Bonnie Fish v. Greatbanc Trust Company

HAMILTON, Circuit Judge. The central issue in this appeal is the application of the statute of limitations for claims for breach of fiduciary duty under the Employee Retirement Income Security Act of 1974 (ERISA), 29 U.S.C. § 1001 et *674 seg. 1 The presumptive limitation period for violations is six years from the date of the last action constituting part of the breach or violation, but the statute provides a limited exception. The time is shortened to just three years from the time the plaintiff gained “actual knowledge of the breach or violation.” 29 U.S.C. § 1113. (The six-year limit can also be extended in cases of fraud or concealment, but neither is at issue here.) The plaintiffs in this case were employees of The Antioch Company who participated in an employee stock ownership plan or ESOP. Their claims arise from a buyout transaction at the end of 2003 in which Antioch borrowed money to buy all stock except the stock owned by the employee stock ownership plan. The buy-out ended badly, leaving Antioch bankrupt and the employee stock ownership plan worthless. The plaintiffs have sued under ERISA for breach of fiduciary duties in the buy-out. The district court granted summary judgment for the defendants under the three-year limit of § 1113(2), finding that proxy documents given to plaintiffs at the time of the buy-out transaction and their knowledge of Antioch’s financial affairs after the transaction gave them actual knowledge of the alleged ERISA violations more than three years before suit was filed. Fish v. GreatBanc Trust Co., 890 F.Supp.2d 1060 (N.D.Ill.2012). We reverse. The plaintiffs’ claims for breach of fiduciary duty do not depend solely on the disclosed substantive terms of the 2003 buy-out transaction. Their claims also depend on the processes that defendant GreatBanc Trust used to evaluate, to negotiate, and ultimately to approve the ill-fated transaction. The plaintiffs’ knowledge of the substantive terms of the buy-out transaction itself therefore did not give them “actual knowledge of the breach or violation” alleged in this case. See Maher v. Strachan Shipping Co., 68 F.3d 951, 956 (5th Cir.1995). Summary judgment on the statute of limitations defense must therefore be reversed. I. Undisputed Facts for Summary Judgment In this appeal from a grant of summary judgment, we consider the factual record in the light most favorable to the plaintiffs and give them the benefit of all conflicts in the evidence and reasonable inferences that may be drawn from the evidence. See Lesch v. Crown Cork & Seal Co., 282 F.3d 467, 471 (7th Cir.2002). We do not necessarily vouch for the objective accuracy of all factual statements here, but defendants moved for summary judgment, which requires that we view the evidence in this harsh light. A. The Parties and the Buy-Out Plaintiffs Bonnie Fish, Christopher Mino, Monica Lee Woosley, and Lynda Hardman were employees of Antioch, which made and sold scrapbooks and related accessories. They were also participants in Antioch’s Employee Stock Ownership Plan (“the Plan”). Before the buy-out transaction closed on December 16, 2003, the Plan owned 43 percent of Antioch’s common stock. The remainder was held primarily by members of the extended Morgan family, which had founded and still controlled Antioch. The Morgan family decided to pursue a major transaction that would accomplish several goals: (a) allow the Morgan family and other shareholders to cash out their Antioch stock holdings at a favorable price; (b) leave the Morgan family in control of the company *675 as fiduciaries of the Plan; and (c) gain tax advantages by converting Antioch to a sub-chapter S corporation with just one shareholder (the Plan). We simplify the details of the transaction, but it was structured so that Antioch would make a tender offer of $850 per share for all shares of its stock. The expectation was that the Morgan family and all other shareholders would sell all their stock, but an express condition of the transaction was that the Plan was required to decline the tender offer so that it would be left as the sole shareholder. To pay the Morgan family and the other shareholders the $850 per share, the relatively small employee-owned company would have to pay more than $150 million in cash, much of it newly borrowed. The Antioch Plan was governed by ERISA. The buy-out transaction was what ERISA treats as a prohibited transaction between an employee benefit plan and parties in interest. The economic substance of the transaction was that the Plan would buy Antioch stock (indirectly) from the Morgan family and other shareholders. The individual defendants — Lee Morgan, Asha Morgan Moran, and Chandra Attiken — were Plan fiduciaries under ERISA, which prohibits transactions between plans and persons in interest (including fiduciaries) unless, among other exceptions, the purchase was for fair market value determined in good faith by the fiduciary. See 29 U.S.C. §§ 1106(a), 1108(e). Antioch and the individual defendants agreed with the other defendant, GreatBanc Trust, to have it become the Plan trustee on a temporary basis for purposes of evaluating the proposed tender offer and making an independent decision about whether to agree to it (by agreeing not to tender the Plan’s shares). GreatBanc Trust became the Plan trustee on August 21, 2003, and remained the trustee until after the buy-out transaction closed. B. The Process Leading to the BuyOut Plaintiffs contend the defendants breached their fiduciary duties to use a sound process to evaluate the fairness of the proposed buy-out. GreatBanc Trust’s role was to serve temporarily as a trustee independent of Antioch and the Morgan family to evaluate the fairness of the transaction for the Plan participants. Great-Banc Trust was to negotiate with defendants on behalf of Plan participants and to keep them informed, and ultimately to approve or reject the buy-out transaction. The plan was for the individual defendants to retain control of Antioch by returning to their fiduciary positions with the Plan after the buy-out. For help in evaluating the transaction, GreatBanc Trust hired Duff & Phelps, a financial advice firm. In early October 2003, Duff & Phelps described the proposed transaction as “the most aggressive deal structure in the history of ESOPs.” (That comment led the Morgans and other Antioch management to contemplate firing GreatBanc Trust and Duff & Phelps.) GreatBanc Trust began negotiating amendments to the proposed transaction. In late October, Antioch agreed to GreatBanc Trust’s request for a so-called Put Protection Price (sometimes called a PPP) for employees who “cashed out” in the three years following the transaction. In an ESOP where the stock is not publicly traded, the plan must provide a “put option” that obliges the company