Luna et al v. Marvell Technology Group, Ltd. et al

Filing 202

ORDER RE MOTION FOR CLASS CERTIFICATION by Judge William Alsup re 160 Motion to Certify Class.(whalc1, COURT STAFF) (Filed on 10/27/2017)

1
2
3
4
5
6

IN THE UNITED STATES DISTRICT COURT

7
FOR THE NORTHERN DISTRICT OF CALIFORNIA
8
9

11
For the Northern District of California

United States District Court

10

DANIEL LUNA, individually and on behalf
of all others similarly situated,
No. C 15-05447 WHA
Plaintiff,

12
13
14
15
16
17
18
19
20
21

v.
MARVELL TECHNOLOGY GROUP, LTD.,
and SEHAT SUTARDJA,

ORDER RE MOTION FOR
CLASS CERTIFICATION

Defendants.
/
INTRODUCTION
In this PSLRA action, lead plaintiff moves for class certification. Defendants oppose.
To the extent stated below, the motion is GRANTED.
STATEMENT
Defendant Marvell Technology Group, Ltd. was and remains a publicly-traded company

22

holding stakes in subsidiaries that produced and sold various semiconductor products.

23

Defendant Sehat Sutardja served as the chief executive officer of Marvell throughout the

24

class period (February 19, 2015 through December 7, 2015).

25

Marvell’s fiscal years ended on January 31, so fiscal year 2015 ended on January 31,

26

2015, and fiscal year 2016 began on February 1, 2015. In a press release on September 11,

27

2015 (soon after the second quarter of fiscal year 2016 closed), Marvell disclosed that its audit

28

committee had begun an independent investigation of certain accounting and internal control

1

matters, and that its quarterly report for the second quarter of 2016 would be delayed (Cons.

2

Amd. Compl. ¶¶ 1 n.1, 88, 162–63).

3

One of the concerns that led to the investigation was an accounting practice engaged in

4

by Marvell of pulling forward sales that would have otherwise been consummated in a future

5

quarter to the end of the current quarter — so-called “pull-in” transactions. In some instances,

6

in order to induce customers to make these purchases earlier than they otherwise would,

7

Marvell offered concessions such as rebates and the extension of payment terms. By

8

recognizing these sales before they were otherwise scheduled to be completed, Marvell was

9

able to recognize the revenue in the earlier quarter, which lead plaintiff, Plumbers and
Pipefitters National Pension Fund, alleges falsely inflated its revenue numbers and in turn its

11
For the Northern District of California

United States District Court

10

stock prices (id. ¶¶ 36–37, 63–82, 124–125).1

12

Following the September 11 disclosure regarding this allegedly improper sales and

13

accounting practice, Marvell’s share price declined by over eighteen percent. That same day

14

Daniel Luna commenced this action (id. ¶ 164).

15

In October 2015, a second press release announced that Marvell’s auditor of over fifteen

16

years, Pricewaterhouse-Coopers LLP, resigned, though it did not elaborate on the reasons for its

17

resignation. Following this disclosure, Marvell’s share price declined by almost another fifteen

18

percent (id. ¶¶ 165–166).

19

On December 7, 2015, Marvell disclosed in a third press release that its audit committee

20

continued to investigate which of the company’s transactions constituted pull-ins, now

21

revealing that the pull-in transactions impacted not only the second quarter of fiscal year 2016,

22

but also the previous two quarters (quarter four of fiscal year 2015 and quarter one of fiscal year

23

2016) (id. ¶ 167).

24

In February 2016, Plumbers and Pipefitters National Pension Fund was appointed lead

25

plaintiff in this action, and Robbins Geller Rudman & Dowd LLP was appointed lead counsel.

26
27
28

1
Marvell also revealed in a September 11 press release that the audit committee was reviewing
Marvell’s internal controls and whether Marvell had sufficient reserves for litigation losses. Lead plaintiff
initially included allegations regarding these controls and litigation loss issues in their complaint, but those
claims have since been dismissed (see Dkt. No. 98).

2

1

In October 2016, an order dismissed all of lead plaintiffs claims, though it granted lead plaintiff

2

leave to amend its claims related to pull-in transactions. Lead plaintiff subsequently amended

3

its complaint and, following a second motion to dismiss, was permitted to proceed with its

4

claims against defendants Marvell and Sutardja only as they relate to Marvell’s allegedly

5

fraudulent pull-in transactions (Dkt. Nos. 53, 98, 138).

