In re: High-Tech Employee Antitrust Litigation

Filing 716

Omnibus Declaration of Christina J. Brown in Support of #715 Reply re Joint Motion to Exclude the Expert Testimony of Edward E. Leamer, Ph.D. , #714 Reply to Joint Motion to Strike the Improper Rebuttal Testimony in Dr. Leamer's Reply Expert Report or, in the Alternative, MOTION for Leave to Submit a Reply Report of Dr. Stiroh filed by Apple Inc.. (Attachments: #1 Exhibit A, #2 Exhibit B, #3 Exhibit C, #4 Exhibit D, #5 Exhibit E, #6 Exhibit F, #7 Exhibit G, #8 Exhibit H, #9 Exhibit I, #10 Exhibit J, #11 Exhibit K, #12 Exhibit L, #13 Exhibit M, #14 Exhibit N, #15 Exhibit O, #16 Exhibit P, #17 Exhibit Q)(Related document(s) #715 , #714 ) (Brown, Christina) (Filed on 2/27/2014) Modified text on 2/28/2014 (dhmS, COURT STAFF).

EXHIBIT B

OMNIBUS BROWN DECLARATION

HIGHLY CONFIDENTIAL

1

UNITED STATES DISTRICT COURT

2

NORTHERN DISTRICT OF CALIFORNIA, SAN JOSE DIVISION

3
4

---------------------------

5

IN RE: HIGH-TECH EMPLOYEE

)

6

ANTITRUST LITIGATION

) No. 11-CV-2509-LHK

7

---------------------------

8
9

HIGHLY CONFIDENTIAL

10
11

VIDEOTAPED DEPOSITION OF EDWARD LEAMER

12

San Francisco, California

13

Friday, October 26, 2012

14

Volume I

15
16
17
18
19
20

Reported by:

21

ASHLEY SOEVYN

22

CSR No. 12019

23

Job No. 1545691

24
25

PAGES 1 - 476
Page 1
Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

Q.

Compensation for job titles.

12:35:14

2

A.

But there are title indicators in here, so

12:35:15

3

that's going to absorb everything that's title

12:35:19

4

specific.

So the thing about that coefficient on

12:35:21

5

that title indicator, it's going to be likely the

12:35:23

6

average compensation within that title adjusted for

12:35:24

7

these other variables.

12:35:24

8
9

Q.

Are the coefficients in your analysis the

same year to year?

12:35:31
12:35:33

10

A.

No, they are not.

12:35:35

11

Q.

So what conclusion do you draw from that?

12:35:37

12

A.

Well, their conclusion is that they're

12:35:40

13

similar.

12:35:43

14

Q.

By --

12:35:46

15

A.

Similar enough to suggest that there's a

12:35:47

16

fairly rigid salary structure in place on a

12:35:49

17

year-by-year basis.

12:35:52

18
19

Q.

Similar in a statistic -- statistically

significant way?

12:35:55
12:35:57

20

A.

I have not explored that possibility.

12:35:59

21

Q.

Haven't tested that, have you?

12:36:01

22

A.

No.

12:36:02

23

Q.

So you just eyeballed it?

12:36:03

24

A.

I guess that's correct.

12:36:14

25

Q.

Now, there is a way to test that, isn't

12:36:18
Page 219

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

there?

2

A.

12:36:20
You're using the word "testing."

The word

12:36:20

3

"statistical testing" is talking about measurability

12:36:24

4

and we're really here about a consequence -- we

12:36:26

5

really should be talking about consequentiality.

12:36:30

6

There are meaningful differences in the wage

12:36:31

7

structures over time.

12:36:34

8

suggesting I should do a formal hypothesis test

12:36:36

9

using econometric power and accept or reject this

12:36:41

10

idea that there isn't any change.

12:36:43

11

consider relevant.

12

changes are consequential, and the consequential

12:36:50

13

changes that are statistically reliable.

12:36:54

14

So you're -- you're

That isn't what I

What's relevant is whether the

So I -- I haven't carried out that

12:36:43

12:36:54

15

exercise, but I would object to what I think would

12:37:00

16

be the target of your hypothesis testing.

12:37:05

17
18

Q.

Let me ask you, please, to look at

paragraph 130 on page 55.

12:37:07
12:37:08

19

A.

Okay.

12:37:10

20

Q.

The third sentence of that paragraph

12:37:10

21

states, "Furthermore, the fact that the coefficients

12:37:12

22

and the regressions did not vary substantially over

12:37:14

23

time, suggests the compensation structures were

12:37:17

24

relatively stable over time."

12:37:20

25

Now, you did not do a statistical test to

12:37:23
Page 220

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1
2
3

draw that conclusion, correct?
A.

12:37:28

I'm -- I'm -- what paragraph are you

referring to?

12:37:30
12:37:31

4

Q.

130.

12:37:32

5

A.

Yeah.

12:37:34

6

Q.

Third sentence.

12:37:34

7

A.

I -- I -- I did not make any formal attempt

12:37:39

8

to determine their instability over time.

9

wisdom to explore the coefficients and came to the

12:37:45

10

conclusion that variability was not consequential.

12:37:53

11

Q.

12:37:42

You eye -- you eyeballed it?

12:37:56

MR. GLACKIN:

12
13

I used my

12:37:57

I'm sorry, he wasn't

finished.

14

12:37:57
THE WITNESS:

And that additional test were

12:37:59

15

carried out in this hypothesis testing that you

12:38:01

16

described, unless it's done in a way that deals with

12:38:05

17

consequentiality, it's not going to be informative

12:38:09

18

to the task that I was assigned.

12:38:15

19

BY MR. PICKETT:

12:38:18

20
21

Q.

Right.

So -- so you eyeballed them,

correct?

