Sacerdote v. New York University

9 F.4th 95

• Court: Court of Appeals for the Second Circuit
• Decided: August 16, 2021
• Docket: 18-2707-cv
• Status: Published

Opinion

18-2707-cv Sacerdote v. New York University

In the United States Court of Appeals For the Second Circuit ________

AUGUST TERM, 2020

ARGUED: OCTOBER 14, 2020 DECIDED: AUGUST 16, 2021

No. 18-2707-cv

DR. ALAN SACERDOTE, DR. HERBERT SAMUELS, MARK CRISPIN MILLER, MARIE E. MONACO, DR. SHULAMITH LALA STRAUSSNER, DR. JAMES B. BROWN, individually and as representatives of a class of participants and beneficiaries on behalf of the NYU SCHOOL OF MEDICINE RETIREMENT PLAN FOR MEMBERS OF THE FACULTY, PROFESSIONAL RESEARCH STAFF AND ADMINISTRATION AND THE NEW YORK UNIVERSITY RETIREMENT PLAN FOR MEMBERS OF THE FACULTY, PROFESSIONAL RESEARCH STAFF AND ADMINISTRATION, Plaintiffs-Appellants,

v.

NEW YORK UNIVERSITY, Defendant-Appellee,

THE TRUSTEES OF COLUMBIA UNIVERSITY IN THE CITY OF NEW YORK, Intervenor.* ________

*The Clerk of Court is respectfully directed to amend the caption as set forth above. 2 18-2707-cv

Appeal from the United States District Court for the Southern District of New York. ________

Before: NEWMAN **, WALKER, and MENASHI, Circuit Judges. ________

The plaintiff-appellant class participates in retirement plans administered by New York University (NYU) and the NYU School of Medicine. Plaintiffs brought this suit against NYU in its capacity as the fiduciary of plaintiffs’ retirement plans, alleging a number of breaches of NYU’s fiduciary duties under the Employment Retirement Income Savings Act (ERISA). Following a bench trial in the United States District Court for the Southern District of New York (Katherine B. Forrest, J.) and post-trial motions (Analisa Torres, J.), they appeal from the entry of judgment in defendant-appellee NYU’s favor and the denial of post-trial motions. On appeal, plaintiffs challenge: (1) the dismissal of their claim that NYU breached its duty of prudence by offering particular share classes of mutual funds in the retirement plans, (2) the denial of leave to amend their complaint to name additional defendants, (3) the striking of their demand for a jury trial, (4) the use of written declarations rather than live examination for direct testimony in the bench trial, (5) some of the district court’s findings in NYU’s favor after the bench trial, and (6) the denial of their motion for a new trial, which argued that the judge presiding over the trial (Forrest, J.) should have been disqualified. We find merit in the first two of these challenges, but

**Circuit Judge Ralph K. Winter, originally a member of this panel, died on December 8, 2020. Circuit Judge Jon O. Newman has replaced Judge Winter on the panel for this appeal. See 2d Cir. IOP E(b). 3 18-2707-cv

none in the remainder. Accordingly, we AFFIRM in part, VACATE in part, and REMAND for further proceedings consistent with this opinion.

Judge Menashi dissents in part in a separate opinion.

________

JEROME J. SCHLICHTER (Sean E. Soyars, on the brief), Schlichter Bogard & Denton, LLP, St. Louis, MO, for Plaintiffs-Appellants Sacerdote, et al.

SETH P. WAXMAN (David M. Lehn, Amy C. Lishinski, on the brief), Wilmer Cutler Pickering Hale and Dorr, LLP, Washington, DC; Alan Schoenfeld, Ryan M. Chabot, on the brief, Wilmer Cutler Pickering Hale and Dorr, LLP, New York, NY; Mark Muedeking, Ian C. Taylor, on the brief, DLA Piper LLP, Washington, DC; Brian Kaplan, on the brief, DLA Piper LLP, New York, NY; for Defendant-Appellee NYU.

Todd M. Schneider, James A. Bloom, on the brief, Schneider Wallace Cottrell Konecky Wotkyns LLP, Emeryville, CA; Todd S. Collins, Eric Lechtzin, on the brief, Berger Montague PC, Philadelphia, PA; for Amici Curiae Law Professors in Support of Plaintiffs-Appellants Sacerdote, et al.

Nancy G. Ross, on the brief, Mayer Brown LLP, Chicago, IL; Brian D. Netter, on the brief, Mayer 4 18-2707-cv

Brown LLP, Washington, DC; for Amici Curiae American Council on Education and Other Higher Education Associations in Support of Defendant- Appellee NYU.

JOHN M. WALKER, JR., Circuit Judge:

The plaintiff-appellant class participates in retirement plans administered by New York University (NYU) and the NYU School of Medicine. Plaintiffs brought this suit against NYU in its capacity as the fiduciary of plaintiffs’ retirement plans, alleging a number of breaches of NYU’s fiduciary duties under the Employment Retirement Income Savings Act (ERISA). Following a bench trial in the United States District Court for the Southern District of New York (Katherine B. Forrest, J.) and post-trial motions (Analisa Torres, J.), they appeal from the entry of judgment in defendant-appellee NYU’s favor and the denial of post-trial motions. On appeal, plaintiffs challenge: (1) the dismissal of their claim that NYU breached its duty of prudence by offering particular share classes of mutual funds in the retirement plans, (2) the denial of leave to amend their complaint to name additional defendants, (3) the striking of their demand for a jury trial, (4) the use of written declarations rather than live examination for direct testimony in the bench trial, (5) some of the district court’s findings in NYU’s favor after the bench trial, and (6) the denial of their motion for a new trial, which argued that the judge presiding over the trial (Forrest, J.) should have been disqualified. We find merit in the first two of these challenges, but none in the remainder. Accordingly, we AFFIRM in part, VACATE 5 18-2707-cv

in part, and REMAND for further proceedings consistent with this opinion.

BACKGROUND

The plaintiffs represent a class of NYU and NYU School of Medicine employees who are suing the University for breach of fiduciary duty in its administration of their retirement plans under ERISA. Plaintiffs participate in either the NYU Retirement Plan for Members of the Faculty, Professional Research Staff, and Administration (the Faculty Plan) or the NYU School of Medicine Retirement Plan for Members of the Faculty, Professional Research Staff, and Administration (the Medical Plan). The Faculty Plan covers most of NYU’s faculty, research staff, and administrative staff, while the Medical Plan serves employees of the School of Medicine.

The NYU Retirement Plan Committee (the Committee) is the nine-member fiduciary entity responsible for administering both plans, having been designated as the Plan Administrator by NYU’s Board of Trustees. The Committee is made up of senior University and Medical Center administrators, including NYU’s Chief Investment Officer, the Senior Vice Presidents of Finance of NYU and the Medical Center, the Medical Center’s Controller, the Vice Presidents of Human Resources of NYU and the Medical Center, the Directors of Benefits of NYU and the Medical Center, and NYU’s Provost (or its designee).

Both the Faculty Plan and Medical Plan (the Plans) are defined contribution plans, as set forth in 29 U.S.C. § 1002(34), and are tax- 6 18-2707-cv

qualified under 26 U.S.C. § 403(b). Defined contribution plans are retirement plans in which the employee contributes directly to her individual account, and the benefits that will ultimately accrue to the employee are a function of the amount she contributes to investments in the plan and the market performance of those investments, minus the expenses of plan administration. 1 Plans that operate under § 403(b)’s beneficial tax scheme are retirement plans administered by certain qualifying non-profits, including universities, that offer mutual fund and annuity investment options to participants. 2

Participants in NYU’s Plans had a range of investment options offered by either TIAA-CREF or the Vanguard Group, the two retirement investment firms under contract with NYU. The Faculty Plan offered 103 investment options (25 from TIAA-CREF; 78 from Vanguard) to plan participants during the class period. The Medical Plan offered 84 options (11 from TIAA-CREF; 73 from Vanguard). Both Plans offered investment options that included fixed annuity contracts (meaning the investment returns at a contractually specified minimum interest rate), variable annuities (returns at a variable interest rate), and mutual funds. Participants could also choose from both actively and passively managed index funds, with actively managed funds charging higher fees for that service.

TIAA-CREF and Vanguard are referred to in the industry as the Plans’ “recordkeepers.” They provide investment and administrative services, for which they charge investment fees and recordkeeping

1 29 U.S.C. § 1002(34). Defined contributions plans stand in contrast to defined benefit plans, in which the benefits ultimately accruing to the employee are fixed rather than dependent on market performance. 2 26 U.S.C. § 403(b). 7 18-2707-cv

fees, respectively. For mutual funds, the investment fees are charged as a percentage of each fund’s assets (the “expense ratio”). The fees can differ depending on the share class of the fund: a “retail” share (the share class that is marketed to individuals with small amounts to invest) typically has a higher expense ratio than an “institutional” share (the share class that is available to institutional investors, including large retirement plans, with large amounts to invest) of the same fund. These fees are measured in “basis points,” with each basis point equaling 0.01% of the fund’s assets. The administrative (recordkeeping) fees are charged either (1) as a flat fee, in which case each fund participant pays a set amount, or (2) by revenue sharing. Under the revenue-sharing model, a fund pays the recordkeeper a set portion of the fund’s expense ratio.

In 2016, plaintiffs brought this suit under 29 U.S.C. § 1132(a)(2), 3 alleging that NYU breached its fiduciary duties of loyalty and prudence and engaged in prohibited transactions, which caused the Plans to incur excessive costs and unreasonable performance losses. The breach allegedly occurred because the defendants: permitted TIAA-CREF to mandate inclusion of specific proprietary accounts, requiring use of TIAA-CREF as the recordkeeper, in the Plans (Counts I and II); incurred unreasonable recordkeeping fees (Counts III and IV); incurred unreasonable investment fees, unnecessary marketing and distribution fees and mortality and

3 Section 1132(a)(2) empowers plan participants and beneficiaries, among others, to sue plan fiduciaries for relief under 29 U.S.C. § 1109. Section 1109(a) makes fiduciaries who breach their fiduciary duties personally liable for resulting losses to the plan and the return of profits that flowed to the fiduciaries, and subject to equitable relief. 8 18-2707-cv

expense risk fees, and thus caused unreasonable performance losses (Counts V and VI); and failed to monitor the investments (Count VII).

On August 25, 2017, the district court granted in part and denied in part NYU’s motion to dismiss, dismissing Counts I, II, IV, VI, and VII in their entirety and Counts III and V in part. 4 The district court’s order dismissed all claims alleging that NYU breached its duty of loyalty under § 404(a)(1)(A); that NYU engaged in prohibited transactions under § 406(a)(1)(A), (C), and (D); and that NYU failed to monitor the investments. 5 The order also dismissed some of the plaintiffs’ claims alleging a breach of the duty of prudence under § 404(a)(1)(B). First, the court dismissed the imprudence claim under Count I, which alleged that NYU mandated inclusion of specific accounts and required the use of TIAA-CREF as recordkeeper. 6 Second, the court dismissed in part the imprudence claims under Count V to the extent they arose from allegations that NYU offered more expensive retail class shares rather than the lower-cost institutional class shares of the same mutual funds (the share-class claim), or incurred unnecessary and unreasonable layers of fees. 7

The only claims that survived dismissal were the imprudence claims in Count III and one of the imprudence claims in Count V. Specifically, Count III survived dismissal on the grounds of imprudence regarding incurring excessive recordkeeping costs (the

4 Sacerdote v. NYU, No. 16-cv-6284 (KBF), 2017 WL 3701482 (S.D.N.Y. Aug. 25, 2017). 5 Id. at *15. 6 Id. at *8. 7 Id. at *11. 9 18-2707-cv

recordkeeping claim); employing a revenue-sharing method to pay recordkeepers (the revenue-sharing claim); and failing to consolidate to a single recordkeeper for each Plan (the recordkeeper- consolidation claim). 8 Count V survived on the ground of imprudence in continuing to include the underperforming CREF Stock Account and TIAA Real Estate Account as investment options (the investment-retention claim). 9 Thus, those portions of Counts III and V were permitted to proceed to trial.