6

Lead plaintiff now seeks to certify the following class:

7

All persons and entities who purchased or otherwise acquired the
common stock of Marvell Technology Group, Ltd. (“Marvell”
or the “Company”) during the period from November 20, 2014
through December 7, 2015, inclusive (the “Class Period”), and
were damaged thereby. Excluded from the Class are Defendants,
present or former executive officers of Marvell and their
immediate family members (as defined in 17 C.F.R. §229.404,
Instructions (1)(a)(iii) and (1)(b)(ii)).

8
9

11
Defendants oppose certification, contesting typicality and predominance, and in the event
For the Northern District of California

United States District Court

10

12
that a class is certified, seek to narrow the class definition.
13
This order follows full briefing and oral argument.
14
ANALYSIS
15
Federal Rule of Civil Procedure 23(a) provides, “One or more members of a class may
16
sue or be sued as representative parties on behalf of all members only if: (1) the class is so
17
numerous that joinder of all members is impracticable; (2) there are questions of law or fact
18
common to the class; (3) the claims or defenses of the representative parties are typical of the
19
claims or defenses of the class; and (4) the representative parties will fairly and adequately
20
protect the interests of the class.” Rule 23(b) sets forth three conditions under which, if the
21
prerequisites of Rule 23(a) are satisfied, a class action may be maintained. Class certification is
22
appropriate if a plaintiff meets all the prerequisites of Rule 23(a) and at least one condition of
23
Rule 23(b). Abdullah v. United States Sec. Assocs., Inc., 731 F.3d 952, 956–57 (9th Cir. 2013).
24
1.

RULE 23(A).

25
A.

Numerosity.

26
Numerosity is satisfied if “the class is so numerous that joinder of all members is
27
impracticable.” Rule 23(a)(1). Given that there were over 511,000,000 shares of Marvell
28
3

1

common stock traded during the class period, with likely thousands of stockholders, the

2

numerosity requirement is satisfied, and defendants do not contend otherwise.

3

B.

Commonality.

FRCP 23(a)(2). The party seeking class certification must show that their claims depend on

6

a common contention “capable of classwide resolution — which means that determination of

7

its truth or falsity will resolve an issue that is central to the validity of each one of the claims

8

in one stroke.” Stockwell v. City & Cty. of San Francisco, 749 F.3d 1107, 1112 (9th Cir. 2014)

9

(quoting Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 350 (2011)). “All questions of fact and

10

law need not be common to satisfy the rule. The existence of shared legal issues with divergent

11
For the Northern District of California

Commonality is satisfied if “there are questions of law or fact common to the class.”

5

United States District Court

4

factual predicates is sufficient, as is a common core of salient facts coupled with disparate legal

12

remedies within the class.” Hanlon v. Chrysler Corp., 150 F.3d 1011, 1019 (9th Cir. 1998).

13

Here, lead plaintiff’s allegations that investors were defrauded by the same misleading

14

statements over the same period of time, and suffered similar losses as a result are sufficient to

15

fulfill Rule 23(a)’s commonality requirement.

16
17

C.

Typicality.

Typicality is satisfied if “the claims or defenses of the representative parties are typical

18

of the claims or defenses of the class.” Rule 23(a)(3). “The test of typicality is whether other

19

members have the same or similar injury, whether the action is based on conduct which is not

20

unique to the named plaintiff[ ], and whether other class members have been injured by the same

21

course of conduct.” Wolin v. Jaguar Land Rover N. Am., LLC, 617 F.3d 1168, 1175 (9th Cir.

22

2010) (quotations omitted). “Under the rule’s permissive standards, representative claims are

23

‘typical’ if they are reasonably co-extensive with those of absent class members; they need not

24

be substantially identical.” Hanlon v. Chrysler Corp., 150 F.3d 1011, 1020 (9th Cir. 1998).

25

Defendants argue that lead plaintiff is atypical because (1) lead plaintiff’s Marvell

26

holdings have actually resulted in a gain over the long term, so, allegedly, they have not been

27

harmed, and (2) Marvell’s investment manager, who controlled their investment strategy,

28
4

1

expressed the opinion that the pull-in transactions were not of great concern and likely not

2

fraudulent.

3
4
5

As explained below, these theories are unavailing and do not defeat typicality.
(1)

Lead Plaintiff’s Allegations of Harm.