12:38:18
12:38:19

22

MR. GLACKIN:

Objection.

12:38:19

23

THE WITNESS:

I examined the coefficients

12:38:20

24

to see if they were what I regarded to be major

12:38:22

25

changes in the structure over time.

12:38:26

And in order to

Page 221

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1
2
3

which one clusters their observations.
Q.

15:31:22

And do you think that your analysis is

subject to that criticism?

15:31:25
15:31:27

4

MR. GLACKIN:

Objection.

15:31:30

5

THE WITNESS:

Well, maybe I misunderstood.

15:31:31

6

I'm not sure what clustering standard -- clustering

15:31:32

7

standard errors are.

15:31:34

8

BY MR. PICKETT:

15:31:35

9

Q.

So you -- you don't think -- well, let me

10

go back.

I'll reask the question.

Are you familiar

11

with the term "clustering standard error"?

15:31:35
15:31:37
15:31:41

12

A.

Well, I know what clustering is.

15:31:44

13

Q.

Uh-huh.

15:31:46

14

A.

And I know what standard of errors are in

15:31:46

15

the context of your question, which I thought is

15:31:49

16

what we were talking about.

15:31:52

17

talking about some kind of standard of errors that

15:31:54

18

apply to the Figure 23 that you're calling

15:31:57

19

clustering standard of errors, which is something

15:31:59

20

that I'm not familiar with.

15:32:01

21
22
23

Q.

But I think you're

Would it be appropriate to use clustering

in performing your regression analysis?
A.

I don't see why.

I don't see why.

15:32:02
15:32:05

It's

15:32:18

24

possible that some argument can be made that it

15:32:22

25

doesn't come to the front of my brain, but off the

15:32:24
Page 328

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

top of my head -- by "clustering," I think you mean

15:32:27

2

using subsets of the data.

15:32:32

3

consequence of nothing in the air that would rather

15:32:34

4

be built into the model, rather than during

15:32:37

5

progression of subsets.

15:32:39

6
7
8
9
10
11

Q.

And I think that's a

Are any of the data series you used

correlated?
A.

15:32:42
15:32:43

Every one of these variables is correlated.

15:32:44

Every variable in the equation has some degree of

15:32:46

correlation.

15:32:52

Q.

Well, why -- okay.

And -- did your conduct

15:33:16

12

regression pick up any lawful agreement -- by

15:33:16

13

"lawful," I mean talking about that unilateral

15:33:20

14

policy, for example, or some joint corroboration

15:33:23

15

that started in 2005?

15:33:26

16

A.

Yes, it will pick up anything that is

15:33:32

17

applicable to that period of time when the thing is

15:33:34

18

turned on.

15:33:37

19

little different for Pixar and Lucasfilm.

20

but unless you have controls in this equation to

15:33:43

21

eliminate the effects of these other material

15:33:48

22

issues --

15:33:51

So it's turned on for 2005, 2009, a
But --

15:33:41

23

Q.

Okay.

15:33:51

24

A.

-- they are going to be picked up by that

15:33:51

25

conduct variable.

15:33:53
Page 329

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1
2
3
4

Q.

And you haven't eliminated those effects,

have you?
A.

15:33:54
15:33:56

I think I have.

I've got quite a few

controls in the equation and --

15:33:56
15:33:58

5

Q.

But the --

15:33:58

6

A.

-- you'll have to suggest to me exactly

15:33:58

7
8
9
10

what is not included.
Q.

15:34:01

If the Google arrangement, vis-a-vis,

15:34:02

Intel, turns out to be unilateral, how have you

15:34:07

controlled for that in 2005?

15:34:10

11

A.

Yeah, I have not done that.

15:34:18

12

Q.

Are you assuming that the error terms in

15:34:20

13

your conduct equation are independent across

15:34:25

14

individuals?

15:34:27

15

A.

Independent across individuals?

I think --

15:34:36

16

are you thinking about intertemporal dependence?

15:34:55

17

I'm not sure -- so there's two -- maybe there's a

15:34:56

18

clustering that you're getting at or maybe it's

15:35:00

19

intertemporal dependence.

15:35:02

20

Let me interpret your equation as -- your

15:35:02

21

-- your question, as if you were referring to

15:35:04

22

intertemporal -- inter -- inter -- inter --

15:35:08

23

intertemporal dependence.

15:35:08

24

Intertemporal dependence.

25

I know what to answer.

So if you look at these live variables,

15:35:13
15:35:13
Page 330

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

agreements.

2

control for things that are going on during that

15:55:08

3

period of time.

15:55:10

4

BY MR. PICKETT:

15:55:10

5
6

Q.

And these other variables are meant to

But the attribution to the cold calls is

not supported by any data, correct?

15:55:04

15:55:10
15:55:15

7

MR. GLACKIN:

Objection, vague.

15:55:18

8

THE WITNESS:

I disagree with that --

15:55:20

9
10

BY MR. PICKETT:
Q.

15:55:22

What data do you have -- (Cross-talking.)

15:55:22

11

MR. GLACKIN:

Let him finish his answer.

15:55:23

12

THE WITNESS:

The contact variable is being

15:55:26

13

turned on during the period of time which the

15:55:28

14

anti-cold calling agreements were in place and

15:55:31

15

turned off when those anti-cold calling agreements

15:55:33

16

were not in place.

15:55:36

17

And that's a sense of it picking up the

15:55:37

18

impact of the anti-cold calling agreements.

15:55:39

19

BY MR. PICKETT:

15:55:44

20

Q.

Without reference to the actual number of

15:55:44

21

cold calls, without reference to the actual

15:55:46

22

information flow that's informing the price

15:55:48

23

discovery process?

15:55:52

24
25

MR. GLACKIN:

Objection, compound and

vague.

