DEPARTMENT OF JUSTICE

Agency decision

Ask Donna

What actually matters in this document.

Text

FEDERAL TRADE COMMISSION

MERGER WORKSHOP

WEDNESDAY, FEBRUARY 18, 2004

9:00 a.m.

FTC Conference Center

601 New Jersey Avenue, N.W.

Washington, D.C.

Reported by:

Rita M. Hemphill, C.V.R.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

C O N T E N T S

PAGE

WELCOME - J. Elizabeth Callison

NON-PRICE COMPETITION/INNOVATION

Moderator:

Panel:

Mark W. Frankena, FTC

Ann Malester

Richard T. Rapp

Steven C. Sunshine

UNILATERAL EFFECTS

Moderator:

Panel:

Luke M. Froeb, FTC

Joseph Kattan

Gregory K. Leonard

Abbott "Tad" Lipsky, Jr.

Gregory J. Werden

Valerie Rabassa

COORDINATED EFFECTS

Moderator:

Panel:

Michael H. Knight, FTC

Jonathan B. Baker

Andrew R. Dick

Paul Yde

Deborah P. Majoras

Steven C. Salop

David T. Scheffman

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

127

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

PAGE

UNCOMMITTED ENTRY

Moderator:

Panel:

197

Theodore A. Gebhard, FTC

Timothy Daniel

A. Douglas Melamed

Frederick "Rick" Warren-Boulton

Mark D. Whitener

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

P R O C E E D I N G S

WELCOME

MS. CALLISON:

This is the second day of the

Joint DOJ/FTC Workshop on Merger Enforcement.

contained a lot of interesting information and useful

panels.

well.

Yesterday

I'm sure today we will follow that up quite

I want to let you know that presentation

materials from our panelists' papers and slide shows that

they have sent in will be available on our respective web

sites.

be there shortly.

hearings of the workshop will be available on the web

sites at some future date.

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

Thank you.

Likewise, the transcripts of the

NON-PRICE COMPETITION/INNOVATION

MR. FRANKENA:

This morning we will be focusing

on the discussion of the effects of mergers on

innovation.

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

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

discussion between the panelists and the audience, with

audience participation and so forth.

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

my first thought was to seek input from the investor

community which didn't seem to be represented.

So I asked my friend Tom, who after all has

about 1,000 patents, whether he could address the

relationship between competition and innovation this

morning.

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

someone proposed that his firm, Edison General Electric

and its competitor, Westinghouse Electric call a truce in

their war to sell electric systems to cities and towns

around the country.

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

Tom explained that he refused to go along with

the proposal because, and I now quote from a contemporary

letter from Edison:

usefulness as an inventor is gone.

be worth a penny.

incentive.

"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,

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

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

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

important inventor.

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

this morning.

speaker.

Professor Davis.

therefore, could not be with us.

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

night consider that.

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

like everything else, the inspiration came to me because

of something my 12-year-old son did.

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

you all have heard.

it for you.

Don't worry.

I was listening to

I'm not going to sing

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

the sharpest tool in the shed."

between that and innovation analysis were just

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

here in the room.

To me, the parallels

The point that I will make generally today is

that innovation analysis has its uses.

I wouldn't call

it the sharpest tool in the shed.

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

out of the shed?

That leads to the

Just as an overview, I hope that these

propositions here are pretty non-controversial, at least

the first few ones are.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

Innovation is an important dimensional rivalry

and intervention of progress.

I don't think there's a

lot of dispute for this point.

I think the studies that

have been done over time have said that innovation has

been responsible for the vast majority of economic

progress that has been made over the last century.

Then when we try to actually apply it to merger

analysis, can we actually predict the effect of a merger

on innovation?

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

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

effects of market structure on goods and on the

relationship between market performance on goods is not

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

We will talk more about this later.

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

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

different views, the legal and economic issues.

are legal issues here, too.

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

that legal and economic issues require a showing -- I

called it "probable effect on output."

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

output in the goods market.

There

We have been focusing mainly

When I say

I mean

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

have our views on it, but perhaps Genzyme is an example

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

where the chairman and the rest of the Commission

basically followed the first three bullet points, thought

in theory there could be an effect on innovation, but

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

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

that's what they did, count me in.

we have to talk about.

Of course, the facts,

Lastly, just in terms of what does this mean

practically?

that the practical application of innovation markets is

infrequent.

market cases.

industries, pharmaceuticals being the obvious example,

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

will go into.

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

markets is that they are really hard to define.

you know an innovation market?

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

it when I see it.

How do

We'd like to see a

We have to account for trying to identify all

the conceivable sources of innovation.

account for how do we know what the strengths and

significance of the population of innovators are.

We have to

Also, the type of innovation that's going on in

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

the market may be relevant to our confidence in market

boundaries.

What do I mean?

Well, are there regulatory barriers that are

going to make innovation much more structured?

innovation tied to certain sets of production assets,

where we know that people without production assets

really are not effective innovators?

For that, I would refer to the Department of

Justice's complaint in the General Motors/ZF

Friederichshafen case.

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

innovate new truck and bus transmissions unless they had

those set of processes.

That case of innovation was tied

Then another way or type of innovation is

patents, but patents are really tricky.

patents, patents may mean there are a necessary set of

assets.

up so that the mergers are complimentary.

really careful.

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

Gilbert and I wrote, but we identify certain sets of

factors that you need to find in order to have some

confidence that you have an innovation market.

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

should go home at that point.

If you

If you can't define the

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

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

point in going on.

work to do.

If you can, of course, there's more

Defining the boundaries of the market is hard.

This part is even harder.

How can you actually predict

the competitive effects?

How can you assess the merger's

effect on innovation and then translate that into an

output effect?

We start off with, I think, a very obvious

question, structures as a means of predicting

performance.

Who is right, Schumpeter or Arrow?

Schumpeter basically says you need monopoly

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

monopoly rents to actually provide incentives for

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

cannibalization concerns, expected return, that perhaps

monopolists do not have the same incentive to innovate at

the same rate as others.

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

literature thinks this is completely an unaddressed

question.

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

inferences.

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

after time, monopolists really have a high incentive to

innovate?

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

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

that question is.

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

We might have an intuition.

We may

That leads us right into the second issue,

which is the problem of the necessity of case specific

evidence.

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

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

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

plaintiff.

innovation competition going on between the two, and that

each have high incentives to innovate.

I think, given the uncertainty in the

You have to believe there is important

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

familiarity with, was a case where the evidentiary record

showed years and years of the two companies actively

targeting each other's innovation activities, actively

pursuing incremental process improvements, and then with

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

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

efficiency.

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

is right sometimes.

we are now in the Arrow camp with these facts.

Maybe Arrow is right sometimes, but

I touched on incremental versus revolutionary.

This is an idea of how big the innovation is.

ways, incremental innovations may be the ones that are a

little bit easier to understand how to protect.

be easier to define where the sources of that innovation

are going to come from.

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

incentive to pursue them, and sometimes they are harder

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

source of that innovation is going to be.

In some

It may

These revolutionary innovations,

In the incentives, I don't think you could

understand the incentives to innovate until you

understand the nature of competition between the

innovations.

downstream level.

That's going to require looking back at the

Are two pharmaceutical companies going to be

manufacturing exactly the same chemical compound that,

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

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

drug status.

will be in competition with one another.

You know, once they get to the market, they

That may lead you to one set of conclusions.

If on the other hand these products are so sufficiently

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

different they are going to be attacking different

markets and there will be a high expected return, maybe

there will be an incentive for both products to be

developed.

going to open up so many new applications and increase

demand that there will still be a strong incentive to

invest.

I don't know.

Maybe the second invention is

Confidence that the reduction in innovation

will lead to an output effect.

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

take into account downstream competition.

are downstream products that are competing against your

upstream products, and those are going to continue to

spur innovation.

That goes to all the

What if there

I think that what all this is saying is that

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

this level of confidence.

the effect outweighed by innovation efficiencies?

Lastly, I put down here, is

Let me just throw in a point just as a matter

of personal opinion.

I know this is not necessarily

accepted by everybody.

whole idea of merger specific efficiencies.

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

these guys probably could have done a contract research

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

view.

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

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

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

engineer efficiencies seems to me to be a misguided

approach.

What I've done by putting efficiencies as part

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

of the efficiencies to the defendants.

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

Eventually, this really belongs in the camp of the

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

efficiencies don't outweigh the other benefits.

Enough about economics.

I think it is

What about the law?

I'm running behind.

There have been a lot of consent decrees.

There has been a complaint that the Department of Justice

filed that did not go to trial.

other cases where innovation is sort of talked about but

not a true innovation market.

There have been some

Does Section 7 really allow us to have an

innovation market?

The few cases that exist on this, I

think, say that without actual sales and a line of

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

that was true, that would doom Section 7 analysis.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

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

our articles is that it would be misplaced for Section 7

to not recognize an innovation market theory.

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

easy proposition that says that merger analysis should be

designed to identify those mergers that lead to reduction

in welfare, to lower output.

The reason

The analysis that I just went through, if you

can go through that analysis and come to the conclusion

that output is reduced, you have satisfied what is the

basic condition.

market, I think, is the innovation market essentially

identifies the place of competitive interaction where the

problem arises, and then the output market actually

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

of commerce and the section of the country that the law

is trying to protect.

Then the question on an innovation

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

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

actual output market, there's no direct effect.

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

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

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

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

wrote, there are some courts that have gone other ways,

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

In the

You

because they are thinking cookie cutter, define a market,

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

blah.

That's one way of doing it.

The second point, can plaintiff prove a

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

think that proposition stands for itself.

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

individual market case.

importance of being able to actually have the documents,

the market evidence, and everything else to show that

this is a real problem.

This would be

This just goes back to the

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

time for some conversation.

this.

I won't spend much time on

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

to say burdens, where's the beef?

why all the fuss?

plaintiff.

Hughes say that upon the establishment of a prima facie

case, that the burden of production shifts over to the

defendant, but the burden of persuasion always remains

with the plaintiff.

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

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

that the defendants satisfied a burden of coming forward

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

with evidence.

It seems to me that any proper application of

this kind of analysis would lead to the burden going back

to the plaintiff to persuade they were actually right.

If Commissioner Thompson's dissenting opinion

means something else, with a presumption, then he means

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

case law, perhaps even for output markets, but certainly

not for innovation markets for the reasons that we have

talked about.

Problems of the investigator.

The first point,

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

The investigator has a duty to evaluate the transaction

before it.

serious competition in innovation, what are you going to

do?

just let the transaction go if you think there is going

to be a problem.

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

is no easy path here, is there?

We have just been

through it.

could be some adjustments, depending on who is involved,

the burden of proof is clearly going to be on the

plaintiff, and the plaintiff and the investigator will

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

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

innovation, the manner in which output will be reduced,

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

the output effect will be felt.

Not easy stuff, right?

We are talking about doing this in at least two

different markets, two different spaces.

Again, as sort of the last resort of the

investigator, the importance of empirical evidence from

the merging parties in the markets cannot be overstated.

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

you all the way there.

evidence.

You are going to need the

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

I just thought through questions on how I might think

about doing an innovation analysis.

Is innovation an important dimension of

rivalry?

for price effects.

beer.

The answer is no, forget it.

Why bother?

Look

Go out with your colleagues for a

Second question, will innovation effect

existing product markets in a reasonable amount of time?

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

innovation markets.

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

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

We toyed with bringing a case in an innovation market.

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

Innovation is an important dimension of competition.

