger simulation is the

16

application of sound methods necessarily.

17

There is also the empirical sense of soundness,

18

which can be very important, and as Luke mentioned in his

19

opening, we don't know very much about the predicted

20

accuracy of merger simulation.

21

very much about the predicted accuracy of anything else,

22

so in a relative evaluation, merger simulation certainly

23

doesn't do worse in predicting than anything else anybody

24

has ever come up with.

25

However, we don't know

I won't offer some of my favorite pieces of
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

77
1

dicta from Supreme Court decisions here.

2

Daubert itself, which held that expert testimony is

3

admissible only if it is sufficiently tied to the facts

4

of the case and will aid the jury in resolving a factual

5

dispute, i.e., only if there is a good fit between the

6

testimony and the inquiry.

7

First, from

Daubert didn't come up with this fit concept.

8

It was borrowed from a Third Circuit decision.

9

it's a very good and precise way of describing what we

10

are looking for, and it's something that certainly

11

antitrust cases are looking for.

12

I think

The same theme expressed in a different and

13

also useful way in the Joiner case.

14

admit opinion evidence that is connected to the existing

15

data only by the ipse dixit of the expert.

16

conclude there is simply too great an analytic gap

17

between the data and the opinion proffered.

18

The court should not

A court may

In my view, it is not uncommon to have expert

19

economic testimony that is merely ipse dixit.

The

20

economist doesn't closely tie his theories to the facts

21

of the case, and I think Joiner is right on track saying

22

this is the kind of thing that courts should not allow.

23

As an example, a very well known one from an

24

antitrust case, the testimony of Robert Hall was excluded

25

in Concord Boat because his oligopoly model, the kind
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

78
1

that you might use for simulation, although hardly

2

anybody does, was not grounded in the economic reality of

3

the industry.

4

perhaps, although I happen to read the case as saying

5

there were like six different ways in which it wasn't

6

grounded in the economic reality of the industry, so I

7

think all of the critiques were fit critiques.

8

This wasn't the only knock on the model

The court said there was nothing wrong with the

9

model as a matter of theory, it was a sound method, but

10

it wasn't the right method for that case.

11

exactly the kind of inquiry a court should do, and I

12

think in the case of Concord Boat, the court was exactly

13

right, it wasn't the right model for that case, and in

14

particular, at least the way the model was calibrated by

15

Hall, it predicted that the defendant would not have more

16

than a 50 percent share without engaging in the

17

challenged practices.

18

I think that's

The fact was the defendant had a 75 percent

19

share before engaging in the challenged practices, and I

20

would say that's got to be improper calibration.

21

only problem with the model the court pointed out, and

22

some of the other ones I think were important as well.

Not the

23

A less well known case but perhaps a more well

24

known economist, Frank Fisher's testimony was excluded in

25

the Booksellers case for purposes of at least determining
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

79
1

damages, which was pretty important in that case, because

2

it contained too many assumptions and simplifications

3

that were not supported by real world evidence.

4

I will not comment on the merits of Fisher's

5

testimony or the court's analysis of it, because I don't

6

know enough about either one, but I will comment on the

7

language used by the court here in excluding the

8

testimony.

9

think about some economic evidence and if it is the right

10

way to think about that evidence, and the right

11

conclusion is that it should be excluded.

12

on too many assumptions and simplifications that are not

13

supported by real world evidence, then it doesn't fit the

14

facts and it isn't admissible evidence.

I think this is exactly the right way to

If it depends

15

What are the key elements of fitting the facts,

16

and I will emphasize these are key elements because this

17

has to be a case by case determination, depending on what

18

the facts are, what the model is, and what you are trying

19

to do with the model.

20

An oligopoly model used in simulation has to

21

reflect critical aspects of competition in the short term

22

at least.

23

aspects of competition.

24

have a basis for saying it reflects the critical ones.

25

That doesn't mean it has to reflect all
It's not going to.

You have to

Some fact based analysis by your experts has to
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

80
1

lead to this conclusion, and he has to be prepared to

2

convince the trier of fact that he is right in thinking

3

about the industry and the model in this way.

4

The model also has to explain the recent past

5

at a fairly high level of generality, especially the

6

intensity of competition, as I like to think about it, as

7

reflected in price cost margins.

8

be explained by models, how the industry has responded to

9

shocks, the level of prices as compared to costs.

10

model should be able to explain these things, not

11

necessarily with exquisite precision.

12

why a model should explain day to day price movements.

13

We don't care about day to day price movements.

14

There is a lot that can

A

There is no reason

The ultimate test, another way of restating the

15

fifth requirement, is that every modeling choice should

16

be justified on some basis.