15:55:53
15:55:54
Page 338

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

THE WITNESS:

The answer -- already, we

15:55:56

2

obviously -- if we had the cold calling data, it

15:55:57

3

would be very useful, very interesting, and we could

15:56:00

4

pursue the price discovery model in an interesting

15:56:02

5

and intervening way.

15:56:06

6

very best that can be possibly be done, which is to

15:56:08

7

use the conduct variable.

15:56:10

8

BY MR. PICKETT:

15:56:13

9

Q.

Absent that, we've done the

And so your conduct regression would pick

15:56:13

10

up any unilateral agreements -- I'm sorry, any

15:56:15

11

unilateral policies that started in 2005 not to cold

15:56:20

12

call, correct?

15:56:24

13
14

MR. GLACKIN:

Objection, asked and answered

again.

15

15:56:26
15:56:27

THE WITNESS:

To the extent that those

15:56:28

16

unilateral agreements suppress wages, the answer is

15:56:29

17

yes.

15:56:33

18

BY MR. PICKETT:

15:56:33

19

Q.

Well, wouldn't they, by your hypothesis?

15:56:33

20

A.

Well, you had this hypothesis that the cold

15:56:36

21

call would be going to somebody else.

15:56:38

22

Q.

I do, yes.

15:56:40

23

A.

So the point is that -- (Cross-talking.)

15:56:42

24

Q.

In your world, wouldn't your conduct

15:56:44

25

regression pick up unilateral policies starting in

15:56:47
Page 339

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

2005?

2

A.

15:56:52
I'll repeat my answer, which is it picks up

15:56:53

3

suppression of compensation during the period of

15:56:58

4

time which the cold call dummy -- the conduct

15:57:01

5

dummies turn it on.

15:57:03

6

Q.

And if Google had a unilateral policy not

15:57:05

7

to cold call Oracle employees, it would pick that

15:57:08

8

up, too?

15:57:11

9

A.

To the extent that these are coincident in

15:57:12

10

time with the agreements that they had, these

15:57:14

11

bilateral agreements they had, and to the extent

15:57:17

12

that they suppress wages during that period of time,

15:57:20

13

it's going to be picked up by the conduct variable

15:57:21

14

unless there's some other control in the equation

15:57:25

15

that accounts for that availability.

15:57:28

16

So do you know how much of this unilateral

15:57:30

17

activity within the defendants -- or outside of the

15:57:32

18

group of seven defendants is being picked up or

15:57:35

19

not?

15:57:38

20

Q.

A.

I do not know how much.

I've indicated

15:57:41

21

that unless that -- that unilateral -- what you're

15:57:45

22

calling unilateral websites is put in place exactly

15:57:49

23

the same period of time that the bilateral

15:57:53

24

agreements were put in place, called off the same

15:57:54

25

time, if it satisfied that requirement, it's going

15:57:57
Page 340

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

A.

Well, these --

16:00:00

2

Q.

-- during the class periods?

16:00:00

3

A.

Those would be anecdotes, rather than a

16:00:00

4
5

general statement, but -Q.

16:00:00

It's relevant data, isn't it?

16:00:00

6

MR. GLACKIN:

Please finish your answer.

16:00:00

7

THE WITNESS:

It's relevant, but not

16:00:00

8

decisive.

16:00:00

9

BY MR. PICKETT:

16:00:00

10

Q.

Do you have decisive data?

16:00:00

11

A.

No, I do not.

16:00:00

12

Q.

Let me ask you about paragraph 76 on page

16:00:00

13

32 of the --

14
15

THE REPORTER:
record.

16
17

16:00:00
One more time, off the

Sorry.

16:00:00

THE VIDEOGRAPHER:

Okay.

Off the record.

It's 5:47.

16:00:00
16:00:00

18

(Recess taken.)

19

16:00:00

THE VIDEOGRAPHER:

Back on the record.

17:49:52
17:49:52

20

It's 5:50.

17:50:08

21

BY MR. PICKETT:

17:50:14

22

Q.

So looking at paragraph 76, you state that

17:50:14

23

"Agreements that reduce the number of bilateral

17:50:18

24

bargains further slow the price discovery process

17:50:21

25

and effect the whole sequence of actual

17:50:24
Page 412

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

transactions."

2

17:50:26

Is your opinion that the challenged

17:50:27

3

agreements reduced the number of bilateral bargains

17:50:29

4

during the class period?

17:50:32

5

A.

Well, I include bargains, conduct in

17:50:35

6

general.

7

calling agreements did reduce the number of

17:50:43

8

contacts.

17:50:45

9
10

Q.

17:50:38

So a bargain is not an agreement, it's a --

it's a discussion about potential agreements?

11
12

And my opinion is that the anti-cold

MR. GLACKIN:

17:50:49

Objection, argumentative,

mischaracterizes.

13

THE WITNESS:

17:50:46

17:50:54
17:50:55

It's a communication -- I

17:50:57

14

want it to be defined as a communication that

17:50:57

15

reveals information about possibilities.

17:51:00

16

BY MR. PICKETT:

17:51:04

17

Q.

And --

17:51:04

18

A.

And the more that that goes on, the more

17:51:08

19

rapidly will be the finding of the equilibrium

17:51:09

20

market.

17:51:12

21
22
23
24
25

Q.

How do you know that other cold calls to

other employers and employees didn't substitute?
A.

This one we've been on before, too.

So the

answer is, I -- I don't have evidence on that.
Q.

So you don't know whether the price

17:51:12
17:51:16
17:51:23
17:51:26
17:51:30
Page 413

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

discovery process was further slowed or not?

2

A.

But we will leave that to the regression.

17:51:34
17:51:37

3

This is -- lays all down the framework, the same

17:51:39

4

thing we've been saying over and over.