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

potential competition case.

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

a potential competition case is easier than an innovation

market case?

Well, there you have it.

Can the boundaries of the innovation market be

determined?

We went through that.

It's not child's

play.

Again, I put the efficiencies in with the obligations of

the plaintiff.

show there is an effect, then the plaintiff should also

be comfortable showing the efficiencies don't outweigh

the other effects.

Does the merger provide incentives us to innovate?

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

And then last, the harm to the output market,

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

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

With that, I know it's all painstakingly clear.

Thank you very much.

MR. FRANKENA:

Thank you.

Ann?

MS. MALESTER:

Well, I must say that I'm really

surprised that this many people showed up this early in

the morning to hear us talk about innovation markets, but

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

this must mean all of you are certainly aware that we

have been having a debate in the antitrust world for well

over a decade about what the antitrust agencies should do

about the concept of innovation markets, and essentially,

should there be any enforcement actions in this area?

As you know, there are those who believe that

because the empirical economic data has not been

conclusive in showing a correlation between increased

concentration and reduced innovation, that really the

antitrust agencies should just throw up their hands and

walk away and not think about bringing enforcement

actions in this area.

There are others, and I count myself in this

group, who believe that preserving competition at the

research and development stages is really important, but

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

real problems and a lot of difficulties in really

assessing how innovation markets should be defined and

what the antitrust agencies' role really should be there.

First, let me just give you two reasons why I

think that it's vital for the agencies to preserve

competition and be vigilant in protecting competition at

the R&D stages.

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

harmful to consumers than a merger that results in a

price increase of quality decrease.

Second, because in the real world, where

companies compete every day, it's very clear that

competition is the primary incentive that spurs increased

level of innovation.

In preparing for this workshop, I re-read the

testimony that the FTC received during the 1995 hearings

on global competition.

the economists could only agree that it was really

inconclusive and they weren't sure what the evidence

showed, the business officials who testified during those

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

report, "Unanimous and emphatic in their view that

competition is the primary incentive for innovation."

I spent well over a decade at the FTC as an

What struck me is that although

assistant director in the merger division.

hundreds of documents, and talked to dozens of business

officials, who told us that they supported that

conclusion.

I have seen

When you look at the reality of the way

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

typically businesses invest more resources, work harder

and work more quickly in their research and development

efforts when they are faced with the possibility that

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

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

working to develop a competing product.

We saw evidence in our investigations that

companies often spend a great deal of time tracking the

research and development projects of other companies that

are working in their general area.

information about the schedule and progress of competing

R&D programs, and they make investment and priority

decisions based on the level of competition they believe

they face.

Companies seek

They also confirmed that being the first to

introduce a new or improved product can be critically

important to many industries.

that gives you increased market acceptance, lock in of

customers, there are many reasons why in some industries

reaching the market first is very important for future

success.

A first mover advantage

Having said all that, I really do agree with

Steve when he explained that it is much more difficult

for the plaintiffs to predict under what circumstances a

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

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

merger will result in a price increase.

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

whether the merged firm will have both the ability and

incentive to reduce the pace of innovation after the

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

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

achieve in combining their R&D programs.

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

the past 10 years in enforcement at both the Justice

Department and the FTC, you will see first a relatively

small number of cases where the agencies alleged only a

reduction in innovation, not tied to an existing current

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

that there are generally in the pharmaceutical and

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

to identify generally the companies that are

participating in research and development, and where the

barriers to new companies beginning development are

extremely high.

Let me turn for a moment to the Genzyme

investigation, which is the most recent pronouncement by

the Commission, and which brought forth three separate

statements by the Chairman and Commissioners Thompson and

Harbour.

That decision could be the subject of an entire

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

First, really the question about whether in innovation

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

markets, there should ever be a presumption of

anti-competitive effects at any level of competition, and

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

future Commission enforcement decisions.

On the first point, I was frankly puzzled by

the seeming importance that all the statements placed on

whether or not there should be a presumption of

anti-competitive effects.

feeling that in existing goods markets, where

presumptions are raised at much lower concentration

levels than was the case in Genzyme, but the antitrust

agencies bring an enforcement action every time a

presumption of legality is raised.

nothing could be further from reality.

Somehow, that left me with the

We all know that

I think that the data that the Justice

Department and the FTC released two weeks ago on

horizontal mergers and the concentration levels and which

ones resulted in enforcement versus closing, make it very

clear that even in existing goods markets, there are

highly concentrated markets where merger may reduce the

number of competitors from three to two and even two to

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

action or the Justice Department did not bring an action.

What does the presumption mean?

that the merger deserves a really close scrutiny, and

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

It only means

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

has to develop an intensive fact based analysis.

I don't think there is anyone at the Commission

who believes that Genzyme did not deserve that kind of

close scrutiny.

Commissioners spend a great deal of time and effort in

analyzing the large investigatory record, and in

assessing what the facts were and what the analysis

should be, and that is really what I think the

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

kind of scrutiny.

Both the agency staff and all of the

Finally, let me just talk for a minute about

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

the Commission and its likely enforcement in the

innovation markets.

First, I think clearly the Commission's

decision indicates that the current Commission is likely

to approach any innovation market analysis with a lot of

caution, but there are four reasons why I think you

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

in other cases based on the Genzyme decision.

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

that not one of the three statements that were issued was

signed onto by anyone other than the author of that

statement.

Two of the Commissioners, Commissioners

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

Swindle and Leary, although they voted to close the

investigation, did not join in any of the statements that

were the public statements that were issued.

Second, in the Genzyme case, both companies,

Genzyme and Novazyme, were actually in the late clinical

stages of drug development, which is earlier than the

other innovation market cases where the Commission has

investigated and taken action before, and that means that

the likelihood of both products making it to market is

less than in the other cases the Commission has looked at

in the pharmaceutical area.

Third, the drugs in question were covered by

the Orphan Drug Act, which grants a seven year market

exclusivity to the first drug that makes it to market.

think that fact also makes it less likely than in a

traditional pharmaceutical market that both drugs would

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

the Genzyme acquisition was not reportable under the

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

closed well before the Commission was in a position to

assess its legality.

Trying to design a remedy in an innovation

market in a consummated merger where the core assets are

scientific personnel, know how, access to academic

researchers, and many other human factors, really

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

presents quite a daunting task, and the vast majority of

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

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

the same level of difficulty in designing and

implementing a remedy.

In sum, I think Genzyme was a very important

case because it gave the Commission an opportunity to

look at a really complete investigatory record, and to

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

Scott, to really debate the issues.

Obviously, different people at the agency came

to different conclusions on what the facts were and what

the conclusions should be.

exercise at the Commission.

I think it was an important

I also think the specifics of that case and of

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

majority opinion on what the appropriate analysis should

be, at least should leave everyone with some caution in

predicting what the Commission will do in future cases.

MR. FRANKENA:

Thank you, Ann.

Dick?

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

speakers are in.

I have to issue a disclaimer before I

begin, because there are many of my colleagues at NERA

that disagree with my extremist positions.

over there, the man with his fingers in his ears, has the

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

Tim Daniel

look of participatory mortification on his face, is one

of those.

not shared by everybody at my firm.

I want you to know that my point of view is

I do find myself once again as the extremist in

the group, wishing to repeal entirely the use of

innovation market analysis.

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

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

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

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

that is a poor characterization of my views about almost

everything else.

I am an opponent of the use of innovation

market analysis frankly for an arcane epistemological

reason, and that is there is an absence of underlying

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

works well when it is based upon economics.

a science.

explanation, a hierarchy of general laws and law like

statements.

Economics is

What that means is that it uses scientific

The horizontal merger guidelines are a

wonderful example of economic science for policy

purposes.

demand elasticity, which is right in the mainstream of

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

law of demand.

Cournot's welfare triangle, Stigler's theory of

oligopoly, all underpinnings of the merger guidelines are

inherent, and the guidelines have a strong theoretical

basis.

facts as well.

concerned.

market analysis is basically reasoning by analogy.

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

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

social gain from letting it go through is predictable,

that you can tell which way that inequality will run.

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

perceive is an analogy to the relationship between

competition, quantity and prices in goods markets, and

the analogy is false.

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

views about that subject.

I won't dwell on that.

I wrote an

Looking now in the five minutes I have, because

I will not impinge on the discussion period, where you

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

three quick points, two of them relating to Genzyme and

then an attempt to be a little constructive instead of

being a naysayer.

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

can't do something better than that.

Whose baby is that?

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

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

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

admirer of Chairman Muris' statement in Genzyme.

remarkable to me for the fact that it changes the basic

policy rule.

with peculiar circumstances of the case, but one should

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

What we are talking about now is a race to

It is

It has to do, as Ann described very well,

monopoly.

It has nothing to do with the kind of

competition that is output expanding, price reducing

competition, the likes of which we talk about when we

talk about horizontal mergers.

to do with this.

Market power has nothing

The underlying characteristic of -- let me say

the most basic finding in the majority statement is even

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

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

not who is going to come in first.

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

than later.

It's whether or not

The fact that there is no possibility for delay

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

the merger means the merger has no negative impact.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

That's my quick two second reading of the majority

viewpoint without all of the reasoning that follows.

All I wish to say is that as an analysis of

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

quite novel.

The lawyers in the group can consider how it would play

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

it's unusual.

It's

That doesn't bother me as an economist.

My real focus is on Commissioner Thompson's

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

extremists and those who favor the innovation market

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

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

concerned, my reading of Commission Thompson's dissent

led me to say game over, I win.

This is a litany of what can go wrong if merger

analysis is applied using its very premise, that is to

say reason by analogy to the goods market.

You talk about a merger, a monopoly, and speak

about market power, even though there are no goods around

and market power is not the issue.

I know about presumption of anti-competitive

effect without either theoretical or empirical basis.

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

things that I won't dwell on so we can get to the

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

conversation part -- this struck me as extraordinary.

There is a difference to be expected in the

resource allocation of Genzyme, arising from the

difference between trying to get the first approval with

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

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

approval with orphan drug exclusivity for a longer period

of time for the sake of first mover advantage.

I can go along with all of those who say

competition spurs innovation.

It makes a certain amount

of sense.

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

misused.

It represents to me a believer in competition,

certainly an oddly crabbed view of human nature.

There is a further litany of errors, errors in

economics and logic, but I believe here, in Commissioner

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

my worry is that some day President Kerry might appoint a

Commission with this frame of mind, and that is the

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

unobjectionable -- to me, understandable decisions both

by the Pitofsky Commission and the Muris Commission, are

not good evidence that we are okay with the innovation

market concept.

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

compromise in my extremist position.

At the very end of that panel, he put the

following question to me.

He said, look, just imagine

that Boeing is building the next airframe.

jet engine makers.

cam systems and start designing the new jet engine that

would go along with that.

There are two

They are about to sit down with their

Do you propose that the Federal Trade

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

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

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

My answer to him was no, sir.

What I wish I

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

No, sir, because

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

good, the design and character of either good has not

been invented.

research effort and whether or not they will pour it on

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

merge, because you can't predict that.

You have the means to analyze not

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

case where there are going to be two goods competing with

one another versus one, if the merger or the joint

venture takes place.

guidelines reasoning to deal with that, and that makes it

okay with me.

We can use horizontal merger

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

My attempt at a constructive end to my

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

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

wrote us a series of possible questions to discuss.

I think the distinction between innovation

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

know very little and predict very little, the distinction

between that and future goods markets, which to me means

taking only those cases where we can actually foresee

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

number of pricing voices will be, and make some attempt

at doing horizontal merger guidelines analysis.