On September 8, 2017, plaintiffs moved both (1) for reconsideration of the district court’s dismissal of the share-class and failure to monitor claims and (2) for leave to amend the complaint to add seventeen individuals who were Committee members during the class period as named defendants and to replead the dismissed claims. On October 17, the district court denied the motion for leave to amend and deferred consideration of the request to replead the dismissed claims until resolution of the pending motion for reconsideration. The district court denied the motion for reconsideration two days later, relying on different reasoning from that supporting the dismissal of the share-class claim. 10

As the parties were preparing for trial, NYU successfully moved to strike plaintiffs’ jury demand. The district court also established trial management rules that specified that all direct testimony would be taken by written declarations (the court’s

8 Id. at *8–9. 9 Id. at *10. 10 Sacerdote v. NYU, No. 16-cv-6284 (KBF), 2017 WL 4736740, at *1–4 (S.D.N.Y. Oct. 19, 2017). 10 18-2707-cv

standing practice for bench trials) and that each side would have 25 hours of trial time to present its case.

The district court held a bench trial on the surviving claims from April 16–26, 2018. On July 31, 2018, the district court issued its written decision finding in favor of NYU on all remaining claims. 11

On August 14, 2018, plaintiffs filed a motion for amended or additional trial findings under Federal Rule of Civil Procedure 52(b) and to alter or amend the judgment under Rule 59(e), seeking findings that individual Committee members had failed to adequately perform their fiduciary duties and removal of those individual Committee members as fiduciaries, despite the overall judgment for NYU. Plaintiffs also appealed to this court on September 11, 2018, but we held the appeal in abeyance pending the district court’s resolution of the post-trial motions.

Meanwhile, by mid-July 2018, it had become public knowledge that Judge Forrest would be leaving the bench. She resigned from the bench effective September 11, 2018, and returned to her prior firm, Cravath, Swaine & Moore LLP, the following day. On October 1, 2018, plaintiffs moved for a new trial pursuant to Rule 60(b) on the ground that Judge Forrest should have been disqualified from the case based on a connection to NYU through a colleague at Cravath. On July 1, 2019, Judge Torres, to whom the case had eventually been reassigned, denied plaintiffs’ various post-trial motions. 12

11 Sacerdote v. NYU, 328 F. Supp. 3d 273, 317 (S.D.N.Y. 2018). 12 Sacerdote v. NYU, No. 16-cv-6284 (AT), 2019 WL 2763922, at *7 (S.D.N.Y. July 1, 2019). 11 18-2707-cv

DISCUSSION

On appeal, plaintiffs argue that: (1) the district court erred in dismissing the share-class claim; (2) the district court erred in denying the motion to amend the complaint to add individual Committee members as defendants, an error that later prejudiced two of their post-trial motions; 13 (3) they were entitled to a jury trial under the Seventh Amendment; (4) the use of written declarations for all direct testimony violated the Federal Rules of Civil Procedure and denied them a fair trial; (5) the district court’s trial findings in NYU’s favor on the recordkeeper-consolidation claim and the investment- retention claim were clearly erroneous; and (6) Judge Forrest should have been disqualified from presiding over this case.

We agree with respect to the first two challenges, and accordingly vacate the dismissal of the share-class claim, vacate the denial of leave to amend, and vacate the denial of the prejudiced post- trial motions. We otherwise affirm.

I. Dismissal of the share-class claim was error

Although the district court granted NYU’s motion to dismiss a number of claims for failure to state a claim under Rule 12(b)(6), the only such claim relevant on appeal is plaintiffs’ allegation in Count V

Plaintiffs appeal the denial of their post-trial Rule 52(b) and 59(e) motions 13

separately from their appeal of the denial of leave to amend. For the reasons explained in Part II, infra, we decline to review the denial of plaintiffs’ post-trial motions because we find antecedent error in the district court’s denial of leave to amend, which prejudiced the review of those motions. 12 18-2707-cv

that the Plans’ fiduciary breached its duty of prudence by offering retail-class shares of certain mutual funds rather than lower-cost institutional-class shares of the same funds (i.e., the share-class claim).

Plaintiffs allege that “the only difference between the various share classes is fees,” 14 and that large investors like the Plans “can obtain [institutional] share classes with far lower costs than retail mutual fund shares.” 15 They allege that “[e]ven if a jumbo plan does not meet the minimum investment thresholds for an institutional share class, fund companies will routinely waive those minimums for billion dollar plans if merely requested.” 16 Supported by a lengthy and detailed chart, plaintiffs make specific allegations regarding the basis point differences in costs between retail and institutional shares of each of dozens of mutual funds offered in the Faculty and/or Medical Plans. They allege that fiduciaries can readily obtain this data on cost-differentials from the prospectus for each fund.

In granting the motion to dismiss, the district court found that “prudent fiduciaries may very well choose to offer retail class shares over institutional class shares . . . because retail class shares necessarily offer higher liquidity than institutional investment vehicles.” 17 It also found that plaintiffs’ allegations of imprudence in this respect were insufficient because “the fees offered for the sixty- three identified retail funds included in NYU’s Options ranged from

14Sacerdote v. NYU, No. 16-cv-6284, ECF No. 39, ¶ 139 (S.D.N.Y. Nov. 9, 2016) (amended complaint). 15 Id. ¶ 141. 16 Id. ¶ 142. 17 Sacerdote, 2017 WL 3701482 at *11. 13 18-2707-cv

4-77 basis points—a lower range than that permitted by the Third, Seventh, and Ninth Circuits.” 18

In responding to plaintiffs’ motion for reconsideration of the share-class claim, the district court changed its reasoning for dismissing the claim. 19 It affirmed the dismissal on the basis that “the ‘prudence of each investment is not assessed in isolation but, rather, as the investment relates to the portfolio as a whole,’” 20 and therefore “it must consider the mix rather than the prudence of any individual option when assessing a prudence claim.” 21 It found the allegations to be deficient, noting that “plaintiffs do not allege that, taken as a whole, the mixes of options in the Plans were imprudent because of the inclusion of these retail class shares.” 22 To withstand dismissal, the district court stated that “[t]he retail class shares would have to be so prevalent that an entire Plan was tainted.” 23 It found that, in this case, plaintiffs’ allegations that the Plans offered retail rather than institutional shares in 63 funds—out of 103 offered by the Faculty Plan

18Id. (citing Tibble v. Edison Int’l, 729 F.3d 1110, 1135 (9th Cir. 2013), vacated, 575 U.S. 523 (2015); Renfro v. Unisys Corp., 671 F.3d 314, 319 (3d Cir. 2011); Loomis v. Exelon Corp., 658 F.3d 667, 669 (7th Cir. 2011); Hecker v. Deere & Co., 556 F.3d 575, 586 (7th Cir. 2009)). 19 Sacerdote, 2017 WL 4736740 at *1–3. Id. at *1 (emphasis in original) (quoting Pension Benefit Guar. Corp. v. 20

Morgan Stanley Inv. Mgmt. Inc., 712 F.3d 705, 717 (2d Cir. 2013) (hereinafter “PBGC”)). 21 Id. at *3. 22 Id. (emphasis in original). 23 Id. 14 18-2707-cv

and 84 offered by the Medical Plan—were insufficient to meet that standard as a matter of law. 24

On appeal, plaintiffs argue that their allegations are sufficient to generate a plausible inference of imprudence, and that the district court misconstrued our precedent in finding otherwise. 25 NYU disagrees on the merits, but it argues principally that, even if dismissal was error, the claim should not be reinstated because the district court’s later trial findings rendered the dismissal harmless.26

For the reasons we now explain, we find that the share-class claim was adequately pled and that we cannot conclude, on the present record, that its dismissal was harmless.

A. Standard of review

We review the grant of a motion to dismiss a claim under Rule 12(b)(6) de novo. 27 We apply the well-established pleading standard articulated in Bell Atlantic Corp. v. Twombly 28 and Ashcroft v. Iqbal 29: “To survive a motion to dismiss, a complaint must contain sufficient factual matter, accepted as true, to ‘state a claim to relief that is plausible on its face.’” 30 In assessing the complaint, we must construe it liberally, accepting all factual allegations therein as true and

24 Id. 25 Pls.-Appellants’ Br. at 30. 26 Def.-Appellee’s Br. at 43. 27 Palin v. N.Y. Times Co., 940 F.3d 804, 809 (2d Cir. 2019). 28 550 U.S. 544 (2007). 29 556 U.S. 662 (2009). 30 Id. at 678 (quoting Twombly, 550 U.S. at 570). 15 18-2707-cv

drawing all reasonable inferences in the plaintiffs’ favor. 31 However, we disregard conclusory allegations, such as “formulaic recitation[s] of the elements of a cause of action.” 32

We have cautioned that “the nature of . . . allegations under ERISA calls for particular care in applying this . . . inquiry in order to ensure that the . . . [c]omplaint alleges nonconclusory factual content raising a plausible inference of misconduct and does not rely on the vantage point of hindsight.” 33 On the other hand, we are cognizant that “ERISA plaintiffs generally lack the inside information necessary to make out their claims in detail unless and until discovery commences.” 34 So, as is true in many contexts, a claim under ERISA may withstand a motion to dismiss based on sufficient circumstantial factual allegations to support the claim, even if it lacks direct allegations of misconduct. 35

31 Palin, 940 F.3d at 809. 32 Twombly, 550 U.S. at 555; see also Iqbal, 556 U.S. at 678. 33 PBGC, 712 F.3d at 718 (emphases in original) (internal quotation marks and citation omitted). 34 Id. (internal quotation marks and citation omitted). 35 See id. (discussing that when an ERISA complaint “contains no factual allegations referring directly to [the defendant’s] knowledge, methods, or investigations at the relevant times,” the claim “may still survive a motion to dismiss if the court, based on circumstantial factual allegations, may reasonably infer” that the defendant acted unlawfully (emphasis in original) (internal quotation marks and citation omitted)). 16 18-2707-cv

B. The share-class claim for breach of the fiduciary duty of prudence was adequately pled

ERISA imposes a “prudent man standard of care” on retirement plan fiduciaries in order “to protect beneficiaries of employee benefits plans.” 36 As relevant to the share-class claim, fiduciaries must “discharge [their] duties . . . 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.” 37

The prudence of a fiduciary “is measured according to the objective prudent person standard developed in the common law of trusts.” 38 ERISA instructs us to assess a fiduciary’s prudence “under the circumstances then prevailing,” so we must “judge a fiduciary’s actions based upon information available to the fiduciary at the time of each investment decision and not from the vantage point of hindsight.” 39 “[T]his standard focuses on a fiduciary’s conduct in arriving at an investment decision, not on its results, and asks

36 Id. at 715 (internal quotation marks and citation omitted); see also 29 U.S.C. § 1104(a)(1). 37 29 U.S.C. § 1104(a)(1)(B). 38 Katsaros v. Cody, 744 F.2d 270, 279 (2d Cir. 1984) (internal quotation marks and citation omitted); see also Tibble, 575 U.S. at 528 (“We have often noted that an ERISA fiduciary’s duty is ‘derived from the common law of trusts.’” (quoting Cent. States, Se. & Sw. Areas Pension Fund v. Cent. Transp., Inc., 472 U.S. 559, 570 (1985))). 39 PBGC, 712 F.3d at 716 (internal quotation marks and citation omitted). 17 18-2707-cv

whether a fiduciary employed the appropriate methods to investigate and determine the merits of a particular investment.” 40

A claim for breach of the duty of prudence will “survive a motion to dismiss if the court, based on circumstantial factual allegations, may reasonably infer from what is alleged that the process was flawed” or “that an adequate investigation would have revealed to a reasonable fiduciary that the investment at issue was improvident.” 41 That is the situation here.