17:51:44

5

the framework that suggests that there will be an

17:51:46

6

impact on price formation and will go to the data to

17:51:47

7

decide whether it's actually there.

17:51:52

8
9

MR. GLACKIN:

It's to set

So I -- since you seem to be

17:51:54

moving on, I want to raise something that came up

17:51:56

10

over the break, which is hat Dr. Leamer needs to

17:51:59

11

leave here at 7:00 o'clock, to go on an airplane and

17:52:02

12

go.

17:52:05

13

beyond 7:00.

And I don't think this deposition needs to go

14

17:52:07

MR. PICKETT:
guaranty.

16

17:52:07

And if we need to adjourn, we can

15

I'll try, but I can't

17:52:09

reconvene.

17
18
19

17:52:10

MR. GLACKIN:

I just want to be clear,

we're going to oppose reconvening after that.

17:52:13

I want to tell you that if I

17:52:15

20

have questions, I'll go to court to try and get the

17:52:17

21

right to do that.

17:52:20

22
23

MR. PICKETT:

17:52:11

(Cross-talking.)

MR. GLACKIN:

Well, we made --

(Cross-talking.)

17:52:21
17:52:21

24

MR. PICKETT:

Let's not waste the time now.

17:52:22

25

MR. GLACKIN:

We made him available for

17:52:22
Page 414

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL
1

more than seven hours as a courtesy, okay, to try to

17:52:24

2

accommodate you.

17:52:26

3

this over two days -- (Cross-talking.)

4

We gave you the option of doing

MR. PICKETT:

Let's not argue on the

17:52:30

5

record.

6

less likely I finish.

17:52:30

7

BY MR. PICKETT:

17:52:30

8
9

Q.

(Cross-talking.) The longer you argue, the

17:52:29

Paragraph 77 of your report is my next

series of questions.

There you state that, "A new

17:52:30

17:52:30
17:52:38

10

employer" -- it's really in paragraph 78, the point

17:52:48

11

I need.

17:53:25

12

neither party to the new employment contract is

17:53:29

13

incented to worry about the destruction," and you

17:53:31

14

talk about a form of creative destruction, "there

17:53:34

15

will be too much destruction, the consequence of

17:53:37

16

which is too little creation."

17:53:38

17

statement?

17:53:40

18

A.

I do see that.

17:53:41

19

Q.

Do you agree that a new employer would be

17:53:44

You state about five lines down that, "If

Do you see that

20

concerned about the destruction of a partner's

17:53:47

21

asset, if they are a partner in a joint

17:53:49

22

collaboration?

17:53:53

23

A.

It could be, yes.

17:53:54

24

Q.

And if they were collaborating on a

17:53:55

25

project, the destruction of the partner's asset

17:53:57
Page 415

Veritext National Deposition & Litigation Services
866 299-5127

HIGHLY CONFIDENTIAL

1

STATE OF CALIFORNIA

) ss:

2

COUNTY OF MARIN

)

3
4
5
6

I, ASHLEY SOEVYN, CSR No. 12019, do hereby
certify:
That the foregoing deposition testimony was

7

taken before me at the time and place therein set

8

forth and at which time the witness was administered

9

the oath;

10

That the testimony of the witness and all

11

objections made by counsel at the time of the

12

examination were recorded stenographically by me,

13

and were thereafter transcribed under my direction

14

and supervision, and that the foregoing pages

15

contain a full, true and accurate record of all

16

proceedings and testimony to the best of my skill

17

and ability.

18

I further certify that I am neither counsel for

19

any party to said action, nor am I related to any

20

party to said action, nor am I in any way interested

21

in the outcome thereof.

22

IN THE WITNESS WHEREOF, I have transcribed my
name this 29th day of October, 2012.

23
24
____________________________
25

ASHLEY SOEVYN, CSR No. 12019
Page 476
Veritext National Deposition & Litigation Services
866 299-5127

1

UNITED STATES DISTRICT COURT

2

NORTHERN DISTRICT OF CALIFORNIA, SAN JOSE DIVISION

3
4

IN RE:

HIGH-TECH EMPLOYEE

No. 11-CV-2509-LHK

5

ANTITRUST LITIGATION

6

____________________________________

7
8

CONFIDENTIAL PORTIONS DESIGNATED

9
10

Continued Videotaped Deposition of EDWARD E.

11

LEAMER, PH.D., Volume 3, taken at the offices

12

of O'Melvey & Myers LLP, Two Embarcadero Center,

13

Suite 2800, San Francisco, California commencing

14

at 9:03 a.m., on Monday, November 18, 2013,

15

before Leslie Rockwood, RPR, CSR No. 3462.

16
17
18
19
20
21
22
23
24

JOB No. 1765129

25

PAGES 857 - 1169
Page 857
Veritext National Deposition & Litigation Services
877-955-3855

1
2
3
4
5
6

Adobe employees?
A.

Well, that's a different question than the one I

was trying to answer.
Q.

Since I couldn't get an answer to that one, I

changed the question.
A.

13:43:17

So the question I was trying to answer, and in

7

order to answer it, I need more information, is my damage

8

estimate inappropriately dependent upon the legal

9

decision on the part of the firm X.

And in order to

10

answer that, I need to know what the hypothetical has to

11

say about the before-and-after period.

12

13:43:34

You gave me explicit hypothetical for the during

13

period, but the statistics is a way of contrasting what

14

happened to the during period with the before and after.

15

In other words, if there were agreements in

16

play, legal agreements at play before and after as well

17

as during, then the during effect of that legal agreement

18

is going to be absorbed by the statistical analysis.

19

What you're identifying is what's different in that

20

period of time.

13:43:51

21

Q.

Right.

22

A.