That distinction between innovation markets on

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

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

well, and the other should quietly go away.

Thank you.

MR. FRANKENA:

Thank you.

Just following up on

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

in 1995, the antitrust agencies had completely listened

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

forward, we won't be defining innovation markets.

think innovation is important and we think competition

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

delineate innovation markets, so we don't want our staff

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

Let's assume back

to be doing that.

My question for Ann is what difference would

that have made in the types of investigations, things

like that?

MS. MALESTER:

I think there are maybe two

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

approach presents enormous dangers to competition, that

would allow mergers to proceed, that I think would have

really, really serious anti-competitive impact.

The first might be similar to the case that the

Justice Department brought, and I think it was during the

time Steve was there, Lockheed/Northrop Grumman.

It was really a step further back from the

Boeing example that Dick and Chairman Pitofsky talked

about, where the Justice Department raised the concern

that there were only three companies left in this country

that really had the capabilities of designing military

aircraft, and if Lockheed were permitted to buy Northrop

Grumman, that would be reduced to two.

There were no specific plans for a fighter

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

there are eventualities that arise that very quickly turn

the Defense Department into needing something new to meet

a new military challenge.

Even though there wasn't a future goods, a

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

specific program or product on the drawing board, both

the Justice Department and the Defense Department felt

very strongly that one of the reasons to block that

merger was to preserve the capabilities of designing and

producing a particular weapons platform.

I think that is a step sort of removed from

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

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

produced yet.

proceed would have very serious anti-competitive effects

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

security.

I think it's a case that if allowed to

The second point I want to make is moving even

further back away from the future goods markets, and this

may be something that is specific to the pharmaceutical

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

in, but may be applicable to others.

I'm not sure.

That is apart from looking at what the impact

of eliminating the merger is on the future goods market,

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

selling a product at some point.

There is actual competition going on during the

research and development phase when two companies know

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

to get approval first.

Those companies will, for

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

example, do clinical trials that involve many more

patients, to make sure they get a broader range of

patients with different kinds of seriousness of disease.

The kinds of clinical trials they do, the

amount of effort and outreach to patients they do, there

are benefits right now in the present of having two

companies competing to develop a drug, completely

separate from the actual benefits, potential benefits,

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

know yet if both companies will successfully develop the

drug.

I think that may not be completely responsive

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

the different kind of scenarios that gives me concern, if

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

good that companies are actually producing or are about

to produce, the agency should simply step away.

MR. SUNSHINE:

Let me just jump in here, too.

I was actually tremendously comforted by Dick's last

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

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

situation where Dick sees a potential enforcement action,

what we are talking about is the area of competitive

interaction between the two companies is in research and

development.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

Yes.

There will be a goods market, but that

goods market is a long way off.

we normally consider to be within the reach of Section 7

of the Clayton Act.

Usually outside of what

Innovation market is an immediate step to

identify where the activity of the two companies come

together and where there might be an actual reduction.

In the case that Dick put forward, there will

be less innovation competition in developing this new

engine some years off, which will translate itself into

output.

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

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

Now we are talking semantics.

MR. RAPP:

Are we talking semantics then if we

rule out the application of the innovation market concept

to the Genzyme case?

we could make about how many goods would otherwise go to

market, what the future market would look like.

are no such goods.

There, there is no prediction that

There

I see that as a sharp distinction.

In other

words, my primary criteria is that the analysis not deal

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

cut back or reduced.

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

That's the distinction I want to draw.

MR. SUNSHINE:

How do you get there?

How do

you get to the output effect on the goods market without

understanding what's going on at the research level?

MR. RAPP:

It maybe that what I'm talking about

comes very, very close to the concept of potential

competition, where we are already very close to home.

You can do that even if the design of the goods

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

a closed conversation.

MR. SUNSHINE:

This is

But just on the potential

competition, you would then relax the two year time

frame.

MR. RAPP:

MR. SUNSHINE:

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

example a lot.

Suppose one flows into that going into

the production process that Boeing and everyone else

knows that there is some probability that this engine

simply will not be a success, that the project they have

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

on the table is for an airplane that requires advances in

science and advances in material science or something

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

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

airplane in need of such an engine.

I would be curious if the panelists generally

could just speak to how that analysis might change if

that were thrown in.

MR. SUNSHINE:

I think it's an essential part

of the analysis.

on what exactly are the incentives of the parties, and

what effect is the combination going to have, factors in

terms of are there efficiencies that come from it, the

necessary investments, because of the size of the project

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

instead of two.

Again, the idea has got to be a focus

All those considerations go into the question

of whether this makes sense or not.

we have talked about before, applying straight horizontal

merger guideline standards to this innovation question is

misplaced.

For all the reasons

If this project is so risky that two firms need

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

answers the question.

Let me stop there.

I'm sure Dick doesn't feel

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

differently about it.

different view.

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

MS. MALESTER:

It seems to me it is not only

the question of is the standard that Dick puts forward

requiring that we know that a particular product be

successful, that they will in fact design an engine,

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

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

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

the technology and know how and expertise to design

aircraft engines, then we should be concerned about

allowing that merger for future aircraft engine

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

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

Stealth technology.

That was another reason

why the Defense Department and Justice Department felt it

was critical to keep a number of companies that knew how

to --

MR. RAPP:

The quick mental analysis that you

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

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

efforts.

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

to be available at the end.

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

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

whether that's a better or a weaker thing.

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

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

This slides

big drug companies.

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

effort have arisen because of the combination of those?

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

connection between concentration and innovation --

These days, we have seven of them.

MS. MALESTER:

I think that's a very, very

different proposition than what we have really been

talking about, which is competition or the lack of

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

don't think in any current goods market that would

concern us.

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

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

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

going to get a better product, we are not.

Those are really not the issues on the table.

The question is whether having competition in designing a

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

purchaser with a better product in the end, a more

innovative, a more radical breakthrough.

I think there are quite a number of historical

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

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

MR. HOVEN:

I'm John Hoven, Justice Department.

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

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

many others, and yours are relevant for some purposes and

not for others.

a good illustration of that.

I think the Lockheed/Northrop merger is

The presumption of the innovation markets

analysis is that innovation is a very predictable

process, just like building bricks.

and you get a product.

You spend the R&D

A good example is the expectation that the kind

of innovation we are looking at is innovation that has a

known identifiable product at the end.

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

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

industries is a process of ideas coming together in

unexpected ways and you want to preserve an industry

structure that allows that to continue to happen, and in

particular, in some cases, an innovation structure that

generates ideas, products that nobody can think of yet.

I think to some

I think the general process of innovation

analysis ought to be one in which one inquires how does

innovation take place in this particular industry, and

maybe it's the kind of approaches you are discussing, but

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

there are a lot of others that should be examined as

well.

MR. SUNSHINE:

I think certainly in the

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

that actually Muris puts forward in the statements, is

the recognition of what you said is exactly right.

The question is should we examine the

innovation market and under what circumstances can we

actually accomplish innovation effects.

minority of cases.

predict it, then you should go home.

It's a small

If you can't have the confidence to

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

innovative activity, there are three categories of what

can happen.

the future product.

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

be used at all.

It can be innovations that will be used in

It can be innovations that are used

That last category I don't think antitrust

enforcement should care about.

For the first two

categories, now we are talking about legal theories, is

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

potential competition case.

to that is.

have to have developed your legal theory and put it

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

To me, that's the connection.

Find the

innovative effect, put it with confidence under the given

set of facts, if you can.

in the business.

there's a case.

If you can't, you shouldn't be

Identify an output effect, and then

MR. RAPP:

May I just add a sentence with that

in mind?

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

troubling part from my standpoint, that it is also

consistent with the dissent, in that Commission Thompson

may have used the same reasoning and had sufficient

confidence in his ability and that of his advisors and

staff to see through these issues.

example of what can happen when innovation market

analysis goes wrong.

It is perfectly consistent with the majority

MR. SUNSHINE:

I think it's an

I agree with your criticism of

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

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

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

that is improper.

Commission, I think it would be misguided.

what Chairman Muris says.

If that were the policy of the

MR. FRANKENA:

That's not

I have a question for Dick.

Based on Chairman Muris' statement, and if you just

accept the facts, are you comfortable with that word of

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

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

incentives and so forth?

MR. RAPP:

Yes.

I'm comfortable with

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

functionally the equivalent of not making innovation

market policy.

One of the reasons that I admire what the

Chairman's statement does is that it reasons sensibly by

analyzing the incentives to an outcome that says let's

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

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

a great deal of respect for our former chairman who would

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

the parents, but I cannot imagine a well reasoned

economic statement -- a statement that is well reasoned

in economics that says and now let's intervene.

MR. FRANKENA:

Taking a broader view here, one

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

and merger enforcement in particular have not focused

sufficiently on the consequences of market power for the

pace of industrial innovation.

I'm just wondering broadly whether you think

the agencies are doing sort of the right amount, too

much, and so forth.

MR. SUNSHINE:

Those words sound strikingly

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

familiar.

I don't know where they came from.

Again, those words were written in 1994 and

1995.

there has been a whole set of investigations by both

agencies where there has been real focus on innovation.

There are cases where I would certainly agree

with Dick that innovation analysis may have been carried

to the extreme or inappropriately applied or applied

without basis.

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

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

is a tool in the shed.

the shed.

circumstances.

sure.

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

directly but raise an issue that arose out of the global

marketplace report, the 1996 report, and it was

specifically quoted in Chairman Muris' statement, which

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

one.

That is this business of saying we recognize

the potential infirmities with this approach, so we are

going to be very cautious and conservative and apply it

in circumstances where only small number mergers are

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

involved.

Is that a fair statement, Ann?

MS. MALESTER:

Do you mean small number of

competitors?

MR. RAPP:

Small number of competitors.

that is not conservative.

one that thinks that?

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

market analysis into every seven into six merger, you are

receiving resources in some sense, but in those settings,

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

the likelihood is that the merger is going to be allowed

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

false positive that arises in your innovation market

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

to be that consequential because somebody else is going

to get to the finish line.

That's radical.

To me,

Am I the only

It's conservative in the sense of

My definition of non-conservative radical

enforcement is to take this controversial concept to look

only to those mergers where it will be consequential

because you are intervening in a merger that is going to

determine the output of innovation, if you will, and

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

result.

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

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

mergers, so it must be benign.

opposite is true.

MS. MALESTER:

At least in my view, the

I think really the reading of

what that global report meant and what the Chairman, I

think, alluded to, is simply the consensus that an

unilateral theory is by far the most likely if we are

going to apply an innovation market at all, and generally

speaking, you are looking at a very small number of

companies that have the specialized assets, so to speak,

to be innovating in the market before you are even

starting to think about unilateral theories.

was the impact of that.

MR. RAPP:

MS. MALESTER:

That really

I don't disagree.

In terms of your point of yes,

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

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

from my point of view, because I think protecting

competition where it makes a difference is important, I

think it is the right place to put our efforts.

MR. RAPP:

If that statement were always

prefaced by the prior statement, that the reason to be

conservative is that we are dealing in a realm of

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

decisions that we make will be the correct ones, if it

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

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

combination of those two things together strikes me as

problematic, my only point.

MR. SUNSHINE:

The

Let me ask you one question tied

to Genzyme.

Suppose that the drug in question did not

qualify for orphan drug status.

outcome of the case?

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

all situation.