to buy back an employee-participant’s stock when the employee retires, leaves employment, or otherwise cashes out. See 26 U.S.C. § 409(h). A PPP is a mechanism to protect ESOP participants against a short- *676 term drop in stock value, such as in the wake of a highly-leveraged transaction. The PPP in this deal imposed a floor price for 2004 cash-outs and set a fixed amount to add to the appraised fair market value of Antioch stock for cash-outs in 2005 and 2006. The PPP created significant additional liability and risk for Antioch and the Plan since the company was contractually obliged to pay the agreed-upon price premium. The PPP was binding no matter how many employees decided to cash out and no matter what the appraised fair market value of the stock might be at the time. In November 2003, Antioch also adopted a new Plan distribution policy that further increased the incentive for Plan participants to “cash out” with the benefit of the PPP after the buy-out. A Plan participant who retired early under the old distribution policy had to wait five years for payments to begin. (That’s the maximum time allowed by federal tax law. See 26 U.S.C. § 409(o).) Under the new distribution policy, payment would begin immediately and the full value would be paid within five years. This change further increased Antioch’s potential repurchase liability after the transaction. As they had begun their work on the Antioch transaction, GreatBanc Trust and Duff & Phelps had asked Antioch to provide repurchase liability projections for twenty-five years after the proposed transaction. The projections compared Antioch’s then-current repurchase obligations to the obligations expected after the buyout. To fulfill this request, Antioch provided GreatBanc Trust with one page from the report that Antioch’s chief financial officer had prepared to assess its liability before and after the proposed transaction. The record does not indicate that Great-Banc Trust or Duff & Phelps ever reviewed or even requested the full report. Without the full report, GreatBanc Trust and Duff & Phelps were unable to verify the key assumptions. They simply took Antioch at its word, according to plaintiffs. These key assumptions, which included the projected retirement age of Plan participants, were made back in July 2003, before the addition of the PPP and the new distribution policy. In other words, according to plaintiffs, GreatBanc Trust’s final approval of the buy-out was based on obsolete and incomplete information. Prior to the final version of the PPP agreement and new distribution policy, Duff & Phelps had provided GreatBanc Trust with a 22-page report summarizing the proposed transaction and a 79-page preliminary report reviewing its impact. These were supplemented in December 2003 by a four-page update to the original review and a final five-page fairness letter. These documents give no indication that GreatBanc Trust or Duff & Phelps considered the potential negative impact of the PPP or the new distribution policy in their fairness analysis. This omission lies at the core of plaintiffs’ claims. It implicates both GreatBanc Trust’s willingness to negotiate with Antioch and the defendants, at least as to the critical price term, and GreatBanc Trust’s consideration of the long-term interests of the Plan. None of the documents prepared by Duff & Phelps were provided to plaintiffs at the time of the transaction. C. The Information Available to Plan Participants Because the three-year limitations period under 29 U.S.C. § 1113(2) runs from the time the plaintiffs had “actual knowledge of the breach or violation,” this appeal depends in large part on the information they had about the transaction more than three years before they filed suit. Antioch sent a proxy statement regarding the tender offer to all Plan participants *677 and shareholders in November 2003, a month before the transaction closed. The proxy statement described the transaction and provided a fairness analysis for non-Plan participants, who had to act independently to tender their shares. The cover letter told Plan participants that Great-Banc Trust had been hired for the sole purpose of ensuring that the transaction was fair, prudent, and in the best interest of the Plan and its participants. Defendants’ motion for summary judgment relied primarily on the disclosures in the proxy materials to show plaintiffs’ early “actual knowledge” of the alleged breaches. The cover letter for the proxy materials said that GreatBanc Trust had determined that “it is prudent and in the best interests of the ESOP participants and beneficiaries not to sell the ESOP’s shares of Antioch’s common stock in the Tender Offer.” When discussing the purchase price of the shares, the proxy materials said: “A condition to the Closing is [GreatBanc Trust’s] receipt of an opinion from Duff & Phelps that the Transaction, as a whole, is fair to the ESOP from a financial point of view.” The proxy letter further noted that Great-Banc Trust “has received a preliminary opinion from Duff & Phelps to that effect.” These bare-bones references to Duff & Phelps’s preliminary report were all the information the Plan participants received about the fairness analysis conducted on their behalf. The proxy materials also included a one-page section titled “Risks Related to the Transaction,” which acknowledged some potential dangers of the highly-leveraged transaction. The materials addressed in bland terms the risks if the tax benefits were overestimated or the purchased shares were overvalued. They also noted that ESOP repurchase obligations could be higher than expected if the fair market value of stock “increases substantially.” The proxy materials provided reassurance, however: “The Company has projected the potential repurchase liability through the year 2013 under a set of assumptions that the Company believes to be reasonable.” The section concluded, though, that repurchase obligations could be unexpectedly higher and could leave the company “insolvent.” No part of the proxy materials provided to the Plan participants disclosed the processes that GreatBanc Trust and Duff & Phelps used to exercise due diligence and to conduct the fairness analysis. Duff & Phelps ultimately provided the required fairness opinion. GreatBanc Trust approved the transaction, which closed on December 16, 2003, making the Plan the sole shareholder of Antioch. D. Antioch’s Collapse After the closing, things were calm for a few months, but Plan participants began cashing out in the summer of 2004. In 2004, seventy employees under the age of fifty resigned and cashed out, taking advantage of the high stock value and the new distribution policy. According to plaintiffs, the many resignations in 2004 depleted Antioch’s remaining cash reserves, and tax savings could not fully offset declining sales. According to plaintiffs, these events set off a downward cycle as liabilities increased and revenues decreased, forcing Antioch into bankruptcy by 2008. Antioch shares and the Plan were worthless, representing a total loss of roughly $60 million to the named plaintiffs and several hundred of their co-workers. See Fish v. GreatBanc Trust Co., 830 F.Supp.2d at 429. 