Steve Harris v. Amgen, Inc.

• Court: Court of Appeals for the Ninth Circuit
• Decided: May 26, 2015
• Docket: 10-56014
• Status: Published

Opinion

FOR PUBLICATION

UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT

STEVE HARRIS; DENNIS F. RAMOS, No. 10-56014 AKA Dennis Ramos; DONALD HANKS; JORGE TORRES; ALBERT D.C. No. CAPPA, On Behalf of Themselves 2:07-cv-05442- and All Others Similarly Situated, PSG-PLA Plaintiffs-Appellants,

v. ORDER AND AMENDED AMGEN, INC.; AMGEN OPINION MANUFACTURING, LIMITED; FRANK J. BIONDI, JR.; JERRY D. CHOATE; FRANK C. HERRINGER; GILBERT S. OMENN; DAVID BALTIMORE; JUDITH C. PELHAM; KEVIN W. SHARER; FREDERICK W. GLUCK; LEONARD D. SCHAEFFER; CHARLES BELL; JACQUELINE ALLRED; AMGEN PLAN FIDUCIARY COMMITTEE; RAUL CERMENO; JACKIE CROUSE; FIDUCIARY COMMITTEE OF THE AMGEN MANUFACTURING LIMITED PLAN; LORI JOHNSTON; MICHAEL KELLY, Defendants-Appellees,

DENNIS M. FENTON; RICHARD NANULA; THE FIDUCIARY COMMITTEE; AMGEN GLOBAL 2 HARRIS V. AMGEN

BENEFITS COMMITTEE; AMGEN FIDUCIARY COMMITTEE, Defendants.

On Remand From The United States Supreme Court

Filed October 30, 2014 Amended May 26, 2015

Before: Jerome Farris and William A. Fletcher, Circuit Judges, and Edward R. Korman, Senior District Judge.*

Order; Concurrence to Order by Judge W. Fletcher; Dissent to Order by Judge Kozinski; Opinion by Judge W. Fletcher

SUMMARY**

ERISA

The panel filed (1) an order amending and replacing its prior opinion and denying, on behalf of the court, a petition for rehearing en banc, and (2) an amended opinion.

* The Honorable Edward R. Korman, Senior United States District Judge for the Eastern District of New York, sitting by designation. ** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. HARRIS V. AMGEN 3

In the amended opinion, on remand from the United States Supreme Court for reconsideration in light of Fifth Third Bancorp v. Dudenhoeffer, 134 S. Ct. 2459 (2014), the panel reversed the district court’s dismissal of a class action brought by current and former employees of Amgen, Inc., and an Amgen subsidiary under the Employee Retirement Income Security Act, alleging breach of fiduciary duties regarding two employer-sponsored pension plans.

The plans were employee stock ownership plans that qualified as “eligible individual account plans,” or “EIAPs.” All of the plaintiffs’ EIAPs including holdings in the Amgen Common Stock Fund, which held only Amgen common stock.

The Supreme Court held in Fifth Third that there is no presumption of prudence for employee stock ownership plan fiduciaries beyond the statutory exemption from the otherwise applicable duty to diversify. The panel held, therefore, that the plaintiffs were not required to satisfy the criteria of Quan v. Computer Sci. Corp., 623 F.3d 870 (9th Cir. 2010), in order to show that no presumption of prudence applied.

The panel held that the plaintiffs stated a claim that the defendants acted imprudently, and thereby violated their duty of care, by continuing to provide Amgen common stock as an investment alternative when they knew or should have known that the stock was being sold at an artificially inflated price. The panel concluded that there was no contradiction between defendants’ duty under the federal securities laws and ERISA. 4 HARRIS V. AMGEN

The panel held that the plaintiffs sufficiently alleged that the defendants violated their duty of loyalty and care by failing to provide material information to plan participants about investment in the Amgen Common Stock Fund. Agreeing with the Sixth Circuit, the panel held that the defendants’ preparation and distribution of summary plan descriptions, including their incorporation of Amgen’s SEC filings by reference, were acts performed in their fiduciary duty.

The panel also reversed the dismissal of derivative claims, as well as a claim that the defendants caused the plans directly or indirectly to sell or exchange property with a party-in interest. Because the Amgen Plan contained no clear delegation of executive authority, the panel reversed the district court’s dismissal of Amgen from the case as a non- fiduciary. The panel remanded the case for further proceedings consistent with its opinion.

Concurring in the denial of rehearing en banc, Judge W. Fletcher wrote that, contrary to the dissent from the denial of rehearing en banc, the panel’s opinion did not hold that as a general matter, when previously concealed material information about a company is eventually revealed, the stock price will inevitably decline by more than the amount it would have declined as a result of merely withdrawing the fund as an investment option. The opinion also did not impose on fiduciaries an obligation to act when they only suspect that there has been a violation of the federal securities laws. Finally, the opinion did not impose on ERISA fiduciaries greater disclosure obligations than those imposed under the federal securities laws. HARRIS V. AMGEN 5

Dissenting from the denial of rehearing en banc, Judge Kozinski, joined by Judges O’Scannlain, Callahan, and Bea, wrote that the opinion failed to give effect to the creation in Fifth Third of stringent new requirements for plaintiffs who sue fiduciaries under ERISA for imprudent investment in an employer’s stock. Judge Kozinski wrote that the opinion created almost unbounded liability for ERISA fiduciaries and subjected corporations to novel, judicially-fashioned disclosure requirements that conflict with those of the securities laws.

COUNSEL

Stephen J. Fearon, Jr. and Garry T. Stevens, Jr., Squitieri & Fearon, LLP, New York, New York; Stephen M. Fishback and Daniel L. Keller, Keller, Fishback & Jackson, LLP, Tarzana, California; Francis M. Gregorek, Betsy C. Manifold, and Rachele R. Rickert, Wolf Haldenstein Adler Freeman & Herz, LLP, San Diego, California, Mark C. Rifkin (argued), Wolf Haldenstein Adler Freeman & Herz, LLP, New York, New York; and Thomas James McKenna, Gainey & McKenna, New York, New York, for Appellants.

Emily Seymour Costin, Sheppard Mullin Richter & Hampton, LLP, Washington, D.C.; Steven Oliver Kramer and Jonathan David Moss, Sheppard Mullin Richter & Hampton, LLP, Los Angeles, California; Jonathan Rose, Alston & Bird, LLP, Washington, D.C.; John Nadolenco, Mayer Brown, LLP, Los Angeles, California; Brian David Netter, Mayer Brown, LLP, Washington, D.C.; and Robert P. Davis (argued), Mayer Brown, LLP, New York, New York, for Appellees. 6 HARRIS V. AMGEN

ORDER

The opinion filed on October 30, 2014, and published at 770 F.3d 865, is hereby amended and replaced by the amended opinion filed concurrently with this order. With these amendments, Judge W. Fletcher has voted to deny the petition for rehearing en banc and Judges Farris and Korman so recommend.

The full court was advised of the petition for rehearing en banc. A judge requested a vote on whether to rehear the matter en banc. The matter failed to receive a majority of the votes of the nonrecused active judges in favor of en banc reconsideration. Fed. R. App. P. 35.

The petition for rehearing en banc is DENIED. No further petitions for rehearing or rehearing en banc will be entertained.

Judge W. Fletcher’s concurrence in the denial of rehearing en banc and Judge Kozinski’s dissent from the denial of rehearing en banc are filed concurrently with this order. HARRIS V. AMGEN 7

W. FLETCHER, Circuit Judge, concurring in the denial of rehearing en banc:1

The panel’s opinion speaks for itself, and I will not repeat our analysis, much of which is directly responsive to concerns expressed by the Supreme Court in Fifth Third Bancorp v. Dudenhoeffer, 134 S. Ct. 2459 (2014).

I write only to correct three ways in which the dissent misrepresents what is in our opinion.

1. Impact of Withdrawal

The dissent characterizes our opinion as holding that withdrawing a fund as an investment option is appropriate because, “as a general matter, ‘when the previously concealed material information about [a] company is eventually revealed . . . the stock price will inevitably decline, almost certainly by more than the amount it would have declined as a result of merely withdrawing the [f]und as an investment option.’” Dissent at 20 (emphasis in original) (quoting Opinion at 46). Based on that characterization, the dissent claims that we ignore the Court’s instruction in Fifth Third to consider whether there will be a net harm to plan participants resulting from withdrawal of a fund. The dissent contends that our reasoning is circular because, under the reasoning it ascribes to us, “withdrawing the fund will always be the better option, because any stock price decline it may

1 Senior Circuit Judge Farris and Senior District Judge Korman were not eligible to vote on whether the appeal in this case should have been reheard en banc, and therefore cannot concur in the denial of rehearing en banc. However, Judge Farris and Judge Korman both agree with what is written here. 8 HARRIS V. AMGEN

precipitate will be deemed ‘inevitable.’” Dissent at 20. (emphasis in original).

Our opinion contains no such general, all-purpose holding. We addressed only the situation where “the previously concealed material information about the company is eventually revealed as required by the securities laws.” Opinion at 46 (emphasis added). As we wrote in the opinion:

In a separate class action simultaneously pending before the same district judge, investors in Amgen common stock claimed violations of federal securities laws based on the same alleged facts as in the ERISA action now before us. In a careful thirty-five page order, the district court concluded that the investors had sufficiently alleged material misrepresentations and omissions, scienter, reliance, and resulting economic loss to state claims under Sections 10(b) and 20(a) of the 1934 Exchange Act. See 15 U.S.C. §§ 78j(b), 78t(a). The district court certified a class based on the facts alleged in the complaint. We affirmed the district court’s class certification in Conn. Ret. Plans & Trust Funds v. Amgen, Inc., 660 F.3d 1170 (9th Cir. 2011). The Supreme Court affirmed in Amgen, Inc. v. Conn. Ret. Plans & Trust Funds, 133 S. Ct. 1184 (2013).

Opinion at 37. We therefore assumed, under Federal Rule of Civil Procedure 8(a) and Ashcroft v. Iqbal, 556 U.S. 662 (2009), that there was material information that had been HARRIS V. AMGEN 9

withheld in violation of the securities laws. Our analysis is based on that assumption.

Withdrawal of the fund as an investment option might indeed “do more harm than good to the fund,” Fifth Third, 134 S. Ct. at 2473, where the securities laws do not independently require disclosure. But where the securities laws do require disclosure of previously withheld material information, as in this case, the impact of the eventual disclosure of that information must be taken into account in assessing the net harm that will result from the withdrawal of the fund. In such a case, as we wrote in our opinion, it is plausible to conclude that the withdrawal of the fund will result in a net benefit, rather than a net harm, to plan participants.