The complaint sets forth cost differentials of specified basis points for the dozens of mutual funds as to each of which, they claim, NYU should have offered lower-cost institutional shares instead of higher-cost retail shares. Plaintiffs allege that this information was included in fund prospectuses and would have been available to inquiring fiduciaries when the fiduciaries decided to offer the funds in the Plans. In sum, plaintiffs have alleged “that a superior alternative investment was readily apparent such that an adequate investigation”—simply reviewing the prospectus of the fund under consideration—“would have uncovered that alternative.” 42 On review of a motion to dismiss, we must draw reasonable inferences

40 Id. (internal quotation marks, citation, and alterations omitted). 41 Id. at 718 (internal quotation marks and citations omitted). 42 Id. at 719 (describing allegations that would be sufficient to raise a plausible inference of imprudence and withstand a motion to dismiss); see also Sweda v. Univ. of Pa., 923 F.3d 320, 331 (3d Cir. 2019), cert. denied, 140 S. Ct. 2565 (2020) (reversing dismissal of claim alleging “that despite the availability of low- cost institutional class shares, [the fiduciary] selected and retained identically managed but higher cost retail class shares,” where the complaint included “a table comparing options in the Plan with the readily available cheaper alternatives”). 18 18-2707-cv

from the complaint in plaintiffs’ favor. 43 Upon doing so, with respect to the share-class allegations, we believe that plaintiffs have sufficiently alleged that NYU acted imprudently in offering the number of retail-class shares identified in the complaint.

Although the district court abandoned its initial rationale for dismissing this claim, we note two problems in its order. First, the notion that “prudent fiduciaries may very well choose to offer retail class shares over institutional class shares” because retail shares offer greater liquidity provides no basis to dismiss pleadings that otherwise generate plausible inferences of the claimed misconduct. Such an argument “goes to the merits and is misplaced at this early stage.” 44 While the plausibility standard requires that facts be pled “permit[ting] the court to infer more than the mere possibility of misconduct,” 45 we do not require an ERISA plaintiff “to rule out every possible lawful explanation for the conduct he challenges.” 46 To do so “would invert the principle that the complaint is construed most favorably to the nonmoving party” on a motion to dismiss. 47

Second, we caution against overreliance on cost ranges from other ERISA cases as benchmarks. While such comparisons may

43 Palin, 940 F.3d at 809. 44 Sweda, 923 F.3d at 333. 45 PBGC, 712 F.3d at 718 (internal quotation marks and citation omitted). 46 Braden v. Wal-Mart Stores, Inc., 588 F.3d 585, 597 (8th Cir. 2009). 47 Id. (internal quotation marks and citation omitted); see also Sweda, 923 F.3d at 326 (rejecting application of Twombly’s heightened antitrust pleading standard to ERISA complaints and noting that, on a motion to dismiss, a district court may not “require[] [an ERISA plaintiff] to rule out lawful explanations for [the defendant’s] conduct”). 19 18-2707-cv

sometimes be instructive, their utility is limited because the assessment of any particular complaint is a “context-specific task.” 48 We cannot rule out the possibility that a fiduciary has acted imprudently by including a particular fund even if, for example, the fees that fund charged are lower than a fee found not imprudent in another case.

The district court’s order denying the motion for reconsideration similarly offers no compelling ground for affirmance. In it, the court faulted plaintiffs for not alleging (1) that “the mixes of options in the Plans were imprudent,” or (2) that the Plans were tainted in their entirety because the retail shares were included. Neither ground for dismissal is persuasive.

As for the “mix” of funds, we agree with the general principle that “the prudence of each investment is not assessed in isolation but, rather, as the investment relates to the portfolio as a whole.” 49 But this principle alone cannot support the district court’s dismissal of the share-class claim in this case. As we have suggested previously, allegations concerning the mix of investments are more centrally relevant to claims of imprudence based on the riskiness of funds or the risk-profile of a portfolio as a whole. 50 Here, with respect to the

48 PBGC, 712 F.3d at 718. 49 Id. at 717; see also Sweda, 923 F.3d at 331 (“employ[ing] a holistic approach” to evaluate ERISA complaint); Braden, 588 F.3d at 598 (noting that ERISA’s “remedial scheme . . . counsel[s] careful and holistic evaluation of an ERISA complaint’s factual allegations before concluding that they do not support a plausible inference that the plaintiff is entitled to relief”). 50See, e.g., PBGC, 712 F.3d at 721 (discussing adequacy of allegations that fiduciary “exposed the [p]lan to excessive risk due to an egregious over- concentration in high-risk mortgage securities”). 20 18-2707-cv

share-class claim, the alleged imprudent choice has nothing to do with the funds’ risk profiles; the choice was simply between higher- or lower-cost shares of the same fund. In some cases where the ‘mix’ analysis is appropriate, there is uncertainty as to what investment option would have been included if the questioned investment option had not been included. Here, however, there was a binary choice between the retail shares and the institutional shares; had the funds not included the former, they would have included the latter, to some extent. Even if this were not the case, the principle that a portfolio should be assessed holistically does not preclude critical assessment of individual funds. 51 Fiduciaries cannot shield themselves from liability—much less discovery—simply because the alleged imprudence inheres in fewer than all of the fund options. If the prudence of a particular investment offering will become clear only in the context of the portfolio as a whole, that argument cannot resolve a motion on the pleadings; it goes to the merits.

As to whether the Plans were tainted in their entirety, we do not suggest that a holistic assessment of the Plans is irrelevant to the share-class claim—we simply think that plaintiffs have pled enough on that claim to withstand dismissal at the pleading stage. They allege that 63 of the funds included in the 103-fund and 84-fund Plans charged excessive (retail share) fees, each of which plaintiffs set forth with specificity. The district court appears to have faulted plaintiffs for failing to calculate what 63/104 or 63/84 would be as a percentage

51 See Sweda, 923 F.3d at 330 (“We did not hold . . . that a meaningful mix and range of investment options insulates plan fiduciaries from liability for breach of fiduciary duty. Such a standard would allow a fiduciary to avoid liability by stocking a plan with hundreds of options, even if the majority were overpriced or underperforming.”). 21 18-2707-cv

of each plan, and then to allege that those percentages were high enough to taint each plan as a whole. But plaintiffs’ non-conclusory allegations plainly pointed the way to these obvious inferences in plaintiffs’ favor.

Drawing all reasonable inferences in their favor, plaintiffs have plausibly alleged that offering these retail shares rather than institutional shares was imprudent. This claim should have been, and now must be, litigated on the merits.

C. The share-class claim’s dismissal was not harmless

NYU urges us to decline to reinstate the share-class claim on the basis that any error in dismissing it was harmless because of two findings the district court made at trial. The first finding was that plaintiffs failed to prove a breach of fiduciary duty in using revenue sharing to fund recordkeeping costs. The second finding was that plaintiffs failed to prove loss resulting from either of the revenue- sharing or recordkeeping claims. NYU’s reliance on these findings is premised on its argument that selecting higher-cost retail shares was necessary to pay recordkeeping fees through revenue sharing. For the reasons that follow, however, we are not persuaded that these findings compel the conclusion that dismissal of the share-class claim on the pleadings was harmless.

i. The district court’s revenue-sharing finding

We first address NYU’s argument that because the use of revenue sharing was found not imprudent at trial, and because the difference in costs between retail and institutional shares supplies the funds for the revenue-sharing arrangement to pay the recordkeepers, 22 18-2707-cv

the dismissal of the share-class claim on the pleadings was harmless. We think NYU’s argument takes the district court’s trial findings too far.

The revenue-sharing claim presented at trial was not concerned with the specific expense ratios of each fund that generated revenue for the recordkeepers. Rather, the district court was asked to focus on two other issues: (1) whether it was imprudent not to cap the per- participant cost of revenue sharing, and (2) whether it was imprudent to use revenue sharing at all instead of employing a flat fee billed to each participant. 52 The court’s rejection of this claim relied only on evidence pertaining to these general questions in relation to the Plans as a whole. 53

We cannot be sure what would have happened at trial (or on summary judgment) had the share-class claim survived dismissal. Importantly, the district court’s rejection of the revenue-sharing claim was specific to plaintiffs’ claim that the cost should have been capped per-participant; the district court did not explicitly find that the revenue-sharing costs themselves were prudent. When pressed at argument to direct us to such a finding in the district court opinion, NYU’s counsel could point only to the district court’s brief rejection of plaintiffs’ separate recordkeeping claim. But in rejecting the recordkeeping claim, the district court simply rejected the plaintiffs’

52 Sacerdote, 328 F. Supp. 3d at 305. 53 Id. at 305–06. 23 18-2707-cv

expert opinion on what would have been a reasonable per-participant amount to charge for recordkeeping services. 54

We decline to foreclose the reinstatement of a wrongly dismissed claim on the basis of a record that was developed with respect to a different question and that is underdeveloped in the context of the harmlessness argument that defendants now press. For example, there is a set of Committee meeting minutes in the record, upon which the dissent heavily relies, that appears relevant to the relationship between retail-share costs and revenue sharing. 55 But the document is unaccompanied by any testimony, is lacking in specificity, and does not compel the dissent’s conclusion that the erroneous dismissal of the share-class claim was necessarily harmless. The fact that one document purports to memorialize a discussion about whether or not to offer retail shares does not establish the prudence of that discussion or its results as a matter of law.

The dissent states that “if revenue sharing is prudent, so too is offering retail shares.” 56 We have no quarrel with the general concept of using retail shares to fund revenue sharing. But, there was no trial finding that the use here of all 63 retail shares to achieve that goal was not imprudent. Simply concluding that revenue sharing is appropriate does not speak to how the revenue sharing is implemented in a particular case. We do not know, for example,

54 Id. at 306. 55 See App’x at 959–62 (NYU Retirement Committee Meeting Minutes, Jan. 10, 2011); see also Dissent at 7–8. 56 Dissent at 4. 24 18-2707-cv

whether revenue sharing could prudently be achieved with fewer retail shares.