The contention that lead plaintiff was not harmed by Marvell’s alleged fraud because

6

their stock prices bounced back, even after significant declines, is flatly incorrect. Defendants’

7

argument fails to account for conditions in the market generally, and simply assumes that a

8

stock’s eventual recovery from an over forty percent decline means that no recovery is available.

9

This, of course, is not the measure of damages under the Exchange Act.
Under the Exchange Act, damages are measured by the negative effect the alleged fraud

11
For the Northern District of California

United States District Court

10

had on the value of a company’s stock. While share price is the starting point for this value, a

12

damage calculation requires courts to adjust for any share price gains that are unrelated to the

13

fraud. See Arrington v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 651 F.2d 615, 621 (9th Cir.

14

1981); Green v. Occidental Petroleum Corp., 541 F.2d 1335, 1344 (9th Cir. 1976) (Judge Sneed

15

concurring) (explaining in detail widely accepted out-of-pocket damages model). Later market

16

gains unrelated to the fraud do not wipe out losses that resulted from the fraud.

17

The court of appeals for the Second Circuit provided a succinct and elucidating

18

description of the rationale for this rule in Rosado v. China North East Petroleum Holdings, Ltd.,

19

692 F.3d 34, 41 (2d Cir. 2012) (internal citations and quotations omitted):

20
21
22
23

[I]t is improper to offset gains that the plaintiff recovers after the
fraud becomes known against losses caused by the revelation of
the fraud if the stock recovers value for completely unrelated
reasons. Such a holding would place the plaintiff in a worse
position than he would have been absent the fraud . . . . If we
credit an unrelated gain against the plaintiff’s recovery for the
inflated purchase price, he has not been brought to the same
position as a plaintiff who was not defrauded.

24
This rationale applies with equal force here. Lead plaintiff argues that Marvell’s share
25
prices declined as a result of its fraudulent revenue recognition practices. The mere fact that at
26
some later time Marvell’s stock returned to its purchase price does not cure the alleged fraud.
27
Moreover, in a PSLRA action, events occurring more than ninety days after the
28
corrective disclosures are not relevant to the damage calculation, which limits consideration
5

1

to “the difference between the purchase . . . price paid . . . by the plaintiff for the subject security

2

and the mean trading price of that security during the 90-day period beginning on the date on

3

which the information correcting the misstatement or omission that is the basis for the action

4

is disseminated to the market.” 15 U.S.C. Section 78u-4(e) (emphasis added); In re Mego

5

Fin. Corp. Sec. Litig., 213 F.3d 454, 461 (9th Cir. 2000), as amended (June 19, 2000).

6

The Exchange Act’s ninety-day damages period further prevents exactly the defense espoused

7

here — that eventual return to purchase price wipes out interim losses.

8

Lead plaintiff is not rendered atypical because there was an eventual recovery in

9

Marvell’s stock price. It, like absent class members, is entitled to damages that it can prove are
the result of misleading statements, even if stock prices rose in later months and years for

11
For the Northern District of California

United States District Court

10

unrelated reasons.

12
13

(2)

Investment Manager Statements.

Defendants further argue that lead plaintiff is atypical because its investment manager,

14

Wedge Capital Management, stated on several occasions that Marvell’s pull-in transactions

15

did not appear to be fraudulent, and/or did not appear to have a significant effect on stock price

16

(Opp. at 8–10; Dkt. Nos. 182-1–182-119).

17

Even if comments made by lead plaintiff’s investment manager could defeat typicality,

18

a proposition for which defendants provide no legal authority, Wedge Capital’s commentary

19

would not do so here. While defendants have cobbled together a handful of selective quotations

20

to make it appear that Wedge Capital was unconcerned with pull-in transactions, a more

21

comprehensive reading shows that Wedge Capital was wary of the practices that form the basis

22

of this action.

23

Defendants point to several internal communications among Wedge Capital employees

24

around the time of the corrective disclosures. For example, in a September 2015 email, a Wedge

25

Capital analyst indicated that “what has been disclosed thus far does not seem to indicate a large,

26

widespread accounting fraud” (Dkt. No. 182-11 at 1 (emphasis added)). In this same email,

27

however, the analyst states that he will “continue to investigate the situation” and observes that

28

there have been “issues with senior management’s operating style and whether that impeded
6

1

effective internal controls,” which might necessitate a change in management or, at the very

2

least, more stringent oversight (ibid.).

