Thompson v. Retirement Plan for Employees of S.C. Johnson & Son, Inc.

CUDAHY, Circuit Judge. The plaintiffs, former members of the S.C. Johnson and JohnsonDiversey cash balance pension plans, appeal from an order of the district court dismissing some of their claims as untimely. They also appeal from the district court’s method for calculating the plaintiffs’ recovery. The Plan defendants cross-appeal, contending that all the plaintiffs’ claims are untimely, and also taking issue with the district court’s damages calculation method. For the reasons that follow, we affirm the district court in most respects but reverse in part and remand for it to reconsider the method of damages calculation. I. Background A. Facts In 1998 S.C. Johnson & Son amended its ERISA plan, converting it from a traditional defined benefit plan into a “cash balance” plan. Cash balance plans are formally classified as defined benefit plans, but they function more like defined contribution plans, in particular by providing an account balance for each participant. But *602 a cash balance plan participant’s balance is only a “notional” tool for estimating pension benefits — not an actual account containing money. As amended, the S.C. Johnson Plan provided that each participant’s notional account balance would be increased by annual “interest credits.” The Plan calculated interest at the greater of 4%, or 75% of the Plan’s rate of return on its investments. Further, the Plan provided that, if a participant left the Plan before reaching age 65, the participant could take a lump-sum distribution of the value of the account. However, the provisions of the Plan ensured that any lump-sum distribution would be only the current account balance. No upward adjustment would be made for the future interest credits the participant would earn by staying in the Plan. The ERISA statute has something to say about early lump-sum distributions: they must be the “actuarial equivalent” of the value of the account at age 65. 29 U.S.C. § 1054(e)(3); see Berger v. Xerox Corp. Ret. Income Guar. Plan, 338 F.3d 755, 759 (7th Cir.2003). The drafters of the present Plans were obviously aware of this rule, because they included § 5.2, which states: The Cash Balance Account is the Actuarial Equivalent of the projected annuity at normal retirement because the Plan deems the return on 30 year Treasuries to be the reasonable rate of return to assume for purposes of that projection .... This section created a wash calculation designed to add zero interest to lump-sum distributions. This is because during the relevant period ERISA prescribed that, when calculating the present value of lump-sum distributions, plans should use the 30-year Treasury rate as the discount rate. 1 So if a participant was leaving at age 40, the Plan would calculate interest out to age 65 at the 30-year Treasury rate, as prescribed in § 5.2 of the Plan — then discount it back to age 40 at the exact same rate, as prescribed by the statute. The participant would therefore receive as a net amount only his current account balance (without future interest). This provision was concededly unlawful. The 30-year Treasury rate, despite the Plan’s ipse dixit, did not produce the “actuarial equivalent” of what the Plan provided to ongoing participants — interest calculated at the greater of 4% or 75% of the Plan’s rate of return. The Plans effectively penalized lump-sum distributees by voiding their future interest credits, and this violated ERISA. See Berger, 338 F.3d at 761; Esden v. Bank of Boston, 229 F.3d 154, 168 (2d Cir.2000). B. Procedural History The plaintiffs, participants in the S.C. Johnson Plan 2 who received lump-sum dis *603 tributions, filed this suit in the Eastern District of Wisconsin on November 27, 2007. The Plan defendants moved to dismiss on the ground that the plaintiffs had not exhausted the internal Plan remedies, but the district court denied the motion in November of 2008. Sometime thereafter, the Plans conceded the unlawfulness of the lump-sum provisions. The parties filed cross-motions for summary judgment. The Plan defendants argued inter alia that the plaintiffs’ claims were time-barred. In March of 2010, the district court partially resolved the summary judgment motions. At the outset, the court held that the applicable statute of limitations was Wisconsin’s six-year contract limitations period. Wis. Stat. § 893.43. Next, the court had to determine when the plaintiffs’ claims accrued, a determination governed by federal law. See Young v. Verizon’s Bell Atl. Cash Balance Plan, 615 F.3d 808, 816 (7th Cir.2010). The court was faced with three possible accrual dates. First, it could hold that the claims accrued in 1998 or 1999, when the Plans distributed to participants SPDs and other informational material about the new cash balance Plan. Second, the court could hold that the claims accrued at the time the plaintiffs received their deficient lump-sum distributions. Third, it could hold that even the receipt of the lump-sum distributions did not start the limitations period, and so the plaintiffs’ claims accrued at some later, unspecified time. Wisely, the court had divided the plaintiffs into two subclasses: subclass A, plaintiffs who had received their lump-sum distributions after November 27, 2001 (six