The dissent also insists that this “numerical claim” is nonetheless foreclosed by the findings at trial because NYU arrived at this number through “a deliberative process for adopting the revenue-sharing model.” 57 We cannot agree. While the absence of a deliberative process may be enough to demonstrate imprudence, the presence of a deliberative process does not, contrary to the dissent’s suggestion, suffice in every case to demonstrate prudence. Deliberative processes can vary in quality or can be followed in bad faith. In assessing whether a fiduciary fulfilled her duty of prudence, we ask “whether a fiduciary employed the appropriate methods to investigate and determine the merits of a particular investment,” not merely whether there were any methods whatsoever. 58

Of course, because this claim has not been litigated on the merits, we offer no opinion on the precise deliberations at issue here. Discovery should take place—and it may turn out to be minimal— before the claim is dispensed with. At the same time, any incentive to future parties to seek discovery as to dismissed claims will be avoided.

ii. The district court’s no-loss findings

NYU’s second argument in favor of harmlessness is that the district court’s findings that plaintiffs failed to prove loss on two of the tried claims—the revenue-sharing and recordkeeping claims—

57 Id. at 1. 58 PBGC, 712 F.3d at 716 (emphasis added). 25 18-2707-cv

foreclose a showing of loss on the share-class claim. We are unpersuaded.

First, regarding the revenue-sharing claim, the district court’s no-loss finding was at most implied; the court made no explicit findings about why plaintiffs had failed to prove loss on that claim. In contrast to the district court’s detailed discussion of the evidence supporting the finding that NYU had not breached its fiduciary duty by using revenue sharing, the district court’s only statement regarding plaintiffs’ failure to show loss relevant to the revenue- sharing claim is a blanket statement in a footnote rejecting their efforts to show loss generally. 59 That the district court did not discuss loss in the specific context of the revenue-sharing claim is understandable— it separately concluded there was no breach of fiduciary duty. However, that absence of reasoning leaves us with no way to assess whether the share-class claim is foreclosed in the way NYU argues.

Second, as to the recordkeeping claim, the district court’s discussion of its no-loss finding was unpersuasive. The full extent of the district court’s findings on this front was a brief rejection of plaintiffs’ proposed alternative range for recordkeeping fees. 60 This finding that plaintiffs had not come up with a credible alternative is distant from saying that the fees charged affirmatively resulted in no loss, and a further distance still from saying that each of the retail-

59 See Sacerdote, 328 F. Supp. 3d at 285 n.25 (explaining that because plaintiffs had not proven loss, the district court would not determine whether to shift the burden to NYU to disprove damages). 60 Id. at 306. 26 18-2707-cv

class shares selected was necessary to pay the recordkeeping costs and none of them resulted in lost opportunity costs.

Moreover, we are hard-pressed to rely on the discussion of loss that the district court did undertake because the discussion was somewhat unclear in several respects. It conflated loss with damages, appeared to answer a question the court claimed to leave undecided, and effectively misallocated the burden of proof on damages.

The first thing that perplexes us is the district court’s conflation of “loss” with “damages.” The court stated expressly that, because it found that plaintiffs had not shown loss, it had no occasion to confront the subsequent question of damages. 61 However, the court then proceeded to describe the evidence exclusively in terms of damages, crediting NYU’s “damages rebuttal expert” 62 and concluding that plaintiffs “ha[d] not met their burden of proof as to damages for excessive recordkeeping fees.” 63 To be clear, these terms are not interchangeable. Loss is measured in this context by “a comparison of what the [p]lan actually earned on the . . . investment with what the [p]lan would have earned had the funds been available for other [p]lan purposes. If the latter amount is greater than the former, the loss is the difference between the two.” 64 The question of how much money should be awarded to the plaintiffs in damages is distinct from, and subsequent to, whether they have shown a loss.

61 See id. at 285 n.25. 62 Id. at 306 n.76 (emphasis added). 63 Id. at 307 (emphasis added). 64 Donovan v. Bierwirth, 754 F.2d 1049, 1056 (2d Cir. 1985). 27 18-2707-cv

The district court’s conflation of the two concepts saps our confidence in its analysis on this subject.

We are further puzzled because, in stating that it had no need to address damages, the district court explicitly declined to resolve which party would bear the burden of proof during a damages analysis 65—but nevertheless it went on to resolve that exact question, and did so incorrectly. It stated that “plaintiffs fail to demonstrate by a preponderance of the evidence that their proposed fee ranges were the only plausible or prudent ones,” and so “[p]laintiffs thus have not met their burden of proof as to damages for excessive recordkeeping fees.” 66 These statements indicate that the district court believed the plaintiffs would, in addition to proving loss, bear the burden of proving the amount of damages. That allocation of the burden was erroneous.

Although plaintiffs bear the burden of proving a loss, 67 the burden under ERISA shifts to the defendants to disprove any portion of potential damages by showing that the loss was not caused by the breach of fiduciary duty.68 This approach is aligned with the Supreme

65 Sacerdote, 328 F. Supp. 3d at 285 n.25. 66 Id. at 307. 6729 U.S.C. § 1109(a); see also Silverman v. Mut. Benefit Life Ins. Co., 138 F.3d 98, 104 (2d Cir. 1998). 68N.Y. State Teamsters Council Health & Hosp. Fund v. Estate of DePerno, 18 F.3d 179, 183 (2d Cir. 1994); see also Donovan, 754 F.2d at 1056 (“[T]he court should presume that the funds would have been used in the most profitable” prudent fashion, and “[t]he burden of proving that the funds would have earned less than that amount is on the fiduciaries found to be in breach of their duty.”). But see Silverman, 138 F.3d at 105 (Jacobs, J., and Meskill, J., concurring) (“Causation of 28 18-2707-cv

Court’s instruction to “look to the law of trusts” for guidance in ERISA cases. 69 Trust law acknowledges the need in certain instances to shift the burden to the trustee, who commonly possesses superior access to information. 70 Even in the context of the share-class claim, where plaintiffs have alleged the known cost-differentials between retail and institutional shares, “it makes little sense to have the plaintiff hazard a guess as to what the fiduciary would have done had it not breached its duty in selecting investment vehicles, only to be told [to] guess again.” 71 In considering the potential opportunity cost to the plaintiff of the investment, “[i]t makes much more sense for the fiduciary to say what it claims it would have done and for the plaintiff to then respond to that.” 72

damages is therefore an element of the claim, and the plaintiff bears the burden of proving it.”). 69Tibble, 575 U.S. at 529; see also LaRue v. DeWolff, Boberg & Assocs., Inc., 552 U.S. 248, 253 n.4 (2008) (“[T]he common law of trusts . . . informs our interpretation of ERISA’s fiduciary duties . . . .”). 70 Restatement (Third) of Trusts, § 100 cmt. f (“When a plaintiff brings suit against a trustee for breach of trust, the plaintiff generally bears the burden of proof. This general rule, however, is moderated in order to take account of the trustee’s . . . superior (often, unique) access to information about the trust and its activities . . . .”). 71 Brotherston v. Putnam Invs., LLC, 907 F.3d 17, 38 (1st Cir. 2018), cert. denied, 140 S. Ct. 911 (2020) (internal quotation marks omitted); see also id. at 39 (joining the Fourth, Fifth, and Eighth circuits in “hold[ing] that once an ERISA plaintiff has shown a breach of fiduciary duty and loss to the plan, the burden shifts to the fiduciary to prove that such loss was not caused by its breach, that is, to prove that the resulting investment decision was objectively prudent”) (citing Tatum v. RJR Pension Inv. Comm., 761 F.3d 346, 363 (4th Cir. 2014); McDonald v. Provident Indem. Life Ins. Co., 60 F.3d 234, 237 (5th Cir. 1995); Martin v. Feilen, 965 F.2d 660, 671 (8th Cir. 1992)). 72 Id. at 38. 29 18-2707-cv

By requiring the plaintiffs here to prove that the alternative fee ranges proposed by their expert were “the only plausible or prudent ones,” 73 the district court failed to shift the burden onto the defendant. Had plaintiffs been able to prove that the charged fees were imprudent, and had the plaintiffs shown a prudent alternative, the burden would have shifted to the defendant to disprove that the entire amount of loss should be awarded as damages. Put differently, if a plaintiff proved that it was imprudent to pay $100 for something but that it would have been prudent to pay $10, it is not the plaintiff’s burden to prove that it would also have been imprudent to pay every price between $11 and $99. It is on the defendant to prove that there is some price higher than $10 that it would have been prudent to pay. 74

Against this backdrop, we decline to foreclose the share-class claim on the basis of the district court’s loss findings. Accordingly, we vacate dismissal of that claim in Count V and order its reinstatement for further proceedings.

II. Leave to amend was denied under the wrong legal standard, and denial was not harmless

Plaintiffs argue that the district court erred when it denied their motion to amend the complaint to add the Committee members as named defendants. We agree, because the district court denied the

73 Sacerdote, 328 F. Supp. 3d at 307 (emphasis added). Cf. LaScala v. Scrufari, 479 F.3d 213, 221 & n.4 (2d Cir. 2007) (noting, in a 74

case where a fiduciary breached his fiduciary duty by giving his son salary raises without trustee approval, that it would be his burden to disprove damages from the salary raises by demonstrating that his son’s services were reasonably necessary and the value of those services equaled the sums paid). 30 18-2707-cv

motion to amend with reference to the wrong legal standard. We therefore vacate the denial of leave to amend and remand for consideration under the correct legal standard. Plaintiffs further argue that this first error led to another, when the district court refused to order the removal of two of the members as fiduciaries. We agree that the outcome of the motion to amend may have affected the outcome of plaintiffs’ post-trial motions for removal of specific Committee members, and therefore we also vacate the relevant rulings on those post-trial motions.

The district court, on December 5, 2016, entered a scheduling order in which the dates seem to have been proposed by the parties that provided, in relevant part: “Amended pleadings may not be filed, and no party may be joined, without leave of Court more than 10 days after the filing of this Order or the filing of a responsive pleading, whichever occurs first.” 75 Nine months later, on September 8, 2017, shortly after resolution of the motion to dismiss and nearly three months before fact discovery closed, plaintiffs sought leave of court to amend the complaint to add the individual Committee members as named defendants.

The district court denied the motion with citations to Federal Rule of Civil Procedure 16(b) and the scheduling order. Specifically, it found that “[t]he time for amending the complaint as of right has passed” and so “without [plaintiffs] demonstrating good cause, the Court may dismiss this untimely motion. Plaintiffs have not

75 Sacerdote v. NYU, No. 16-cv-6284 (KBF), ECF No. 43 (S.D.N.Y. Dec. 5, 2016) (scheduling order). 31 18-2707-cv

demonstrated good cause for their failure to include the defendants whom they now propose to add.” 76

We review denial of leave to amend for abuse of discretion, 77 and will find it when the district court’s “decision rests on an error of law (such as the application of the wrong legal principle).” 78 The district court here applied the wrong Federal Rule of Civil Procedure to the motion to amend, so we must vacate the ruling.