2. Knowledge of Fiduciaries

The dissent contends that we impose on fiduciaries an obligation to act when they “only . . . suspect” there has been a violation of the federal securities laws, and that under our opinion a fiduciary would have an obligation to act whenever there is “any arguable violation” of those laws. Dissent at 21 (emphasis in original). That is not what we wrote. Our opinion nowhere requires a fiduciary to act based on mere suspicion or arguable violation of the federal securities laws. Under well-established circuit precedent, “[a] violation [of ERISA’s prudent person standard] may occur where a company’s stock . . . was artificially inflated during that time by an illegal scheme about which the fiduciaries knew or should have known, and then suddenly declined when the scheme was exposed.” In re Syncor ERISA Litig., 516 F.3d 1095, 1102 (9th Cir. 2008) (emphasis added); see also 29 U.S.C. § 1105(a)(3) (imposing liability on a plan fiduciary 10 HARRIS V. AMGEN

for another fiduciary’s breach of fiduciary responsibility “if he has knowledge of a breach by such other fiduciary, unless he makes reasonable efforts under the circumstances to remedy the breach”). We wrote repeatedly and consistently that a fiduciary’s obligation to act is triggered only when he or she “knew or should have known” of a violation of the securities laws.

For example, we wrote that the fiduciaries in this case were obliged to act only when they “knew or should have known that material information was being withheld from the public.” Opinion at 46 (emphasis added). We concluded that the plaintiffs in this case had shown that it was “plausible,” under Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009), that at least some fiduciaries “knew or should have known that the Amgen Common Stock Fund was purchasing stock at an artificially inflated price due to material misrepresentations and omissions by company officers.” Opinion at 44 (emphasis added). And we held that, on remand, the defendants were entitled to argue “that their liability, or the extent of their liability, should depend upon the extent to which they knew, or should have known, that material information was being withheld from the public in violation of the federal securities laws.” Opinion at 49 (emphasis added). See also id. at 39, 41, 54, 55, 56.

3. Disclosure Obligations Under ERISA

Finally, the dissent contends that our opinion imposes on ERISA fiduciaries greater disclosure obligations than those imposed under the federal securities laws. It writes:

The panel also disregards the Court’s second key instruction, that we carefully HARRIS V. AMGEN 11

consider how ERISA-based obligations may conflict with disclosure requirements under the securities laws. The panel reasons that such a conflict simply can’t occur because “if defendants had revealed material information in a timely fashion to the general public . . . they would have simultaneously satisfied their duties under both the securities laws and ERISA.” But the panel fails to appreciate the Court’s concerns in Fifth Third. The Court was not only concerned that fiduciaries would be forced to violate the securities laws to comply with ERISA, it was also worried that “ERISA-based obligations” would be broader than the disclosure requirements under the securities law and would therefore interfere with the compromise Congress struck when enacting those laws.

The securities laws do not require continuous disclosure of all information that may bear on a stock price. Congress . . . enacted a comprehensive and tessellated statutory scheme for corporate disclosure that imposes obligations on certain corporate officers to reveal information at specific times. See, e.g., 15 U.S.C. §§ 78m, 78o(d). There is no allegation that 17 of the 19 defendants here violated the securities laws, or that they even had disclosure obligations under those laws. Yet under the panel’s holding, they are liable under ERISA for failing to do precisely what the securities law do not require of them: immediately disclose 12 HARRIS V. AMGEN

inside information at the moment they “should have known” it was material.

Dissent at 22–23 (emphases in original).

The dissent is mistaken. We nowhere wrote that ERISA fiduciaries, including defendants in this case, have broader disclosure obligations than those imposed under the federal securities law. In response to Fifth Third (and to arguments made by defendants before Fifth Third was decided), we carefully considered whether “ERISA-based obligations may conflict with disclosure obligations under the securities laws.” We also carefully restricted our description of defendants’ disclosure duties under ERISA to those disclosure obligations that complied with, but did not exceed, obligations under the securities laws. We agree with the dissent that “the securities laws do not require continuous disclosure of all information that may bear on a stock price,” and we nowhere wrote that ERISA requires any such “continuous disclosure.”

We wrote:

Compliance with ERISA would not have required defendants to violate [federal securities] laws; indeed, we interpret ERISA to require first and foremost that defendants not violate those laws. That is, if defendants had revealed material information in a timely fashion to the general public (including plan participants), thereby allowing informed plan participants to decide whether to invest in the Amgen Common Stock Fund, they would have simultaneously satisfied their duties HARRIS V. AMGEN 13

under both the securities laws and ERISA. . . . Alternatively, if defendants had made no disclosures but had simply not allowed additional investments in the Fund with the price of Amgen stock was artificially inflated, they would not thereby have violated the prohibition against insider trading, for there is no violation absent purchase or sale of stock.

Opinion at 48–49 (emphasis in original).

In response to defendants’ argument that they “owe no duty under ERISA to provide material information about Amgen stock to plan participants who must decide whether to invest in such stock,” we wrote that defendants’ “fiduciary duties of loyalty and care to plan participants under ERISA, with respect to company stock, are [not] less than the duty they owe to the general public under the securities laws.” Id. at 51 (emphasis added). But we never wrote, or even suggested, that defendants owe a greater disclosure duty than that imposed under the securities laws. We summarized, “[T]here is no contradiction between defendants’ duty under the federal securities laws and ERISA. Indeed, properly understood, these laws are complementary and reinforcing.” Id. 14 HARRIS V. AMGEN

Judge KOZINSKI, with whom Judges O’SCANNLAIN, CALLAHAN and BEA join, dissenting from the denial of rehearing en banc:

The Supreme Court has previously admonished us for ignoring a grant, vacate and remand (GVR) order and “reinstating [our] judgment without seriously confronting the significance of the cases called to [our] attention.” Cavazos v. Smith, 132 S. Ct. 2, 7 (2011). We’re at it again. In Fifth Third Bancorp v. Dudenhoeffer, 134 S. Ct. 2459 (2014), the Supreme Court created stringent new requirements for plaintiffs who sue fiduciaries under ERISA for imprudent investment in an employer’s stock. Here, in response to a GVR, the panel not only fails to give effect to those requirements, but also insulates our circuit law from important aspects of the Supreme Court’s holding.

The panel’s decision creates almost unbounded liability for ERISA fiduciaries, plainly at odds with what the Court instructed. Worse still, the panel’s rule will have grave consequences for corporations across America, leaving them acutely vulnerable to meritless lawsuits and subjecting them to novel, judicially-fashioned disclosure requirements that conflict with those of the securities laws. I sincerely regret that a majority of our court did not see fit to take this case en banc. I expect the Supreme Court will promptly correct our error.

1. Congress has long viewed employee ownership of employer stock as “a goal in and of itself.” Moench v. Robertson, 62 F.3d 553, 568 (3d Cir. 1995). To further this goal, Congress has given companies numerous incentives to create retirement plans that permit investment in their own stock. Under such plans, employees choose the proportion of HARRIS V. AMGEN 15

their retirement savings to be placed in a “fund” consisting entirely of company stock, and the proportion to be placed into other funds that contain a more diversified portfolio. Corporate officers typically administer these plans and serve as fiduciaries with certain obligations under ERISA. However, plan fiduciaries typically don’t have discretion to decide how an employee’s savings are to be apportioned between the funds in a plan. So, for example, when an employee says he wants 25% of his monthly retirement savings placed in the employer-stock fund, 25% of those savings are invested in employer stock. The fiduciary is effectively an intermediary: He must take the savings the employee apportions to the employer fund and buy the company’s stock with it.

So far, so good. The trouble occurs when a fiduciary has reason to believe that employer stock might be overvalued. Though a fiduciary can’t elect to diversify employee savings of his own accord, he can remove company stock as an investment option by withdrawing the fund, thereby preventing employees from continuing to invest in what he suspects might be overpriced shares. But removing company stock as an investment option is a radical step. It may violate the terms of a plan’s written instruments, it can send a signal to the market that something is seriously wrong with the company and it certainly undermines employees’ investment autonomy. Therefore, whenever a fiduciary fears an employer’s stock is overvalued, he is, in the Supreme Court’s words, “between a rock and a hard place: If he keeps investing and the stock goes down he may be sued for acting imprudently . . . but if he stops investing and the stock goes up he may be sued for disobeying the plan documents” or otherwise harming the fund. Fifth Third, 134 S. Ct. at 2470. 16 HARRIS V. AMGEN

Recognizing the uniquely vulnerable position of ERISA fiduciaries, many courts, including ours, had previously held that a fiduciary’s investment in employer stock should be given a “presumption of prudence.” See, e.g., Quan v. Computer Scis. Corp., 623 F.3d 870, 881 (9th Cir. 2010). Under this presumption, a fiduciary was liable only if he continued to invest in employer stock when the company was facing collapse or catastrophic decline. In Fifth Third, the Supreme Court considered whether fiduciaries are owed such a presumption. The plaintiffs there argued that, far from being presumed prudent, fiduciaries should be liable whenever they possessed inside information suggesting company stock was overvalued, and failed to either publicly disclose that information or remove the stock as an investment option. Id. at 2464.

The Court’s decision in Fifth Third was a compromise. While the Court rejected the presumption of prudence as inconsistent with ERISA’s text, it recognized that, without such a presumption, fiduciaries were at acute risk of liability. The Court therefore stressed the special importance of the motion to dismiss to “weed out meritless lawsuits.” Id. at 2470. To facilitate a rigorous 12(b)(6) inquiry, the Court crafted new and daunting liability requirements that plaintiffs must plausibly allege are met in order to state a claim. Two of them are relevant to this case. First, the Court held that there is no liability if any “prudent fiduciary in the defendant’s position could [] have concluded that stopping purchases . . . or publicly disclosing negative information would do more harm than good to the fund by causing a drop in the stock price and a concomitant drop in the value of the stock already held by the fund.” Id. at 2473. Second, the Court stated that lower courts should carefully “consider the extent to which an ERISA-based obligation either to refrain HARRIS V. AMGEN 17

on the basis of inside information from making a planned trade or to disclose inside information to the public could conflict with the complex insider trading and corporate disclosure requirements imposed by the federal securities laws or with the objectives of those laws.” Id.

2. Plaintiffs’ underlying legal theory in this case is functionally identical to that in Fifth Third. Plaintiffs allege that Amgen, a large pharmaceutical company, concealed the negative results of a clinical trial for an anemia drug and also marketed a risky off-label use for that drug. After the results of the trial came to light and the off-label use of the drug was restricted by the FDA, Amgen’s stock dropped by approximately 30%. Plaintiffs claim that fiduciaries of Amgen’s stock-ownership plans knew or should have known that the stock was overvalued based on inside information, and should have either removed the Amgen stock as an investment option or revealed to the general public the test results and the alleged riskiness of the off-label use.

The panel initially decided this case before Fifth Third and reversed the district court’s dismissal. Harris v. Amgen, Inc., 738 F.3d 1026 (9th Cir. 2013). Amgen supplemented its petition for certiorari after Fifth Third was decided, specifically pointing out the panel’s inconsistency with the two requirements discussed above. The Court vacated the panel’s decision and remanded for reconsideration in light of Fifth Third, obviously expecting the panel would impose the two new liability requirements relevant to this case.