13:44:06

So if the hypothetical that you're imagining is

23

magically there was one other agreement that started on

24

exactly that day and ended that day, and absent other

25

agreements, legal or otherwise, that were impacting

13:44:20
Page 1025

Veritext National Deposition & Litigation Services
877-955-3855

1

information flow, then yes, that's going to be absorbed

2

by the conduct variable.

3

Q.

So just to back off then, your model cannot

4

distinguish the impact of a do not cold-call agreement

5

from a unilateral decision by a company not to cold-call

6

a defendant?

7

MR. GLACKIN:

Object to the form.

8

THE WITNESS:

13:44:38

I don't agree with that for the

9

reasons I indicated, which is that if there are similar

10

agreements put in place before and after, that

11

establishes the comparison that you're going to use for

12

deciding whether the conspiracy period is abnormal or

13

not.

14

Q.

BY MR. RILEY:

Okay.

15

A.

If your hypothetical I agree, but your

13:44:49

Let's go back --

16

hypothetical's a very strange one in which there were no

17

other agreements before, after, or during, except for

18

this one that happened to be exactly coterminous with all

19

13:45:03

the other agreements.

20
21

Q.

13:45:15

And again, I'm trying to understand your theory here.

22
23

Well, let's not make it exactly coterminous.

The Apple-Adobe agreement, for purposes of your
analysis, was in effect from 2005 to 2009, I think.

24

A.

That's correct.

25

Q.

Okay?

So during that period, I want you just to

13:45:29
Page 1026

Veritext National Deposition & Litigation Services
877-955-3855

1

assume that from 2007 to 2008, another company,

2

company X, made a unilateral decision not to cold-call

3

Adobe employees.

4

In your model, the impact of that decision by

5

company X not to cold-call Adobe employees would be

6

reflected in the conduct variable that you associate with

7

the agreement between Apple and Adobe.

8
9

A.

13:45:58

I tried to explain why that's not necessarily

the case.

It is the case if you have absolutely that one

10

agreement.

11

that's not quite coterminous but almost coterminous with

12

the period from 2005 to 2009.

13

If you hypothetically have a single agreement

13:46:17

But if there were other agreements that were

14

present in the before and after period, that's going to

15

be absorbed within the analysis in which you compare the

16

during period with the before and after.

17

13:46:32

So if we're going to go down that route, we're

18

going to have to collect all the legal agreements, not

19

just ones that happen to be in the midst of the

20

conspiracy period from 2005 to 2009.

21

Q.

13:46:46

You keep saying we will have to collect all the

22

legal agreements.

23

to a unilateral decision.

24
25

A.

And what do you mean?

I'm referring

I'm sorry, I didn't mean an agreement.

Unilateral decisions.

13:46:57
Page 1027

Veritext National Deposition & Litigation Services
877-955-3855

1

Q.

And you made no effort to control for the

2

unilateral decisions of firms not to cold-call a

3

defendant company?

4

A.

Well, I think the better way of saying it is the

5

assumption that underlies my regression is that the

6

activity of unilateral actions was present in the before

7

period, present in the during, and present in the after

8

period at about the same frequency, except in the sense

9

of the variables that are controlling for differences in

10
11

market conditions.
Q.

13:47:10

13:47:28

But you didn't critically examine that

12

assumption that there was similar activity before,

13

during, and after the conduct period?

14

MR. GLACKIN:

Object to the form.

15

THE WITNESS:

I was provided no information with

16

regard to these other agreements or other unilateral

17

13:47:43

actions.

18

Q.

BY MR. RILEY:

So you didn't receive any

19

material about other unilateral decisions not to

20

cold-call into defendant companies?

21

A.

13:47:56

I -- I saw what I would consider to be anecdotes

22

and what I need -- what I would need is a data set, not a

23

couple of anecdotes.

24
25

Q.

Are you aware that during this period, there

were alleged agreements, for example, between Intel and

13:48:16
Page 1028

Veritext National Deposition & Litigation Services
877-955-3855

1

Apple which are not being challenged in this case?

2

A.

Yes, I am aware.

3

Q.

And those agreements occurred during the conduct

4

period; correct?

5

A.

That's correct.

13:48:34

6

Q.

But your model makes no allowance for the fact

7

that there may have been an agreement between Apple and

8

Intel that had an impact on the flow of information to

9

Apple employees?

10

MR. GLACKIN:

Object to the form.

11

THE WITNESS:

13:48:50

Except in the sense that I've

12

already indicated, which is if there were comparable

13

agreements struck in place prior to the conspiracy period

14

and after the conspiracy, then that's all absorbed in the

15

statistical analysis.

16

Q.

BY MR. RILEY:

13:49:04
But as between Intel and Apple, I

17

want you to assume that that agreement didn't come into

18

effect until during the conduct period.

19

A.

No, I didn't mean the specific agreement.

I

20

meant the sets of agreement that are not captured in my

21

model.

22

were present before and after.

23

saying that what you want to do is to somehow make the

24

comment that the model inappropriately absorbs these

25

legal actions.

13:49:15

Not agreements, but unilateral decisions that
In other words, I'm just

And I'm saying that might be, but it

13:49:36
Page 1029

Veritext National Deposition & Litigation Services
877-955-3855

1

might not be, because you'd have to look at the before

2

and after periods.

3

Q.

And that's something you didn't do.

You didn't

4

look at the before and after periods with regard to other

5

legal actions that may have restricted the flow of

6

information to these companies.

7

A.

13:49:53

Well, I didn't see any evidence that these other

8

agreements were specific to the periods of -- in which

9

the conspiracy occurred.

10

Q.

In your report, your first report, you actually

11

include a diagram that shows this agreement between Apple

12

13:50:09

and Intel.

13

A.

That's correct.

14

Q.

And that's at page 10, Figure 2.

15

A.

I'm aware that that was part of that display.