The ability of a plaintiff to amend the complaint is governed by Rules 15 and 16 of the Federal Rules of Civil Procedure which, when read together, set forth three standards for amending pleadings that depend on when the amendment is sought. At the outset of the litigation, a plaintiff may freely amend her pleadings pursuant to Rule 15(a)(1) as of right without court permission. 79 After that period ends—either upon expiration of a specified period in a scheduling order or upon expiration of the default period set forth in Rule 15(a)(1)(A)—the plaintiff must move the court for leave to amend, but the court should grant such leave “freely . . . when justice so requires” pursuant to Rule 15(a)(2). This is a “liberal” and “permissive” standard, and the only “grounds on which denial of leave to amend has long been held proper” are upon a showing of “undue delay, bad

76 Sacerdote v. NYU, No. 16-cv-6284 (KBF), ECF No. 100 (S.D.N.Y. Oct. 17, 2017) (order denying motion to amend) (citations omitted). 77 Shimon v. Equifax Info. Servs. LLC, 994 F.3d 88, 91 (2d Cir. 2021). 78 Zervos v. Verizon N.Y., Inc., 252 F.3d 163, 169 (2d Cir. 2001). 79 Fed. R. Civ. P. 15(a)(1). 32 18-2707-cv

faith, dilatory motive, [or] futility.” 80 The period of “liberal” amendment ends if the district court issues a scheduling order setting a date after which no amendment will permitted. It is still possible for the plaintiff to amend the complaint after such a deadline, but the plaintiff may do so only up a showing of the “good cause” that is required to modify a scheduling order under Rule 16(b)(4). 81

The language of the scheduling order in this case set the deadline (ten days) for amending without leave of court. It set no expiration date after which all amendments were prohibited, which would have triggered the stricter Rule 16(b)(4) “good cause” standard thereafter. Thus, if plaintiffs wanted to amend after the stated deadline, they only needed the court’s leave—under Rule 15(a)(2)— which they sought by filing their motion for leave to amend. By considering plaintiffs’ motion to amend under Rule 16, the district court here committed legal error and thus abused its discretion.

The dissent, to find grounds for affirmance on this point, looks beyond the plain language of the order and speculates that what the district court really intended when it set the deadline to amend “without leave of Court” was to also set a deadline after which even amendments with leave of Court would not be permitted. 82 But

80Loreley Fin. (Jersey) No. 3 Ltd. v. Wells Fargo Sec., LLC, 797 F.3d 160, 190 (2d Cir. 2015) (internal quotation marks and citation omitted). 81 Parker v. Columbia Pictures, 204 F.3d 326, 340 (2d Cir. 2000). 82 Dissent at 10–11. 33 18-2707-cv

litigants are entitled to rely on the meaning suggested by the plain language of a court order, as these plaintiffs did here.

The dissent also suggests that the scheduling order at issue was a “pro forma” order, and that our analysis should therefore be affected by the possibility that similar orders were entered in other cases. 83 Based purely on the formatting of the document, we do not necessarily disagree that it originated as a form: there appear to have been blank spaces in which the parties filled in their proposed dates for the various scheduling deadlines. 84 But we are unpersuaded to change our legal analysis as a result.

First of all, we have found nothing defective in the order itself, only in the district court’s ensuing ruling on the motion for leave to amend, which applied the wrong legal standard. Other district courts that have issued an identical scheduling order, and evaluated a motion for leave to amend under that order pursuant to Rule 15(a)(2), would have no cause for concern. District courts wishing to evaluate motions for leave to amend under Rule 16 after a particular date need only write their scheduling orders consistent with that intent, and state that no amendment will be permitted after that date in order to trigger the Rule 16 standard. And secondly, even if the dissent’s speculation that other district courts may have committed similar interpretive missteps is true, that provides no reason for us to review this decision more deferentially. The efficacy of our appellate review

83 Id. at 12. 84We note, however, that there is no evidence in the record about other scheduling orders entered in cases before this district court that would inform this speculation. 34 18-2707-cv

should not be affected by the possible existence of other similarly worded scheduling orders in other cases.

We cannot find this error harmless because the resulting denial of leave to amend may have later affected plaintiffs’ post-trial motions. In the district court’s trial findings, it had harshly criticized as incompetent the performance of two Committee members— Margaret Meagher and Nancy Sanchez—who were among the fiduciaries that plaintiffs had sought (through their motion to amend) to name as defendants. 85 Thereafter, in their proper role of policing

85 Meagher was the Committee co-chair who served as the Senior Director of Benefits for NYU and later the Senior Director of Benefits for the Medical Center. The district court described her testimony as “concerning” and found “that Meagher does not have the depth of knowledge appropriate to oversee a plan the size of the NYU Faculty and Medical Plans.” Meagher’s testimony “made it clear that she viewed her role as primarily concerned with scheduling, paper movement, and logistics,” and “she displayed a surprising lack of in-depth knowledge concerning the financial aspects of managing a multi-billion dollar pension portfolio and a lack of true appreciation for the significance of her role as a fiduciary.” The court further noted that Meagher “appeared to believe it was sufficient for her to have relied rather blindly on [a retained investment advisor firm]’s expertise.” See Sacerdote, 328 F. Supp. 3d at 291 & nn. 35–37. Sanchez was Meagher’s supervisor and the Senior Vice President and Vice Dean for Human Resources and Organizational Development and Learning at the Medical Center. The district court found that Sanchez “was similarly unfamiliar with basic concepts relating to the Plans” and “d[id] not view herself as having adequate time to serve effectively on the Committee.” In one notable portion of testimony, when she was asked to identify the plan administrator, Sanchez responded, “I don’t review the plan documents. That’s what I have staff for.” Sanchez said that she relied upon Meagher to review the materials for her but, of course, Sanchez has her own full vote on the Committee. See id. at 291 & n.37. 35 18-2707-cv

the Plans’ fiduciaries, plaintiffs pressed for Meagher and Sanchez’s removal. 86

Plaintiffs’ effort was thwarted, in major part, because Meagher and Sanchez had not been named as defendants. Judge Torres denied plaintiffs’ post-trial motions on the ground that Judge Forrest had previously considered and rejected ordering their removal from the Committee. 87 However, we see no such ruling by Judge Forrest, even implicitly, in the trial findings. The only question put to the district court was the Committee’s performance as a whole because the Committee, in NYU’s shoes, was the only defendant. 88 In answering that question, the district court found that although “the level of involvement and seriousness with which several Committee members treated their fiduciary duty [was] troubling, [the court] does not find that this rose to a level of failure to fulfill fiduciary obligations. Between [the investment advisor firm’s] advice and the guidance of the more well-equipped Committee members . . . , the Court is persuaded that the Committee performed its role adequately.” 89 In our view, this passage means only that, while

86Such equitable relief is available under 29 U.S.C. § 1109(a), which provides that “[a]ny person who is a fiduciary with respect to a plan who breaches any of the responsibilities, obligations, or duties imposed upon fiduciaries” by ERISA “shall be subject to . . . equitable or remedial relief as the court may deem appropriate, including removal of such fiduciary.” See also Katsaros, 744 F.2d at 281 (“Since the trustees here acted imprudently . . . , it was not an abuse of discretion for the court to remove the trustees pursuant to its equitable power.”). 87 Sacerdote, 2019 WL 2763922 at *3. 88 See, e.g., Sacerdote, 328 F. Supp. 3d at 279 (describing the issue at trial as plaintiffs’ “claim that NYU, through its Retirement Plan Committee . . . failed to fulfill certain of its fiduciary obligations under ERISA). 89 Id. at 293. 36 18-2707-cv

certain constituent members of the Committee were incompetent, their colleagues’ diligence saved the Committee itself from failing to fulfill its fiduciary obligations.

Accordingly, although the decision of whether to order removal of ERISA fiduciaries would be a matter committed to the discretion of the trial court, 90 here, there is no such exercise of discretion to which we must defer. Had Meagher and Sanchez been named in the complaint as defendants, the district court would have had to enter judgments specific to each of them after trial, finding whether each had breached her fiduciary duty as an individual member of the Committee. Given the district court’s harsh assessment of Meagher and Sanchez’s performance as fiduciaries, it is hardly inevitable that the district court would have found in their favor and declined to remove them as fiduciaries had it been required to enter those judgments.

We therefore vacate the denial of leave to amend and remand for consideration under the correct legal standard. We also vacate the denial of plaintiffs’ Rule 52(b) and 59(e) post-trial motions; those motions sought findings specific to Meagher and Sanchez, and so, depending on how the motion to amend is disposed of on remand, those motions may require further consideration as well.

III. Plaintiffs waived their jury demand

Plaintiffs argue that they had a Seventh Amendment right to a trial by jury, and therefore that the district court erred in striking their jury demand. We disagree. The record of proceedings before the

90 Katsaros, 744 F.2d at 281. 37 18-2707-cv

district court makes clear that plaintiffs waived their jury demand.91 Accordingly, we need not address the substance of their Seventh Amendment argument.

On December 4, 2017, NYU moved to strike plaintiffs’ jury demand. Under the Southern District of New York’s local rules, plaintiffs’ opposition to NYU’s motion was due within fourteen days. 92 On December 19, one day after the deadline for plaintiffs to file a response had expired, the district court granted NYU’s motion to strike. Plaintiffs now argue that we should excuse their failure to respond to NYU’s motion as “inadvertent[].” 93 But they offer no justification for their admitted inadvertence—let alone a sympathetic one—and no explanation for their failure to subsequently raise this issue before the district court.

Plaintiffs did not, upon being alerted by the district court’s December 19 order, move for reconsideration of the order denying a jury trial. 94 Nor did they object at the pretrial conference, at which the

91 See Royal Am. Managers, Inc. v. IRC Holding Corp., 885 F.2d 1011, 1018 (2d Cir. 1989) (“[T]he right to jury trial may be waived by conduct of the parties.”). 92See S.D.N.Y. Local Rule 6.1(b)(2), available at https://www.nysd.uscourts.gov/sites/default/files/local_rules/rules-2018-10- 29.pdf. 93 Pls.-Appellants’ Br. at 46. 94 Although “[t]he standard for granting such a motion is strict, and reconsideration will generally be denied unless the moving party can point to controlling decisions or data that the court overlooked,” Shrader v. CSX Transp., Inc., 70 F.3d 255, 257 (2d Cir. 1995), plaintiffs could have filed such a motion to litigate their claimed constitutional right in good faith. Indeed, we have recognized that motions for reconsideration are appropriate vehicles to “correct a clear error or prevent manifest injustice.” Kolel Beth Yechiel Mechil of Tartikov, Inc. 38 18-2707-cv

parties and court turned their attention to the forthcoming bench trial. And finally, the plaintiffs thereafter “participat[ed] in a bench trial without objection[, which alone] constitutes waiver of the jury trial right.” 95 Under these circumstances, “[i]t would be patently unfair and, in effect, an ambush of the trial judge on appeal if appellant were allowed to lodge an early demand for a jury, participate in a bench trial without objection, and then assign as error the failure to honor the jury demand.” 96

IV. The district court’s use of written direct testimony was not an abuse of discretion

The district court followed its bench-trial practice of taking direct testimony by written submissions, followed by live cross- examination and live redirect. Plaintiffs argue that this practice violated the Federal Rules of Civil Procedure and denied plaintiffs a fair trial. These arguments lack merit.

Plaintiffs claim to have lodged this objection with the district court, but the letter they point to as evidence of that objection contains no such argument. 97 In that letter, they argued only that they needed more trial time for oral cross-examination of witnesses in light of the court’s practice of taking direct testimony by written declaration.

v. YLL Irrevocable Tr., 729 F.3d 99, 104 (2d Cir. 2013) (internal quotation marks and citation omitted). 95 Royal Am. Managers, 885 F.2d at 1018. 96 Id. (internal quotation marks, citation, and alterations omitted). 97 Pls.-Appellants’ Reply Br. at 24. 39 18-2707-cv

Having acknowledged the practice without objection, they effectively consented to it.