Unsurprisingly, given that it was filed before Fifth Third was decided, the existing complaint fails to adequately plead those two requirements. A complaint may survive a motion to dismiss only “when the plaintiff pleads factual content that 18 HARRIS V. AMGEN

allows the court to draw the reasonable inference that the defendant is liable for the misconduct alleged.” Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009) (citing Bell Atl. Corp. v. Twombly, 550 U.S. 544, 556 (2007)). The Supreme Court held in Fifth Third that a defendant is only “liable for the misconduct alleged” if no reasonable fiduciary in his position could conclude that withdrawing the fund or disclosing inside information would do more harm than good to the fund. When, as here, the Supreme Court changes—or more precisely defines—what constitutes “misconduct,” it inescapably follows that the “factual content” that must be pled also changes. Yet, the panel holds the complaint here survives simply because it recites the conclusion that fiduciaries could have withdrawn the fund or disclosed inside information. Nowhere does the complaint even allege that defendants could have done so without doing more harm than good to the fund, let alone plead sufficient facts to make such an allegation plausible. Nor do plaintiffs allege that defendants could have disclosed inside information without conflicting with the securities laws—Fifth Third’s other novel liability requirement.

Sure, the complaint is long and contains plenty of background information regarding the alleged inflation of Amgen stock. But a complaint’s sufficiency no longer depends merely on its length or level of detail. In the Twiqbal era, plaintiffs must state facts that “plausibly suggest an entitlement to relief.” Iqbal, 556 U.S. at 681. A complaint that fails to state sufficient facts to plausibly suggest how Fifth Third’s new requirements have been met must be dismissed, no matter how extensive its other allegations may be. HARRIS V. AMGEN 19

After all, how can meritless ERISA fiduciary suits be “weeded out” at the motion to dismiss stage, if a complaint can survive through no more than an unadorned conclusion that fiduciaries could have withdrawn the fund or disclosed information? Any complaint filed by minimally competent counsel will surely do that. By “unlock[ing] the doors of discovery for [those] armed with nothing more than conclusions,” Iqbal, 556 U.S. at 678–79, the panel’s holding not only conflicts with Fifth Third’s special emphasis on Rule 12(b)(6), it fundamentally undermines Iqbal and Twombly in our circuit. Future litigants in our court will now be able to inflict massive discovery costs on defendants by reciting liability requirements, without furnishing any of the facts necessary for us to plausibly infer that those requirements have been met.

3. It’s not just the panel’s failure to remand that’s suspect, it’s the reasoning it employs to get there. Quite aside from its ramifications for pleading standards, the panel’s reasoning renders meaningless crucial language in Fifth Third, in open disregard for the intent behind the Supreme Court’s GVR order.

Let’s start with the Court’s requirement that liability will attach only if no “prudent fiduciary” could “conclude[] that stopping purchases . . . or publicly disclosing negative information would do more harm than good to the fund.” The panel first asserts that, “given the relatively small number of Amgen shares that would not be purchased by the Fund in comparison to the enormous number of actively traded shares, it is unlikely that the decrease in the number of shares that would otherwise have been purchased, considered alone, would have an appreciable negative impact on the share price.” How does the panel know that, you ask? I’m not 20 HARRIS V. AMGEN

sure—it’s not an allegation that was pled in the complaint. So, the panel’s view can only be based on some extra-record speculation, the sort of thing we are neither permitted nor equipped to engage in.

What the complaint does allege is that, “If Company Stock were eliminated as an investment option under the Plan, [it] would have sent a negative signal to Wall Street analysts, which in turn would result in reduced demand for Amgen Stock and a drop in the stock price.” First Amended Complaint ¶ 330. As the complaint appears to acknowledge, withdrawal of the fund as an investment option is the worst type of disclosure: It signals that something may be deeply wrong inside a company but doesn’t provide the market with information to gauge the stock’s true value. Of course, there may be exceptional circumstances where such extreme action is compelled by ERISA, and Fifth Third calls for a careful parsing of the particular allegations in a complaint to decide when that is so. But, instead of engaging in that fact-sensitive inquiry, the panel holds that withdrawing the fund was appropriate because, as a general matter, “when the previously concealed material information about [a] company is eventually revealed . . . the stock price will inevitably decline, almost certainly by more than the amount it would have declined as a result of merely withdrawing the [f]und as an investment option.”

Under that theory, withdrawing the fund will always be the better option, because any stock price decline it may precipitate will be deemed “inevitable.” But, for Fifth Third’s requirement to mean anything at all, the Supreme Court must have contemplated situations where a fiduciary could permissibly balance the long and short run effects of withdrawal on the share price, or account for the fact that a HARRIS V. AMGEN 21

badly timed withdrawal could cause the stock value to drop below its efficient-market level. The panel’s holding washes those possibilities away. It blesses a complaint that does nothing more than allege the hypothetical capability of withdrawing the fund, without requiring a single allegation regarding the probable effects of that withdrawal. In our circuit, a fiduciary now can never be safe from a lawsuit if he fails to withdraw the fund based on the reasonable belief that it will “do more harm than good to the fund by causing a drop in the stock price.” Fifth Third, 134 S. Ct. at 2473. The panel’s reasoning renders that crucial language in Fifth Third utterly without meaning.

That holding implicates a far broader range of situations than just those in which an actual securities violation has occurred. Remember, at the time of acting, a fiduciary won’t know whether there was a securities violation; he’ll only have reason to suspect there was one. Under conditions of uncertainty, the only way a fiduciary can avoid the risk of liability is by disclosing any arguable violation. For example, a fiduciary might believe that a company’s financial performance is being overstated by senior officials. Or he might believe that a piece of information needs to be disclosed immediately under the securities laws, when senior officials think only periodic disclosure is required. Such differences of opinion are a common occurrence in most corporations. A fiduciary—often a mid-level administrator with no independent legal counsel and limited information about the company’s overall situation—may well be egregiously wrong in his assessment. Yet, under the panel’s holding, he risks liability every time he fails to act on his impulses, even when any proposed course of action would have disastrous consequences for the share price. And, don’t forget, such share-price drops—when they inevitably result— 22 HARRIS V. AMGEN

will punish all those employees who had previously chosen to invest in the company.

The panel also disregards the Court’s second key instruction, that we carefully consider how ERISA-based obligations may conflict with disclosure requirements under the securities laws. The panel reasons that such a conflict simply can’t occur because “if defendants had revealed material information in a timely fashion to the general public . . . they would have simultaneously satisfied their duties under both the securities laws and ERISA.” But the panel fails to appreciate the Court’s concerns in Fifth Third. The Court was not only concerned that fiduciaries would be forced to violate the securities laws to comply with ERISA, it was also worried that “ERISA-based obligations” would be broader than the disclosure requirements under the securities laws and would therefore interfere with the compromise Congress struck when enacting those laws. Fifth Third, 134 S. Ct. at 2473.

The securities laws do not require continuous disclosure of all information that may bear on a stock price. Congress specifically rejected that route because of the enormous transaction costs and inefficiencies such disclosures would create. Instead, it enacted a comprehensive and tessellated statutory scheme for corporate disclosure that imposes obligations on certain corporate officers to reveal information at specific times. See, e.g., 15 U.S.C. §§ 78m, 78o(d). There is no allegation that 17 of the 19 defendants here violated the securities laws, or that they even had disclosure obligations under those laws. Yet, under the panel’s holding, they are liable under ERISA for failing to do precisely what the securities laws do not require of them: immediately disclose inside information at the moment they “should have known” HARRIS V. AMGEN 23

it was material. The panel has a duty, following Fifth Third, to assess whether compelling such disclosures might conflict with the securities laws. Instead, the panel acts as if the Supreme Court hadn’t spoken.

4. It makes matters worse that the panel’s adventurism occurs in a matter of exceptional importance that drastically impacts thousands of companies and millions of employees who participate in stock-ownership plans. Every company that offers such a plan now faces the chaotic prospect of its plan fiduciaries releasing a disparate array of half-truths and incomplete data to the market; or worse, the incessant withdrawal and reinstatement of its fund as fiduciaries are forced to act upon every tidbit of inside information they fear might make them the target of a lawsuit. What conceivable benefit flows from having a company’s “VP of human resources” publicly explain that he disagrees with a CEO’s financial projection? What virtue is there in triggering a stock price collapse by withdrawing the fund, simply because the “director of benefits” is worried that an erroneous statement was made? I understand the impulse to deter securities fraud. But it’s hardly rational to require every blind man to report on the shape of the whole elephant.

Let’s also not forget that many ERISA fiduciary suits are as bad for employees as they are for companies. Settling meritless lawsuits is a costly endeavor and the money will no doubt come out of workers’ pockets sooner or later, whether that be through diminished salaries, layoffs or reductions in employer benefit contributions.

And a proliferation of ERISA fiduciary suits will surely have the long-term effect of forcing companies to permanently withdraw company stock as an investment 24 HARRIS V. AMGEN

option, even though the presence of such an option has been shown to enhance employee satisfaction, reduce the propensity for layoffs and increase an employer’s likelihood to directly contribute to its employees’ retirement benefits. Even if none of that were so, Congress has made the considered policy judgment to encourage the creation of employee stock-ownership plans and has specifically instructed courts to refrain from “regulations and rulings [that] block the establishment and success of [such] plans.” See Tax Reform Act of 1976, Pub. L. No. 94–455, § 803(h), 90 Stat. 1590 (1976). Leaving aside the litany of practical problems the panel opinion creates, its promiscuous liability standard flies in the face of Congress’s unmistakable will.

* * *

As an intermediate court, our role is to faithfully apply the law as announced by the Supreme Court. The Court in Fifth Third plainly intended to offer fiduciaries robust protection against litigation at the motion to dismiss stage. The Court devoted multiple pages of its opinion to liability requirements that are genuinely novel. The Court then granted a petition for certiorari that specifically directed us to re-examine our prior holding in light of those new liability requirements. Eschewing the simple and expedient solution of a remand, the panel substituted its own judgment for that of the Supreme Court. That decision evinces an impermissible disregard for controlling authority and will have dire consequences for corporations and employees alike. It’s a decision we will come to regret. HARRIS V. AMGEN 25

OPINION

W. FLETCHER, Circuit Judge:

Plaintiffs, current and former employees of Amgen, Inc. (“Amgen”) and its subsidiary Amgen Manufacturing, Limited (“AML”), participated in two employer-sponsored pension plans, the Amgen Retirement and Savings Plan (the “Amgen Plan”) and the Retirement and Savings Plan for Amgen Manufacturing, Limited (the “AML Plan”) (collectively, “the Plans”). The Plans were employee stock-ownership plans that qualified as “eligible individual account plans” (“EIAPs”) under 29 U.S.C. § 1107(d)(3)(A). All of the plaintiffs’ EIAPs included holdings in the Amgen Common Stock Fund, one of the investments available to plan participants. The Amgen Common Stock Fund held only Amgen common stock.

After the value of Amgen common stock fell, plaintiffs filed a class action under the Employee Retirement Income Security Act (“ERISA”) against Amgen, AML, Amgen’s board of directors, and the Fiduciary Committees of the Plans (collectively, “defendants”), alleging that defendants breached their fiduciary duties under ERISA. The district court dismissed the complaint against Amgen under Federal Rule of Civil Procedure 12(b)(6) on the ground that Amgen was not a fiduciary. It dismissed the complaint against the other defendants, who were fiduciaries, after applying the “presumption of prudence” articulated in Quan v. Computer Sciences Corp., 623 F.3d 870 (9th Cir. 2010). Alternatively, even assuming the absence of the presumption, the district court dismissed the complaint on the ground that defendants had not violated their fiduciary duties. 26 HARRIS V. AMGEN

In an earlier opinion, we reversed the district court’s dismissal of the complaint. Harris v. Amgen, Inc., 738 F.3d 1026 (9th Cir. 2013). Applying Quan, we held that the presumption of prudence did not apply. We held, further, that, in the absence of the presumption, plaintiffs had sufficiently alleged violation of the defendants’ fiduciary duties. Finally, we held that Amgen was an adequately alleged fiduciary of the Amgen Plan.