16

Q.

Why did you put that in there if, in fact, there

17

is no challenge to the agreement between Apple and Intel?

18
19

13:50:30

A.

Say that again?

Why did I put it in if there's

no challenge to that agreement?

20

Q.

Yes.

21

A.

Well, this is attorneys telling me what they

22

wanted me to study.

This chart didn't come from me.

I

23

think it came from the Department of Justice, but perhaps

24

not.

25

consider.

But anyway, this is what the attorneys told me to
13:50:58
Page 1030

Veritext National Deposition & Litigation Services
877-955-3855

1

Q.

And so when you constructed your model, you

2

assumed a contract between Apple -- a no cold-call

3

agreement between Apple and Intel?

4

A.

No, I would not have done anything differently

5

because I've got -- I still have got the chain.

6

break the chain, then maybe there's an issue.

7

there's a chain of interlinked firms, then this thing is

8

going to leak out to all the firms involved in that

9

conspiracy.

10

Q.

If you

13:51:10

As long as

So, for example, one could eliminate several of

11

the actual bilateral agreements, and in your view, the

12

13:51:26

impact would be the same --

13

A.

No.

14

Q.

-- of the class?

15

A.

No, the impact would be less because the

16

information flow is -- is not being -- it's being legally

17

reduced, not illegally reduced, according to your

18

13:51:36

hypothetical.

19

Q.

But if you, for example, were to eliminate the

20

agreement between Intel and Pixar, that would have no

21

impact on your damages analysis?

22

A.

13:51:58

Well, the damage analysis that I've made doesn't

23

refer at all to this bilateral relationships.

It treats

24

the conspiracy as a single overriding fact, and I'm just

25

making a presumption that that overriding fact requires

13:52:23
Page 1031

Veritext National Deposition & Litigation Services
877-955-3855

1

STATE OF CALIFORNIA

) ss:

2

COUNTY OF MARIN

)

3
4
5
6

I, LESLIE ROCKWOOD, CSR NO. 3452, do hereby
certify:
That the foregoing deposition testimony was

7

taken before me at the time and place therein set forth

8

and at which time the witness was administered the oath;

9

That testimony of the witness and all objections

10

made by counsel at the time of the examination were

11

recorded stenographically by me, and were thereafter

12

transcribed under my direction and supervision, and that

13

the foregoing pages contain a full, true and accurate

14

record of all proceedings and testimony to the best of my

15

skill and ability.

16

I further certify that I am neither counsel for

17

any party to said action, nor am I related to any party

18

to said action, nor am I in any way interested in the

19

outcome thereof.

20
21

IN WITNESS WHEREOF, I have subscribed my name
this 20th day of November, 2013.

22
23
24

__________________________________

25

LESLIE ROCKWOOD, RPR, CSR NO. 3462
Page 1169
Veritext National Deposition & Litigation Services
877-955-3855

1

UNITED STATES DISTRICT COURT

2

NORTHERN DISTRICT OF CALIFORNIA

3

SAN JOSE DIVISION

4
5
6

IN RE:

HIGH-TECH EMPLOYEE

)
) No. 11-CV-2509-LHK

7

ANTITRUST LITIGATION

)

______________________________)
8
9
10
11
12
13
14
15
16

VIDEOTAPED DEPOSITION OF EDWARD E. LEAMER Ph.D.

17

San Francisco, California

18

Thursday, December 19, 2013

19

Volume IV

20
21

Reported by:
CARLA SOARES

22

CSR No. 5908

23

Job No. 1784254

24
25

Pages 1170 - 1489
Page 1170
Veritext National Deposition & Litigation Services
877-955-3855

1

reflection of employment by the firm that she's

2

studying?

3

MR. GLACKIN:

07:54:06

Object to the form.

4

BY MR. RILEY:

5

Q

Is that right?

6

A

That is not entirely right because there's

07:54:11

7

another variable, but it's a share available and an

8

absent variable.

9

Q

Which variable is that?

10

A

So we'll look at Exhibit 3.

Row 27 has a

11

variable that represents the rate of hiring by this

12

particular employer, which is the number of new

13

hires in the firm divided by the number of employees

14

07:54:27

in the previous year.

15
16

Q

And that's variable 27 which is in the

07:54:48

original Leamer model, correct?

17

A

That's correct.

18

Q

Now, you criticized Dr. Stiroh for

19

misunderstanding your new hire variable, which is

20

the total hiring by the seven defendants, as

21

reflecting an industry effect as opposed to an

22

employer effect, right?

23

A

I guess that's correct, yes.

24

Q

Well, let's not guess.

07:55:12

25

paragraph 121.

Let's turn to

You say, "Dr. Stiroh admitted at

07:55:26
Page 1194

Veritext National Deposition & Litigation Services
877-955-3855

1

deposition that she has made a fundamental error by

2

believing that the total new hires variable controls

3

for effects of the industry on compensation when, in

4

fact, it only contains data from the defendants and

5

is identified in my prior work as an employer effect

6

variable."

7

07:55:43

07:55:58

So again, you're saying that in your prior

8

work, variable 28, the total number of new hires,

9

was identified as an employer effect variable; is

10
11

that right?
A

07:56:14

Well, it's between the two obviously.

12

It's not an industry effect because it's specific to

13

the seven defendants, but it's not a defendant

14

effect because it applies to all seven defendants.

15

Q

16

That's not my question.

07:56:29

My question is, you're claiming that in

17

your prior work, you identified the total new hire

18

variable as an employer effect variable.

19

what you say in that sentence, correct?

20
21
22

A

That's correct.

That's

That's what it says in

07:56:46

the sentence, yes.
Q

But, in fact, you identified the total

23

number of new hires in your October report as an

24

industry effect, didn't you?