3

Other emails defendants’ point to likewise express concern over “management stretching

4

to inflate revenue and/or earnings” and indicate that Wedge Capital is not yet sure of the full

5

scope of the problem (Dkt. No 182-12 at 2–3). Despite some analysts expressing guarded

6

optimism that they did not “think there [was] a significant accounting fraud” (id. at 1), there

7

was a general expression of concern among Wedge Capital personnel following the corrective

8

disclosures.

9

In any event, Wedge Capital’s opinion at the time these pull-ins were coming to
light does not defeat typicality. At the time it made its assessments, it had no access to any

11
For the Northern District of California

United States District Court

10

internal Marvell documents or memorandum, and was simply basing its opinion on the public

12

announcements Marvell had released — which announcements did not disclose the full scope

13

of improper incentives such as extended payment terms of rebates related to the pull-ins (Dkt.

14

No. 187-5 at 224). That lead plaintiff’s investment advisor vacillated between concern over

15

Marvell’s revenue recognition practices, and optimism that there would not ultimately be a

16

finding of fraud is not dispositive of any of the legal issues here. The opinions of Wedge Capital

17

do not render lead plaintiffs’ claims so unique from other class members as to overcome the

18

“permissive standards” of the typicality requirement. See Hanlon, 150 F.3d at 1020.

19
20

D.

Adequacy.

Defendants next contend that lead plaintiff has an interest adverse to other class members

21

because it continues to hold large amounts of Marvell stock and therefore has an interest in

22

minimizing rather than maximizing damages in order to avoid a drop in Marvell’s stock price.

23

Defendants rest this assertion on In re Seagate Tech. II Sec. Litig., 156 F.R.D. 229, 230

24

(N.D. Cal. 1994) (Judge Vaughn Walker), which held that “plaintiffs with continuing equity

25

interests have an incentive to minimize the aggregate liability of the [stock] issuer” and therefore

26

“may attempt to shape the evidence in such a fashion as to limit, or eliminate, the recoveries of

27

other class members.” Ultimately, that decision held that this “equity conflict” alone did not

28

defeat class certification.
7

1

The alleged “equity conflict” that Seagate II observed, but ultimately found did not

2

defeat adequacy, has since been rejected by virtually every court to consider it. Courts have

3

noted that “equity conflict” is “present in almost every large, complex securities case” and would

4

therefore be an inappropriate basis for denial of class certification. Connecticut Ret. Plans & Tr.

5

Funds v. Amgen, Inc., No. CV07-2536PSG, 2009 WL 2633743, at *6 (C.D. Cal. Aug. 12, 2009),

6

aff'd on other grounds, 660 F.3d 1170 (9th Cir. 2011), aff'd, 568 U.S. 455 (2013); see also In re

7

Intelligent Elecs., Inc. Sec. Litig., No. CIV. A. 92-1905, 1996 WL 67622, at *5 (E.D. Pa. Feb.

8

13, 1996); In re Honeywell Int'l Inc. Sec. Litig., 211 F.R.D. 255, 261 (D.N.J. 2002) (collecting

9

cases).
Moreover, defendants here have failed to show that lead plaintiff intends to maximize

11
For the Northern District of California

United States District Court

10

damages for itself (and others that have continued to hold Marvell common stock) while

12

minimizing damages for those putative class members that sold their shares following the

13

disclosures. Indeed, the damages model lead plaintiff has proposed measures lost value as

14

stock prices declined following corrective disclosures occurring between September and

15

December 2015, and would therefore result in a damage calculation that would provide

16

equally whether purchasers sold or held their stock following the final corrective disclosure.

17

Defendant’s “equity conflict” argument does not defeat lead plaintiff’s showing of adequacy.

18

2.

19

Rule 23(b)(3) requires that “questions of law or fact common to class members

20

predominate over any questions affecting only individual members.” This requirement “tests

21

whether proposed classes are sufficiently cohesive to warrant adjudication by representation.”

22

Amchem Prods., 521 U.S. at 623. Class certification under Rule 23(b)(3) is proper when

23

common questions represent a significant portion of the case and can be resolved for all

24

members of the class in a single adjudication. Comcast v. Behrend, 569 U.S. 27 (2013).

25
26

RULE 23(b) PREDOMINANCE.

A.

Damages.