Defendants petitioned for a writ of certiorari. The Supreme Court deferred ruling on the petition while it considered Fifth Third Bancorp v. Dudenhoeffer, 134 S. Ct. 2459 (2014), another ERISA case in which the presumption of prudence was at issue. In Quan, we had held that the presumption of prudence was available to ERISA fiduciaries for both EIAPs and employee stock ownership plans (“ESOPs”) “when the plan terms require or encourage the fiduciary to invest primarily in employer stock.” Quan, 623 F.3d at 881. Overruling Quan and similar decisions by our sister circuits, the Supreme Court held in Fifth Third that there was no presumption of prudence for ESOP fiduciaries beyond the statutory exemption from the otherwise applicable duty to diversify. Fifth Third, 134 S. Ct. at 2467; 29 U.S.C. § 1104(a)(2). After deciding Fifth Third, the Court granted certiorari, and vacated and remanded for reconsideration in light of its decision. Amgen, Inc. v. Harris, 134 S. Ct. 2870 (2014).

On reconsideration in light of Fifth Third, we again reverse the district court’s dismissal. HARRIS V. AMGEN 27

I. Background

The following narrative is taken from the complaint and documents that provide uncontested facts. On a motion to dismiss, we assume the allegations of the complaint to be true. See Tellabs, Inc. v. Makor Issues & Rights, Ltd., 551 U.S. 308, 322 (2007).

Amgen is a global biotechnology company that develops and markets pharmaceutical drugs. AML, a wholly owned subsidiary of Amgen, operates a manufacturing facility in Puerto Rico. To provide retirement benefits to their employees, Amgen set up the Amgen Plan on April 1, 1985. AML set up the AML Plan in 2002 and it became effective on January 1, 2006.

The Plans are covered by the Employee Retirement Income Security Act (“ERISA”). Both qualify as “individual account plans.” See 29 U.S.C. § 1002(34). Plan participants contribute a portion of their pre-tax compensation to individual investment accounts. They receive benefits based solely upon their contributions, adjusted for any gains and losses in assets held by the Plans. Participants may contribute up to thirty percent of their pre-tax compensation. They may select from a number of investment funds offered by the Plans. One of those is the Amgen Common Stock Fund, which holds only Amgen stock. Amgen stock constituted the largest single asset of both Plans in 2004 and 2005.

This litigation arises out of a controversy concerning Amgen drugs used for the treatment of anemia. Anemia is a condition in which blood is deficient in red blood cells or hemoglobin. Causes of anemia include an iron-deficient diet, excessive bleeding, certain cancers and cancer treatments, 28 HARRIS V. AMGEN

and kidney or liver failure. In the early 1980s, Amgen scientists discovered how to make artificial erythropoietin, a protein formed in the kidneys that stimulates erythropoiesis, the formation of red blood cells. After this discovery, Amgen commercialized the manufacture of a class of drugs known as erythropoiesis-stimulating agents (“ESAs”) to treat anemia.

In 1989, the Federal Drug Administration (“FDA”) approved Amgen’s first commercial ESA, epoetin alfa, for the treatment of anemia associated with chronic kidney failure. Amgen marketed epoetin alfa for approved uses under the brand name EPOGEN (“Epogen”), and licensed patents to Johnson & Johnson (“J&J”) to develop additional marketable uses. J&J obtained FDA approval between 1991 and 1996 to market epoetin alfa under the brand name PROCRIT (“Procrit”) for anemia associated with chemotherapy and HIV therapies, for chronic kidney diseases, and for pre-surgery support of anemic patients. J&J had exclusive marketing rights for Procrit under its licensing agreement with Amgen.

Sometime before 2001, Amgen developed a new ESA, darbepoetin alfa, whose sales by Amgen were not restricted by J&J’s exclusive marketing rights for Procrit. Darbepoetin alfa, marketed as Aranesp, lasts longer in the bloodstream than epoetin alfa. The FDA approved Aranesp for treatment of anemia associated with chronic kidney failure and cancer chemotherapy. Aranesp has taken significant market share from J&J’s Procrit. At the time the complaint was filed, Aranesp “control[led] half the market” for non-dialysis ESA. Sales of EPOGEN and Aranesp have been “core to [Amgen’s] survival and success,” making up roughly half of Amgen’s $14.3 billion in revenue in 2006. HARRIS V. AMGEN 29

In the late 1990s and early 2000s, several clinical trials raised safety concerns regarding the use of ESAs for particular anemic populations. In 1998, the Normal Hematocrit Study tested the efficacy of ESAs on anemia patients with pre-existing heart disease. The study was terminated because the test group experienced statistically significant higher rates of blood clotting. In 2003 and early 2004, two trials — ENHANCE and BEST — tested ESAs on cancer patients in Europe. The ENHANCE trial showed shorter progression-free survival and shorter overall survival of head and neck cancer patients for the ESA group than the placebo group. The BEST trial was terminated after four months because breast cancer patients in the group taking epoetin alfa had a higher rate of death than those in the placebo group.

ENHANCE and BEST did not test the safety of ESAs for the specific uses and doses for which they had been approved in the United States. In March 2004, the FDA published notice in the Federal Register that the Oncology Drug Advisory Committee (“ODAC”), an FDA-sponsored group of oncology experts, would convene in May 2004 to discuss safety concerns about Aranesp. In April, before the ODAC meeting, an Amgen spokesperson stated during a conference call with investors, analysts, and plan participants that “the focus [of the ODAC meeting] was not on Aranesp” and that “the safety for Aranesp has been comparable to placebo.”

During its two-day meeting with ODAC, the FDA urged Amgen to conduct further clinical trials to test the safety of ESAs for uses that had already been approved by the FDA. Amgen made a presentation at the meeting outlining what it called the “Amgen Pharmacovigilance Program,” consisting of five ongoing or planned clinical trials testing Aranesp “in 30 HARRIS V. AMGEN

different tumor treatment settings.” Amgen’s Vice President for Oncology Clinical Development described the Amgen program as the “responsible and credible approach to definitively resolv[e] the questions raise[d]” by the FDA.

One of the trials under Amgen’s program was the Danish Head and Neck Cancer Group (“DAHANCA”) 10 Trial. The DAHANCA 10 Trial tested whether high doses of Aranesp could help shrink tumors in patients receiving radiation therapy for head and neck cancer. On October 18, 2006, DAHANCA investigators temporarily halted the study “due to information about potential unexpected negative effects.” Amgen was informed of the temporary halt of the study on or near that day. Amgen did not disclose that the DAHANCA 10 Trial had been temporarily halted.

An analysis of the halted DAHANCA 10 Trial was completed on November 28, 2006. The principal investigator reported that “[b]ased on these outcome results the DAHANCA group concluded that the likelihood of a reverse outcome, i.e. that Aranesp would be significantly better than in control[,] was almost non-existing.” The DAHANCA 10 Trial was permanently terminated on December 1, 2006. DAHANCA investigators concluded that “there is a small but significant poor outcome in the patients treated with Aranesp” in that tumor growth was worse for patients who took Aranesp compared to patients who did not. Amgen was informed in December 2006 that the study had been permanently terminated.

Another clinical trial, CHOIR, raised additional safety concerns about ESAs. The CHOIR trial investigated the safety of epoetin alfa (EPOGEN) when used to treat chronic kidney disease patients. The safety monitoring board for HARRIS V. AMGEN 31

CHOIR terminated the trial when a higher incidence of death and cardiovascular hospitalization was observed among epoetin alfa users. Yet another clinical trial, CREATE, tested the benefit provided by Roche Pharmaceuticals’s ESA in raising hemoglobin levels in patients with chronic kidney disease. On November 16, 2006, Roche announced that the results of the CREATE trial “clearly show that there is no additional cardiovascular benefit from treating to higher hemoglobin levels in this patient group.”

On November 20, Amgen posted a public statement responding to the CHOIR and CREATE trials. Amgen wrote, “A very substantial body of evidence, developed over the past 17 years, demonstrates that anemia associated with chronic kidney disease can be treated safely and effectively with EPOGEN and Aranesp when administered according to the Food and Drug Administration (FDA)-approved dosing guidelines.” Two weeks later, Amgen issued a press release to correct “what the company believes are misleading and inaccurate news reports regarding the use of its drugs.” Amgen reiterated, “EPOGEN and Aranesp are effective and safe medicines when administered according to the Food and Drug Administration (FDA) label.”

Amgen also conducted its own clinical trial, the “103 Study.” The 103 Study tested Aranesp in 939 patients with anemia secondary to cancer. The FDA later described the 103 Study as “demonstrat[ing] significantly shorter survival rate[s] in cancer patients receiving ESAs as compared to th[o]se receiving transfusion support.” However, during a January 2007 conference call, an Amgen representative described the 103 Study as not demonstrating a “statistically significant adverse [e]ffect of Aranesp on overall mortality in this patient population.” He said that “the risk benefit ratio 32 HARRIS V. AMGEN

for Aranesp in these extremely ill patients with anemia secondary to malignancy is, at best, neutral and perhaps negative.” During what may have been the same conference call, discussing Amgen’s fourth-quarter earnings on January 25, an Amgen representative stated, in response to concerns expressed about the 103 Study, that “we have a well established risk benefit profile.”

During a February 16, 2007, investor conference call, defendant Kevin Sharer, Amgen’s President, Chief Executive Officer, and Chairman of the Board, stated, “We strongly believe, as we have consistently stated, that Aranesp and EPOGEN are safe and effective medicines when used in accordance with label indications.” During a March conference call, defendant Sharer reiterated, “When we look at the totality of data, we believe our products are safe and effective when used on-label.” On March 9, 2007, Amgen posted a statement on the company website available to plan participants under the title “Amgen’s Statement on the Safety of Aranesp (darbepoetin alfa) and EPOGEN (Epoetin alfa)”:

Aranesp (darbepoetin alfa) and EPOGEN (Epoetin alfa) have favorable risk/benefit profiles in approximately four million patients with chemotherapy-induced anemia or CKD when administered according to the FDA- approved dosing guidelines.

Amgen engaged in extensive marketing, encouraging both on- and off-label uses of its ESAs. Amgen trained its sales representatives to ask questions that steered doctors to discussions about off-label uses. In an Amgen sales personnel manual, Amgen gave an “expanded list” of “excellent questions” to ask doctors in order to move the HARRIS V. AMGEN 33

discussions toward off-label uses. Examples include, “What is keeping you from using Aranesp in all your MDS/HIV/CIA patients?” MDS is myelodysplastic syndrome, an illness often resulting in anemia. The FDA has never approved Aranesp to treat MDS or HIV patients.