2 *678 According to plaintiffs, Antioch collapsed because the buy-out transaction was far too generous to the Morgan family and other shareholders, and because the transaction included ill-advised promises to Plan participants about their ability to receive comparable stock prices in cash if they retired or left the company within a few years. Saddled with excessive debt incurred to pay the Morgan family in the 2003 buy-out, Antioch was vulnerable to such a “stampede” to cash out. II. Analysis Defendants GreatBane Trust Co., Lee Morgan, Asha Morgan Moran, and Chandra Attiken all owed fiduciary duties to the Plan and its participants. Plaintiffs allege that GreatBane Trust violated its fiduciary duties under ERISA by failing to take reasonable steps to evaluate the fairness of the Morgan family’s proposed buy-out before agreeing to the transaction. Plaintiffs contend that the other defendants failed to monitor GreatBane Trust sufficiently, failed to disclose material information to GreatBane Trust, and acted under a conflict of interest where they would benefit from the transaction regardless of its effect on the employees in the Plan. (We have simplified the theories considerably; plaintiffs have identified a number of more specific procedural failings in GreatBane Trust’s evaluation of the proposed transaction.) The plaintiffs’ claims focus on the fairness analysis performed by GreatBane Trust and the individual defendants’ actions prior to the 2003 transaction. Plaintiffs contend that all defendants breached the fiduciary duty of prudence, see 29 U.S.C. § 1104(a)(1)(B), and engaged in a prohibited transaction without adequate consideration, see §§ 1106(a) and 1108(e). The defendants argued, and the district court agreed, that plaintiffs’ claims are time-barred because they had actual knowledge of the alleged breaches of fiduciary duty more than three years before filing suit. We begin by analyzing the “actual knowledge” standard under § 1113(2) and then turn to the plaintiffs’ claims and the relevant evidence. A. “Actual Knowledge” Under § 1113(2) ERISA provides its own statute of limitations. The generally applicable rule bars an action brought more than six years after the end of the fiduciary breach, violation, or omission. 29 U.S.C. § 1113(1). The statute also bars an action if it is commenced more than “three years after the earliest date on which the plaintiff had actual knowledge of the breach or violation.” § 1113(2). The application of the three-year exception to the six-year default rule turns on the meaning of “actual knowledge,” which must be distinguished from “constructive” knowledge or inquiry notice. Martin v. Consultants & Administrators, Inc., 966 F.2d 1078, 1086 (7th Cir.1992). The different circuit courts of appeals currently apply different tests for actual knowledge. See generally Wright v. *679 Heyne, 349 F.3d 321, 327-29 (6th Cir.2003). The strictest test applies the three-year bar only when the plaintiff knows not only the facts underlying the alleged violation but also that those facts constitute a violation under ERISA. See International Union v. Murata Erie N. Amer., 980 F.2d 889, 900 (3d Cir.1992). A strong textual argument supports this position because the text phrases the three-year limit in the unusual terms of “actual knowledge of the breach or violation” rather than merely knowledge of facts or knowledge of injury. See 29 U.S.C. § 1113(2) (emphasis added). 3 Other circuits do not require knowledge that the law was violated but still demand “actual knowledge of all material facts necessary to understand that some claim exists, which facts could include necessary opinions of experts, knowledge of a transaction’s harmful consequences, or even actual harm.” Maher v. Strachan Shipping Co., 68 F.3d 951, 954 (5th Cir.1995) (also quoting but not adopting the Third Circuit’s decision in International Union, 980 F.2d at 900, which requires knowledge of the law) (quotations omitted); see also Caputo v. Pfizer, Inc., 267 F.3d 181, 193 (2d Cir.2001). We have observed that “it is difficult to say in the abstract precisely what constitutes ‘actual knowledge.’ ” Consultants & Administrators, 966 F.2d at 1086. Our most concise definition is “knowledge of ‘the essential facts of the transaction or conduct constituting the violation,’ ” with the caveat that “it is ‘not necessary for a potential plaintiff to have knowledge of every last detail of a transaction, or knowledge of its illegality.’ ” Rush v. Martin Petersen Co., 83 F.3d 894, 896 (7th Cir.1996), quoting Consultants & Administrators, 966 F.2d at 1086. This court’s precedent seems consistent with the Fifth Circuit’s approach in Maher, which requires knowledge of “all material facts” but not knowledge of every detail or knowledge of illegality. 68 F.3d at 954. And a court applying § 1113(2) must take care to resist the temptation to slide toward reliance upon constructive knowledge or imputed knowledge, neither of which is actual knowledge. B. Plaintiffs’ Claims for Breach of Fiduciary Duties ERISA imposes general standards of loyalty and prudence that require fiduciaries to act solely in the interest of plan participants and to exercise their duties with the “care, skill, prudence, and diligence” of an objectively prudent person. 29 U.S.C. § 1104(a)(1); Eyler v. Comm’r of Internal Revenue, 88 F.3d 445, 454 (7th Cir.1996). In addition, § 1106 supplements the general fiduciary duty provisions by prohibiting ERISA fiduciaries from causing a plan to enter into a variety of transactions with a “party in interest.” See Keach v. U.S. Trust Co., 419 F.3d 626, 635 (7th Cir.2005). As a general rule, a fiduciary may not engage in a direct or indirect transaction constituting a “sale or exchange, or leasing, of any property between the plan and a party in interest.” *680 29 U.S.C. § 1106(a)(1)(A). A plan fiduciary is a party in interest, as are officers, directors, and major shareholders of a plan sponsor like Antioch. 29 U.S.C. § 1002(14)(A) & (H). Section 1106 begins, though, by saying “Except as provided in section 1108,” which provides numerous exceptions to the prohibited transaction rule. The most relevant exception for this case is for plan purchases of employer securities. Section 1106(a) does not apply to such purchases if, among other conditions, the transaction “is for adequate consideration.” § 1108(e). ERISA defines adequate consideration as “the fair market value of the asset as determined in good faith by the trustee....” § 1002(18)(B). Plaintiffs contend that by carrying out the Antioch buy-out transaction in 2003, all the defendants violated the general duty of prudence under § 1104 and engaged in a transaction prohibited by § 1106(a). We have before us not the merits of those claims but only the statute of limitations defense. To decide when the plaintiffs gained actual knowledge of the alleged breaches of fiduciary duty, we must examine the nature of the alleged breaches. See, e.g., Maher, 68 F.3d at 956. 