Does that change the

Orphan drug status, for those of

MS. MALESTER:

that go into coming to a decision.

orphan drug situation adds a complication, but in and of

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

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

factors in assessing whether or not the Commission should

take action.

MR. FRANKENA:

I think there are a lot of facts

In my own view, the

Just one final question.

there anything that you think the antitrust agencies

could or should try to clarify about their approach to

analyzing the effect of mergers on innovation?

anything left they could actually do?

MR. RAPP:

Is there

I think conversations like this are

illuminating.

I can't think of anything other than that.

You know, seeing Greg Werden and Luke over there sitting

side by side leads me to ask a question.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

If it's not

fair, forgive me.

Are there or have there been interactions

between the agencies about the differences in the

application of it?

the comparison.

drugs gets a lot more cases where entry barriers are high

and things like that.

or not there is interaction between the agencies on

enforcement of innovation markets.

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.

MR. WERDEN:

MR. SUNSHINE:

None that I participated in.

I thought Greg and Luke were

working on the 2004 innovation market merger guidelines.

That's

MR. FRANKENA:

Thank you very much.

I'd like

to thank both our panelists and the audience.

(Applause.)

UNILATERAL EFFECTS

MR. FROEB:

unilateral effects analysis.

today.

Welcome to the session on

We have five speakers

We have Greg Leonard.

He's here.

I know he's

here.

We will start with Valerie Rabassa from the

European Commission, followed by Joe Kattan, Greg Werden,

and Tad Lipsky, and then Greg Leonard, if he gets here in

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

time.

He will bring up the rear.

There he is.

I want to thank the panelists for agreeing to

come on relatively short notice.

I also want to thank

the staff who has put together this conference, and also

worked on the enforcement data, getting the enforcement

data together, and releasing that as part of our desire

to be more transparent to the external bar and the

external community.

I also want to acknowledge the efforts of my

predecessor, Dave Scheffman, who began much of the work

that is now just coming to fruition.

By 1999, they had displaced the structure

conduct/performance paradigm in industrial organization

economics, thinking on pricing and output coordination in

oligopolies had evolved considerably from the view that

it made coordination almost inevitable.

While economists never entirely rejected

coordinated effects theories for mergers, they did reject

exclusive reliance on them, and they had more plausible

theories for many cases.

that unilateral effects analysis appeared prominently in

the 1992 horizontal merger guidelines, which were jointly

promulgated by the FTC and the Department of Justice.

Thus, it was not surprising

The unilateral effects analysis satisfied the

attorneys' demand for simple intuition that they could

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

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

it satisfied the economists' demand for more rigorous

analysis.

structural theoretic models to make quantitative

predictions of unilateral competitive effects.

It wasn't long before economists began using

In a price setting model, price goes up.

In an

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

win.

bargaining power.

we learned that mergers aren't usually profitable with

equality setting models.

In a bargaining model, the merged firm gains

In the equality setting model, well,

The controversy surrounding unilateral effects

analysis has focused on the application of structural

models to individuals cases, their main virtue is they

force assumptions to be made explicit and they provide a

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

the merger.

The shortcoming is that the models are

necessarily unrealistic and abstract away from the

important features of the industry. As such, the results

may be quite misleading.

Unfortunately, we have little evidence on

whether these models can accurately predict the effects

of real mergers.

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

theories.

What follows is we will hear from Valerie

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

Leonard on this and more general topics.

with Valerie.

Let's start off

MS. RABASSA:

Thank you, Mr. Chairman.

Let me talk about the Lagardere case.

It's a

European case which took place last year in France.

was analyzed by the European Commission.

the leading 2003 cases of the media industry.

Let me talk about the transaction.

It is one of

It was very

simple.

the first and second player of the book industry.

book industry in Europe is characterized by a very high

level of vertical integration.

industry.

publishing right, and who sell the rights to the

publisher, and distribute the books to the retailers who

distribute or sell the books to the final consumer.

given the particularity of these cases was an econometric

study carried out for the Commission by Professor Marc

Ivaldi, and it studied basically the effects of trash and

trees in the downstream segment.

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

a final decision.

In this study, they will in effect

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

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

consumers who produce and choose a profit of the firm.

So this was the launch with VUP, if Lagardere

had decided to increase price obviously unilaterally,

some of the final consumers would turn to the other

competing publisher, who may reach VUP.

thought of the concentration with VUP has shared a

subsidiary of Lagardere to accept part of this

competitive pressure, and so can recover part of these

customer.

So I've always

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

nested logic model, which is quite adapted to the book

industry, which is characterized by differentiated

products.

family of the discrete choice model, which was very

interesting in this case because obviously you cannot buy

too seminal a book.

The logic model, as you know, is from the

The consumer in this case made a discrete

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

choose a book on the concerned list.

consistent with the Bertrand Model of competition and the

estimation was three stage least-squared.

Simulation is

So the reasons why it's quite interesting is

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

because we find that the end price of the books were

significantly increased in the downstream segment.

Obviously, we were looking at the consumer surplus and we

find that consumer surplus will also fall significantly,

which was equivalent to a very negligible part of the

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

What was wrong obviously there is that the

price increase was linked with the market size, and we

were able to reconstitute the market size in this case.

Again, we decided to receive -- the results

were quite robust to incorporate it into the final

decision, and we have used a boot strap method to

construct confidence intervals that are quite often used

in the econometrics field and we find that there is only

five percent variability that the price rise to the

concentration that could be including significant a

significant interval of plus or minus one percent, with

the mean value of the price change.

Results were quite robust because of the very

high number of observation to at least 10,000.

different statistical traits were quite significant, and

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

book and the intra-brand correlation were quite stable.

Altogether I feel that's reasonable to quite robust as

for when we decide to incorporate this econometric study

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

The

into the final decision.

Thank you very much.

MR. FROEB:

Joe, you're next.

MR. KATTAN:

I want to talk about, I think,

differences between the ways that lawyers tend to look at

unilateral effects and the way economists do, and

particularly with reference to the new models, the merger

simulations that are being done increasingly in

unilateral effects cases.

I want to begin by just talking a little bit

about the 1992 guidelines and what they say, and to

contrast that a little bit with the new simulation models

and pose the question whether we are asking the same

questions, that is do the guidelines give guidance or do

these simulation models answer the same questions that

the guidelines are asking.

I think it's pretty clear that the 1992

guidelines really changed the way that people look at

unilateral effects before 1992.

basically thought of in terms of monopoly or dominance,

and the insight of the guidelines was that unilateral

effects can arise outside the monopoly context.

Unilateral effects were

The merger of Daimler and Chrysler does not

exactly have the same effect as the merger of Daimler and

BMW, and if you were just looking at HHIs and market

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

shares, you may get a misleading answer.

Some people thought we were returning to

submarkets, which was a term that had been discredited at

least within the Beltway.

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

on the submarket nomenclature simply because it is

something that the courts are familiar with and are

comfortable with, regardless of whether that nomenclature

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

unilateral effects analysis.

Certainly, if I were

I think it is certainly the case that inclusion

of unilateral effects was the most important change in

merger law since the 1982 guidelines, and had a very

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

very long ago that one could hear people talking about

whether unilateral effects was in fact the only valid

theory for looking at mergers.

swung back a little bit.

I think the pendulum has

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

the guidelines is a synthesis of good economic theory,

but also practical judgment about the qualities of the

tools we have available to us in merger investigations,

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

To repeat very briefly what the guidelines say,

not to state the obvious, but to give a benchmark for

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

what we are looking at, the guidelines focus on spacial

separation between products and product space.

is on localized competition where certain sellers compete

more closely with one another than with other sellers who

are in the same product market.

The focus

According to the guidelines, that requires that

there be a significant set of shares of consumers who

regard the product of the merging parties as their first

and second choices, and certainly the way that lawyers go

about applying the guidelines is by asking very often

about whether we are dealing with first and second

choices.

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

That's one of the first screens one applies

We have a 35 percent screen for the merged

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

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

where the market shares were lower than that that have

been brought.

I think these reflect a pragmatic tradeoff

between quality of the analytical tools that are

available in merger investigations and the theory.

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

by Greg Werden and others, that you can have an

anti-competitive effect in a merger where the merging

parties are not the first and second choices of

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

It is

customers, and you can certainly draw the theoretical

picture that would be supported and have numbers to

support it.

The question certainly in 1992, and not

accounting for all developments and the analytical tools

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

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

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

look at next best substitutes.

Now, the way that the lawyers tend to approach

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

most people here.

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

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

More broadly, I think doing an antitrust case

You look at markets, first of all.

without defining what the market is is a recipe for

disaster.

There was a recent case involving Rambus

Company that is in the news, where Infinion brought both

antitrust and fraud claims against Rambus.

claims were thrown out because Infinion forgot to define

a geographic market for memory chip technology, something

that probably would have taken two paragraphs in the

expert reports.

The antitrust

I think economists often take the position that

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

defining markets is superfluous if they can get to the

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

but certainly from the viewpoint of the courts, if you

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

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

definition of the geographic market was clearly a red

herring.

Once you define the market, I think the lawyers

focus on the issue of next best substitutes.

Why?

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

is driven by interviews and documents.

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

boxes and boxes of documents stacked up in the hallways.

I think there is a critique that one hears from

A lawyer's fetish

the economists that lawyers use models but don't

articulate them with sufficient particularity.

Certainly, the legal method does not require articulation

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

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

assumptions are and if

the limitations are.

the models are.

assumptions are not made, what

Lawyers don't tell you exactly what

The model effectively if we look for

significant market share.

We look for next best

substitutes, and that embodies certain assumptions on the

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

effects of the internalization of loss sales.

That is

sales that would have been lost to the merger partner

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

effects of that on the merged firm's incentives.

It's certainly true that the economic models

spell out their assumptions with greater particularity.

Lawyers have an advantage.

courts are lawyers.

thrown out on Daubert grounds, you know, the economic

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

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

the court.

The decision makers and the

You cannot have a legal argument

It's a different issue at the agency.

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

less bound by the guidelines because the methodology of

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

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

times from various economists in various contexts,

mergers and otherwise, that the whole process of defining

markets and driving the analysis that way is an

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

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

There is certainly an institutional bias

against the approach that the lawyers use.

I think the

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

real answer."

The tools that the lawyers use are fairly

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

crude in many cases, making predictions about the future,

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

which are unstated, and have various degrees of support,

and obviously, if through the application of the data

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

answer, why not try to do that.

It is certainly the case that the economic

approach has the advantage that the models have well

articulated specifications, which the legal approach does

not.

One of the things that we see now is that

economists have an insatiable appetite for data.

lawyers want documents.

a belief that virtually everything can be solved through

data exercises.

Economists want data.

The

There is

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

Tim made the comment to the effect that the analysis not

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

questions about whether the constraints of Bertrand

models are realistic in the context of differentiated

markets because they allow one point for differentiation,

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

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

variables that affect the positioning of different

competitors in product space.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

The insinuations are with us, and I think they

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

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

it.

Modeling simulations can be mutual if we are

all comfortable that they are attuned to the market

realities, that they are based on defensible assumptions,

and that's obviously where the debate has been.

One criticism has been that a small difference

in the assumptions can make very big differences in the

results.

economists is well, we have different ways of doing

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

impose some kind of reality check on your analysis and

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

from the standpoint of the party that you are.

The answer that I think you will hear from the

I'll pose the question whether there is a

conservative assumption from the standpoint of the

plaintiff, as all these models seem to be designed to

show a price increase.