Defendants argue that lead plaintiff has not shown that damages can be calculated on a

27

classwide basis as required under Comcast Corp. v. Behrend, 569 U.S. 27, 34 (2013), instead

28

proffering a “generic approach” that fails to isolate the particular fraud theories remaining in
8

1

this case from other potentials causes of share price declines. This is particularly problematic,

2

defendants argue, because here “major confounding news” — such as the news of inadequate

3

litigation reserves — “was disclosed on the same ‘corrective disclosure’ dates” lead plaintiff

4

identifies as resulting in pull-in-related price declines (Opp. at 12–13).

5

Defendants’ arguments, however, misperceives the holding of Comcast. There, the

6

district court certified only one of four theories of antitrust liability alleged by the plaintiffs,

7

and yet permitted a damages model that encompassed all four theories. The Supreme Court

8

found that the proposed damages model did not satisfy Rule 23(b)(3)’s predominance

9

requirement because it did not measure damages on a classwide basis, instead providing a model
three quarters of which could not apply to class members, with no apparent way to isolate the

11
For the Northern District of California

United States District Court

10

one quarter that could. 569 U.S. at 35–37.

12

Unlike Comcast, in which a damage figure was calculated assuming theories of liability

13

that had since been eliminated from the case, lead plaintiff’s proposed damages model relies

14

on but one theory of liability — that Marvell’s misstatements related to pull-in transactions

15

artificially inflated prices, resulting in price declines when the true nature of those transactions

16

was revealed. Lead plaintiff’s damages expert, Professor Steven Feinstein, has provided an

17

event study on market efficiency, and explained that he will calculate classwide damages based

18

on price declines tied to each allegedly fraudulent disclosure using an similar event study (Dkt.

19

No. 160-2 ¶¶ 172–173).

20

Defendants’ argument that Professor Feinstein has not shown how he will disaggregate

21

price inflation attributable to confounding events is not, as defendants would have it, an attack

22

on his damages model, but is rather an inquiry into loss causation. Loss causation, however,

23

need not be analyzed at the class certification stage. Erica P. John Fund, Inc. v. Halliburton

24

Co., 563 U.S. 804 (2011) (“The question presented in this case is whether securities fraud

25

plaintiffs must also prove loss causation in order to obtain class certification. We hold that they

26

need not.”).

27

Nor do any of defendants’ other authorities undermine predominance here. In

28

Werdebaugh v. Blue Diamond Growers, No. 12-CV-02724-LHK, 2014 WL 7148923, at *11
9

1

(N.D. Cal. Dec. 15, 2014) (Judge Lucy Koh), for example, the court found that the regression

2

model used by the plaintiff’s expert necessarily “render[ed] the damages model insufficient

3

under Comcast” because it combined multiple variables (some of which did not align with the

4

plaintiff’s theory of liability) making it impossible to “isolat[e] the damages attributable to

5

Defendant’s alleged wrongdoing.” Plaintiff in In re POM Wonderful LLC, No. ML 10-02199

6

DDP RZX, 2014 WL 1225184, at *4 (C.D. Cal. Mar. 25, 2014) (Judge Dean Pregerson)

7

likewise relied upon a damages model that did not find support in its theory of liability. Ibid.

8

These decisions are inapplicable to lead plaintiff’s damages model for the same reason that

9

Comcast does not apply. In each, the plaintiff proposed a damages model that would have
necessarily encompassed damages not falling within the theory of liability.

11
For the Northern District of California

United States District Court

10

Such is not the case here. Lead plaintiff has proposed a model capable of classwide

12

application. The use of an event study to isolate damages stemming from a particular cause is

13

not unique to this action. It is a feature of virtually every securities action, which must account

14

for stock fluctuations unrelated to the particular theory of liability asserted in the case. That lead

15

plaintiff has not yet provided a loss-causation model does not defeat predominance.

16

Lead plaintiff has satisfied the requirements of Rule 23(a) and (b), and has therefore

17

shown that class certification is appropriate. Nevertheless, questions remain regarding the

18

proper scope of the class in this action.

19

3.

20

Defendants argue that the class definition proposed by lead plaintiff is unsupportable

CLASS DEFINITION.

21

because it encompasses a time period that predates any allegedly false statements, and continues

22

beyond the period that lead plaintiff could have reasonably relied upon Marvell’s allegedly

23

misleading statements. They propose a class including only investors who acquired Marvell

24

common stock between February 19, 2015 (the date fourth quarter 2015 revenues were made

25

public) and September 11, 2015 (the date of the press release announcing the internal audit

26

committee investigation), and who did not sell their stock before September 11, 2015 (Opp.

27

at 17).