1. Substantive and Procedural Elements of Plaintiffs’ Claims Whether an ERISA fiduciary has acted prudently requires consideration of both the substantive reasonableness of the fiduciary’s actions and the procedures by which the fiduciary made its decision: “In reviewing the acts of ESOP fiduciaries under the objective prudent person standard, courts examine both the process used by the fiduciaries to reach their decision as well as an evaluation of the merits.” Eyler v. Comm’r, 88 F.3d at 455. This is true when determining whether an act was prudent under the general standard of § 1104 and whether an otherwise prohibited transaction under § 1106 is saved by “adequate consideration” under § 1108(e). Keach, 419 F.3d at 636. In Reach we explained that this combination of substantive and procedural aspects of the fiduciary’s duties was consistent with a proposed Department of Labor regulation. Id. at 636 & n. 5. The Department of Labor has identified two requirements for a transaction to be considered supported by adequate consideration: a substantive requirement that the value assigned reflect the fair market value of the asset, and a procedural requirement that the fiduciary actually determine the value assigned in good faith. See Prop. DOL Reg. § 2510.3-18(b); 53Fed.Reg. 17,632-33 (May 17, 1988); see also Chao v. Hall Holding Co., 285 F.3d 415, 437 (6th Cir.2002) (endorsing test); Donovan v. Cunningham, 716 F.2d 1455, 1467-68 (5th Cir.1983) (describing standard before proposed regulation). In this case, Duff & Phelps provided GreatBanc Trust with financial advice about the proposed buy-out. That advice is highly relevant, of course, but we agree with our colleagues in the Fifth Circuit: “An independent appraisal is not a magic wand that fiduciaries may simply wave over a transaction to ensure that their responsibilities are fulfilled.” Donovan v. Cunningham, 716 F.2d at 1474. We said in Reach that “an independent assessment from a financial advisor ... is not a complete defense against a charge of imprudence.” 419 F.3d at 636-37 (internal quotation omitted). Whether the transaction is exempted under § 1108 by adequate consideration depends in part on whether GreatBanc Trust performed sufficient due diligence, including reasonable investigation into Duff & Phelps’s process and independent scrutiny of materials from Antioch. When determining whether a fiduciary’s process is sufficient, “ ‘the *681 degree to which a fiduciary makes an independent inquiry is critical.’ ” Keach, 419 F.3d at 636, quoting Eyler, 88 F.3d at 456. A fiduciary’s reliance on a financial advisor is evidence of prudence, but some inquiry into the advisor’s qualifications and methods is still required. Id. at 637. Whether GreatBanc Trust properly approved the buy-out transaction despite the prohibition in § 1106 depends on whether its process was sufficient to fulfill the procedural requirement of adequate consideration. GreatBanc Trust received Duff & Phelps’s evaluation of the fairness of the transaction. While GreatBanc Trust could rely on the fairness analysis of an expert, it must still demonstrate that its reliance on the advice from Duff & Phelps for this particular transaction was justifiable. That means a plaintiff asserting a process-based claim under § 1104, § 1106(a), or both does not have actual knowledge of the procedural breach of fiduciary duties unless and until she has actual knowledge of the procedures used or not used by the fiduciary. The Fifth Circuit has held that to trigger the “actual knowledge” statute of limitations clock under § 1113(2) for a process-based claim, the plaintiffs “must have been aware of the process utilized by [the fiduciary] in order to have had actual knowledge of the resulting breach of fiduciary duty.” Maher, 68 F.3d at 956 (reversing summary judgment for fiduciaries on process-based claim that had been based on three-year limit). We could not affirm here without creating a circuit split with Maher, as defendants acknowledged in oral argument, and we see no good reason to do so. The reasoning of Maher is sound. The Ninth Circuit has made the same point for process-based claims: the three-year limit cannot be triggered merely through disclosure of the terms of an imprudent investment when a claim “hinge[s] on the infirmities in the selection process.” Tibble v. Edison Int’l, 729 F.3d 1110, 1121 (9th Cir.2013) (affirming judgment for beneficiaries and rejecting statute of limitations defense). Thus, for process-based claims under §§ 1104 and 1106(a), the three-year limit is not triggered by knowledge of the transaction terms alone. 4 Our disagreement with the district court centers on this procedural aspect of plaintiffs’ claims. The district court rejected plaintiffs’ arguments targeting the process GreatBanc Trust used to evaluate the transaction: “Their true complaint is about the substance of the decision, not about the process undertaken in reaching the decision, for no matter how much process GreatBanc undertook, plaintiffs would still be complaining that the ultimate decision that set the redemption price too high was imprudent.” Fish, 890 F.Supp.2d at 1067 (emphasis in original). There is no doubt that the harm alleged by plaintiff was based on the substantive terms of the buyout, but knowledge of an unwise decision does not amount to “actual knowledge” of an imprudent process, which is an independent breach of fiduciary duty. The district court’s conclusion overlooked the procedural dimension of a fiduciary’s duties un *682 der ERISA and the ability of a plaintiff to show she was harmed by a fiduciary’s substantive decision precisely because the fiduciary violated ERISA by failing to comply with its procedural obligations. 