17

economic theory, for example, the assumption of profit

18

maximization, and needs no further justification.

19

is what economists do.

20

assume that, don't call an economist.

21

It could be dictated by

That

If you don't want somebody to

It can be supported by industry data, for

22

example, if the model is calibrated properly, that fits

23

the price and shares of industry, that kind of data I

24

have in mind.

25

ratios or demand elasticities.

Also, the data may speak to diversion
There are lots of

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

81
1

different ways in which models could be supported or

2

refuted by data.

3

It should be consistent with stylized facts of

4

the industry.

5

important elements of competition?

6

How does competition work?

What are the

If you have a typical differentiated products

7

case and it involves a retail sector, how does the

8

manufacturing sector interact with the retail sector and

9

how does the retail sector behave?

10

These are the kinds of stylized facts that have

11

to be studied before you can properly model an industry,

12

and you should be prepared to explain how the facts

13

support the particular modeling choices.

14

Fourth, some modeling choices may turn out to

15

be unimportant.

16

different choice.

17

choices.

18

model doesn't really matter.

19

justification.

20

You may be able to try a particularly
There may be a major difference in

You may find out for that particular thing, the
That is a perfectly fine

Finally, particularly when you can't do any of

21

the above, it may be in addition to doing some of the

22

above, you can justify a choice by doing a sensitivity

23

analysis and showing that over the range of plausible

24

assumptions, you have picked one that if your plaintiff,

25

for example, leads to relatively small price increase
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

82
1

projections.

2

There are some modeling choices in merger

3

simulation that you are not going to be able to justify,

4

and that's what you have to do.

5

A couple of illustrations.

This is, I think,

6

the easiest and best one of some kind of modeling choice

7

that you are never going to be able to justify based on

8

the facts of the data, and that is what is the assumed

9

functional form for the demand curve?

10

Here I have plotted between the competitive

11

prices over in the lower right and the monopoly prices in

12

the upper left, four demand curves that have been used in

13

merger simulation.

14

competitive price to the monopoly price, it involves a

15

vastly larger price increase for some demand curves than

16

others.

As you can easily see, going from the

17

Where you are going to see one of these or

18

perhaps some other functional form is when you do a

19

merger simulation, and it's going to affect in a very

20

substantial way the price increase predictions.

21

for the plaintiff, I would be using linear demand, which

22

produces small price increases, absent some strong

23

evidence I can't imagine ever having, that one of these

24

other functional forms fits better.

25

If I'm

A final illustration has to do with demand
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

83
1

elasticities.

In differentiated products, merger

2

simulation is a way of translating the demand

3

elasticities into price increase predictions.

4

those predictions are sensitive to the demand

5

elasticities, and it should be obvious that we never know

6

exactly what those elasticities are.

Obviously,

7

Well, so you should consider what the price

8

increase predictions are for a range of elasticities.

9

Here is a very simple illustration of how that might work

10

for the WorldCom/Sprint merger, where a very strong

11

assumption about demand is made so that we can place

12

parameters on this model just based on two elasticities.

13

One is the aggregate elasticity demand for

14

residential long distance service, and the other is

15

WorldCom's firm level elasticity for its long distance

16

service.

17

considers a very wide range of plausible values.

18

The academic literature is estimated in

For each of these elasticities, this plot

19

elasticity demand for residential long distance of about

20

one, and this gives quite a wide range around one, from a

21

half to one and a half.

22

elasticity here is from 1.25 clear up to 4, which is a

23

huge range of elasticities.

24

truth lies somewhere in that interval.

25

The range for WorldCom demand

It's highly likely that the

What this plot shows is that for some
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

84
1

combinations of parameters, a 4/10ths of a price increase

2

on up to 1.2 percent price increase for the average price

3

over the whole residential long distance, which of course

4

includes AT&T and all these other little guys which

5

collectively had, I think, around 70 percent a few years

6

ago, so WorldCom and Sprint were at least a relatively

7

small portion of the industry, significant, but

8

relatively small.

9

other residential long distance service companies that

10

are going to have vastly smaller price increases than the

11

merging firms.

12

So we are averaging out over these

That is why these numbers are so small.

If you down it to .4 percent, it may be very

13

easy for merger specific efficiencies to swamp that, and

14

for the net price increase to be negative.

15

percent, that's not really so likely, and that's probably

16

a price increase big enough that the agencies would worry

17

about it, and you might say very plausibly that WorldCom

18

and Sprint were closer substitutes than this model

19

assumes, if so, the price increases would be even bigger,

20

and the contrary is also correct.

21
22

MR. FROEB:

Thank you, Greg.