years prior to filing suit), and subclass B, plaintiffs who had received their lump-sum distributions before November 27, 2001. The court held that the claims accrued when the plaintiffs received their lump-sum distributions; therefore, the subclass A plaintiffs were timely and the subclass B plaintiffs were untimely. Since the Plan had admitted its wash calculation was unlawful, the court next had to consider subclass A’s recovery. The subclass A plaintiffs were entitled to the value, at the time of their lump-sum distributions, of future interest payments of 4% or 75% of the Plan’s investment return rate. Of course, there is no formula to capture that value conclusively since there is no way of knowing ex ante what the Plan’s annual investment returns will be. Both the plaintiffs and the Plan defendants asked for summary judgment in favor of their proposed method of calculating the wrongly deprived future interest credits. But the district court did not select a method. Instead, it “order[ed] that [the Plans] recalculate lump sum distributions pursuant to the requirements of the law.” The court further provided that if the parties “are unable to reach an agreement ... they remain free to resubmit the issue to the court....” The court relied on Durand v. Hanover Ins. Group, Inc., 560 F.3d 436, 442 (6th Cir.2009), as authority for delegating the recovery issue to the parties. Unsurprisingly, the parties could not agree on a method of calculating future interest credits. The district court received further briefing on how to calculate the recovery, 3 and issued an order resolving the matter in August of 2010. In so doing, the court credited the Plan defendants’ contention that they were entitled to deference in choosing the appropriate method. The court stated in relevant part, *604 “[t]he court believes that Corikrigkft 4 supports referral of the interest crediting rate question to the Plans and compels a grant of deference to the Plans’ proposed method for recalculating lump sum distributions.” Then, the court selected a modified version of the Plan defendants’ second proposed method, which provided that future interest should be calculated by using the Plans’ average return rate over the five years leading up to each participant’s lump-sum distribution. The plaintiffs timely appealed, and the Plans timely cross-appealed. We perceive the questions presented in this appeal to be as follows: 1) Timeliness a) Did the plaintiffs’ injuries accrue when the Plan circulated information about the conversion to a cash balance plan in 1998 and 1999? b) If not, did the plaintiffs’ injuries accrue when they received lump-sum distributions affected by the unlawful Plan provision? 2) The Plaintiffs’ Recovery a) Should the district court determine how future interest credits should be valued for purposes of determining subclass A’s recovery, or should the parties? b) Is the Plan defendants’ proposed method of calculating future interest credits entitled to deference? c) How should the plaintiffs’ future interest credits be calculated? II. Analysis A. Statute of Limitations Neither party is completely satisfied with the district court’s statute of limitations ruling. The Plan defendants argue that the court was correct to rule that subclass B was untimely, since their lump-sum distributions occurred over six years before they filed their complaint. But the Plans would have us go farther and find subclass A untimely as well, because they believe all participants were informed of the relevant Plan provisions in 1999 (and the lawsuit was filed after 2005). The plaintiffs argue that the court was correct to rule that subclass A was timely, and that subclass B should also have been treated as timely because the lump-sum distributions did not start the statute of limitations clock. As the district court appreciated, accrual of ERISA claims is governed by federal law, although the statute of limitations itself is borrowed from state law. See Young, 615 F.3d at 816. We have held that “[t]he general federal common law rule is that an ERISA claim accrues when the plaintiff knows or should know of conduct that interferes with the plaintiffs ERISA rights.” Id. at 817. Further, “a claim to recover benefits under § 502(a) accrues ‘upon a clear and unequivocal repudiation of rights under the pension plan which has been made known to the beneficiary.’ ” Id. (quoting Daill v. Sheet Metal Workers’ Local 73 Pension Fund, 100 F.3d 62, 65 (7th Cir.1996)). 5 *605 We affirm the district court’s rulings on the statute of limitations questions. Although it is a very close question, we disagree with the Plan defendants that certain SPDs and informational material distributed to Plan participants in 1998 and 1999 amounted to an “unequivocal repudiation” sufficient to trigger the statute of limitations. The record contains several Plan informational communications circulated around 1999 that touch on lump-sum distributions. The 1999 SPD stated, “[y]ou can choose from several payment options including a lump sum payment and several types of ... annuities. You can also choose to leave your money in the plan and continue to earn investment credits.” The SPD then described the lump-sum payment option as follows: “[t]he entire value of your account is paid in one payment. No further