Part of the issue that I have is price

increases are assumed in these models based on structural

assumptions, even though we are trying to run away from

structural analysis.

The question is what are we trying to show

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

here?

Are we trying to show what the magnitude of the

price increase is.

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

where it comes out.

I think everybody will tell you no,

When you get somebody telling you that I can

predict with 95 percent degree of confidence that you

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

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

you can debate that.

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

also before I take into account issues like repositioning

of product entry and efficiencies.

Of course, there's a footnote that

There is a risk, I think, of false empiricism

where you have these results which are so precisely given

with high degrees of confidence and decimal points, you

know, somebody is telling you this is the price increase

they are going to get.

I think this is fairly paradoxical, but the

economists' approach is more likely to identify problems

with a merger than the more traditional way the lawyers

do, simply because the models are designed to predict a

price increase, and because these models will predict a

price increase, even in relatively unconcentrated

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

the models have a problem or the guidelines are wrong

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

because the models will predict a price increase in

situations where the merger would not even get a second

look under the guidelines, things that are below 1500.

The economists' answer is well, we are not

really trying to predict the price increase, we have to

take into effect entry repositioning and efficiencies,

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

you have done is really shifted the burden of proof to

the merging parties, because you have said on the basis

of some data analysis that you have a price increase

unless these other factors come into play, so now please

prove to me that this price increase will not happen.

In fact, to really say to the parties in most

cases prove repositioning, because anybody that has done

a lot of mergers will know that most mergers, the

efficiencies that we see are not efficiencies that affect

the incremental costs in a meaningful way.

efficiencies that I've seen tend to be overhead

synergies.

real.

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

you can expect to translate into a beneficial price

effect.

very hard to show entry, particularly when you are

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

have large investments integrated with differentiation,

that is promotion, advertising, and also distribution

arrangement.

The parties are now in a position of being told

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

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

is repositioning.

common.

implicitly the burden of proof to the merging parties by

effectively assuming a price increase and really saying

the only issue is how big a price increase.

Repositioning is pervasive.

It's

At the same time, we are in effect shifting

Another question to ask is what happens to the

next best substitutes analysis.

We talked at the

beginning about what the guidelines say.

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

parties are first or second choice of the large share of

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

and really instead of being differentiated by product

attributes or by promotional issues, products are

differentiated by their sales level.

differentiation that the analysis assumes because that is

really what we are looking at.

The guidelines

That is the form of

Of course, we are assuming identical cross

elasticities of all products with respect to a given

product, which given the assumption that some products

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

are closer substitutes to each other than others, it

seems to at least raise some issues.

The more complex models, they impose great

cost.

They have an insatiable appetite for data.

There

are issues with using retail level data as a proxy for

wholesale competition.

Dave Scheffman and others have put together a

very detailed analysis of that, and it deals both with

implementation issues, what do you do when a price

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

collects prices on a weekly basis, as well as conceptual

issues, that most wholesale prices now really are two

part tiered and relatively fixed and a variable element,

and obviously consumer prices are an one part tier.

The issue that I guess I will end with is

whether these models follow the guidelines.

source of anti-competitive unilateral effects that the

guidelines are talking about has to do with the spacial

separation of product, are we really obscuring that issue

by looking at models that are defining differentiation

effectively based on the levels of sales, are we also

shifting the burden of proof to the merging parties

implicitly by doing that.

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

cost of doing these types of analyses.

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

that as a defensive measure because the economists are

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

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

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

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

endless boxes of documents.

The fact is

My feeling is that the current models are going

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

cases are decided in front of the agencies, and parties

will be foolhardy to ignore these models because they

have become part of that decision making process, and

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

don't have the investigative tools to look at consummated

mergers and see what happens and calibrate the

assumptions in these models to actual market results, is

to try to refine them and make them better.

MR. FROEB:

Thank you, Joe.

Valerie, just

because Joe raised this question, did you look at the

retail sector to see what kind of contracts they had with

publishers?

tiers?

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

books to retailers and wholesale was the sale of books to

the final consumer.

right data.

problem of getting data.

We had many problems getting the

We can talk about this problem later, the

MR. FROEB:

Thank you.

Greg?

MR. WERDEN:

Before I get to my prepared

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

said.

The first one was a serious misquotation of my

views.

He paraphrased the guidelines, which I will

actually quote to you, and then said I disagreed.

don't.

The sentence I will quote says "Substantial

unilateral price elevation in a market for differentiated

products requires it to be a significant share of sales

in the market accounted for by consumers who regard the

products of the merging firms as their first and second

choices."

I totally agree that is a very nice way of

putting the point, notice it said significant share, not

a large share,

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

wants to then translate this into saying there is some

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

and the merging products have to be closest in this

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

The crucial thing to know is that typically in

differentiated products, different consumers rank the

product differently.

product and some people will buy another one.

That's why some people buy one

What the guidelines say, which I totally agree

with, is are you using a significant number that is

viewed as the first and second choices.

Secondly, he suggested that modeling, the kind

that economists do, merger simulation and other things,

is more likely to find mergers problematic than whatever

it is that lawyers do instead.

statement because that's just totally contrary to my

experience.

I find this a remarkable

One of the reasons Luke and I got into this 12

years ago was the lawyers had tremendously exaggerated

notions of likely anti-competitive effects of mergers,

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

earth is with economic models, and it has worked.

In a lot of cases in which the models have

shown that it was implausible there would be significant

price effect.

enough, even with very small efficiencies, that the net

price effect would be positive.

didn't bring because of the modeling.

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

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

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

single one of those cases that I know of.

I think modeling has probably kept the agencies

from bringing some cases, although it might have caused

them to bring some other ones they should have brought,

and in both cases, that was for the better.

Starting with my prepared remarks, the concept

of merger simulation.

If the same well specified

oligopoly model reasonably describes the outcome of the

competitive process, both before and after the merger, a

lot of conditions, then that model can be used to

generate a quantitative prediction of the merger's

unilateral competitive effects after it is first

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

would prevail but for the merger.

All it says is if we run a model in our

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

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

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

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

model.

and shares and elasticities or whatever we observe for

the industry, and to generate what the post-merger

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

internalization of the competition between the merging

firms.

We can put numbers on what the guidelines talk

about in some cases.

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

worked for the Department of Justice, that's the

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

in several respects.

to tell the judge that there is a significant number of

customers who view the merging products as first and

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

say the likely implication is they will raise price 10

percent.

In the first place, it's one thing

I think you are much more likely to win the

case if you can make the latter statement than if you

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

course.

Plus, there is a way to put things in

perspective, like cost reductions.

off if you do a quantitative analysis.

otherwise.

You can trade them

You can't

That's really my second point here in the

slide.

There are certainly cases in which mergers may

have significant efficiencies or in which we think even

if we don't have specific evidence of efficiencies, it's

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

very plausible the merger will have some efficiencies.

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

simulation and very easy in some cases to show that

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

increase.

We don't bring those cases.

Merger simulation can be an useful way to focus

an investigation or try by identifying which factors or

assumptions really matter.

modeling exercise that one assumption gives you one

result, another assumption gives you a very different

result, then it becomes a very crucial fact which is

which of those assumptions is the right one or the better

one.

If you can show in the

That can be a very useful process.

You can also find some things really don't

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

figuring those things out.

Limitations.

The fundamental limitation is no

economic model is going to capture every nuance of

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

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

usefully predict.

Second.

Price increase predictions can't be

any better than rough estimates.

There is a whole lot

going on that isn't in the model.

All kinds of things

are going to happen in the world that you can't possibly

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

anticipate.

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

no idea of the likely effect of the merger.

You are only given some idea of the likely

Third.

Merger simulation is basically designed

to predict price effects, possibly other short term

effects of the merger.

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

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

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

tools in law or economics to really predict very well in

most cases, so what merger simulation gives us is

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

give us.

It's never going to predict the

The basic theme of my talk is that before you

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

before you use it for any other purpose for which you

give significant weight to the actual predictions, it

should have to pass an admissibility screen from the

rules of evidence.

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

less prominence.

Rule 702 interpreted by cases like

What I take away from Rule 702 and the case law

applying it as it applies to economic testimony in

antitrust cases is that testimony is admissible if the

witness is an expert in the relevant field of economics,

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

if the testimony employs said methods from the relevant

field of economics, and if the testimony reliably applies

those methods to the facts of the case.

I'll elaborate these a little, mostly the

third.

I think the required expert knowledge is a fairly

low hurdle.

Heisenberg, a famous physicist, who defines an expert who

is someone who knows the worse mistakes that can be made

in his subject and who manages to avoid them.

kind of expertise we need in any area, including merger

simulation.

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

What merger simulation does is use absolutely

standard, well accepted tools from economics, so in that

sense, theoretical sense, merger simulation is the

application of sound methods necessarily.

There is also the empirical sense of soundness,

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

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

accuracy of merger simulation.

very much about the predicted accuracy of anything else,

so in a relative evaluation, merger simulation certainly

doesn't do worse in predicting than anything else anybody

has ever come up with.

However, we don't know

I won't offer some of my favorite pieces of

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

dicta from Supreme Court decisions here.

Daubert itself, which held that expert testimony is

admissible only if it is sufficiently tied to the facts

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

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

testimony and the inquiry.

First, from

Daubert didn't come up with this fit concept.

It was borrowed from a Third Circuit decision.

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

are looking for, and it's something that certainly

antitrust cases are looking for.

I think

The same theme expressed in a different and

also useful way in the Joiner case.

admit opinion evidence that is connected to the existing

data only by the ipse dixit of the expert.

conclude there is simply too great an analytic gap

between the data and the opinion proffered.

The court should not

A court may

In my view, it is not uncommon to have expert

economic testimony that is merely ipse dixit.

The

economist doesn't closely tie his theories to the facts

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

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

As an example, a very well known one from an

antitrust case, the testimony of Robert Hall was excluded

in Concord Boat because his oligopoly model, the kind

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

that you might use for simulation, although hardly

anybody does, was not grounded in the economic reality of

the industry.

perhaps, although I happen to read the case as saying

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

grounded in the economic reality of the industry, so I

think all of the critiques were fit critiques.

This wasn't the only knock on the model

The court said there was nothing wrong with the

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

it wasn't the right method for that case.

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

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

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

particular, at least the way the model was calibrated by

Hall, it predicted that the defendant would not have more

than a 50 percent share without engaging in the

challenged practices.

I think that's

The fact was the defendant had a 75 percent

share before engaging in the challenged practices, and I

would say that's got to be improper calibration.

only problem with the model the court pointed out, and

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

Not the

A less well known case but perhaps a more well

known economist, Frank Fisher's testimony was excluded in

the Booksellers case for purposes of at least determining

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

damages, which was pretty important in that case, because

it contained too many assumptions and simplifications

that were not supported by real world evidence.

I will not comment on the merits of Fisher's

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

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

language used by the court here in excluding the

testimony.

think about some economic evidence and if it is the right

way to think about that evidence, and the right

conclusion is that it should be excluded.

on too many assumptions and simplifications that are not

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

facts and it isn't admissible evidence.

I think this is exactly the right way to

If it depends

What are the key elements of fitting the facts,

and I will emphasize these are key elements because this

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

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

to do with the model.

An oligopoly model used in simulation has to

reflect critical aspects of competition in the short term

at least.

aspects of competition.

have a basis for saying it reflects the critical ones.

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

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

convince the trier of fact that he is right in thinking

about the industry and the model in this way.