Amgen created a speakers program in which Amgen paid for dinners at which “expert” speakers talked to physicians and other providers about off-label uses for Aranesp. Speakers program events were not accredited as continuing medical education seminars conducted by an independent medical association. Amgen paid not only the speakers but also the doctors and other medical providers who attended the events. The $1,000 payments to physician attendees were “paid from [Amgen’s] marketing budget.”

Amgen educated medical providers about the profit they could obtain by prescribing its ESAs. Before January 1, 2005, Medicare calculated drug reimbursement rates based on the average wholesale price (“AWP”) of drugs. Medical providers could purchase Amgen’s ESAs at a price lower than the AWP, but could charge Medicare the AWP. Amgen created spreadsheets and other tools to help providers calculate the profit. Amgen also encouraged doctors to use its ESAs inefficiently. For example, it encouraged doctors to deliver Epogen intravenously rather than subcutaneously, because an intravenous delivery of the drug requires a substantially larger dose to achieve the same effect.

Amgen marketing efforts were successful. For example, Amgen’s worldwide sales of Aranesp increased fourteen percent during the first quarter of 2007 compared to the same quarter in 2006. Amgen told investors on several occasions that its marketing practices were proper. In public SEC 34 HARRIS V. AMGEN

filings, Amgen stated that it marketed its products only for on-label uses. In December 2006, in response to negative publicity about off-label uses, Amgen issued a press release “intended to clarify Amgen’s position on the use of EPOGEN and Aranesp and to correct what the company believes are misleading and inaccurate news reports regarding the use of its drugs.” The company clarified that “Amgen only promotes the use of EPOGEN and Aranesp consistent with the FDA label.” On a January 2007 conference call, Amgen stated that “our promotion [of EPOGEN] has always been strictly according to our label, we do not anticipate a major shift in clinical practice.”

In February 2007, The Cancer Letter published an article entitled “Amgen Didn’t Tell Wall Street About Results of [DAHANCA] Study.” The article reported that the DAHANCA trial had been temporarily halted due to the “significantly inferior therapeutic outcome from adding Aranesp to radiation treatment of patients with head and neck cancer.” On February 23, the Associated Press announced that the USP DI, an influential drug reference guide, had delisted Aranesp as a treatment for anemia in cancer patients not undergoing chemotherapy. On February 27, the New York Times published an article stating:

New studies are raising questions about whether drugs that have been used by millions of cancer patients might actually be harming them. The drugs, sold by Amgen, Roche, and Johnson & Johnson, are used to treat anemia caused by chemotherapy and meant to reduce the need for blood transfusions and give patients more energy. But the new results suggest that the drugs may make the cancer HARRIS V. AMGEN 35

itself worse. . . . [S]ome cancer specialists and securities analysts say the new information may make doctors more cautious in using the drugs, which have combined sales for the three companies exceeding $11 billion and have been heavily promoted through efforts that include television commercials.

On March 9, the FDA mandated a “black box” warning for off-label use of Aranesp and Epogen. A black box warning is the strongest warning the FDA can require. Cf. 21 C.F.R. § 201.57(c)(1) (2012). The black box warning read:

Recently completed studies describe an increased risk of death, blood clots, strokes, and heart attacks in patients with kidney failure where ESAs were given at higher than recommended doses. In other studies, more rapid tumor growth occurred in patients with head and neck cancer who received these higher doses. In studies where ESAs were given at recommended doses, an increased risk of death was reported in patients with cancer who were not receiving chemotherapy and an increased risk of blood clots was observed in patients following orthopedic surgery.

On March 21, 2007, two House of Representatives subcommittees opened an investigation into the safety profile of Aranesp and EPOGEN as well as into Amgen’s off-label marketing practices. The Chairs of those two subcommittees “ordered” Amgen to halt direct-to-consumer advertising and 36 HARRIS V. AMGEN

physician incentives pending further FDA action. On May 8, the FDA noted on its website that Aranesp and EPOGEN “were clearly demonstrated to be unacceptable” in high doses. On May 10, ODAC reconvened and voted to restrict the use of ESAs, to expand existing warnings, and to require ESA manufacturers to conduct further studies.

Defendant Sharer, Amgen’s President and CEO, told a Wall Street Journal reporter in an interview that 2007 was the “most difficult [year] in [Amgen’s] history.” According to Sharer, there was an “unexpected $800 million to $1 billion hit to operating income due to safety concerns” about Aranesp. Sales of Aranesp decreased by fifty percent.

Amgen stock, and thus the Amgen Common Stock Fund, lost significant value as a result of these safety concerns. The class period runs from May 4, 2005, to March 9, 2007. Amgen common stock was at its high of $86.17 on September 19, 2005. On February 16, 2007, when The Cancer Letter published its article revealing that Amgen had not been forthcoming about the result of the DAHANCA 10 Trial, Amgen stock sold for $66.73. When ODAC voted to restrict the use of ESA drugs, on or shortly after May 10, the price of Amgen stock dropped to $57.33, the class period low. Between September 19, 2005 and the ODAC vote, the price of Amgen stock dropped $28.83, or thirty-three percent.

On August 20, 2007, plaintiffs Steve Harris, a participant in the Amgen Plan, and Dennis Ramos, a participant in the AML Plan, filed a complaint alleging that defendants breached their fiduciary duties under ERISA. The district court dismissed Harris’s claims for lack of standing, on the ground that Harris no longer owned assets in the Amgen Plan on the date he filed his complaint. Harris v. Amgen, Inc., HARRIS V. AMGEN 37

573 F.3d 728, 731 (9th Cir. 2009). The court dismissed Ramos’s claims without leave to amend on the ground that he had failed to identify the proper fiduciaries of the AML Plan. Id. We reversed, holding that Harris had standing as a “participant” of the Amgen Plan during the Class Period, and that Ramos should have been allowed to amend the complaint. Id.

The complaint now at issue is the First Amended Class Action Consolidated Complaint (“FAC”), filed on March 23, 2010, by five plaintiffs, including Harris and Ramos. The FAC alleges six counts of violation of fiduciary duty under ERISA against Amgen, AML, nine Directors of the Amgen Board (“the Directors”), and the Plans’ Fiduciary Committees and their members. The district court dismissed the FAC against Amgen on the ground that it was not a fiduciary. It dismissed the FAC against the remaining defendants under Rule 12(b)(6) for failure to state a claim.

In a separate class action simultaneously pending before the same district judge, investors in Amgen common stock claimed violations of federal securities laws based on the same alleged facts as in the ERISA action now before us. In a careful thirty-five page order, the district court concluded that the investors had sufficiently alleged material misrepresentations and omissions, scienter, reliance, and resulting economic loss to state claims under Sections 10(b) and 20(a) of the 1934 Exchange Act. See 15 U.S.C. §§ 78j(b), 78t(a). The district court certified a class based on the facts alleged in the complaint. We affirmed the district court’s class certification in Conn. Ret. Plans & Trust Funds v. Amgen, Inc., 660 F.3d 1170 (9th Cir. 2011). The Supreme Court affirmed in Amgen, Inc. v. Conn. Ret. Plans & Trust Funds, 133 S. Ct. 1184 (2013). 38 HARRIS V. AMGEN

For the reasons that follow, we reverse the district court’s decision in the ERISA case before us.

II. Standard of Review

“We review de novo the district court’s grant of a motion to dismiss under Rule 12(b)(6), accepting all factual allegations in the complaint as true and construing them in the light most favorable to the nonmoving party.” Skilstaf, Inc. v. CVS Caremark Corp., 669 F.3d 1005, 1014 (9th Cir. 2012). “[C]ourts must consider the complaint in its entirety, as well as other sources courts ordinarily examine when ruling on Rule 12(b)(6) motions to dismiss, in particular, documents incorporated into the complaint by reference, and matters of which a court may take judicial notice.” Tellabs, Inc., 551 U.S. at 322. We then determine whether the allegations in the complaint and information from other permissible sources “plausibly suggest an entitlement to relief.” Ashcroft v. Iqbal, 556 U.S. 662, 681 (2009); Starr v. Baca, 652 F.3d 1202, 1216 (9th Cir. 2011) (quoting Iqbal).

III. Discussion

Congress enacted ERISA to provide “minimum standards . . . assuring the equitable character of [employee benefit] plans and their financial soundness.” 29 U.S.C. § 1001(a). These minimum standards regulate the “conduct, responsibility, and obligation for fiduciaries of employee benefit plans . . . .” Id. § 1001(b). “Congress painted with a broad brush, expecting the federal courts to develop a ‘federal common law of rights and obligations’ interpreting ERISA’s fiduciary standards.” Bins v. Exxon Co. U.S.A., 220 F.3d 1042, 1047 (9th Cir. 2000) (en banc) (citation omitted). HARRIS V. AMGEN 39

The Supreme Court has established certain interpretive rules specific to ERISA’s fiduciary duties. These duties, including those governing fiduciary status, “draw much of their content from the common law of trusts, the law that governed most benefit plans before ERISA’s enactment.” Varity Corp. v. Howe, 516 U.S. 489, 496 (1996). ERISA reflects a “congressional determination that the common law of trusts did not offer completely satisfactory protection.” Id. at 497. The law of trusts “often . . . inform[s]” but does “not necessarily determine the outcome of” an interpretation of ERISA’s fiduciary duties. Id. The common law of trusts offers “only a starting point” that must yield to the “language of the statute, its structure, or its purposes,” if necessary. Id.

We first address the sufficiency of the FAC against each properly named fiduciary. We then address whether the plaintiffs have adequately alleged that Amgen is a fiduciary.

A. Sufficiency of the FAC

The district court dismissed all six counts of the FAC under Rule 12(b)(6). Plaintiffs have appealed only the dismissal of Counts II through VI.

1. Count II

Plaintiffs allege in Count II that defendants acted imprudently, and thereby violated their duty of care under 29 U.S.C. § 1104(a)(1)(B), by continuing to provide Amgen common stock as an investment alternative when they knew or should have known that the stock was being sold at an artificially inflated price. Defendants originally contended that they were entitled to a “presumption of prudence” under Quan v. Computer Sci. Corp., 623 F.3d 870 (9th Cir. 2010). 40 HARRIS V. AMGEN

In our earlier opinion, we held that plaintiffs had satisfied the criteria of Quan, such that the presumption of prudence did not apply. The Supreme Court’s opinion in Fifth Third has now made clear that an ERISA plaintiff does not need to satisfy the criteria we articulated in Quan. The Court wrote in Fifth Third:

[T]he law does not create a special presumption favoring ESOP fiduciaries. Rather, the same standard of prudence applies to all ERISA fiduciaries, except that an ESOP fiduciary is under no duty to diversify the ESOP’s holdings.

134 S. Ct. at 2467. Defendants are EAIP fiduciaries rather than ESOP fiduciaries, but they do not dispute that Fifth Third applies equally to them, and they do not contend that they enjoy a presumption of prudence. However, defendants contend that their actions were prudent even if the presumption of prudence does not apply.