25

A

I don't recall exactly the answer to that

07:56:59
Page 1195

Veritext National Deposition & Litigation Services
877-955-3855

1

question.

2

paragraph 121.

3
4
5

Q

It says what it says here in the -- in

07:57:01

Are you not comfortable with what it says

in that paragraph?
A

Well, as I said before, there are three

07:57:08

6

different kinds of variables.

7

be careful in using the word "employer" to -- in a

8

way which distinguishes the seven employers from the

9

individual employers.

10

I think we ought to

So the language is a little

ambiguous here.

07:57:29

11

Q

But this is your language.

12

A

Correct.

13

Q

You wrote this.

14

A

Yeah.

15
16

As I reread this, I saw the

ambiguity in that sentence.
Q

07:57:38

In fact, that sentence is false because in

17

your October 2013 reply report, you identified the

18

hiring variable, the total new hiring variable, as

19

an industry effect, didn't you?

20

A

Are we talking about language or --

21

Q

Yes, language.

22

report.

23

A

07:57:56

Language you used in your

I don't recall the specific language.

24

the reality is there are three different kinds of

25

But

variables.

07:58:04
Page 1196

Veritext National Deposition & Litigation Services
877-955-3855

1

disruptive cold calls can be.

2

BY MR. RILEY:

3

Q

08:49:44

Dr. Leamer, you rely on your conduct

4

regression to show that the alleged do-not-cold-call

5

agreements had an impact on the price discovery

6

process and therefore compensation, correct?

7

A

08:50:38

I think that's correct, although you

8

are -- you're describing kind of a chain of logic

9

that I'm not sure that I need to agree to.

10

So I'm using the regression to identify

11

the amount of undercompensation that occurred during

12

the period of time when these agreements were in

13

08:51:05

place.

14

Q

But you're using the regression to show

15

that the do-not-cold-call agreements had an impact

16

on compensation, correct?

17
18
19
20
21

A

08:51:23

I'm using it to measure the impact.

That's correct.
Q

You're using your regression to show that

the agreements in your view, in fact, had an impact?
A

08:51:37

Well, I tried to make clear in my

22

discussion that there's a distinction between

23

hypothesis testing and estimation.

24
25

And I don't know if that's where you're
going with this question, but I think of my job as

08:51:50
Page 1236

Veritext National Deposition & Litigation Services
877-955-3855

1

primarily an estimation job, which is to determine

2

to the best of my ability the amount of

3

undercompensation that has occurred here and to rely

4

primarily on the documentary evidence that suggests

5

that these agreements were in place and intended to

6

have an impact on compensation, and that the

7

hypothesis that there was absolutely no impact on

8

compensation with these secret agreements had -- let

9

me put it back.

08:51:54

08:52:06

The hypothesis that these secret

10

agreements had absolutely no impact on compensation,

11

I don't regard that as a very plausible thing.

12

08:52:31

So my job is not to use the data to make a

13

determination of innocence or guilt but rather to

14

estimate the amount of damages that were created by

15

these illegal agreements.

16

Q

08:52:47

So you're testifying you do not rely on

17

your conduct regression to show that these

18

agreements had an impact on compensation?

19

A

Well, that's an overstatement.

I'm just

20

trying to say I pursued both of these tasks, both

21

the hypothesis testing task that you're referring to

22

now and the estimation task.

23

08:53:00

as primarily an estimation task.

But I think of my task

24

On the other hand, I've done -- studied

25

the hypothesis testing as well and have a lengthy

08:53:15
Page 1237

Veritext National Deposition & Litigation Services
877-955-3855

1

discussion of that in my report, suggesting that

2

even if you had no other evidence about the

3

existence of these agreements, that still the data

4

would be supportive of the conclusion that the

5

agreements had a suppressive effect on compensation.

6

Q

08:53:17

08:53:31

In your prior deposition at page 413, line

7

21, through 414, line 7, you were asked the

8

following questions:

9

"Question:

How do you know that other

10

cold calls to other employers and employees

11

didn't substitute?

12

"Answer:

08:53:49

This one we've been on before,

13

too.

14

evidence on that.

15

So the answer is I -- I don't have

"Question:

So you don't know whether the

16

price discovery process was further slowed

17

down or not.

18

"Answer:

08:54:03

But we will leave that to the

19

regression.

20

framework.

21

over and over.

22

that suggests that there will be an impact on

23

price formation, and we'll go to the data to

24

decide whether it's actually there."

25

This is -- lays all down the
The same thing we've been saying

08:54:18

It is to set the framework

Is that your testimony?

08:54:33
Page 1238

Veritext National Deposition & Litigation Services
877-955-3855

1
2

A

That sounds like what I might have said,

Q

08:54:34

So in that testimony, you're saying you

yes.

3
4

will go to the data in your regression analysis to

5

determine whether there was an actual impact on

6

compensation?

7

A

8

That sentence sounds like hypothesis

testing.

9

Q

08:54:43

I would admit that, yes.
So you, in fact, used your regression, at

10

least originally used your regression to do a

11

hypothesis testing?

12

A

08:54:55

Well, I still do use the regression in

13

support of the conclusion that there are damages

14

here.

15

exercise and the estimation exercise.

16

So I'm doing both the hypothesis testing

Q

08:55:10

But in your reply report, you testified

17

that, in fact, there is no hypothesis testing

18

problem presented by this case, correct?

19
20
21

A

Well, I don't think that hypothesis

testing is a critical issue here.
Q

That's correct.

08:55:28

So on the one hand you're saying you used

22

the regression to do hypothesis testing, and then on

23

the other hand you're saying there really is no

24

hypothesis testing problem here.

25

A

So the -- by the words "hypothesis

08:55:42
Page 1239

Veritext National Deposition & Litigation Services
877-955-3855

1

testing," let's be clear what that's a reference to.