It is a “well-established general rule that a court of appeals will not consider an issue raised for the first time on appeal.” 98 In any event, we have approved of the practice of taking direct testimony by written submissions in bench trials. 99 While in certain cases this practice might exceed a district court’s discretion, there is nothing in this record indicating that the district court abused its discretion to manage trials efficiently. 100

V. The district court did not err in ruling for NYU on the tried claims

Plaintiffs appeal the district court’s entry of judgment for NYU after trial, and specifically make arguments that the trial court erred with respect to: (A) the recordkeeper-consolidation claim, and (B) the investment-retention claim. 101 We reject these arguments.

98 Otal Invs. Ltd. v. M/V CLARY, 673 F.3d 108, 120 (2d Cir. 2012) (per curiam). See Ball v. Interoceanica Corp., 71 F.3d 73, 77 (2d Cir. 1995) (per curiam) 99

(“[W]e approve the procedure allowing the parties to produce direct evidence from their witnesses in writing while permitting subsequent oral cross- examination—particularly when the parties agree to that procedure in advance.”). See United States v. Yakobowicz, 427 F.3d 144, 150 (2d Cir. 2005) (reviewing 100

“trial management issue . . . for abuse of discretion”); Manley v. AmBase Corp., 337 F.3d 237, 249 (2d Cir. 2003) (“District courts have considerable discretion in the management of trials . . . .”). Because we find no error in the district court’s determinations after trial 101

that NYU did not breach its fiduciary duties, we have no occasion to consider plaintiffs’ argument about the district court’s loss analysis in the context of their appeal from the trial findings. 40 18-2707-cv

“We review the district court’s findings of fact after a bench trial for clear error and its conclusions of law de novo.” 102 The clear error standard permits us to set aside a district court’s factual findings only if we are “left with the definite and firm conviction that a mistake has been committed.” 103 The district court’s determination as to whether NYU breached its fiduciary duty rests on an “application of those facts to draw conclusions of law, including a finding of liability, [and so] is subject to de novo review.” 104 Because we discern no clear error in the district court’s factual findings, we have little difficulty agreeing that NYU did not breach its fiduciary duties in the ways argued by plaintiffs at trial.

Turning first to the recordkeeper-consolidation claim: plaintiffs argued that it was imprudent for NYU to use multiple recordkeepers for the Plans rather than consolidating to one

102 L.I. Head Start Child Dev. Servs., Inc. v. Econ. Opportunity Comm'n of Nassau Cnty., Inc., 710 F.3d 57, 65 (2d Cir. 2013); see also Fed. R. Civ. P. 52(a)(6) (providing that, in an action tried without a jury, “[f]indings of fact, whether based on oral or other evidence, must not be set aside unless clearly erroneous, and the reviewing court must give due regard to the trial court’s opportunity to judge the witnesses’ credibility”). 103 United States v. U.S. Gypsum, 333 U.S. 364, 396 (1948). 104 F.D.I.C. v. Providence Coll., 115 F.3d 136, 140 (2d Cir. 1997) (internal quotation marks and citation omitted); see also L.I. Head Start, 710 F.3d at 69–71 (noting that the district court’s factual finding was not clearly erroneous and then finding, as a matter of de novo review on appeal, “that the Administrators breached their fiduciary duties with respect to the . . . [c]laim”); cf. LoPresti v. Terwilliger, 126 F.3d 34, 39 (2d Cir. 1997) (holding that a legal conclusion that a particular individual qualifies as a fiduciary under ERISA is subject to de novo review). To the extent language in Katsaros suggests that the applicable standard of review on this question might be clear error, see 744 F.2d at 279, it plainly has not survived our subsequent precedent. 41 18-2707-cv

recordkeeper. The Medical Plan contracted with TIAA-CREF, Vanguard, and Prudential as recordkeepers until 2013, when it consolidated with TIAA-CREF, while the Faculty Plan contracted with both TIAA-CREF and Vanguard throughout the class period. 105

The district court found that, while plaintiffs were correct that “[c]onsolidation may lead to lower recordkeeping fees[,] . . . [t]he evidence at trial support[ed] defendant’s contention that technical and other requirements prevented immediate consolidation.” 106 Specifically, the district court credited testimony from several Committee members to the effect that a lengthy and resource- intensive change of the computer systems used for payroll, finance, student records, and human resources at the Washington Square campus (where Faculty Plan members work) precluded the Committee from consolidating recordkeepers during that time.107 The court credited NYU’s belief that “any recordkeeper switch could not be completed without risk of significant errors or additional changes prior to completion of this global update of NYU’s systems and technology.” 108 The switchover of the University’s IT systems created this hurdle because “[a] change in recordkeepers would entail significant coordination with and changes to the new systems being implemented,” due to the fact that the recordkeepers’ systems must

105 Sacerdote, 328 F. Supp. 3d at 293–94. 106 Id. at 294. 107 Id. at 295–98. 108 Id. at 298. 42 18-2707-cv

interface with NYU’s systems in order to allow participants to access their account information. 109

Plaintiffs argue on appeal, as they did before the district court, that the IT justification for not consolidating recordkeepers sooner is not credible because the contemporaneous Committee meeting minutes lack references to these technical difficulties. 110 But “clear error review mandates that we defer to the district court’s factual findings, particularly those involving credibility determinations.” 111 In light of the district court’s extensive discussion of witness testimony that persuaded it to credit the IT justification, we find no clear error in the factual findings forming basis of the court’s rejection of the recordkeeper-consolidation claim.

Having accepted the district court’s factual findings, we agree that NYU did not breach its fiduciary duty of prudence by failing to consolidate recordkeepers any faster than it did. In light of the technical challenges NYU was facing, including the risk that participants would suffer disrupted account access, we cannot conclude that a hypothetical prudent fiduciary would have acted any differently.

Turning next to the investment-retention claim: plaintiffs argue that NYU breached its duty of prudence by retaining two particular funds—the CREF Stock Account and the TIAA Real Estate Account—

109 Id. & n.50. 110 Pls.-Appellants’ Br. at 71; Sacerdote, 328 F. Supp. 3d at 296 n.47. 111Phx. Glob. Ventures, LLC v. Phx. Hotel Assocs., Ltd., 422 F.3d 72, 76 (2d Cir. 2005) (per curiam). 43 18-2707-cv

in the Plans beyond when it should have because NYU was using inadequate benchmarks to decide which funds to retain.

The district court found after detailed analysis, however, that the benchmarks used by the Committee to evaluate these two funds were appropriate in light of these funds’ unique and complex characteristics and that the retention of these funds was valuable in diversifying the plans. 112 The district court found that the fact the Committee changed the benchmarks employed to monitor the TIAA Real Estate Account during the relevant period demonstrated careful attention to the fund’s performance. 113 Similarly, the district court found that the Committee “focused on the difficulties with benchmarking that the CREF Stock Account presented,” held “specialized discussions” about it, and was “actively engaged” with monitoring this fund’s benchmarks. 114 With respect to its evaluation of the funds’ performance, the district court found that these funds were performing as well as could be expected from contemporaneous assessments. 115 The district court specifically discredited plaintiffs’ expert testimony regarding these funds’ performance. 116 We see no clear error in these factual findings.

Based on the foregoing findings, we also agree that NYU did not breach its duty of prudence by failing to remove the CREF Stock Account and TIAA Real Estate Account from the Plans. The facts

112 Sacerdote, 328 F. Supp. 3d at 310–16. 113 Id. at 311. 114 Id. at 314 (internal quotation marks omitted). 115 Id. at 311, 314–15. 116 Id. at 311 & n.110, 314–16. 44 18-2707-cv

found at trial demonstrated that the Committee paid special attention to these funds and retained them on the strength of their performance against legitimate benchmarks. We agree with the district court that a hypothetical prudent fiduciary would have made similar choices if presented with these circumstances.

VI. Judge Forrest’s attenuated connection to NYU did not require disqualification

Plaintiffs argue, as they did before Judge Torres, that Judge Forrest was disqualified from presiding over this case due to her connection to one of NYU’s board members, and that Judge Torres erred by denying plaintiffs’ motion for a new trial on that ground following Judge Forrest’s resignation from the bench. We review the district court’s denial of a motion for new trial for abuse of discretion. 117

Federal law provides that: “Any justice, judge, or magistrate judge of the United States shall disqualify himself in any proceeding in which his impartiality might reasonably be questioned.” 118 And we have said: “[P]hrased differently, would an objective, disinterested observer fully informed of the underlying facts, entertain significant doubt that justice would be done absent

117Gomez v. City of New York, 805 F.3d 419, 423 (2d Cir. 2015) (per curiam). We note that, had plaintiffs moved before Judge Forrest for her recusal, rather than moving before Judge Torres for a new trial after Judge Forrest’s resignation, we would similarly be reviewing Judge Forrest’s decision to preside for abuse of discretion. S.E.C. v. Razmilovic, 738 F.3d 14, 30 (2d Cir. 2013), as amended (Nov. 26, 2013). 118 28 U.S.C. § 455(a). 45 18-2707-cv

recusal?” 119 Here, we think not. Judge Forrest’s connection to NYU is the sort of “remote, contingent, or speculative” relationship that “is not the kind of interest which reasonably brings into question a judge’s impartiality.” 120

Plaintiffs argue that Judge Forrest was disqualified because of her employment, both before taking the bench and after leaving the bench, at Cravath, Swaine & Moore LLP. They assert that because Judge Forrest left the bench for Cravath six weeks after issuing her trial findings, and because Cravath’s chairman Evan Chesler, who was a mentor to Judge Forrest before she took the bench and would be a close colleague after she left, serves on the NYU Board of Trustees, she had a “prospective financial relationship” with NYU that called her impartiality into question. 121 Upon a close look, this argument does not hold water.

Chesler was one of NYU’s sixty-one voting Board members, and he was one of more than eighty partners at Cravath. Chesler had no personal financial interest in this case, and his conduct is not at issue. Significantly, he did not sit on the NYU Board’s Retirement Committee. Nonetheless, plaintiffs claim that because Chesler had a “personal strong charitable interest in raising money for NYU’s endowment,” there is an appearance of bias on Judge Forrest’s part. 122 Their theory is that Judge Forrest would want to enter judgment in NYU’s favor in order to protect “donor confidence in NYU’s

119 United States v. Lovaglia, 954 F.2d 811, 815 (2d Cir. 1992). 120 In re Drexel Burnham Lambert Inc., 861 F.2d 1307, 1313 (2d Cir. 1988). 121 Pls.-Appellants’ Br. at 49. 122 Id. at 53. 46 18-2707-cv

treatment of employees and retirees” and therefore benefit Chesler’s personal charitable interest. 123 Plaintiffs also theorize that, as a personal matter, Judge Forrest “would be reluctant to strain [her] relationship” with Chesler “by condemning an institution to which Mr. Chesler has major ties and holds deep affection.” 124

We believe that plaintiffs’ theories of impropriety are too far- fetched to reasonably call Judge Forrest’s impartiality into question. Her prospective financial relationship was with Cravath, not Chesler individually. Cravath was never involved in this case, and Chesler’s involvement is limited to his membership on a large Board of Trustees in his personal capacity.