ERISA requires that a fiduciary perform duties under a plan “with the care, skill, prudence, and diligence under the circumstances then prevailing that a prudent man acting in a like capacity and familiar with such matters would use in the conduct of an enterprise of a like character and with like aims.” 29 U.S.C. § 1104(a)(1)(B). This standard governs a fiduciary’s decision to allow investment of plan assets in employer stock. Quan, 623 F.3d at 878–79. “This is true, even though the duty of prudence may be in tension with Congress’s expressed preference for plan investment in the employer’s stock.” Id. at 879 (internal quotation marks omitted). A “myriad of circumstances” surrounding investments in company stock could support a violation of the HARRIS V. AMGEN 41

prudence requirement. In re Syncor, 516 F.3d at 1102. “‘A court’s task in evaluating a fiduciary’s compliance with this standard is to inquire whether the individual trustees, at the time they engaged in the challenged transactions, employed the appropriate methods to investigate the merits of the investment and to structure the investment.’” Quan, 623 F.3d at 879 (quoting Wright, 360 F.3d at 1097) (alterations and quotation marks omitted).

Count II alleges that defendants knew or should have known about material omissions and misrepresentations, as well as illegal off-label sales, that artificially inflated the price of the stock while, at the same time, they continued to offer the Amgen Common Stock Fund as an investment alternative to plan participants. The district court held that, even without the assistance of the presumption of prudence, defendants were entitled to dismissal of Count II under Rule 12(b)(6). We disagree.

We begin by noting that we held in Syncor that “[a] violation [of the prudent man standard] may occur where a company’s stock . . . was artificially inflated during that time by an illegal scheme about which the fiduciaries knew or should have known, and then suddenly declined when the scheme was exposed.” In re Syncor, 516 F.3d at 1102. In Syncor, the company was a fiduciary that knowingly made cash bribes to doctors in Taiwan in violation of the Foreign Corrupt Practices Act. Upon disclosure of these illegal payments, Syncor’s stock price lost nearly half its value. “Despite these illegal practices, the [fiduciaries] allowed the Plan to hold and acquire Syncor stock when they knew or had reason to know of Syncor’s foreign bribery scheme.” Id. at 1098. We held on appeal from summary judgment that “there is a genuine issue whether the fiduciaries breached the 42 HARRIS V. AMGEN

prudent man standard by knowing of, and/or participating in, the illegal scheme while continuing to hold and purchase artificially inflated Syncor stock for the ERISA Plan.” Id. at 1103.

In their original briefing, filed before the Court decided Fifth Third, defendants made five arguments in favor of dismissal of Count II. None is persuasive. First, defendants argue that investments in Amgen stock during the class period were not imprudent “because Amgen was not even remotely experiencing severe financial difficulties during that time, and remains a strong, viable, and profitable company today.” This argument is beside the point. Amgen was not “experiencing severe financial difficulties” during the relevant time period in part because of the very actions about which plaintiffs are now complaining. That is, Amgen was earning large but unsustainable profits based on improper and unsustainable sales of EPOGEN and Aranesp. Further, Amgen may have been, and may now be, a “strong, viable, and profitable company,” but that does not mean that the price of Amgen stock was not artificially inflated during the class period.

Second, defendants argue that the decline in price in Amgen stock was insufficient to show an imprudent investment by the fiduciaries. They write, “[A]s the District Court correctly held, this ‘relatively modest and gradual decline in the stock price’ does not render the investment imprudent.” As an initial matter, we note that the proper question is not whether the investment results were unfavorable, but whether the fiduciary used “‘appropriate methods’” to investigate the merits of the transaction. Quan, 623 F.3d at 879 (quoting Wright, 360 F.3d at 1097); see also Kirschbaum, 526 F.3d at 254 (explaining that the “test of HARRIS V. AMGEN 43

prudence is one of conduct, not results”); Bunch v. W.R. Grace & Co., 555 F.3d 1, 7 (1st Cir. 2009) (same). But defendants’ argument fails even on its own terms. Their argument is foreclosed by the district court’s decision in the federal securities class action against Amgen based on the same alleged sequence of events. See Conn. Ret. Plans & Trust Funds v. Amgen, Inc., 660 F.3d 1170 (9th Cir. 2011), aff’d Amgen Inc. v. Conn. Ret. Plans & Trust Funds, 133 S. Ct. 1184 (2013). If the alleged misrepresentations and omissions, scienter, and resulting decline in share price in Connecticut Retirement Plans were sufficient to state a claim that defendants violated their duties under Section 10(b), the alleged misrepresentations and omissions, scienter, and resulting decline in share price in this case are sufficient to state a claim that defendants violated their duty of care under ERISA.

Third, quoting Kirschbaum, 526 F.3d at 253, 256, defendants argue that

[w]hen, like here, retirement plans are at issue, courts must be mindful of “the long- term horizon of retirement investing, as well as the favored status Congress has granted to employee stock investments in their own companies.” . . . [H]olding fiduciaries liable for continuing to offer the option to invest in declining stock would place them in an “untenable position of having to predict the future of the company stock’s performance. In such a case, [a fiduciary] could be sued for not selling if he adhered to the plan, but also sued for deviating from the plan if the stock rebounded.” 44 HARRIS V. AMGEN

Defendants’ reliance on Kirschbaum is misplaced. The court wrote in that case, “The Plan documents, considered as a whole, compel that the Common Stock Fund be available as an investment option for employee-participants.” Kirschbaum, 526 F.3d at 249. The concerns expressed in Kirschbaum have little bearing on the case before us. Here, unlike in Kirschbaum, the fiduciaries of the Amgen and AML Plans were under no such compulsion. They knew or should have known that the Amgen Common Stock Fund was purchasing stock at an artificially inflated price due to material misrepresentations and omissions by company officers, as well as by illegal off-label marketing, but they nevertheless continued to allow plan participants to invest in the Fund.

Fourth, quoting In re Computer Sciences Corp., ERISA Litig., 635 F. Supp. 2d 1128, 1136 (C.D. Cal. 2009), aff’d 623 F.3d 870 (9th Cir. 2010), defendants argue that if the Amgen Fund had been “remove[d] . . . as an investment option,” based on nonpublic information about the company, this action “may have brought about ‘precisely the result [P]laintiffs seek to avoid: a drop in the stock price.’” The Court wrote in Fifth Third:

To state a claim for breach of the duty of prudence on the basis of inside information, a plaintiff must plausibly allege an alternative action that the defendant could have taken that would have been consistent with the securities laws and that a prudent fiduciary would not have viewed as more likely to harm the fund than to help it.

134 S. Ct. at 2472. More specifically, the Court wrote: HARRIS V. AMGEN 45

[L]ower courts faced with such claims should also consider whether the complaint has plausibly alleged that a prudent fiduciary in the defendant’s position could not have concluded that stopping purchases — which the market might take as a sign that insider fiduciaries viewed the employer’s stock as a bad investment — or publicly disclosing negative information would do more harm than good to the fund by causing a drop in the stock price and a concomitant drop in the value of the stock already held in the fund.

Id. at 2473.

Defendants’ argument does not take into account the fact that, quite independently of any obligation under ERISA, the federal securities laws require disclosure of material information. Consider, first, a situation in which the Fund is not removed as an investment option until after the material information has been concealed from the public for a substantial period of time, and the stock price has been substantially inflated as a result. In this situation, the adverse consequences of the removal of the Fund would be no greater than, and probably substantially less than, the consequences of the disclosure required by the securities laws. This is so for several reasons. First, removing the Fund as an investment option would not mean liquidation of the Fund. It would mean only that while the share price is artificially inflated, plan participants would not be allowed to invest additional money in the Fund, and that the Fund would therefore not purchase additional shares at the inflated price. Second, given the relatively small number of Amgen shares that would not be purchased by the Fund in comparison to the 46 HARRIS V. AMGEN

enormous number of actively traded shares, it is unlikely that the decrease in the number of shares that would otherwise have been purchased, considered alone, would have an appreciable negative impact on the share price. Finally, if the investing public were to take the removal of the Fund as a negative signal about the value of Amgen stock, any reduction in the stock price would anticipate (and only partially) the inevitable result of Amgen’s eventual compliance with the federal securities laws. That is, when the previously concealed material information about the company is eventually revealed as required by the securities laws, the stock price will inevitably decline, almost certainly by more than the amount it would have declined as a result of merely withdrawing the Fund as an investment option. It is thus quite plausible, in this situation, that defendants could remove the Fund from the list of investment options without causing undue harm to plan participants.

Next, consider a situation in which the Fund is removed as an investment option as soon as the fiduciaries — including fiduciaries without disclosure obligations under the federal securities laws — knew or should have known that material information was being withheld from the public. If the fiduciaries with inside knowledge but without disclosure obligations act to remove the Fund as an investment option as soon as Amgen’s share price begins to be artificially inflated — that is, as soon as those fiduciaries with disclosure obligations begin to violate the securities laws — that action may cause those fiduciaries to comply with their obligations under the securities laws. In that event, there will be no artificial increase in the share price, and no corresponding decline at a later time. Even if removal of the Fund as an investment opinion does not cause those defendants with disclosure obligations to comply with the securities laws, its HARRIS V. AMGEN 47

removal will at least protect plan participants from investing in Amgen stock as artificially inflated prices. Removal of the Fund as an investment option might cause a drop in the share price, perhaps slightly more than the amount of any initial artificial inflation. This very drop in stock price might cause the insider fiduciaries with disclosure obligations to comply with the securities laws. But even if the drop in stock price does not cause these fiduciaries to comply, removal of the Fund as an investment option will prevent the greater harm to plan participants that would result if no disclosure is made, if the stock price continues to inflate artificially, and if plan participants are allowed to make continued investments in the Fund at increasingly inflated prices. In other words, it is quite plausible that in this situation, too, defendants could remove the Fund as an investment option without causing undue harm to plan participants.

We emphasize that any problem created by allowing plan participants to invest in the Fund as it purchased Amgen stock at artificially inflated prices is a problem of the defendants’ own making. Both the insider fiduciaries without disclosure obligations under the federal securities laws and those with such obligations have it within their power to prevent harmful investments by plan participants. Insider fiduciaries without disclosure obligations should act to protect plan participants as soon as they know or should know that information of the kind for which disclosure is required under the securities laws is not being released to the public. Insider fiduciaries with disclosure obligations should act to protect plan participants under ERISA as soon as the federal securities laws require disclosure. The fact that the fiduciaries decide not to act at this early stage does not mean that their ERISA fiduciary duties do not apply thereafter. Quite the opposite. It means 48 HARRIS V. AMGEN

that they are continuing to violate their fiduciary duties by not acting.