The model also has to explain the recent past

at a fairly high level of generality, especially the

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

reflected in price cost margins.

be explained by models, how the industry has responded to

shocks, the level of prices as compared to costs.

model should be able to explain these things, not

necessarily with exquisite precision.

why a model should explain day to day price movements.

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

There is a lot that can

There is no reason

The ultimate test, another way of restating the

fifth requirement, is that every modeling choice should

be justified on some basis.

economic theory, for example, the assumption of profit

maximization, and needs no further justification.

is what economists do.

assume that, don't call an economist.

It could be dictated by

That

If you don't want somebody to

It can be supported by industry data, for

example, if the model is calibrated properly, that fits

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

have in mind.

ratios or demand elasticities.

Also, the data may speak to diversion

There are lots of

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

different ways in which models could be supported or

refuted by data.

It should be consistent with stylized facts of

the industry.

important elements of competition?

How does competition work?

What are the

If you have a typical differentiated products

case and it involves a retail sector, how does the

manufacturing sector interact with the retail sector and

how does the retail sector behave?

These are the kinds of stylized facts that have

to be studied before you can properly model an industry,

and you should be prepared to explain how the facts

support the particular modeling choices.

Fourth, some modeling choices may turn out to

be unimportant.

different choice.

choices.

model doesn't really matter.

justification.

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

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

above, you can justify a choice by doing a sensitivity

analysis and showing that over the range of plausible

assumptions, you have picked one that if your plaintiff,

for example, leads to relatively small price increase

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

projections.

There are some modeling choices in merger

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

and that's what you have to do.

A couple of illustrations.

This is, I think,

the easiest and best one of some kind of modeling choice

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

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

functional form for the demand curve?

Here I have plotted between the competitive

prices over in the lower right and the monopoly prices in

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

merger simulation.

competitive price to the monopoly price, it involves a

vastly larger price increase for some demand curves than

others.

As you can easily see, going from the

Where you are going to see one of these or

perhaps some other functional form is when you do a

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

substantial way the price increase predictions.

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

produces small price increases, absent some strong

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

other functional forms fits better.

If I'm

A final illustration has to do with demand

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

elasticities.

In differentiated products, merger

simulation is a way of translating the demand

elasticities into price increase predictions.

those predictions are sensitive to the demand

elasticities, and it should be obvious that we never know

exactly what those elasticities are.

Obviously,

Well, so you should consider what the price

increase predictions are for a range of elasticities.

Here is a very simple illustration of how that might work

for the WorldCom/Sprint merger, where a very strong

assumption about demand is made so that we can place

parameters on this model just based on two elasticities.

One is the aggregate elasticity demand for

residential long distance service, and the other is

WorldCom's firm level elasticity for its long distance

service.

considers a very wide range of plausible values.

The academic literature is estimated in

For each of these elasticities, this plot

elasticity demand for residential long distance of about

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

half to one and a half.

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

huge range of elasticities.

truth lies somewhere in that interval.

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

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

on up to 1.2 percent price increase for the average price

over the whole residential long distance, which of course

includes AT&T and all these other little guys which

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

ago, so WorldCom and Sprint were at least a relatively

small portion of the industry, significant, but

relatively small.

other residential long distance service companies that

are going to have vastly smaller price increases than the

merging firms.

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

easy for merger specific efficiencies to swamp that, and

for the net price increase to be negative.

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

a price increase big enough that the agencies would worry

about it, and you might say very plausibly that WorldCom

and Sprint were closer substitutes than this model

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

and the contrary is also correct.

MR. FROEB:

Thank you, Greg.

Up above one

Joe, do you want

equal time now or later?

MR. KATTAN:

I'll wait.

MR. FROEB:

Tad?

MR. LIPSKY:

Thanks, Luke.

It's a great honor

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

to be on the same stage with such distinguished

practitioners of these various economic approaches to

antitrust, and I have to say in listening to the

unilateral effects theory, you would almost be grateful,

I suppose, that you have a merger before the Commission

and before the DOJ where the theory is going to be

innovation markets or potential competition where now

this is applicable and you can't do simulations because

you have no current output for both firms.

nice.

That might be

MR. WERDEN:

You need an option model.

MR. LIPSKY:

As I said before, I think we are

focused on the wrong ex-Frenchman.

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

much what passes for antitrust economics right now was

all in the original core notebook back in 1838.

think unilateral effects were in there.

option of the theory of monopoly.

Bertrand was an IUV

I don't

Maybe it's an

I want to go at some specific issues that have

been alluded to but haven't been raised directly.

I can give a somewhat different focus.

Maybe

Who decides and how effective are those

economics in this process.

I'm going to back over the

first principles that I don't remember thinking about

since my first day in econometrics as an undergrad, and

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

probably I should have been thinking about them a lot

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

discussion and reveals my approach to some basic points

here.

First, there are facts, and some facts are

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

build a factory.

mental constructs of mathematical models, oligopoly

models, what have you.

model for merger law.

There are models which are basically

I have here sort of the classical

If an industry is more concentrated, the output

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

concentrated, ceteris paribus.

statistical tests which are propositions that grow out of

the application of a specific econometric method used to

estimate the parameters of a specific model, and I have

no idea what I'm saying here.

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

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

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

you actually needed to apply econometric methods to a

model in a particular industry.

Then there are

Obviously, the silicon-

I want to echo something I heard Joe say

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

think it has do with the idea that we are all decisionFor The Record, Inc.

Waldorf, Maryland

(301)870-8025

makers and we are not all judges, we are essentially

lawyers.

have enough economics or econometrics to understand

really what is contained in an assertion like that, and

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

later in the presentation.

There is a certain assumption that they don't

The value of a statistical test, the values of

the parameters and all those squares and all that neat

stuff that the economists roll around and try to persuade

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

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

something we have heard today.

We had a description of a statistical outcome

for a model estimated in the context of the Booksellers

merger.

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

shares of those two firms in the defined book market are

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

you I will find some problem with the econometric model

that was presented.

shares and you talk about entry and market conditions and

other sort of stylized facts about the way book

publishing works, I would be entirely prepared to accept

that those statistical results are valid, but the key

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

have a significant reason to doubt the validity of the

underlying model.

There is sort of the red letter error type

thing.

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

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

forgot to adjust for elasticity or you have too many

multiple linear variables or whatever it is.

That's where you give your data to Greg Werden

Even assuming that the econometrics is done

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

significant doubt about the truth of the model.

I think this has had a fairly profound impact

in my professional lifetime in the area of merger

analysis, which is when I entered grad school, people I

think we're beginning to have doubts about this long

series of econometric studies showing a correlation

between concentration and prices or concentration and

profits, and looking back, I'm not sure which was the

more popular, based on your original structure conduct

performance paradigm.

Why is it that it was thought to apply

econometrics to these models 30 or 40 years ago, and now,

I think it would be fair to say that those types of

models have been fairly thoroughly discredited.

process that not only involves "better econometrics."

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

It's a

I'm thinking particularly of Porter's 1972 Review of

Economic Statistics Paper, where he said, you can take

these regressions that are being used as evidence that

concentration and profits are positively correlated, and

notice that even though there is a claim for the validity

of that relationship, the statistical tests show that the

coefficients are small and the statistical confidence

intervals are rotten.

I've discovered one reason why that might be.

We divided the samples in two groups, one is sort of

convenience goods group, where it was sold in drug stores

and grocery stores, a small amount purchased at any given

time, very little consumer search behavior for the

characteristics of the product, often very subjective

characteristics of the product.

side.

You put that off to one

The particular model that had been applied

worked pretty well in that set of industries, but it

didn't work at all in a different set of industries,

which was things like cars or consumer durables, where

you do have significant consumer search behavior and a

lot of the other underlying variables represented by the

method of retailing which were thought to be important.

Time marches on.

It's not like you put it in

at one end of the lab and get it out at the other and

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

have a high degree of confidence that it's right.

much more an art, as I think somebody else said.

It's

Economic model is critical to the entire

process of antitrust decision making because you have to

decide a specific case, and every litigant, as I put here

in the middle, every litigant tells a story, and every

story has a model.

which is very bad, and I agree with the earlier remarks

that the focus on the unilateral effects of the merger

simulation approaches sometimes gets you a long ways by

making the assumptions that the underlying model is

explicit, always a good thing to do.

Sometimes the model is unstated,

Then the way the economic model really comes to

the floor in indicia decision making -- it's one of the

interesting phenomena, I think, in recent antitrust

jurisprudence.

In Matsushita, for example, the narrow issue

decided by the Supreme Court was whether an expert report

about the possible existence of a predatory pricing

conspiracy could be admitted, with the majority saying

no, because it's crazy, essentially, and the minority led

by Justice White saying wait a minute, what is the

seventh amendment for?

What is this?

A jury trial?

Expert testimony?

Justice White wanted to pretend, for purposes

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

of his very powerful decision and opinion in Matsushita,

that the validity of the expert's testimony was a fact.

I don't think that testimony was a fact.

something different.

about a model.

be applied to particular facts, but the fact that the

Matsushita majority said we are essentially entitled to

disregard what this expert says.

something of a turning point in anti-trust jurisprudence.

It was a model.

It was a story

It was a story about why a model should

State Oil vs. Khan.

It was

I think it was

There is a very explicit

example of saying reasoning about maximum vertical price

agreements was wrong and, therefore, the per se rule was

wrong.

Brooke Group is kind of interesting because the

oligopolistic disciplinary pricing model in Brooke Group,

the plaintiff's story was this is a big nasty oligopoly,

six firms.

like crazy, and the third largest firm decided to lower

the boom on the smallest firm, confident that the leaders

of the market, R.J. Reynolds with 38 percent of the

market, respectively, would stand by that Williamson

discipline licit, later owned by Brooke Group, and the

court, maybe because they were represented by Phil

Areeda, but the court said that could fly, it could be an

oligopolistic disciplinary pricing theory under the

The smallest firm really started to discount

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

Robinson-Patman Act, and sometimes I wonder and at the

same time feel thankful for the fact that there aren't a

lot more cases.

an oligopolistic disciplinary pricing theory that is

being tried out in this Tyson's case, but it's amazing

that nobody else has tried such a theory.

Maybe the Packers and Stockyard Act has

The Supreme Court says you must be covered

under the Robinson-Patman Act.

case was that doesn't apply here.

on my list and use it for this point, Brooke Group was

essentially the rejection of an expert report.

who said, I observed the factual pre-conditions for and

the theoretical apparatus for appreciating this

oligopolistic disciplinary pricing theory in this case,

and the court said no, that's crazy.

with the facts, not a crazy theory, inconsistent with the

facts.

What they said in that

This is why I put it

An expert

It's inconsistent

A particular fact that the court was dealing

with was that the main underpinning for the oligopolistic

disciplinary pricing theory was that the cigarette

industry was an oligopoly, with high prices, high

profits, sticky prices, sticky shares, all those things

we associate with oligopoly.

Well, it just so happens that I was senior

executive for the plaintiff that testified it wasn't an

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

oligopoly, prices weren't sticky, that profits weren't

abnormal, and I think the court had a hard time

swallowing the expert relying on a theory which was

essentially contradicted by the leaders of the plaintiff.

We have this weird situation where the judges

are deciding economic issues.

They are rejecting expert testimony.

Quartet, the four cases that really form the Daubert

line, are in a sense an extension and may have even been

suggested by Matsushita and Brooke Group, chronologically

it works out, I suppose we will never know.

30 years from now when the papers are disclosed we will

find out, but judges are getting more and more

uncomfortable with economic theorizing.

They are rejecting models.