2. Evidence of Plaintiffs’ Knowledge In support of their motion for summary judgment, the defendants showed that they provided information to the plaintiffs in November 2003 about the terms of the proposed buy-out, and they point to the “stampede” of cash-outs that began Antioch’s slide toward bankruptcy began in 2004, more than three years before plaintiffs filed suit. This evidence does not show, however, that the plaintiffs gained knowledge of the inadequate processes used by GreatBanc Trust to approve the Antioch buy-out more than three years before they filed this suit. Without undisputed proof of such knowledge of inadequate processes, we must reverse summary judgment for the defendants. a. The Proxy Materials The evidence here could support a finding that Duff & Phelps failed to perform an independent assessment because it simply accepted Antioch’s July 2003 assumptions regarding the company’s projected repurchase liability. Furthermore, Duff & Phelps’s work, largely concluded by October, was performed before the PPP agreement and the new distribution policy became key elements of the transaction. Plaintiffs have offered evidence that GreatBanc Trust then committed its own procedural error by relying unreasonably and uncritically on Duff & Phelps’s analysis to justify approval of the transaction. (Recall that we are reviewing a grant of summary judgment based on the statute of limitations, so we assume for now that plaintiffs will be able to prove these allegations on the merits.) Plaintiffs did not have actual knowledge of the violations they allege because the evidence does not show they had any indication that any of these procedural failures had occurred. GreatBanc Trust received four reports prepared by Duff & Phelps, including more than 100 pages of analysis prior to the mailing of the proxy materials, yet none of these were provided to the plaintiffs. Instead, the proxy materials said blandly that some analysis occurred resulting in Duff & Phelps’s determination the transaction was “fair.” The message to the plaintiffs, at least implicitly, was that the Plan trustee had used proper procedures and that the transaction was therefore not a prohibited transaction under § 1106. GreatBanc Trust also provided no explanation of its decision to rely on the financial advisor’s opinion. Without considerable insight into both Duff & Phelps’s analysis and GreatBanc Trust’s reasons for relying upon it, plaintiffs could not determine whether the buy-out transaction was supported by adequate consideration as required by §§ 1106(a) and 1108(e) or whether GreatBanc Trust acted prudently under § 1104. Plaintiffs knew almost no relevant facts, let alone the essential facts constituting the violations, see Consultants & Administrators, 966 F.2d at 1086, and thus could not have had actual knowledge of these alleged procedure-based breaches. Defendants point out that the proxy materials provided some information about risk, including the risks that ultimately doomed Antioch. Instead of highlighting the specific circumstances of the Antioch buy-out, however, the proxy materials simply provided a short list of the risks inherent in any highly-leveraged ESOP transaction. And while the materials explained that Antioch might struggle in the face of high repurchase obligations, they also did *683 not disclose a major substantive risk: that an inflated stock valuation might increase ESOP redemptions beyond the debt-burdened company’s ability to pay. See Armstrong v. LaSalle Bank, N.A., 446 F.3d 728, 731-32 (7th Cir.2006) (explaining a trustee’s duty to avoid a “run” of ESOP redemptions when a company faces a liquidity shortage). Yet even if the proxy materials had disclosed these substantive risks more fully, they would not have provided actual knowledge of the violations alleged in this case. Plaintiffs challenge not only the substantive prudence of the buy-out transaction but also the procedures used by GreatBanc Trust to assess the transaction. Moreover, the proxy materials did not even mention the PPP and new distribution plan in the risk disclosure, let alone that they increased the danger that a “stampede” of cash-outs would occur. Nor did they indicate whether and how Great-Banc Trust considered the possible impact on the Plan’s and employees’ long-term interests when negotiating these amendments. Because the proxy materials did not describe GreatBanc Trust’s methods, they could not have given plaintiffs actual knowledge of their claims based on its alleged failure to use sound processes in deciding whether to approve the buy-out transaction. b. The Stampede Begins Defendants also argue that Antioch’s noticeable decline began during 2004, the first year after the transaction, shown primarily by the large number of employees who resigned or retired to cash out, including seventy ESOP participants under the age of fifty. According to defendants, this unexpectedly high number of cash-outs provided actual knowledge of GreatBanc Trust’s alleged imprudence. After all, say defendants, those employees perceived the extent of Antioch’s troubles. We reject this argument as a basis for summary judgment. First, different employees were always likely to weigh differently the risks and benefits of the choice between quitting to benefit from the high stock price or staying with the company. More fundamental, the increase in cash-outs might have suggested to plaintiffs that “something was awry,” but again, neither inquiry notice nor constructive knowledge triggers the three-year limit of § 1113(2). E.g., Consultants & Administrators, 966 F.2d at 1086. Additional evidence indicates that Antioch’s overall performance in 2004 appeared to be strong based on the company’s annual report. Defendant Lee Morgan’s “Chairperson’s Letter” reported that 2004 had been a good year for the employee-owners, and he noted that the stock price had increased even after the unexpectedly high number of ESOP redemptions. Hindsight reveals that the increased rate of ESOP redemptions in 2004 was the first symptom of Antioch’s downward spiral after the transaction. But the limited evidence of the company’s decline then available to plaintiffs fell far short of providing actual knowledge that GreatBanc Trust had failed to use prudent processes to weigh the risks and benefits of the transaction. As plaintiff Hardman testified, she was aware of the increase in redemptions in 2004, but based on the information she had received, she thought there was “no reason to get concerned. It was so soon after the transaction that, you know, I felt that GreatBanc Trust had done their homework and this was all taken into consideration.” Her reasoning is consistent with the District of Columbia Circuit’s reasoning in Fink v. National Savings and Trust Co., *684 772 F.2d 951, 957 (D.C.Cir.1985) (reversing summary judgment based on three-year statute of limitations for process-based claims: “beneficiaries are entitled to assume that in performing these [fiduciary] acts, the fiduciaries thought about them. If this were not so, the lengthy list of fiduciary duties under ERISA would mean nothing more than caveat emptor. A fiduciary’s independent investigation of the merits of a particular investment is at the heart of the prudent person standard.”). We cannot fault Ms. Hardman or any other plaintiff for having faith in the independent trustee supposedly protecting the Plan participants’ interests. 