Fifth, defendants argue that “they could not have removed the Amgen Stock Fund based on undisclosed alleged adverse material information — a potentially illegal course of action” (emphasis in original). Defendants misunderstand the nature of their duties under federal law. As we noted in Quan, “[F]iduciaries are under no obligation to violate securities laws in order to satisfy their ERISA fiduciary duties.” Quan, 623 F.3d at 882 n.8. The central problem in this case is that Amgen officials, many of whom are defendants here, made material misrepresentations and omissions in violation of the federal securities laws. Compliance with ERISA would not have required defendants to violate those laws; indeed, we interpret ERISA to require first and foremost that defendants not violate those laws. That is, if defendants had revealed material information in a timely fashion to the general public (including plan participants), thereby allowing informed plan participants to decide whether to invest in the Amgen Common Stock Fund, they would have simultaneously satisfied their duties under both the securities laws and ERISA. See Cal. Ironworkers Field Pension Trust v. Loomis Sayles & Co., 259 F.3d 1036, 1045 (9th Cir. 2001) (“ERISA imposes upon fiduciaries a general duty to disclose facts material to investment issues.”); Acosta v. Pac. Enter., 950 F.2d 611, 619 (9th Cir. 1991) (holding that a fiduciary is affirmatively required to “inform beneficiaries of circumstances that threaten the funding of benefits”). Alternatively, if defendants had made no disclosures but had simply not allowed additional investments in the Fund while the price of Amgen stock was artificially inflated, they would not thereby have violated the prohibition against insider HARRIS V. AMGEN 49

trading, for there is no violation absent purchase or sale of stock.

We note that the foregoing analysis presumes that at least some defendants were subject both to ERISA’s duty of prudence and to the requirements of the securities laws. On remand from the Supreme Court, defendants assert for the first time that this is not so for all of the defendants. But no defendant made an argument in the district court based on this ground, and nothing in our opinion forecloses a defendant from making such an argument on remand from this court. That is, nothing in our opinion prevents defendants from arguing on remand from this court that their liability, or the extent of their liability, should depend upon the extent to which they knew, or should have known, that material information was being withheld from the public in violation of the federal securities laws, and the extent that they had, or did not have, an obligation under the those laws to reveal such information to the public.

Finally, defendants argue that Fifth Third announced “new pleading requirements” applicable to ERISA cases such as this one. We disagree. The Court wrote as follows:

We consider more fully one important mechanism for weeding out meritless claims, the motion to dismiss for failure to state a claim. That mechanism . . . requires careful judicial consideration of whether the complaint states a claim that the defendant acted imprudently. See Fed. Rule Civ. Proc. 12(b)(6); Ashcroft v. Iqbal, 556 U.S. 662, 677–680 (2009); Bell Atlantic Corp. v. Twombly, 550 U.S. 5434, 554–563 (2007). 50 HARRIS V. AMGEN

Because the content of the duty of prudence turns on “the circumstances . . . prevailing” at the time the fiduciary acts, § 1104(a)(1)(B), the appropriate inquiry will necessarily be context specific.

134 S. Ct. at 2471.

To the extent defendants are arguing that Fifth Third requires a higher pleading standard of particularity or plausibility, this passage from the Court’s opinion makes clear that they are mistaken. Ashcroft and Twombly had already been decided when this case was first before us on appeal, and the Court’s citation of those two cases indicates that it was not articulating a new pleading standard in this sense. To the extent defendants are arguing that the Court has articulated new standards of liability (as opposed to a new standard of pleading) that we had not previously applied, they are also mistaken. It is true that the Court articulated certain standards for ERISA liability in Fifth Third. But we had already assumed those standards when we wrote our earlier opinion. For example, the Court specified in Fifth Third that a fiduciary is not required to perform an act that will do more harm than good to plan participants. We had assumed that to be so, and had addressed precisely this point in our earlier opinion. See Harris v. Amgen, 738 F.3d at 1041.

We therefore conclude that plaintiffs have sufficiently alleged that defendants have violated the duty of care they owe as fiduciaries under ERISA. HARRIS V. AMGEN 51

2. Count III

Plaintiffs allege in Count III that defendants violated their duty of loyalty and care under 29 U.S.C. §§ 1104(a)(1)(A) and (B) by failing to provide material information to plan participants about investment in the Amgen Common Stock Fund. Defendants contend that they have limited obligations under ERISA to disclose information to plan participants, and that their disclosure obligations do not extend to information that is material under the federal securities laws. Defendants contend, further, that plaintiffs have not alleged detrimental reliance by plan participants on defendants’ omissions and misrepresentations. Finally, defendants contend that their omissions and misrepresentations, if any, were not made in their fiduciary capacity. We disagree.

To some extent, the analysis for Count II overlaps with the analysis for Count III. We have already established that there is no contradiction between defendants’ duty under the federal securities laws and ERISA. Indeed, properly understood, these laws are complementary and reinforcing.

Defendants’ first argument is that they owe no duty under ERISA to provide material information about Amgen stock to plan participants who must decide whether to invest in such stock. In other words, defendants contend that their fiduciary duties of loyalty and care to plan participants under ERISA, with respect to company stock, are less than the duty they owe to the general public under the securities laws. Defendants are wrong, as we made clear in Quan:

We have recognized [that] . . . “[a] fiduciary has an obligation to convey complete and accurate information material to the 52 HARRIS V. AMGEN

beneficiary’s circumstance, even when a beneficiary has not specifically asked for the information.” Barker [v. Am. Mobil Power Corp., 64 F.3d 1397, 1403 (9th Cir. 1995)]. “[T]he same duty applies to ‘alleged material misrepresentations made by fiduciaries to participants regarding the risks attendant to fund investment.’” Edgar [v. Avaya Inc., 503 F.3d 340, 350 (3d Cir. 2007)].

Quan, 623 F.3d at 886. We specifically endorsed the Third Circuit’s definition of materiality in Quan. We wrote, “[A] misrepresentation is ‘material’ if there was a substantial likelihood that it would have misled a reasonable participant in making an adequately informed decision about whether to place or maintain monies in a particular fund.” Id. (quoting Edgar, 503 F.3d at 350) (internal quotation marks omitted).

Defendants’ second argument is that plaintiffs have failed to show that they relied on defendants’ material omissions and misrepresentations. Defendants contend that plaintiffs must show that they actually relied on the omissions and misrepresentations. It is well established under Section 10(b) that a defrauded investor need not show actual reliance on the particular omissions or representations of the defendant. Instead, as the Supreme Court explained in Erica P. John Fund, Inc. v. Halliburton Co., 131 S. Ct. 2179 (2011), the investor can rely on a rebuttable presumption of reliance based on the “fraud-on-the-market” theory:

According to that theory, “the market price of shares traded on well-developed markets reflects all publicly available information, and, hence, any material misrepresentations.” HARRIS V. AMGEN 53

[Basic, Inc. v. Levinson, 485 U.S. 224, 246 (1988)]. Because the market “transmits information to the investor in the processed form of a market price,” we can assume, the Court explained [in Basic], that an investor relies on public misstatements whenever he “buys or sells stock at the price set by the market.” Id.[] at 244, 247.

Erica P. John Fund, 131 S. Ct. at 2185; see also Conn. Ret. Plans & Trust, 133 S. Ct. 1184 (2013). We see no reason why ERISA plan participants who invested in a company stock fund whose assets consisted solely of publicly traded common stock should not be able to rely on the fraud-on-the- market theory in the same manner as any other investor in a publicly traded stock.

Defendants’ final argument is that statements made to the Securities and Exchange Commission in documents required by the federal securities laws were not made in a fiduciary capacity, and that these statements therefore cannot be considered in an ERISA suit for breach of fiduciary duty. Although our circuit has not decided the issue, defendants might be correct if these documents had only been filed and distributed as required under the securities laws, for such acts would have been performed in a corporate capacity. See Lanfear v. Home Depot, Inc., 679 F.3d 1267, 1285 (11th Cir. 2012) (“When the defendants in this case filed the Form S-8s and created and distributed the stock prospectuses, they were acting in their corporate capacities and not in their capacity as ERISA fiduciaries.”); Kirschbaum, 526 F.3d at 257 (“REI was discharging its corporate duties under the securities laws, and was not acting as an ERISA fiduciary.”). However, defendants did more than merely file and distribute the 54 HARRIS V. AMGEN

documents as required by the securities laws. See Varity Corp., 516 U.S. at 504 (fiduciary may be “communicating with [plan participants] both in its capacity as employer and in its capacity as plan administrator”) (emphasis in original).

As they were required to do under ERISA, defendants prepared and distributed summary plan descriptions (“SPDs”) to Plan participants. See 29 U.S.C. § 1022(a) (requiring fiduciaries to provide a summary plan description). In the SPDs for both the Amgen and the AML Plans, defendants explicitly incorporated by reference Amgen’s SEC filings, including “The Company’s Annual Report on Form 10-K for the year ending December 31, 2006,” and “The Company’s Current Reports on Form 8-K filed on January 19, 2007, February 20, 2007, March 2, 2007, and March 12, 2007, respectively.” Plaintiffs allege that the defendants knew or should have known that statements contained in these filings, incorporated by reference into the SPDs, were materially false and misleading.

We hold that defendants’ preparation and distribution of the SPDs, including their incorporation of Amgen’s SEC filings by reference, were acts performed in their fiduciary capacities. In so holding, we agree with the Sixth Circuit, which has held that such incorporation by reference is an act performed in a fiduciary capacity:

Defendants exercised discretion in choosing to incorporate the [SEC] filings into the Plan’s SPD as a direct source of information for Plan participants about the financial health of [the company] and the value of its stock, an investment option under the plan. The SPD is a fiduciary communication to plan HARRIS V. AMGEN 55

participants and selecting the information to convey through the SPD is a fiduciary activity. Moreover, whether the fiduciary states information in the SPD itself or incorporates by reference another document containing that information is of no moment. To hold otherwise would authorize fiduciaries to convey misleading or patently untrue information through documents incorporated by reference, all while safely insulated from ERISA’s governing reach. Such a result is inconsistent with the intent and stated purposes of ERISA . . . and would create a loophole in ERISA large enough to devour all its protections.

Dudenhoefer v. Fifth Third Bancorp, 692 F.3 410, 423 (6th Cir. 2012) (internal citation omitted); see also In re Citigroup ERISA Litigation, 662 F.3d 128, 144–45 (2d Cir. 2011) (noting that SEC filings had been incorporated in the Plans’ SPDs, but dismissing ERISA claim on the ground that plaintiffs had not sufficiently alleged that the defendant fiduciaries knew or should have known that the filings contained false information); Quan, 623 F.3d at 886 (assuming, “without deciding, that alleged misrepresentations in SEC disclosures that were incorporated into communications about an ERISA plan are ‘fiduciary communications’ on which an ERISA misrepresentation claim can be based.”) (citations omitted). The statements made in Amgen’s SEC filings and incorporated in the Plans’ SPDs may therefore be used under ERISA to show that defendants knew or should have known that the price of Amgen shares was artificially inflated, and to show that 56 HARRIS V. AMGEN

plaintiffs presumptively detrimentally relied on defendants’ statements under the fraud-on-the-market theory.

We therefore conclude that plaintiffs have sufficiently alleged that defendants have violated the duty of loyalty and care they owe as fiduciaries under ERISA. We emphasize, however, as to Counts II and III, that we have decided only that the complaint contains allegations with a sufficient degree of plausibility to survive a motion to dismiss under Rule 12(b)(6). A determination whether defendants have actually violated their fiduciary duties requires fact-based determinations, such as the likely effect of the alternative actions available to defendants, to be made by the district court on remand, with the assistance of expert opinion as appropriate.