28
10

1

Lead plaintiff argues that November 20, 2014 (the day third quarter 2015 revenues

2

were made public) is the appropriate start date because third quarter financials were rendered

3

fraudulent as a result of undisclosed pull-in transactions that took place during that quarter.

4

It further contends that its proposed December 7, 2015 end date is appropriate because the full

5

scope of the fraud was not disclosed until that date.

6

A.

7

Start Date.

The parties’ dispute over the appropriate start date for the class period boils down to the

2015. Undoubtedly, lead plaintiff’s case has focused on pull-ins beginning in the fourth quarter

10

of 2015, during which quarter it was later revealed that as much as seven to eight percent of sales

11
For the Northern District of California

question of whether allegedly improper pull-in transactions began in the third or fourth quarter of

9

United States District Court

8

revenue was attributable to pull-ins (Cons. Amd. Compl. ¶ 162).

12

Lead plaintiff, however, also alleges that beginning in the third quarter of 2015, improper

13

pull-in transactions rendered revenue figures materially misleading (id. ¶¶ 6, 26 n.5, 130).

14

Though third-quarter pull-ins amounted to less than one percent of Marvell’s revenue

15

(approximately $6.1 million), this revenue allowed Marvell to meet its earnings-per-share target

16

for the quarter by inflating the share price by one cent. These pull-ins, plaintiff argues, should

17

have been revealed to investors, as they gave the misleading impression that Marvell was able to

18

meet its quarterly targets, when in fact sales were not as robust as the figures made them seem

19

(see id. ¶ 130; Dkt. No. 138 at 7; Dkt. No 187-6).

20

In support of its position that third quarter 2015 pull-ins were improper, lead plaintiff

21

points to text exchanges from high-level Marvell personnel in the final days of the third quarter,

22

in which they urged their sales teams to pull in last minute revenue by convincing customers to

23

advance shipments and purchases (Dkt. No. 187-7). In particular, lead plaintiff points to a text

24

message from Marvell’s CFO to a senior vice president in which the CFO stated that a customer

25

could pull in, but that they needed to “bear in mind that this will impact Q4” (Dkt, No. 187-7 at

26

132).

27

Defendants acknowledge that there were pull-in transactions in the third quarter,

28

but nevertheless argue that there is no evidence that any third quarter 2015 pull-ins were
11

1

improper — e.g. induced by side deals. Instead, they contend that all of the pull-in revenue

2

in the third quarter of 2015 represents properly recognized sales, and therefore did not mislead

3

investors. Marvell further observes that unlike the other three quarters at issue in this action,

4

there was never a corrective disclosure identifying third quarter 2015 pull-ins as misleading or

5

improper.

demonstrating that these pull-ins were the result of any impropriety, does not show that

8

Marvell misled its investors. It is widely accepted that pull-in transactions and channel stuffing

9

are not “inherently fraudulent.” In re Smith & Wesson Holding Corp. Sec. Litig., 669 F.3d 68,

10

76 (1st Cir. 2012). In In re Cypress Semiconductor Sec. Litig., 891 F. Supp. 1369, 1381 (N.D.

11
For the Northern District of California

Simply stating that there were pull-ins during the third quarter of 2015, without

7

United States District Court

6

Cal. 1995) (Judge Robert Aguilar), for example, the court granted summary judgment in favor of

12

a defendant corporation, finding that the use of pull-ins, in and of themselves, did not render

13

revenue or earnings figures misleading. It held that “Pull-ins do not [automatically] result in the

14

improper recognition of revenue under generally accepted accounting principles [because] [t]hey

15

are actual sales which are treated no differently than any other sale, i.e., revenue is recognized

16

upon shipment to the customer.” Ibid.

17

Lead plaintiff’s showing that third quarter 2015 pull-ins allowed Marvell to meet its

18

earnings-per-share targets, or even that they may lessen sales to that same customer in the next

19

quarter, does not suffice to show that these transactions were misleading or fraudulent. Though

20

lead plaintiff need not prove its case at this juncture, it must make some showing, grounded

21

in fact, that these pull-ins were the result of an impropriety, or were otherwise misleading.

22

The scant allegations here do not suffice. Accordingly, the class period shall be limited to a

23

start date of February 19, 2015, when the fourth quarter 2015 revenues were made public.

24
25

B.

End Date.

Defendants next argue that the class period should end on September 11, 2015, the

26

date of the first corrective disclosure. It was not reasonable, defendants argue, for investors

27

to continue to rely on earlier misstatements following these disclosures (Opp. at 19–20).