3. Defendants’ Additional Arguments Defendants offer two additional arguments in support of summary judgment. First, they contend that plaintiffs received sufficient information but were “willfully blind” to it. Second, they contend that because the fiduciary defendants bear the burden of proving that they acted prudently and used sound processes to evaluate the transaction, information about the defendants’ processes was not an element of plaintiffs’ causes of action, so that their lack of knowledge did not prevent them from having “actual knowledge” of the alleged breaches. We reject these arguments. a. Willful Blindness The district court determined that defendants provided plaintiffs with sufficient information of the alleged breaches and that plaintiffs were “willfully blind” to that information such that they should be charged with actual knowledge. See Fish, 890 F.Supp.2d at 1065. Defendants urge that “willful blindness” is a basis for affir-mance because it is equivalent to actual knowledge. As explained, plaintiffs did not have access to materials sufficient to provide actual knowledge of the alleged process-based ERISA violations. Yet there is a more fundamental problem with reliance upon willful blindness to support summary judgment in a civil case. As the district court explained, willful blindness is a concept taken from criminal law and the often-given “ostrich” instruction. See, e.g., United States v. Garcia, 580 F.3d 528, 536-38 (7th Cir.2009); United States v. Ramsey, 785 F.2d 184, 190-91 (7th Cir.1986). Willful blindness is distinct from constructive knowledge (what a party “should have known”), negligence, or even reckless disregard for the facts. It implies a deliberate and conscious decision not to pursue the facts. The district court distinguished willful blindness from carelessness, but the Supreme Court has made clear in a civil case that the doctrine of willful blindness is considerably narrower. See Global-Tech Appliances, Inc. v. SEB S.A., — U.S. -, 131 S.Ct. 2060, 179 L.Ed.2d 1167 (2011). The Supreme Court took pains to distinguish willful blindness from negligence or even reckless or deliberate indifference toward the facts: While the Courts of Appeals articulate the doctrine of willful blindness in slightly different ways, all appear to agree on two basic requirements: (1) the defendant must subjectively believe that there is a high probability that a fact exists and (2) the defendant must take deliberate actions to avoid learning of that fact. We think these requirements give willful blindness an appropriately limited scope that surpasses recklessness and negligence. Under this formulation, a willfully blind defendant is one who takes deliberate actions to avoid confirming a high probability of wrongdoing and who can almost be said to have actually known the critical facts. See G. Williams, Criminal Law § 57, p. 159 (2d ed. 1961) (“A court can properly find *685 wilful blindness only where it can almost be said that the defendant actually knew”). By contrast, a reckless defendant is one who merely knows of a substantial and unjustified risk of such wrongdoing, see ALI, Model Penal Code § 2.02(2)(c) (1985), and a negligent defendant is one who should have known of a similar risk but, in fact, did not, see § 2.02(2)(d). 131 S.Ct. at 2070-71 (emphasis added; footnote omitted) (affirming jury finding that defendant was willfully blind to plaintiffs patent before beginning infringing conduct). If willful blindness has a place in the analysis of the “actual knowledge” three-year statute of limitations under § 1113(2) — a question we do not decide here — it would almost certainly present a genuine issue of material fact to be resolved by the finder of fact at trial. In criminal cases, the ostrich instruction on willful blindness describes an inference that a jury may make, not a rule of law that must be applied even where the party denies actual knowledge. See, e.g., United States v. Carrilo, 435 F.3d 767, 780-81 (7th Cir.2006) (describing issue in terms of what a jury may infer); Seventh Circuit Pattern Criminal Jury Instruction No. 4.10 (2012). 5 Consistent with these observations, we have said that “finding the line between ‘willful blindness’ and ‘reason to know may be like finding the horizon over Lake Michigan in a snowstorm.” Hard Rock Café Licensing Corp. v. Concession Services, Inc., 955 F.2d 1143, 1151 n. 5 (7th Cir.1992); see also Consultants & Administrators, 966 F.2d at 1086 (“in cases near the border the distinction may well be nearly semantic”). In other words, only rarely could that line be drawn as a matter of law. 6 Accordingly, even if we assume willful blindness is relevant under the actual knowledge standard of § 1113(2), and even if defendants had made the relevant information available to the plaintiffs, the willful blindness theory would not be a sufficient basis for summary judgment here. b. The Burden of Proof for Prohibited Transactions Defendants also argue that the burden of proof regarding whether a fiduciary used appropriate processes and exchanged property for adequate consideration means that plaintiffs’ claims are time-barred. Under ERISA, the burden of proof is on a defendant to show that a transaction that is otherwise prohibited under § 1106 qualifies for an exemption under § 1108. See, e.g., Keach, 419 F.3d at 636; accord, e.g., Harris v. Amgen, Inc., 738 F.3d 1026, 1045 (9th Cir.2013), petition for cert. filed on other grounds, (Jan. 10, 2014). We have applied the same burden of proof to the adequacy of a fiduciary’s investigation and processes under the more general fiduciary duty in § 1104. Eyler v. Comm’r, 88 F.3d 445, 455 (7th Cir.1996), citing Donovan v. Cunningham, 716 F.2d 1455, 1467-68 (5th Cir.1983). ERISA plans engage in transactions nomi *686 nally prohibited by § 1106 all the time, while also taking steps to comply with ERISA by relying on one or more of the many exceptions under § 1108. The burden of proof makes good sense as a policy matter because the fiduciary will ordinarily have the information needed to know whether an exception applies under § 1108. Defendants reason that because they have the burden of proving that they used appropriate processes to determine fairness and fair market value in the Antioch buy-out, their use of appropriate processes is an affirmative defense rather than an element of the plaintiffs’ case. Plaintiffs therefore did not need knowledge of inadequate processes, defendants argue, to have “actual knowledge” of the alleged breaches. Defendants’ argument is clever, but it’s not supported by case authority. It’s also not realistic. First, defendants’ theory runs directly contrary to the Fifth Circuit’s decision in another case of process-based fiduciary duty claims, Maher v. Strachan Shipping Co., 68 F.3d at 956, which they urge us to reject. In Maher, the defendants had used retirement