I think the Daubert

Maybe 25 or

The problem that we have is, as has been

previously said in the panel, the lawyers control this

process.

the ones who are appointed by the politicians to hold the

positions at the agencies.

over the place.

The lawyers present the case.

The lawyers are

There are just lawyers all

Whatever economic content has been suffused

into anti-trust, I think there is a great deal of it, it

has to come through the lawyers and judges.

existing institution that filters out the quality of

economic testimony and compares the relative

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

There is no

persuasiveness of different economic views of these

particular cases that is really designed in a way you

would want an institution designed, to get to the truth

of the matter.

We have an advocacy process run by lawyers.

don't have time to sit around for the process to work.

You have some guy who is bringing cases based on the

notion that concentrations and profits are strongly

correlated, we don't have time to sit around and discover

that is wrong.

You have a failure in the antitrust realm.

It's not like the Challenger disaster, where the market

blows up and everybody says whoa, something really went

wrong.

There are antitrust disasters.

I think the United Machinery case is an

antitrust disaster.

case founded, I think, are illegitimate theories of

violation, suing remedy.

but it took about 10 to 20 years to happen.

debate as to whether the machinery industry was really

one that was destined to remain in American business

anyway, given all the things happening in the world.

That company driven out of business,

United Machinery was destroyed,

You can

I have summarized Matsushita and Brooke Group.

I mentioned the Daubert Quartet.

I also want to focus

you on the sources -- judges articulate their

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

dissatisfaction with the process by which economic

expertise is applied during antitrust cases.

in other cases, too.

They do it

Justice Breyer I don't think has ever mentioned

antitrust in his talking about this issue.

he gave a talk at AEI recently on this same thing, and

might have mentioned antitrust.

toxic torts and that kind of scientific evidence.

I understand

His focus is more on

He has basically proposed in a speech to the

AEA a few years ago, we really need to weed out the

cranks.

professional body weed out the people who know what they

are talking about and the people who don't.

We need to have the AEA or some responsible

I'm not so sure that's a great solution for

antitrust.

Association having a qualifications panel for antitrust

economists.

I can't imagine the American Economics

Judge Posner, in HFCS, a very intricate

exploration of the economic elements of that case that

were relevant to the judgment of whether there is

adequate evidence of conspiracy.

the district judge and said, look, when this goes back

down to explore this issue.

summary judgment in favor of the defendants, it would be

a really good idea to use Rule 706 of the Federal Rules

He pretty much beat on

Now that I have reversed the

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

of Evidence and appoint some experts so you can figure

this out at least as well and maybe better than I did.

If you haven't read Posner's decisions in Asahi

Glass vs. Pentech, and there is sort of a cluster of

cases that are related, you really need to read these

decisions.

Richard Posner, sitting as a district judge,

can you imagine, and going through unbelievably complex

cases involving a patent settlement and is that illegal,

there's some infringement issues along with other stuff,

you can learn about disappearing polymorphs and other

chemical curiosities.

It's just wonderful.

The reason it is inserted in my remarks here is

that Posner makes some rather pointed remarks about some

theories that are now popular in the analysis of pioneer

generic patent litigation settlements.

itching to see lower courts be more rigorous and more

explicit about how they confront economic modeling.

He is obviously

Just to treat unilateral effects as essentially

a case study of what I've been talking about, the

fundamental inside unilateral effects is really a

no-brainer. Sure, A's customers could go to B.

merger, they won't be able to go to B.

do they have to climb to get out of a hole.

going to impose any kind of long run pricing and

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

After the

How high a crater

Is that

imposition on consumers?

performance?

Is that going to decrease

We have already heard a lot this morning and

I'm not going into the fact that the analyses by which

you can get at these phenomena quantitatively require a

whole bunch of assumptions.

I conclude with the issue of kind of baseline

credibility.

world in which the decision makers do not have the

expertise to decide whether a model or econometric study

offered is worth a darn, what is the point at which we

are going to allow econometric testimony to influence the

decisions in antitrust cases?

If we are going to continue to live in a

I think it's a problem.

Let me suggest, and

here I want to run straight at an issue that Joe raised,

which is it's the lawyers who decide, to me, the lawyers

shouldn't decide.

review everything that goes into an antitrust case, sort

of certify for the judge, not that the expert is a good

expert, but that he thinks the analysis is pretty

plausible, hasn't found any particular issue with it.

It's not easy to imagine institutions that

Maybe independent economists should

would effectively accomplish what I'm trying to drive at,

but think about that issue.

actually decide?

Should the economist

Should the judges, should the lawyers

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

be bound by what the economists say?

David Scheffman, who has written and I think as

others have mentioned this morning, the judgment on these

economic issues is done simultaneously, you can't

separate them out.

Nabisco Brands case that was decided, where New York

challenged a merger that both Federal agencies had passed

on and found that Hippocrates was not fringed.

Read Fred Kahn's testimony in the

This is from the first sentence of the

Aphorisms, his little sayings about good medical

practice.

Do no harm.

Aphorisms.

Bill Baxter was a great fan of Hippocrates.

That wasn't the first sentence of the

The first sentence is "Life is short, and Art

long, the crisis fleeting, experience perilous, and

decision difficult."

That applies here, too.

These are some criticisms I have of typical

enforcement agency modeling.

Market definitions tend to

be narrow.

There tends to be, I think, heavy discounting of dynamic

effects, and it's wrong for obvious reasons.

necessarily does it consciously, but it tends to

encourage the bringing of enforcement actions where

otherwise somebody a little more relaxed on those

assumptions would not have any concern.

They tend to focus on isolated time periods.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

No one

Also, there is a tendency to go to extremes.

We will often hear arguments in the front office that

make it sound as if the parties have been unable to agree

on anything in terms of the assumptions that are going

into their arguments, their story.

I've mentioned some of these alternatives,

reliance on neutral experts, Rule 706.

Do we need to

have a category of economists who work for the agencies

but don't really have a role in cases, or maybe they

should be outside consultants, people who theoretically

won't have any percentage in the outcome because they are

not going to be involved in either the prosecution or

defense of the case.

idea.

I think we should think about.

I don't think that's such a great

I just raise it because this is the kind of thing

Can we get more peer review, maybe can we get

more industrial organization professors to look at these

things?

I think a lot of them do.

Finally, my last desperate gambit, maybe a

Federal judicial center should be given a pot of money to

study all this stuff.

Just to remind you of why they liked the per se

theory, United States vs. Topco. I loved that footnote

that says if we were to consider the economic

justifications for this joint venture, it would send us

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

100

into the wilds of economic theory, so welcome to the

wilds.

MR. FROEB:

Thank you, Tad.

Greg?

While we

are waiting for Greg, I want to ask Greg Werden a

question.

simulation, it's a seldom useful exercise, too costly,

too time consuming, and it's persuasive only when we are

using it to justify something we already know, like from

the totality of the facts, we know that the merger is not

going to raise price, here's an economic model that says

that.

Both Tad and Joe raised questions about merger

Do you have any reaction to that?

frequently it is used, how frequently it influences

decision making at the DOJ?

MR. WERDEN:

How

In the first place, I don't think

either of them were as negative on merger simulations as

you suggested they were, particularly not Tad.

MR. LIPSKY:

It's helpful to clarify.

I would

agree the outcome is seldom very persuasive.

determine the outcome and often is really not

sufficiently persuasive to have a major influence on the

judgment.

MR. WERDEN:

It doesn't

I think there are two kinds of

major influences.

One, at a very early stage, there

certainly have been cases where merger simulation helped

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

101

decide not to proceed with an investigation.

and dirty merger simulation, confirming what some people

already suspect, it's pretty easy for this merger to not

raise price much, to lower price, next case.

Very quick

Also, there are a number of cases in which I

think sophisticated analysis was given a lot of weight in

the agency decision about challenging or not challenging

a merger.

MR. FROEB:

MR. LEONARD:

Greg, go ahead.

Thanks.

Luke, Greg and I were

part of a group of economists who got the ball rolling

with merger simulation 10 or 12 or 14 years ago, whatever

it was.

It quickly became pretty popular and it happened

quickly.

I think what happens when you have something

that gains popularity very quickly is that a backlash

occurs subsequently, and I think that has happened a

little bit.

problems.

We have been talking about some of the

I thought what I would do today is try to

address some of the questions that have been raised about

merger simulation.

Let me just start by saying, and this has been

mentioned so I'm going to go fast, basically, you have

the fact based inquiry, which are reviewing the

documents, reviewing the depositions, interviewing

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

102

customers, so forth and so on.

simulation.

Then you have merger

They shouldn't be substitutes for each other.

They really should be complements.

something to the party.

They both add

Merger simulation has two pieces to it.

The

first is what I'd call guesstimating consumer demand

functions.

do that possibly by looking at documents, if the correct

ones are out there.

You can do that with econometrics.

You can

The second case is the model of firm behavior

you are going to use.

oligopoly models.

People have been talking about

That's what I mean there.

The two pieces together are what constitute

merger simulation.

mean to include both pieces.

When I talk about it, I'm going to

Let me first say what does merger simulation

add where you have the documents.

like it.

What does merger simulation add?

argue is it helps move merger analysis closer to what I

would call science.

Attorneys certainly

With economists, I would say they have value.

I guess what I would

What do I mean by "science?"

The most important thing is that the analysis

is based on a set of theories or a theory that is

testable.

That's really the essence of what science is

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

103

about.

We come up with theories.

pass the test, we accept them.

them.

with a better theory.

is all about.

We test them.

If they

We keep trying to test

Eventually, they may be rejected, and we come up

That's what the scientific method

Second, and what has been mentioned earlier

today, the underlying assumptions of the analysis are

very clearly delineated.

of the analysis can be replicated by somebody else.

The third thing is the results

Finally, scientific analysis typically will

allow you to calculate the precision of your results.

think Tad mentioned that it is conditional on the

assumptions, and that is certainly true.

Why should merger simulation aspire to be a

science?

I guess the first thing I think as probably the

most important is it gets closer to objectivity and

further from subjectivity.

really about.

That is again what science is

Secondly, it provides some certainty.

know what the rules of the game are, and that's the whole

purpose for the merger guidelines, I think, to lay out

some rules and to say here is how the game is going to be

played, and then gives people an idea about how the

analysis will go and allows them just to know how things

are going to fly.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

If you

104

Next is the sources of disagreement between the

parties, I think, are a lot more easily identified when

you have some scientific approaches on each side, a

larger definition, arguments that are based on documents.

There is a lot of subjectivity.

There is a lot of hidden

assumptions, a lot of judgment.

If you can instead

replace that with an analysis of elasticities, then you

are talking about hard numbers.

about how you get those elasticities.

elasticities, I think, is better than arguing about some

vague ideas that come out of documents.

There may be issues

Arguing about the

This is what I said before.

What we are going

to get is bad methods are going to be replaced by good

methods, and good methods are going to be replaced by

better methods.

process.

That is the essence of the scientific

Obviously, if you come out with a result that

is very uncertain, you are going to put a lot less weight

on it.

In large, merger simulation is in fact a

science or a scientific analysis.

It's based certainly

on well established economic theory.

of consumer demand that drives the demand equations.

have the econometric theory that is behind the estimation

procedures used to get at the demand elasticities.

We have the theory

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

105

have the oligopoly theory, which as someone was pointing

out, is based on infringement from many, many years ago.

It has been around for a long time.

We tend to lay out the assumptions, the

underlying assumptions, very clearly, and this is a very

important point.

tested.

is very important.