3. Counts IV and V

The district court correctly concluded that Counts IV and V are derivative of Counts II and III. Because we reverse the district court’s dismissal of Counts II and III, we also reverse its dismissal of Counts IV and V. See In re Gilead Sciences Sec. Litig., 536 F.3d 1049, 1055 (9th Cir. 2008).

4. Count VI

Count VI alleges that defendants caused the Plans directly or indirectly to sell or exchange property with a party-in- interest, in violation of 29 U.S.C. § 1106(a). Specifically, Count VI alleges that Amgen and AML are parties-in-interest that concealed material information in order to inflate the price of Amgen stock sold to the Plans. In relevant part, 29 U.S.C. § 1106(a)(1) provides, HARRIS V. AMGEN 57

A fiduciary with respect to a plan shall not cause the plan to engage in a transaction, if he knows or should know that such transaction constitutes a direct or indirect –

(A) sale or exchange, or leasing, of any property between the plan and a party in interest; . . .

(D) transfer to, or use by or for the benefit of a party in interest, of any assets of the plan[.]

A party in interest includes “any fiduciary” of a plan or “an employer” of the plan beneficiaries. 29 U.S.C. § 1002(14).

Defendants did not argue in the district court that Count VI fails to state a prohibited transaction claim under § 1106(a)(1). Nor do they raise this argument on appeal. Instead, defendants argue that 29 U.S.C. § 1108(e) exempts the sale of employer stock from the restrictions of § 1106(a)(1).

Section 1108(e) specifies that § 1106 does not prohibit the purchase or sale of employer stock if, as relevant here, (1) the sale price was the “price . . . prevailing on a national securities exchange”; (2) no commission is charged for the transaction, and (3) the plan is an EIAP. 29 U.S.C. §§ 1107(d)(5), (e)(1), 1108(e). In Howard v. Shay, 100 F.3d 1484, 1488 (9th Cir. 1996), we held that because § 1108(e) is an affirmative defense, a defendant has the burden to prove its applicability. We explained, “A fiduciary who engages in a self-dealing transaction pursuant to 29 U.S.C. § [1106(a)] has the burden of proving that he fulfilled his duties of care 58 HARRIS V. AMGEN

and loyalty and that the ESOP received adequate consideration [under § 1108(e)].” Id.; see also Marshall v. Snyder, 572 F.2d 894, 900 (2d Cir. 1978) (“The settled law is that in [prohibited self-dealing transactions] the burden of proof is always on the party to the self-dealing transaction to justify its fairness [under a statutory exception].”). Citing Howard, the Eighth Circuit has held that a plaintiff need not plead in his complaint that a transaction was not exempt under § 1108(e). See Braden v. Wal-Mart Stores, Inc., 588 F.3d 585, 600–01 (8th Cir. 2009); see also Jones v. Bock, 549 U.S. 199, 211–12 (2007) (holding that a plaintiff need not plead the absence of an affirmative defense, even a defense like exhaustion of remedies, which is “mandatory”).

Because the existence of an exemption under § 1108(e) is an affirmative defense, we can dismiss Count VI based on the § 1108(e) exemption only if the defense is “clearly indicated” and “appear[s] on the face of the pleading.” 5B Charles Alan Wright & Arthur R. Miller, Federal Practice & Procedure § 1357 (3d ed. 2004); see also Jones, 549 U.S. at 215 (citing Wright & Miller for rule that affirmative defense must appear on the face of the complaint). Here, we cannot say that the face of the complaint clearly indicates the availability of a § 1108(e) defense.

B. Amgen as Properly Named Fiduciary

Amgen argues that it is not a fiduciary under the Plan because it has delegated its discretionary authority. “To be found liable under ERISA for breach of the duty of prudence and for participation in a breach of fiduciary duty, an individual or entity must be a ‘fiduciary.’” Wright v. Or. Metallurgical Corp., 360 F.3d 1090, 1101 (9th Cir. 2004). In defining a fiduciary, ERISA says, HARRIS V. AMGEN 59

a person is a fiduciary with respect to a plan to the extent (i) he exercises any discretionary authority or discretionary control respecting management of such plan or exercises any authority or control respecting management or disposition of its assets . . . or (iii) he has any discretionary authority or discretionary responsibility in the administration of such plan.

29 U.S.C. § 1002(21)(A). “We construe ERISA fiduciary status ‘liberally, consistent with ERISA’s policies and objectives.’” Johnson v. Couturier, 572 F.3d 1067, 1076 (9th Cir. 2009) (quoting Ariz. State Carpenters Pension Trust Fund v. Citibank, 125 F.3d 715, 720 (9th Cir. 1997)). Whether a defendant is a fiduciary is a question of law we review de novo. See Varity Corp. v. Howe, 516 U.S. 489, 498 (1996).

Under ERISA, a “named fiduciary” is “a fiduciary who is named in the plan instrument.” 29 U.S.C. § 1102(a)(2). The Amgen Plan provides that Amgen is “the ‘named fiduciary,’ ‘administrator[,]’ and ‘plan sponsor’ of the Plan (as such terms are used in ERISA).” ERISA grants a named fiduciary broad authority to “control and manage the operation and administration of the plan.” 29 U.S.C. § 1102(a)(1). “Generally, if an ERISA plan expressly provides for a procedure allocating fiduciary responsibilities to persons other than named fiduciaries under the plan, the named fiduciary is not liable for an act or omission of such person in carrying out such responsibility.” Ariz. State Carpenters, 125 F.3d at 719–20 (citing 29 U.S.C. § 1105(c)(2)). 60 HARRIS V. AMGEN

Amgen argues that it delegated authority to trustees and investment managers. Section 15.1 of the Plan provides, “To the extent that the Plan requires an action under the Plan to be taken by the Company [Amgen], the party specified in this Section 15.1 shall be authorized to act on behalf of the Company.” Section 15.1 says nothing about delegation to trustees and investment managers. Rather, it explains that the Fiduciary Committee has the authority, on behalf of the Company, to “review the performance of the Investment Funds . . . and make recommendations” and to “otherwise control and manage the Plan’s assets.” In the absence of a Fiduciary Committee, the Global Benefits Committee will perform these tasks. Section 14.2 of the Plan governs the relationship between Amgen (“the Company”) and the trustees and managers. It provides:

The Trustee shall have the exclusive authority and discretion to control and manage assets of the Plan it holds in trust, except to the extent that . . . the Company directs how such assets shall be invested [or] the Company allocates the authority to manage such assets to one or more Investment Managers. Each Investment Manager shall have the exclusive authority to manage, including the authority to acquire and dispose of, the assets of the Plan assigned to it by the Company, except to the extent that the Plan prescribes or the Company directs how such assets shall be invested. Each Trustee and Investment Manager shall be solely responsible for diversifying, in accordance with Section 404(a)(1)(C) of ERISA, the investment of the assets of the Plan assigned HARRIS V. AMGEN 61

to it by the Committee, except to the extent that the plan prescribes or the Committee directs how such assets shall be invested.

ERISA requires that a trustee hold plan assets in trust for plan participants. 29 U.S.C. § 1103(a). A trustee has “exclusive authority and discretion to manage and control the assets of the plan” subject to two exceptions. Id. The first exception is that a plan may “expressly provide[] that the trustee or trustees are subject to the direction of a named fiduciary who is not a trustee.” Id. § 1103(a)(1). Under this exception, a named fiduciary with the power to direct trustees is a fiduciary with authority to manage plan assets. The second exception is that an “investment manager,” duly licensed as an investment adviser under federal or state law, may also be appointed to manage plan assets in lieu of the trustee. Id. §§ 1002(38)(B), 1103(a)(2).

There is no question that Amgen appointed a trustee. However, nothing in the record indicates that Amgen appointed an investment manager. Neither ERISA nor the Plan requires that an investment manager be appointed. Even if Amgen had appointed an investment manager, the Plan makes clear that the trustee and any investment manager do not have complete control over investment decisions. See 29 U.S.C. § 1002(21)(A)(i) (defining a person with “any authority or control” over plan assets to be a fiduciary) (emphasis added); cf. Gelardi v. Pertec Comp. Corp., 761 F.2d 1323, 1325 (9th Cir. 1985) (finding delegation where defendant “retained no discretionary control”) (emphasis added), overruled on other grounds in Cyr v. Reliance Standard Life Ins. Co., 642 F.3d 1202, 1207 (9th Cir. 2011). 62 HARRIS V. AMGEN

Section 15.1 of the Plan, which authorizes the Fiduciary Committee to take action on behalf of Amgen, does not preclude fiduciary status for Amgen. In Madden v. ITT Long Term Disability Plan for Salaried Empl., 914 F.2d 1279, 1284 (9th Cir. 1990), we held that the company had delegated authority to an administration committee where the plan provided that the Committee had “‘responsibility for carrying out all phases of the administration of the Plan’” and had the “‘exclusive right . . . to interpret the Plan and to decide any and all matters arising hereunder.’” (emphasis omitted). This language contains two features absent from the language in the Amgen Plan. First, it delegates responsibility for all phases of administering the plan, rather than responsibility “to the extent that the Plan requires an action . . . to be taken by the Company.” Second, and more important, it provides the Committee the exclusive right to make decisions under the plan. The Amgen Plan merely authorizes the Fiduciary Committee to act on behalf of Amgen. It neither provides exclusive authority to the Committee, nor precludes Amgen from acting on its own behalf.

Other courts have found a company’s grant of exclusive authority to a delegate and an express disclaimer of authority to be critical. In Maher v. Massachusetts General Hospital Long Term Disability Plan, 665 F.3d 289 (1st Cir. 2011), the First Circuit held that a hospital had delegated its fiduciary duties when the plan stated, “‘The Hospital shall be fully protected in acting upon the advice of any such agent . . . and shall not be liable for any act or omission of any such agent, the Hospital’s only duty being to use reasonable care in the selection of any such agent.’” Id. at 292. In Costantino v. Washington Post Multi-Option Benefits Plan, 404 F. Supp. 2d 31 (D.D.C. 2005), the district court for the District of Columbia found delegation when the plan granted the plan HARRIS V. AMGEN 63

administrator “‘sole and absolute discretion’” to carry out various Plan duties. Id. at 39 n.8. Given that ERISA allows fiduciaries to have overlapping responsibilities under a plan, a clear grant of exclusive authority is necessary for proper delegation by a fiduciary. See 29 U.S.C. § 1102(a)(1) (“[O]ne or more named fiduciaries . . . jointly or severally . . . have authority to control and manage the operation and administration of the plan”); see also 1 ERISA Practice and Litigation § 6:5 (“Those who wish to avoid liability exposure through allocation of plan responsibilities to others must therefore take pains to ensure that their documents fully authorize the contemplated delegation.”).

Because the Plan contains no clear delegation of exclusive authority, we reverse the district court’s dismissal of Amgen from the case as a non-fiduciary.

Conclusion

We conclude that defendants are not entitled to a presumption of prudence, that plaintiffs have stated claims under ERISA in Counts II through VI, and that Amgen is a properly named fiduciary under the Amgen Plan. We therefore reverse the decision of the district court and remand for further proceedings consistent with this opinion.

REVERSED and REMANDED.