plan assets to buy a single premium annuity contract to pay for benefits. The transaction allowed the defendants to return millions in cash from the retirement plan to the company, but the company that sold the annuity later went into conservatorship and retirement payments were cut substantially. Like the defendants in this case, the defendants in Maher asserted that the plaintiffs had actual knowledge of the alleged breach of fiduciary duty when they learned of the transaction — in Maher the purchase of the annuity from a shaky seller, and here the highly leveraged buyout — and in both cases the district courts granted summary judgment to the defendants. The Fifth Circuit reversed in Maher. The plaintiffs’ knowledge that the seller of the annuity seemed financially shaky indicated that something might be awry, but that did not amount to actual knowledge of the breach. Rather, the plaintiffs were challenging the selection of the seller, and “they must have been aware of the process utilized by [the employer] in order to have had actual knowledge of the resulting breach of fiduciary duty.” 68 F.3d at 956, citing Donovan v. Cunningham, 716 F.2d at 1467. We agree with that conclusion, which is also consistent with Waller v. Blue Cross of California, 32 F.3d 1337, 1341 (9th Cir.1994) (holding that three-year clock did not begin to run on process-based claims under §§ 1104 and 1106 at time plaintiffs learned of purchase of annuities from shaky seller). Whether the transaction here was prohibited depends on the extent of the fiduciaries’ processes used to evaluate it.' Plaintiffs did not know about the alleged inadequacy of those processes more than three years before they filed suit. Both sides cite the Eighth Circuit’s decision in Brown v. American Life Holdings, Inc., 190 F.3d 856 (8th Cir.1999), on this issue. The plaintiff in Brown alleged that fiduciaries breached their duties by investing plan assets too conservatively. The key passage in the opinion said: Therefore, when a fiduciary’s investment decision is challenged as a breach of an ERISA duty, the nature of the alleged breach is critical to the actual knowledge issue. For example, if the fiduciary made an illegal investment — in ERISA terminology, engaged in a prohibited transaction — knowledge of the transaction would be actual knowledge of the breach. But if the fiduciary made an imprudent investment, actual knowledge of the breach would usually require some knowledge of how the fiduciary *687 selected the investment. See Maher v. Strachan Shipping Co., 68 F.3d 951. 955-56 (5th Cir.1995), and cases cited. 190 F.3d at 859 (emphasis in original). The Eighth Circuit ultimately affirmed summary judgment for the defendants, concluding that the plaintiff knew of the alleged failure to diversify investments at the time the transactions were disclosed to him, and finding that the plaintiff had failed to articulate clearly a process-based claim. Id. at 860. We agree with the broad language in Brown, but it does not answer the questions before us, which depend on whether the Antioch buy-out was in fact a prohibited transaction or an imprudent transaction, both of which depend in turn on the processes used by the defendants to approve the buy-out. We agree with Maher and Waller because their analysis fits most comfortably within the overall statutory framework of ERISA as well as the text of § 1113. Under defendants’ unrealistic theory, plaintiffs could have and even should have filed suit immediately after the 2003 buy-out took place, without undertaking any investigation of the affirmative defense that the defendants themselves were invoking at the time. We doubt it would have been prudent or even responsible for plaintiffs to have filed suit at the time, knowing only (a) that the transaction was prohibited under § 1106 unless § 1108 applied, and (b) that defendants claimed it did apply. Consider the situation the plaintiffs faced back in 2003 and 2004. The defendants disclosed to plaintiffs their intention to go forward with a transaction nominally prohibited under § 1106. The disclosures also assured the plaintiffs that the defendants were taking steps to make sure the transaction was for adequate consideration and would be approved only after appropriate procedures had been used to evaluate the fairness of the transaction and the adequacy of the consideration. In other words, defendants were telling the plaintiffs that the transaction was protected under § 1108. That is not providing actual knowledge of a violation or breach of fiduciary duty. In rejecting defendants’ argument, we are well aware of the distinction in civil procedure between the elements of a plaintiffs claims and an affirmative defense. That distinction does not extend to the “actual knowledge” standard under § 1113 when a defendant invokes an exception under § 1108. In deciding whether sanctions should be imposed on plaintiffs who filed unfounded cases, we have said that plaintiffs and their attorneys “may have a responsibility to examine whether any obvious affirmative defenses bar the case.” Matter of Excello Press, Inc., 967 F.2d 1109, 1113 (7th Cir.1992) (finding that Federal Rule of Bankruptcy Procedure 9011 (parallel to Federal Rule of Civil Procedure 11) could impose a duty to investigate an obvious “ordinary course of business” defense, but reversing sanctions) (emphasis in original; quotation omitted); see also Bethesda Lutheran Homes and Services, Inc. v. Born, 238 F.3d 853, 858 (7th Cir.2001) (reversing denial of sanctions as abuse of discretion where claim was barred by affirmative defense of claim preclusion); White v. General Motors Corp., 908 F.2d 675, 682 (10th Cir.1990) (affirming sanctions based on obvious affirmative defense of release); McLaughlin v. Bradlee, 803 F.2d 1197, 1205 (D.C.Cir.1986) (affirming sanctions where plaintiff failed to anticipate affirmative defense of issue preclusion). In the case of an ERISA plan that invokes a § 1108 exception to a § 1106 prohibition, the plaintiff does not have actual knowledge of an alleged violation until she knows that the exception does not apply. These plaintiffs did not have actual *688 knowledge of the violations based on the information defendants provided. That information claimed that defendants had been prudent, had used appropriate procedures to evaluate the Antioch buy-out transaction, and had concluded that the consideration would be adequate. To the extent defendants argue that this approach extends the limitations period too long, the response is that the six-year limit in § 1113(1) remains applicable to protect defendants from stale claims. 