2

It's exploring the hypothetical that these

3

agreements had absolutely no impact on compensation.

4

Exactly zero coefficients in the regression.

5

And there are some settings in which that

6

08:55:56

very simple hypothesis makes sense, but it doesn't

7

08:55:45

make sense in this setting.

8
9

So the issue isn't whether the number is
zero.

The number (sic) is whether it's a small

10

positive, a large positive, or maybe a small

11

negative.

12

Your side would like to think that somehow

13

these agreements had the opposite effect of actually

14

08:56:13

making the employees better off.

15

So the data is primarily -- if you want to

16

talk about a hypothesis that would be relevant, it's

17

that the impact is so small that it can be treated

18

as if it were zero.

19

less than a tenth of a tenth percent perhaps.

20

Q

08:56:25

Say whatever that number was,

Dr. Leamer, in your deposition testimony,

21

your first deposition that I quoted, you implied

22

that you were using the regression -- conduct

23

08:56:43

regression to do hypothesis testing, correct?

24

MR. GLACKIN:

Object to the form.

25

THE WITNESS:

Well, if you're referring to

08:56:55
Page 1240

Veritext National Deposition & Litigation Services
877-955-3855

1

seems so implausible that we don't need to entertain

2

that.

3

Q

08:58:04

So you begin your data analysis with the

4

presumption that the agreements had an impact on

5

compensation?

6

A

08:58:14

That actually is not the case.

I begin

7

with an open mind -- when you say "have an impact,"

8

it could be either positive or negative.

9

restriction on the sign.

10
11

There's no

I'll let the data help me

determine what the number is.
Q

08:58:27

So you actually begin your data analysis,

12

as you say, with an open mind.

13

prejudged the issue about whether these agreements

14

had an impact one way or the other, correct?

15

A

You haven't

Well, make sure you say "one way or the

16

other," because I think positive or negative makes

17

sense to me.

18

the impact is positive and not negative, that's an

19

appropriate hypothesis.

20

it's exactly zero, that's pretty farfetched.

21

08:58:39

inappropriate to the circumstance.

22
23

Q

If you want to test a hypothesis that

But the hypothesis that
It's

08:58:52

In this case, Dr. Leamer, you reject the

use of statistical significance, correct?

24

A

That's not correct.

25

Q

You embrace the use of statistical

08:59:19
Page 1242

Veritext National Deposition & Litigation Services
877-955-3855

1
2
3
4

significance in this case?
A

08:59:21

I describe how it should be done

correctly.
Q

In your reports, you repeatedly use the

5

conventional 5 percent statistical significance

6

level, correct?

7

A

08:59:29

Well --

8

MR. GLACKIN:

Object to the form.

9

THE WITNESS:

What do you mean by "used"?

10

BY MR. RILEY:

08:59:39

11

Q

You report it, correct?

12

A

They are the standard things that come

13

rolling out of computer packages.

14

that you're referring to is a whole bunch of

15

numbers.

16

interpreted with some wisdom.

17

But that table

Every one of those numbers has to be

08:59:50

So the fact that a coefficient is

18

statistically significant, that means something to

19

me, and it's appropriate to have that in the

20

printout.

21

Q

09:00:02
In fact, you relied on statistical

22

significance in your critique of various issues in

23

this case.

24
25

A

A variable with a large T value estimated

with accuracy is different from a variable that

09:00:13
Page 1243

Veritext National Deposition & Litigation Services
877-955-3855

1

doesn't have that feature.

2

09:00:17

In particular, the variable that we were

3

talking about before, the fact that it has a very

4

high T value makes me very reluctant to take it out

5

of the equation.

6

Q

09:00:27

So I think the answer to my question is

7

yes, you did, in various aspects of your work in

8

this case, rely on statistical significance,

9

correct?

10
11

A

But let's make sure that we get exactly

09:00:39

clear what we mean by that.

12

One is it determines whether a coefficient

13

is exactly zero or not.

14

testing.

15

That's a hypothesis

To me, the word "statistically

09:00:53

16

significance" isn't about hypothesis testing.

17

about the measurability of an effect.

18

significant -- statistically significant coefficient

19

doesn't mean that variable is important, which is

20

what you and I think what the word "significant"

21

must mean.

22

It's

A highly

09:01:06

It means its effect is measurable.
I'll use that word always to signify that

23

comment -- that interpretation, not to suggest

24

hypothesis testing.

25

Q

But in that sense, you have reported and

09:01:20
Page 1244

Veritext National Deposition & Litigation Services
877-955-3855

1

relied on statistically significant variables

2

throughout your work in this case, correct?

3

A

Well, certainly in a relative sense.

09:01:22

I

4

indicated that because I've got that old result

5

which says the variable with the biggest T is the

6

most resistant variable in the sense of if you

7

change the model, it's not going to have a big

8

impact on that variable.

9

09:01:33

So in a relative sense, there's no

10

question that the T values mean something.

11

absolute sense, it has to be interpreted very

12

carefully.

13
14

Q

In an

09:01:45

Let's go back to your December 2012 reply

report at paragraph 107.

15

MR. GLACKIN:

This would be Tab B.

16

THE WITNESS:

Which tab?

17

MR. GLACKIN:

Tab B.

18

THE WITNESS:

I don't have a Tab B.

19

MR. GLACKIN:

Sorry.

20

09:02:03

It should be the

fourth document.

09:02:10

21

THE WITNESS:

I see that, yes.

22

MR. GLACKIN:

Sorry, George.

23
24
25

Which

paragraph?
MR. RILEY:

Paragraph 107 in the

December 10th, 2012, reply brief of Dr. Leamer --

09:02:23
Page 1245

Veritext National Deposition & Litigation Services
877-955-3855