Up above one

Joe, do you want

equal time now or later?

23

MR. KATTAN:

I'll wait.

24

MR. FROEB:

Tad?

25

MR. LIPSKY:

Thanks, Luke.

It's a great honor

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

85
1

to be on the same stage with such distinguished

2

practitioners of these various economic approaches to

3

antitrust, and I have to say in listening to the

4

unilateral effects theory, you would almost be grateful,

5

I suppose, that you have a merger before the Commission

6

and before the DOJ where the theory is going to be

7

innovation markets or potential competition where now

8

this is applicable and you can't do simulations because

9

you have no current output for both firms.

10

nice.

That might be

11

MR. WERDEN:

You need an option model.

12

MR. LIPSKY:

As I said before, I think we are

13

focused on the wrong ex-Frenchman.

14

study, and I've said elsewhere many times before, pretty

15

much what passes for antitrust economics right now was

16

all in the original core notebook back in 1838.

17

think unilateral effects were in there.

18

option of the theory of monopoly.

19

Bertrand was an IUV

I don't

Maybe it's an

I want to go at some specific issues that have

20

been alluded to but haven't been raised directly.

21

I can give a somewhat different focus.

22

Maybe

Who decides and how effective are those

23

economics in this process.

I'm going to back over the

24

first principles that I don't remember thinking about

25

since my first day in econometrics as an undergrad, and
For The Record, Inc.
Waldorf, Maryland
(301)870-8025

86
1

probably I should have been thinking about them a lot

2

more than I have, but I think it helps ground the

3

discussion and reveals my approach to some basic points

4

here.

5

First, there are facts, and some facts are

6

really facts, like how much and how long it takes to

7

build a factory.

8

mental constructs of mathematical models, oligopoly

9

models, what have you.

10

model for merger law.

11

There are models which are basically

I have here sort of the classical

If an industry is more concentrated, the output

12

will be lower and price higher than if it's less

13

concentrated, ceteris paribus.

14

statistical tests which are propositions that grow out of

15

the application of a specific econometric method used to

16

estimate the parameters of a specific model, and I have

17

no idea what I'm saying here.

18

based widget market doesn't exist except in my head, but

19

this is the kind of assertion you would find at the end

20

of the articles back in the 1960s and 1970s as to what

21

you actually needed to apply econometric methods to a

22

model in a particular industry.

Then there are

Obviously, the silicon-

23

I want to echo something I heard Joe say

24

earlier. Maybe I won't put the words in your mouth.

25

think it has do with the idea that we are all decisionFor The Record, Inc.
Waldorf, Maryland
(301)870-8025

I

87
1

makers and we are not all judges, we are essentially

2

lawyers.

3

have enough economics or econometrics to understand

4

really what is contained in an assertion like that, and

5

that's one of the problems I'm going to get to a little

6

later in the presentation.

7

There is a certain assumption that they don't

The value of a statistical test, the values of

8

the parameters and all those squares and all that neat

9

stuff that the economists roll around and try to persuade

10

you that what they have done is great or is viable at

11

least, let me give you a very concrete example based on

12

something we have heard today.

13

We had a description of a statistical outcome

14

for a model estimated in the context of the Booksellers

15

merger.

16

but I will tell you right now, if the combined market

17

shares of those two firms in the defined book market are

18

say in the 5, 10, or 15 percent range, I can guarantee

19

you I will find some problem with the econometric model

20

that was presented.

21

shares and you talk about entry and market conditions and

22

other sort of stylized facts about the way book

23

publishing works, I would be entirely prepared to accept

24

that those statistical results are valid, but the key

25

point is that the econometric exercise is useless if you

I happen not to know a thing about that merger,

If you get up to much higher market

For The Record, Inc.
Waldorf, Maryland
(301)870-8025

88
1

have a significant reason to doubt the validity of the

2

underlying model.

3

There is sort of the red letter error type

4

thing.

5

and you tell him what your model was and he comes back

6

half an hour later and says, oh, you screwed up, you

7

forgot to adjust for elasticity or you have too many

8

multiple linear variables or whatever it is.

9

That's where you give your data to Greg Werden

Even assuming that the econometrics is done

10

perfectly, it doesn't mean anything if you have a

11

significant doubt about the truth of the model.

12

I think this has had a fairly profound impact

13

in my professional lifetime in the area of merger

14

analysis, which is when I entered grad

[Text truncated at 120,000 characters. The full text is on the page linked above.]

---

Source: Frix Law Library, https://www.frixlaw.com/law-library/documents/agency%3Aftc%3Af55014301a8e43e9. Public record. Not legal advice.