Although we agree with plaintiffs that the appearance of judicial impartiality is of the utmost importance, parties who dislike court rulings cannot later rely upon first-time assertions of tenuous, preexisting alleged conflicts of interest to avoid those rulings. Plaintiffs had similar arguments before Judge Forrest ruled against them, but never made those arguments. Even though plaintiffs did not know that Judge Forrest would leave the bench for Cravath until she did so, the knowledge that she had come to the bench from Cravath and had previously been close to Chesler at the firm was readily ascertainable at all stages of the litigation. Chesler’s attendance at Judge Forrest’s Senate confirmation hearing was a matter of public record—a fact plaintiffs themselves rely upon in

123 Id. 124 Id. 47 18-2707-cv

arguing for her disqualification on appeal. 125 And yet, they made no attempt to move for Judge Forrest’s disqualification “at the earliest possible moment,” 126 as they are required to do.

Under these circumstances, we discern no reasonable questions about the appearance of Judge Forrest’s impartiality. Judge Torres’s denial of plaintiffs’ motion for a new trial was not an abuse of discretion.

* * *

In sum, we hold that plaintiffs adequately pled a breach of the fiduciary duty of prudence in Count V’s share-class claim, and we cannot find the district court’s dismissal of this claim harmless on the present record. We therefore vacate its dismissal and reinstate the claim for further proceedings. We also find that the district court erred in denying plaintiffs’ motion to amend to name individual Committee members as defendants. We therefore vacate denial of leave to amend and vacate denial of the ensuing Rule 52(b) and 59(e) motions for post-trial findings concerning two of those individuals.

125Pls.-Appellants’ Br. at 55 (citing Confirmation Hearing on Federal Appointments Before the S. Comm. on the Judiciary, 112th Cong. 127 (2011) (statement of Katherine B. Forrest, Nominee to the U.S. District Court for the Southern District of New York)). 126Apple v. Jewish Hosp. & Med. Ctr., 829 F.2d 326, 333 (2d Cir. 1987); see also id. at 334 (“[T]wo concerns prompt this rule. First, judicial resources should not be wasted; and, second, a movant may not hold back and wait, hedging its bets against the eventual outcome.”). 48 18-2707-cv

We reject the remainder of plaintiffs’ arguments on appeal, affirming the trial of their claims without a jury, the use of written direct testimony at that trial, the entry of judgment for NYU on the tried claims, and the denial of their Rule 60 motion for a new trial based upon Judge Forrest’s alleged disqualification. We remand for further proceedings consistent with this opinion.

CONCLUSION

For the foregoing reasons, we AFFIRM in part, VACATE in part, and REMAND for further proceedings consistent with this opinion. 18-2707 Sacerdote v. NYU

MENASHI, Circuit Judge, dissenting in part:

I join the opinion of the court insofar as it affirms the judgment of the district court, and I dissent insofar as the court vacates and remands that judgment. I would not remand for further proceedings on the share-class claim, and I do not believe the district court abused its discretion in denying leave to amend the complaint.

I

The court appears to entertain two versions of the share-class claim: a categorical version (that NYU acted imprudently by including any retail shares) and a numerical version (that NYU acted imprudently by including too many retail shares). Neither version of the claim can prevail based on the trial record. The categorical claim is foreclosed by the district court’s decision that the revenue-sharing model was prudent—a judgment that the plaintiffs do not even appeal. Because retail shares enable revenue sharing, if revenue sharing is not imprudent, then neither is the inclusion of retail shares. The numerical claim is foreclosed by the uncontested determination that NYU followed a deliberative process for adopting the revenue- sharing model that includes the retail shares. ERISA requires prudence, meaning that an employer must follow a deliberative process in making its decisions, even if the decisions are imperfect. Here, NYU followed a deliberative process for deciding which retail shares to offer; therefore, NYU acted prudently even if the plaintiffs or the court could imagine a better ultimate decision. For these reasons, both versions of the share-class claim are foreclosed by the district court’s judgment after trial, and therefore the district court’s purportedly erroneous dismissal of the share-class claim was harmless.

A

Like all retirement plans, NYU’s retirement plans require the service of recordkeepers. Recordkeepers calculate and track account balances and investment performance and prepare and deliver enrollment materials, notices, and other materials to plan participants. For these services, recordkeepers must be paid. There are two ways to pay recordkeepers: based on the number of participants in the plan (a flat per-participant fee) or based on the assets under management (an asset-based fee).

NYU chose the latter. NYU paid the recordkeeping fees through a method called “revenue sharing,” in which NYU would offer retail shares—rather than institutional shares—of investment products as investment options for the plan participants. Many of the investment options NYU offered are available in two classes of shares: retail shares and institutional shares. The only difference between the two classes of shares is the cost; the underlying asset is the same. Retail shares have higher expense ratios. 1 Institutional shares have lower expense ratios. Just as sellers in other industries offer wholesale prices to large purchasers, investment managers offer lower-priced “institutional” shares to large clients. As one of the largest defined- contribution plans in the country, NYU could obtain institutional shares for its plan participants.

1 An expense ratio is the amount that an investment company charges investors to manage an investment portfolio.

2 Yet for sixty-three funds, NYU offered participants the higher- priced retail shares rather than the lower-priced but otherwise identical institutional shares. The court suggests that, based on the pleadings, the “reasonable inference[]” is that the inclusion of retail shares was due to NYU’s neglect and therefore its imprudence. Ante at 17. The court says that an “adequate investigation,” consisting of “simply reviewing the prospectus of the fund under consideration,” would have “uncovered” the “superior alternative investment” of institutional shares—and that the plaintiffs have plausibly alleged that NYU failed to conduct that adequate investigation. Id.

But the trial record reveals that NYU in fact investigated its alternatives and made a considered decision to offer retail shares rather than institutional shares. NYU did so for a perfectly reasonable reason: the excess cost of the retail shares paid for the recordkeeping fees under NYU’s revenue-sharing model. Under revenue sharing, administrative fees are not charged separately—as a flat per- participant fee would be—but are covered by the higher expense ratios of the retail-share offerings. Plan participants who buy the retail shares pay more, and the investment manager then transfers a portion of the excess expense ratios to the recordkeeper. The revenue-sharing model and the retail-share offerings cannot be viewed in isolation because the latter enables the former: revenue sharing works by offering higher-priced retail shares. 2

2 The plaintiffs recognized this interdependency in their argument before the district court. Though the plaintiffs argue on appeal that retail shares might be imprudent even if revenue-sharing is prudent and permissible, the plaintiffs’ complaint alleged that offering retail shares was imprudent precisely because it enabled revenue-sharing. See, e.g., App’x 109-10

3 Consequently, if revenue sharing is prudent, so too is offering retail shares. The district court concluded that the revenue-sharing plan was prudent. It found that revenue-sharing arrangements were “common” and that NYU had “du[ly] consider[ed] … the appropriate pros and cons” in rejecting the plaintiffs’ favored alternative—“a flat per-participant model”—because the flat fee would not be “fair” to participants with “relatively small account balance[s]” and because “flat dollar fees cannot be assessed against the TIAA and CREF annuity account balances in the Plans.” Sacerdote v. New York Univ., 328 F. Supp. 3d 273, 305-06 (S.D.N.Y. 2018). The plaintiffs do not challenge this ruling on appeal. Therefore, because the inclusion of retail shares is the mechanism by which revenue sharing operates, the district court’s determination that the revenue-sharing model was prudent forecloses the categorical version of the share-class claim: if revenue sharing is prudent, the inclusion of retail shares must also be prudent.

B

The court rejects this justification for including retail shares— that “the difference in costs between retail and institutional shares supplies the funds for the revenue-sharing arrangement to pay the recordkeepers”—by suggesting that NYU violated ERISA’s standard of prudence because it could have bargained for a better deal on “the specific expense ratios of each fund that generated revenue for the recordkeepers.” Ante at 22. The court thus shifts its argument from the categorical version of the share-class claim, according to which

(Complaint ¶ 223) (“[T]he use of these funds was tainted by the recordkeepers’ financial interest in including these funds in the Plans, which Defendant failed to consider.”).

4 NYU failed to investigate the possibility of institutional shares, 3 and defends the numerical version instead. According to the numerical version of the share-class claim, even if the inclusion of some number of retail shares would not be imprudent, sixty-three retail shares was too many.

The numerical claim is also foreclosed by the district court’s findings at trial. NYU did not act imprudently by including sixty- three retail shares in its retirement plan because NYU arrived at that number through a deliberative process—the deliberative process through which it adopted the revenue-sharing model. A deliberative process is what ERISA requires.

The court implies that an employer might be imprudent under ERISA if it makes a considered decision but fails to get the best deal possible. See ante at 25-26 (describing the standard for the share-class claim as requiring a showing that “each of the retail-class shares selected was necessary to pay the recordkeeping costs and none of them resulted in lost opportunity costs”). But ERISA’s standard of prudence requires the fiduciary to follow an appropriate process leading to its decision, not to make a perfect decision. See Pension Ben. Guar. Corp. (PBGC) v. Morgan Stanley Inv. Mgmt. Inc., 712 F.3d 705, 718 (2d Cir. 2013) (noting that a claim for breach of fiduciary duty under ERISA depends on showing that “the process was flawed” such that the fiduciaries failed to conduct “an adequate investigation” and that it is not “necessarily sufficient to show that better investment

3 See ante at 17 (arguing that the “plaintiffs have alleged ‘that a superior alternative investment was readily apparent such that an adequate investigation’—simply reviewing the prospectus of the fund under consideration—‘would have uncovered that alternative’”).

5 opportunities were available at the time of the relevant decisions”). The court itself even recognizes that ERISA’s “standard focuses on a fiduciary’s conduct in arriving at an investment decision, not on its results, and asks whether a fiduciary employed the appropriate methods to investigate and determine the merits of a particular investment.” Ante at 16-17 (quoting PBGC, 712 F.3d at 716). ERISA does not require an employer to obtain the best possible result as long as the employer acts prudently by following a deliberative process.

A fiduciary acts imprudently when it fails to follow a deliberative process or fails to conduct an adequate investigation at all. 4 A district court case, Tibble v. Edison Int’l, No. 07-CV-5359, 2010 WL 2757153 (C.D. Cal. July 8, 2010), is instructive. In Tibble, the district court concluded that the employer was imprudent in the selection of retail shares because the employer did not consider the relative benefits of institutional shares at all. In other words, the fiduciary acted imprudently because it undertook no process to investigate its options:

4 The plaintiffs recognized this standard when they filed their suit. Their new argument on appeal—that offering sixty-three retail shares is imprudent even though some lesser number might not be—is not reflected in their complaint. In the complaint, Count V alleged that offering any retail shares was imprudent because no prudent fiduciary would have included retail shares after following a proper process: “The failure to select far lower-cost share classes for the Plans’ mutual fund options that are identical in all respects … except for cost, demonstrates that Defendant failed to consider the size and purchasing power of the Plans when selecting share classes and failed to engage in a prudent process for the selection, monitoring, and retention of those mutual funds.” App’x 71 (Complaint ¶ 147) (emphasis added). In this way, the complaint applied the correct standard by alleging a wholesale failure to investigate the option of institutional shares. But the plaintiffs could not prove those allegations at trial.