7 C. Remaining Issues on Appeal Defendants are not entitled to summary judgment based on § 1113(2). Because we are reinstating all claims, we address the remaining issues only briefly. Plaintiffs challenge the dismissal of later-added plaintiff Evolve Bank, which on January 16, 2008 became the final trustee of the Plan. Using as a launching pad a footnote by the district judge calling Evolve Bank’s addition a “manipulative tactic,” see Fish v. GreatBanc Trust Co., 830 F.Supp.2d 426, 430 n. 7 (N.D.Ill.2010), the parties have debated whether plaintiffs’ knowledge should be imputed to Evolve Bank. The district court stated that “manipulation could be effected by replacement of knowing fiduciaries with new fiduciaries without actual knowledge,” id,., but it was defendants’ choice to make Evolve Bank a new fiduciary in this case. We agree with the Ninth Circuit’s reasoning in Landwehr v. DuPree, 72 F.3d 726, 732 (9th Cir.1995), that knowledge should not be imputed from one party to another for purposes of the “actual knowledge” standard under § 1113(2). Defendants complain that allowing a new fiduciary to avoid the three-year statute of limitations would undermine the statute, but again, defendants would still be protected by the six-year limit of § 1113(1). Plaintiffs also appeal the district court’s dismissal of their alternative claim for defendants’ failure to sue themselves for their own breaches of fiduciary duty. We recognized such a theory could be viable in Consultants & Administrators, 966 F.2d at 1089-90. The theory seems in this case to be only a backstop theory if plaintiffs were to lose under § 1113(2) based on their own knowledge and the dismissal of Evolve Bank. We are reversing summary judgment on both of those grounds. The district court did not explain its reasons for dismissing this alternative failure-to-sue theory. We vacate that dismissal as well, and leave it to plaintiffs to decide whether they wish to continue to pursue the alternative theory on remand. If they do so, we leave it to the district court to evaluate the theory. Plaintiffs also raise a number of objections to the district court’s unusual procedure in granting the defendants’ second motion for summary judgment. The district court limited briefing to just a couple of issues and then proceeded to grant the motion without further briefing, concluding that plaintiffs had been given ample opportunities to present all of their evidence and legal arguments. Because we are reversing on all claims, we need not address these issues. Plaintiffs will have a chance to litigate all issues anew upon remand. We add that under the circumstances here, plaintiffs should be permitted to amend their complaint based on the four intervening years of litigation and the discovery they undertook after first amending their complaint, see Foman v. Davis, 371 U.S. 178, 182, 83 S.Ct. 227, 9 L.Ed.2d 222 *689 (1962) (emphasizing that Rule 15(a)’s command that leave to amend “shall be freely given” should be followed unless apparent interest weighs against amendment, such as undue delay, bad faith, or futility); Barry Aviation, Inc. v. Land O’Lakes Municipal Airport, 377 F.3d 682, 689-90 (7th Cir.2004) (reversing denial of leave to amend), and discovery should no longer be restricted to statute of limitations issues. Finally, we deny plaintiffs’ separate motion for reassignment pursuant to Seventh Circuit Rule 36. We are confident that upon remand the issues will be considered fairly. Because the plaintiffs did not have actual knowledge of the alleged ERISA violations, defendants’ motion for summary judgment should have been denied. We REVERSE the judgment of the district court and REMAND the case for further proceedings consistent with this opinion. Footnotes [1] Citations to ERISA in this opinion are to the sections as codified in Title 29 of the United States Code rather than to the sections in the ERISA legislation as enacted. [2] Antioch's collapse highlights a risk of employee stock ownership plans, especially *678 when an ESOP is a major shareholder of a corporation whose stock is not publicly traded, such as Antioch. Without an efficient market for the stock, the proper valuation of stock for purposes of paying employees who retire or leave the company becomes critical for the company’s financial viability. "If the price is set too low, employees who leave will feel shortchanged. If it is set too high it may precipitate so many departures that it endangers the firm’s solvency.” Armstrong v. LaSalle Bank, N.A., 446 F.3d 728, 730 (7th Cir.2006). The latter prospect can produce an accelerating stampede — initially to take advantage of the high price, but eventually to leave before the company folds under the growing demand for cash payments. [3] Most statutes of limitations run from the time a claim accrues, and the reference to actual knowledge of a violation in § 1113(2) is exceptional. Cf. Cada v. Baxter Healthcare Corp., 920 F.2d 446, 450 (7th Cir.1990) (explaining that limitations period for age discrimination claim starts when claim accrues, meaning when plaintiff discovers he has been injured); 15 U.S.C. § 15b (Sherman Act claims barred “unless commenced within four years after the cause of action accrued”); 15 U.S.C. § 77m (various limitation periods under Securities Act of 1933 based on time “after the discovery of the untrue statement or the omission, or after such discovery should have been made by the exercise of reasonable diligence,” or "after the violation upon which it is based”). [4] Even for a substance-based claim, the terms of a transaction alone would only rarely provide actual knowledge under § 1113(2) since either an expert opinion or actual harm would likely be necessary before an ESOP participant could know of the flaws in the substance of a fiduciary's decision when only the bare terms were disclosed. See Gluck v. Unisys Corp., 960 F.2d 1168, 1177 (3d Cir.1992); see also Brown v. Owens Corning Inv. Review Comm., 622 F.3d 564, 573 (6th Cir.2010) (charging plaintiffs with actual knowledge once allegedly imprudent investment “had lost almost all value”). [5] We do not address issues here about how Global-Tech Appliances may apply to the exact wording of criminal jury instructions about knowledge, as discussed in the committee comments to Pattern Instruction No. 4.10. [6] Global-Tech Appliances illustrates the point. The plaintiff showed that the defendant had deliberately copied a foreign model of its product that did not bear notice of U.S. patents, and then obtained an opinion of nonin-fringement from a U.S. lawyer whom it did not tell it had copied plaintiff's product. 131 S.Ct. at 2064. The Court found this evidence sufficient to support a jury finding of willful blindness and thus actual knowledge, but did not say such a finding was required as a matter of law. Id. at 2071-72. [7] We leave for another day questions that might be raised concerning the scope of a plaintiff's duty to investigate under Federal Rule of Civil Procedure 11 when contemplating a suit based on 29 U.S.C. § 1106.