28
12

1

In support of this argument, defendants observe that the September 11 press release revealed

2

every alleged accounting problem that forms the basis of lead plaintiff’s theory of liability.

3

Defendants’ argument is unavailing. First, it is inaccurate that the September 11 press

4

release revealed all of the accounting issues that form the basis of lead plaintiff’s claims (see

5

Opp. at 20–21). The September 11 disclosure revealed only Marvell’s concerns over pull-in

6

transactions in the second quarter of 2016. As would come to light in later disclosures, a

7

significant proportion of revenues in the fourth quarter of 2015, and the first quarter of 2016

8

was likewise the result of pull-in transactions, which disclosures resulted in further declines in

9

stock price (see Dkt. Nos. 188-8–11).
The newly revealed negative information and the market’s reaction to that information

11
For the Northern District of California

United States District Court

10

demonstrate that only some suspicion had been cast aside by the September disclosure.

12

See Juniper Networks Sec. Litig., 264 F.R.D. 584, 593 (N.D. Cal. 2009) (Judge James Ware).

13

Indeed, the fact that only one quarter’s revenue recognition practices had been called into

14

question in the September disclosures makes it clear that the statement had not completely cured

15

prior misstatements. “If some tentative or partial disclosure of fraud is published but the truth is

16

not fully revealed, there would be no reason to assume that the market fully recovered from the

17

impact of misrepresentation or omission. Investors who purchased after such a disclosure may

18

well have done so at a price still inflated by the same fraud (even if less so) and may suffer

19

losses when the full details of the fraud are exposed.” In re LDK Solar Sec. Litig., 255 F.R.D.

20

519, 528–29 (N.D. Cal. 2009).2

21

Given the partial nature of the September disclosure, it does not make sense to limit the

22

class period so narrowly. Accordingly, this order accepts lead plaintiff’s December 7, 2015 end

23

date, the day on which concerns over fourth quarter 2015 and first quarter 2016 revenue

24

recognition practices came to light.

25
26
27
28

2
Defendants’ back-up argument that, even though more negative news came to light in December
2015, the effect on Marvell’s stock price was statistically insignificant (Opp. at 23–24) is likewise unpersuasive.
Marvell’s price vacillated on the day of the announcement and then fell by approximately eight percent the
following day, and continued to decline in the short-term (Dkt. No. 188-9–10).

13

1

C.

Last Sales Date.

2

Finally, defendants seek to limit the class definition to exclude any purchasers who sold

3

all of their allegedly fraud-inflated shares prior to September 11, 2015, when the first corrective

4

disclosure was made. This is a sensible limitation. Lead plaintiff does not dispute that such

5

class members would be unable to prove damages, but suggests that this should be dealt with

6

during the claims process, and need not alter the class definition.

7

This order disagrees. The class definition is hereby modified to exclude any investors

8

who sold all of their shares prior to September 11, 2015. The class is defined as follows

9

(modifications bolded):

11
For the Northern District of California

United States District Court

10

12
13
14

All persons and entities who purchased or otherwise acquired the
common stock of Marvell Technology Group, Ltd. (“Marvell” or
the “Company”) during the period from February 19, 2015
through December 7, 2015, inclusive (the “Class Period”), and
were damaged thereby. Excluded from the Class are investors
who sold all of their shares prior to September 11, 2015, and
Defendants, present or former executive officers of Marvell and
their immediate family members (as defined in 17 C.F.R.
§229.404, Instructions (1)(a)(iii) and (1)(b)(ii)).

15

CONCLUSION

16

For the reasons stated above, plaintiff's motion for class certification is GRANTED.

17

Plumbers and Pipefitters National Pension Fund is appointed lead plaintiff, and the firm of

18

Robbins Geller Rudman & Dowd LLP is appointed lead counsel. The above-quoted class is

19

hereby CERTIFIED. Within TWENTY-ONE CALENDAR DAYS of the date of entry of this order, the

20

parties shall submit jointly an agreed-upon form of notice, a joint proposal for dissemination of

21

the notice, and the timeline for opting out of the action. Plaintiff must bear the costs of the

22

notice, which shall include mailing by first-class mail.

23
24
IT IS SO ORDERED.
25
26

Dated: October 27, 2017.

WILLIAM ALSUP
UNITED STATES DISTRICT JUDGE

27
28
14