6 The Investments Staff simply recommended adding the retail share classes of these three funds without any consideration of whether the institutional share classes offered greater benefits to the Plan participants. Thus, the Plan fiduciaries responsible for selecting the mutual funds (the Investment Committees) were not informed about the institutional share classes and did not conduct a thorough investigation. Id. at *25. Moreover, “[i]n the one instance in which the Plan fiduciaries actually reviewed the different share classes of one of these three funds, the fiduciaries realized that it would be prudent to invest in the institutional share class rather than the retail share class,” indicating that the presence of retail shares was not the result of a deliberative process. Id. at *26.

By contrast, the trial evidence in this case shows that NYU followed a deliberative process through which it made a considered decision to offer the sixty-three retail shares to finance its recordkeeping fees through revenue-sharing. As an initial matter, the presence of institutional shares in the plan offerings in this case shows that, unlike the employer in Tibble, NYU was not ignorant of the existence of institutional shares—and it is not plausible that NYU failed to “simply review[] the prospectus” to “uncover[] that alternative.” Ante at 17. To the contrary, NYU made a considered decision to maintain a revenue-sharing model that required the sixty- three retail-share offerings. Among other things, the minutes of the Retirement Committee’s meeting of January 10, 2011, illustrate the deliberative process by which the offerings were determined:

The Committee next discussed the availability of lower cost share classes as communicated to NYU by Vanguard, and the possibility of implementing a lower

7 share class in the NYU program. CLC [Cammack LaRhette Consulting, an outside investment adviser] noted that although the share class reduction seems to be an offer of a fee reduction, it actually offers plan sponsors the opportunity to decide how to structure the fees of the plan. That is, Vanguard would allow the plan to utilize the lower share classes as long as Vanguard continues to receive the required revenue necessary to administer the program. A plan sponsor could choose to use a lower cost share class for the program, but, because Vanguard would still require a certain amount of revenue for its services to NYU, Vanguard would require that any revenue lost from the lower cost share class be made up by either a per participant fee or direct payments from NYU. The Committee sought additional clarification as to whether NYU could utilize the lower share classes in its program. CLC confirmed that NYU could choose to utilize the lower share class. However, because the lower share class funds do not return any recordkeeping revenue to Vanguard, they would need to make up this revenue, either by issuing a per participant fee or by the plan sponsor paying to offset the cost. The existing share class in the NYU program provides 13 [basis points] of recordkeeping revenue to Vanguard; this would need to be made up by NYU or plan participants if the switch is made to the lower cost share class. The Committee agreed that since a change in share classes would not result in an actual fee reduction for plan participants, it did not make sense to change share classes at this time. App’x 959-62 (NYU Retirement Committee Meeting Minutes, Jan. 10, 2011). When the employer weighs the relevant variables and arrives

8 at a considered decision—as NYU did here—the employer has not violated its fiduciary duty of prudence even if, in hindsight, the decision could have been better. See Katsaros v. Cody, 744 F.2d 270, 279 (2d Cir. 1984) (noting that the decision must be evaluated “from the perspective of the ‘time of the [challenged] decision’ rather than from the ‘vantage point of hindsight’”). Because NYU made a considered decision by following a deliberative process, the plaintiffs cannot prevail on remand. 5

In short, the plaintiffs have already received a trial on whether NYU acted prudently when it implemented the revenue-sharing model by offering sixty-three retail shares to plan participants. That trial showed that NYU followed a deliberative process that demonstrates prudence. Therefore, even if the district court erred in dismissing the share-class claim, that error was harmless because the trial effectively disposed of the claim. For that reason, I would not remand. See, e.g., Wilson v. Hanrahan, 804 F. App’x 58, 61 (2d Cir. 2020) (affirming dismissal of a claim because, even if the dismissal were an

5 The court writes that “the presence of a deliberative process does not … suffice in every case to demonstrate prudence” because the process might have been “followed in bad faith” or “vary in quality.” Ante at 24. Yet there has been a trial about the process NYU followed to adopt the revenue- sharing model, and “the trial record here reflects due consideration of the appropriate pros and cons,” showing that “the Committee’s choice to employ [revenue-sharing] was [not] imprudent.” Sacerdote, 328 F. Supp. 3d at 306. The trial record reflects a deliberative process, not one “followed in bad faith” or of poor “quality.” Ante at 24.

9 error, “that error was harmless” because the subsequent verdict meant that the claim “would necessarily have failed”). 6

II

The court also vacates the district court’s denial of leave to amend the complaint because the district court relied on Rule 16(b)’s “good cause” standard instead of Rule 15(a)’s liberal standard. Ante at 29-34. That is incorrect. The district court referenced the correct standard when it denied leave to amend.

The court acknowledges that “[t]he period of ‘liberal’ amendment ends if the district court issues a scheduling order setting a date after which no amendment will permitted.” Id. at 32. The district court issued a pro forma scheduling order setting a date after which it would not entertain amendments without leave:

Amended pleadings may not be filed, and no party may be joined, without leave of Court more than 10 days after the filing of this Order or the filing of a responsive pleading, whichever occurs first.

6 In addition, while the district court’s findings might not strictly foreclose a showing of loss on remand, the district court’s findings make such a showing highly unlikely. The district court concluded that the plaintiffs failed to establish that NYU paid “excessive recordkeeping fees.” Sacerdote, 328 F. Supp. 3d at 306-07 & n.76. Perhaps, on remand, the plaintiffs will offer testimony showing that NYU’s fees are higher than other institutions using revenue-sharing. But the plaintiffs could have introduced such evidence at trial to support their challenge to the revenue-sharing model. They did not—and do not now—indicate that they have such evidence. Thus, the plaintiffs have effectively obtained a trial on the issue of excessive fees, and there is no good reason to re-run the trial looking for the same evidence.

10 Special App’x 147. The obvious implication of this order is that the district court would not liberally grant leave to amend after the date it set.

Today’s opinion, however, refuses to acknowledge this obvious implication. Instead of adhering to the ordinary meaning of the order, the court insists that the district court set no deadline to seek leave to amend with leave of court after the ten-day period. 7 According to the court, therefore, the district court intended to cut off amendment as of right but for some reason still intended to liberally and freely grant leave to amend.

We should not read district court orders so tendentiously. Cf. Kanter v. Barr, 919 F.3d 437, 454 (7th Cir. 2019) (Barrett, J., dissenting) (“[J]udicial opinions are not statutes, and we don’t dissect them word- by-word as if they were.”). The meaning of the scheduling order is plain. It indicates that in the normal course, no pleadings may be amended; however, where there is good cause, pleadings may be amended if the court grants leave. We have said that the “lenient standard of Rule 15(a)” does not apply when a party seeks to amend after the deadline set in a scheduling order; under those circumstances, a party must show “good cause” under Rule 16(b). Parker v. Columbia Pictures Indus., 204 F.3d 326, 340 (2d Cir. 2000). The language of the scheduling order effectively communicates that the

7 Ante at 32 (“The language of the scheduling order in this case set the deadline (ten days) for amending without leave of court. It set no expiration date after which all amendments were prohibited, which would have triggered the stricter Rule 16(b)(4) ‘good cause’ standard thereafter.”); see also Reply Br. 12 (“The qualifier ‘without leave of Court’ necessarily means that amendments ‘with leave of Court’ were not subject to the same deadline.”).

11 district court intended to invoke that restriction. 8 “[W]ithout leave of the court” does not indicate that anything goes with leave of the court. Special App’x 147.

This reasonable understanding of the scheduling order accords with the structure of Rules 15(a) and 16(b). Rule 15(a) does not mention scheduling orders. Scheduling orders are issued pursuant to Rule 16(b), which sets out the requirements for issuing scheduling orders. For this reason, Rule 15(a) applies in the absence of a scheduling order, but when a scheduling order is issued, Rule 16(b)

8 The court misconstrues this dissent as “look[ing] beyond the plain language of the order.” Ante at 32. To the contrary, I think the meaning of the scheduling order is straightforward and I follow its plain language. The court, by contrast, reads the order in a highly technical fashion divorced from ordinary meaning. The ordinary meaning of a text includes not only its semantic content—that is, “the meaning of the words and phrases as combined by the rules of syntax and grammar”—but also its “pragmatic enrichment,” or “the contribution that context makes to meaning.” Lawrence B. Solum, Communicative Content and Legal Content, 89 NOTRE DAME L. REV. 479, 488 (2013). Thus, the “full communicative content” results from both the semantic and the pragmatic meaning. Id. As a pragmatic matter, sometimes “what is said implicitly includes something else that is closely related. For example, if I say ‘Jack and Jill are married,’ this frequently communicates some additional information, which could have been stated explicitly as follows: ‘Jack and Jill are married [to each other].’” Lawrence B. Solum, Contractual Communication, 133 HARV. L. REV. F. 23, 28 (2019) (describing “impliciture”). Here, the district court’s order— dictating that amended pleadings may not be filed without leave after ten days—communicates that the district court does not intend to liberally grant leave to amend after that deadline. The court improperly seizes on the order’s literal semantic meaning to the exclusion of its pragmatic meaning. But see Amy Coney Barrett, Assorted Canards of Contemporary Legal Analysis: Redux, 70 CASE W. RES. L. REV. 855, 859 (2020) (“[T]extualism isn’t a mechanical exercise, but rather one involving a sophisticated understanding of language as it’s actually used in context.”).

12 applies. We have explained that “the Rule 16(b) ‘good cause’ standard, rather than the more liberal standard of Rule 15(a), governs a motion to amend filed after the deadline a district court has set for amending the pleadings” because, “if we considered only Rule 15(a) without regard to Rule 16(b), we would render scheduling orders meaningless and effectively would read Rule 16(b) and its good cause requirement out of the Federal Rules of Civil Procedure.” Parker, 204 F.3d at 340 (alteration omitted). Here, the district court issued a Rule 16(b) scheduling order, and this court finds “nothing defective in the order itself.” Ante at 33. Yet the court holds that even though the district court properly issued a scheduling order pursuant to Rule 16(b), it nevertheless abused its discretion by following the “good cause” standard of Rule 16(b) rather than the liberal standard of Rule 15(a). In doing so, the court renders the scheduling order meaningless; in its view, the district court was obliged to freely permit amendment despite its issuance of a Rule 16(b) scheduling order. I would instead adhere to the rule that “a district court does not abuse its discretion in denying leave to amend the pleadings after the deadline set in the scheduling order where the moving party has failed to establish good cause.” Parker, 204 F.3d at 340.

The court’s decision today could have unexpected consequences. The scheduling order in question was one of Judge Forrest’s pro forma scheduling orders. She appears to have used it regularly, and perhaps other judges have used the same or similar language. Are all those orders now defective? In light of today’s opinion, district judges must beware. Instead of reading the scheduling order in the stilted fashion on which the court insists, I would read it reasonably and affirm the judgment. Cf. Lombardo v. City of St. Louis, 141 S. Ct. 2239, 2242 (2021) (Alito, J., dissenting) (“If we

13 expect the lower courts to respect our decisions, we should not twist their opinions to make our job easier.”).

***

The share-class claim is foreclosed by the district court’s judgment after trial. The categorical version of the claim necessarily fails because if revenue sharing is not imprudent, neither is the inclusion of retail shares. The numerical version of the claim necessarily fails because the deliberative process by which NYU adopted its revenue-sharing model satisfies ERISA’s duty of prudence. I therefore would not remand the case because the dismissal of the share-class claim, even if erroneous, was harmless. The district court also did not abuse its discretion in denying leave to amend. Accordingly, I dissent in part.