It says assumptions can often be

I think that is another aspect of all this that

For instance, the demand model you choose can

be tested.

an AIDS model perhaps, under the right conditions.

We can distinguish between a logit model and

I can give another economist the data and I can

say here's what I did, and that person could hopefully

get the same results, replication.

There are going to be choices along the way and

we may disagree about those, but at least we know what

those choices are, and they are fully described.

Finally, you can get a standard error for a

predicted price change.

assumed demand models and assumed oligopoly models, but

hopefully again those have been subject to some testing

along the way.

Again, it is conditional on the

The so-called document approach, it is not

really a science, at least in the same sense.

all, usually not incredibly well linked to economic

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

First of

106

theory.

Lawyers have models in their minds.

think that's true, but they are not very well

articulated, and so the assumptions themselves aren't

laid out very well.

A lot of times, they don't even know what the assumptions

are.

I don't

The assumptions often aren't tested.

I think if you think about what is in these

documents, that really is one of the problems.

at these documents and we assume that the authors of them

are discussing things or analyzing things in the way that

is actually meaningful for the merger analysis we are

trying to perform.

always the case.

We look

I don't think that is necessarily

If you look at the Staples case, there was a

certain set of analyses done.

entirely different purpose.

necessary.

agreed with that and went forward and did an additional

analysis.

They were done for an

I think further analysis was

I think even the economists on the FTC side

Similarly, customer reviews are used a lot.

think there are some serious problems by sampling.

The

customers that care are the ones who come forward.

They

may also tend to be the marginal ones.

of that.

What do we make

Another problem with the documents is the

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

107

results can't be replicated, at least in the same way

that a merger simulation can be, because you are going to

have two different people look at the documents and

review them and come up with very different conclusions.

Finally, in terms of some questions that have

been raised about merger simulation.

technique?

true of almost any scientific inquiry.

complete.

Is it a perfected

The answer to that is clearly no.

It's never final.

That is

It's never

Things are always refined.

It doesn't need to be finalized to be useful.

Whatever the current state is, it can be useful in

certain ways, even though it may be improved upon later.

I think it's a theory of evolution.

In Darwin's original formulation, he thought

that evolution occurred gradually over time.

looked at the fossil record, you would find one organism

that clearly evolved from an earlier one, but they were

very different, and it was talked about as being gaps in

the fossil record, we were missing the fossils for the

organisms in between.

When they

The punctuated equilibrium theory was developed

and basically what it said was wait a minute, evolution

isn't gradual.

no change for long periods of time in between.

explains the facts better, and it is perhaps a better

It happens in short bursts and there is

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

108

theory than the original one.

If you go back to the 1930s and 1940s, it

doesn't mean Darwin's formulation was useless.

it formed a lot of scientific thought.

In fact,

The second question that comes up is does

merger simulation provide an answer.

No, it doesn't.

That's one of the consequences of the fact that in this

science, we have to go out and test the assumptions, and

we might find the assumptions aren't valid.

merger simulation is not going to be -- at least in that

format, is not going to work.

Then the

It would be impossible to capture all the

economic processes in a model that is going to work,

given we are not going to be able to do this merger

simulation in that case.

If you contrast that to the documents approach

where you can't always get an answer, you can get any

answer you want within the bounds of reason.

often get two answers, one for the defendant and one for

the plaintiff.

You can

I would argue that because it's a science, you

are not always going to get an answer, and that's a fact

of life, I think.

Does merger simulation involve choices?

it clearly does.

The good thing is the choices are

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

Yes,

109

clearly articulated, and you can get down to brass tacks

and have the experts on each side argue about what the

right choice is.

testing.

the arguments.

Often, the choices are subject to

Again, very important.

That's a way to resolve

If you contrast that to the documents approach

where the assumptions are often not expressly stated and

often can't be tested or certainly aren't tested, I think

there the choices are so hidden.

different.

among experts about what choices to make.

It doesn't

make merger simulation any less a science.

Scientific

disputes have existed forever.

them.

That makes it very

And the fact that there might be disputes

There are ways to reserve

Is merger simulation the only co-scientific

merger analysis, and I would say no to that.

there are other kinds of analyses that are co-scientific

that are helpful.

industry when one firm's plant blows up and isn't able to

supply the market any more, and see where customers turn

to and what other firms do in terms of prices.

Certainly,

We can look at what happens in the

You can also look at what happens when there is

a new product introduction.

There are obviously

complexities about these types of analyses as well, and

choices and everything else.

In certain cases, they are

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

110

going to be more appropriate.

FTC vs. Staples is an example of that, where

the analyses were done based on looking at markets where

there were a different number of competitors or where the

number of competitors changed.

Should the merger simulation approach replace

the documents approach?

absolutely not.

documents and it is going to give you very useful

qualitative information that allows you to fit the model

better and give you institutional details, and perhaps

even allow you to do some testing of the model, as I

said, is very important.

As I said at the outset,

A lot of times, you can look at the

What I am going to turn to is some of the

problems that I think merger simulation faces.

The first

is it may not appeal to attorneys.

little bit about that earlier.

really does require a fairly high level of economic

expertise, and I think attorneys, it may seem like a

black box to them.

We have heard a

The first thing is it

They not feel as comfortable with it.

Secondly, I think when you end up in a battle

of experts like this for each side, talking about very

sophisticated choices, as we were talking about before,

you just end up in a situation where no one understands

what the heck they are talking about, and the two experts

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

111

basically cancel each other out and the fact finder

doesn't pay any attention to it.

I think there is some of that.

What Tim was

talking about before, perhaps having an outside expert to

resolve some of those things would be helpful in that

regard.

There are some that feel merger simulation is

too new to be attempted in a courtroom.

true that law tends to move more slowly than science.

an example, I was thinking about intellectual property

damages cases.

I think it is

What happened is you basically had to show

there were no acceptable non-infringing substitutes to

get any lost profits at all.

reasonable royalty side of things.

don't get any damages, but you don't get any lost

profits.

Of course, there is this

It doesn't mean you

In terms of the share based approach, it kind

of made things better, but again, it is based on

assumptions that an economist would question, I think, or

would want to test at least.

I think simulation would really help in these

kinds of cases and hopefully will be used sometime soon.

Greg already talked about this stuff, so I'm

going to skip that.

I just want to say one thing about

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

112

the usefulness of comparing predictions from merger

simulations to actual outcomes.

standard scientific practice to do that.

It is absolutely a

I have this paper with Jerry Hausman where we

try to do that a little bit, I think with mixed results,

but fairly positively.

Since every industry is really very different,

I'm not sure how much it would help to say in these five

industries, we have validated the merger simulation

approach.

us when we were applying it to the sixth one.

It's not clear to me how much that would help

Finally, I guess I'll just finish by saying as

I said before, this is a development process, as is all

science, and there are clearly areas where merger

simulation can be improved.

list is the old Darwin model.

beating today.

generally, it would be great to come up with a better

oligopoly model.

of merger simulation lies.

Thanks.

I think number one on the

Bertrand has taken a

It's absolutely true, just in economics

I think that is really where the future

QUESTION AND ANSWER SESSION

MR. FROEB:

Thank you, Greg.

I know we are

heavily weighted towards economists on this panel.

want to give the attorneys some time to rebut or say

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

113

whatever you want.

MR. KATTAN:

I actually find myself in

agreement with most of the things that the economists

have said.

method.

are out there.

articulated, and we can beat up on them by saying, well,

I have a problem with that assumption or this assumption.

The models that are explicitly by lawyers are a

Certainly when they critique the legal

I think they have a valid point.

Their models

Their assumptions are clearly

lot more difficult to discern.

There is certainly a

difference in the level of opacity, and therefore, in the

susceptibility of the model to questioning, and the deck

is stacked institutionally in favor of the lawyers,

largely because they control the process.

To me, the issue is what do I do when I have a

set of choices, each of which is going to predict a price

increase no matter what, and I'm told basically take your

choice, which poison do you want.

I would ask Greg Leonard, how do you sell to a

client the idea that they really have to do this, you

know, however many thousands of dollars it is going to

cost them to pay you, to come with a model that will

predict the price increase, next to which you will have

yes, but if you count for efficiencies, you account for

this, you account for that, that price increase will be

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

114

obviated or it's such a small price increase, that it

really shouldn't matter.

MR. LEONARD:

I think you have to go into it

realizing that the model is going to predict a price

increase, unless those costs and elasticities are zero,

which is unlikely in a lot of these cases.

one percent or two percent, it is just not enough to

worry about, especially because there are probably

offsetting effects.

that is going to wipe it out.

MR. KATTAN:

If you get

If they are even small efficiencies,

There is this language in the

merger guidelines, which I can't recite as well as I

should be able to, due to the fact that the five percent

test is not a tolerance level for price increases, and we

have certainly seen cases where the predicted price

increase was very, very small, and a challenge in fact

took place.

I don't take comfort in the prediction of an

one percent price increase and the argument that well,

you wouldn't need a lot of efficiencies to obviate.

worry about my ability in most cases to show any kind of

efficiency that is going to bring prices down.

show significant efficiencies that are driving the deal,

but as I said before, they are usually going to be more

in the overhead category.

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

You can

115

MR. LEONARD:

If you go back to the guidelines

and say well, that's sort of how we should do things.

you follow the guidelines strictly, next best substitute

or second, if it turns out it's the third one, you can

still have a price increase there, too.

economics of it and I don't know how you get around that.

I don't think following the guidelines versus merger

simulation saves you from that.

assumption in there, that we don't care about it if it's

the third on the list.

This is just the

Again, you are making an

I don't know if that's right.

MR. FROEB:

Tad?

MR. LIPSKY:

I think part of the problem -- I

think there are some respectable roles for simulation,

making explicit the hidden assumptions is a good one.

feel that a lot of these analyses, what triggered the

thought was the mention of the five percent.

the 1982 guidelines were written, one of the principal

reservations about using the five percent SSNIP test was

that it would suggest an unsustainable degree of

precision.

Back when

You know, this is going to become a standard

and the lawyers are going to seize on this, it's a very

specific number used in a very specific way, and sure

enough, that's exactly what happened, to the point that

you may recall that the major feature -- there was

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

116

actually an 1984 set of guidelines issued by Paul

McGrath -- one of the major features of the 1984

guidelines was this big introductory statement that said,

now, wait a minute, you totally have the wrong idea on

this five percent test.

some markets could be 1.5.

the right question and the magnitudes we have no idea

about.

Some markets could be seven and

This is just meant to suggest

Similarly, I think, in the simulation area, if

you can be confident about the sign of the first

derivative for any of the major variables, whether it is

price, output or whatever, someone is going to triumph

from the fact that you can establish it from any modeling

device at all, that would be real progress.

I really liked the analogy to the dispute over

evolution.

equilibrium was popularized as a theory, there are some

modeling gradulists who describe punctuated equilibrium

in a manner intended to be pejorative as evolution by

jerks.

respond by calling the gradulists evolution by creeps,

which suggests something of the flavor of the debate over

unilateral effects theory in antitrust law.

We know that even after punctuated

Of course, there are punctuated equilibrium guys

The question I want to pose is this, and let me

say it very bluntly.

My view is that most of the

For The Record, Inc.

Waldorf, Maryland

(301)870-8025

117

shortcomings in the way that economics is applied to

antitrust decision making, it's not economics isn't

perfect, everybody knows all the stuff we have been

saying about you d

This text is long and has been trimmed here. Open the source document for the complete record.

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.