pension benefit will be available from the Company.” It later stated, “[y]ou can ... choose to leave your money in your retirement account after you leave the Company up until age 65. Each year, your Cash Balance Account will receive an investment credit.” We think that these SPD statements were inadequate to convey the crucial defect in the Plans: that early lump-sum distributions would not be increased to reflect the present value of future interest credits continuing to age 65. True, the 1999 SPD told participants that they would not earn investment credits if they left the Plan. But that would have been true even with a properly-functioning plan, because such a plan would incorporate the projected value of future interest payments into the lump sum and then discontinue those interest payments. And stating that “[t]he entire value of your account [will be] paid in one payment” does not elucidate how that value is decided. The Plans’ “Investing in You” newsletters also touched on lump-sum distributions, although like the SPDs they usually left considerable room for uncertainty. For instance, the statement that participants could think of their cash balance account as “much like a savings account” could be read to suggest that lump-sum distributions would be in the amount of the account balance only. But this was a generic comparison to illustrate the difference between the company’s traditional pension plan and the new cash balance plan. There was no explanation of how far the analogy carried, and a participant obviously would have been incorrect to assume that the cash balance plan was like a bank account in every respect. It is true that several “Investing in You” newsletters stated that a lump-sum distribution would be in the amount of the cash balance account. We conclude, although not without some difficulty, that this too was inadequate to initiate the limitations period, if only because these newsletters were obviously meant to be a simplified explanation of a transition from one complicated plan structure to another. In the context in which it appeared, this incidental statement would not likely alert a participant that he was being deprived of something to which he might be entitled. We do not assume that a participant would have understood it to have a legal effect or to mean that the terms of the Plan itself unlawfully omitted the obscure future interest right. And a participant familiar with the right to future interest might still assume that the future interest credits would somehow be incorporated in the account balance. In all, we think these SPD and newsletter statements are a collection of hints. They are assembled in one place for purposes of this litigation, but from the perspective of Plan participants they ap *606 peared among the pages of various circulations distributed over the course of months. Although it is certainly possible that generic Plan communications can prospectively repudiate unequivocally participant rights, see, e.g., Winnett v. Caterpillar, Inc., 609 F.3d 404, 410 (6th Cir.2010), 6 we note that the more traditional case in which recovery is barred involves some direct communication to a participant who is actually pressing the issue. 7 In the present case, given the relative obscurity of the right at issue, the fact that most of the Plans’ references to lump-sum distributions offered only oblique guidance about the crucial flaw at issue here, and the fact that the most illuminating statements were found in informal Plan newsletters as opposed to the more legally weighty SPDs, we think that there has been no clear and unequivocal repudiation of the participants’ rights to future interest credits under ERISA. Accordingly, the district court properly held that the statute of limitations period did not begin in 1998 or 1999. We further agree with the district court that when the participants received their lump-sum distributions, this served as an unequivocal repudiation of any entitlement to benefits beyond the account balance. As noted above, informational circulars confirmed that after a lump-sum distribution, no additional benefits would be forthcoming. Given that the distributions were calculated consistent with the Plan document and every Plan communication, the lump-sum distributions served as the final step of a clear repudiation of the participants’ entitlement to anything different. We specifically reject the plaintiffs’ argument, in support of the thesis that the lump-sum distributions did not start the running of the statute of limitations, that they could not have understood their injury without seeing the full Plan document. Contrary to the plaintiffs’ argument, the Plan defendants did not improperly conceal the wash calculation in the Plan document; they never mentioned it to the participants because it was designed to have no effect. Moreover, the plaintiffs did not need to see the wash calculation language in the Plan to understand that they had received their account balance and nothing more. Beginning in 2003 the denial of future interest credits was unlawful under squarely applicable precedent from this court. See Berger, 338 F.3d at 763. 8 *607 The plaintiffs point to this court’s 2010 opinion in Young for the proposition that a lump-sum distribution does not start the statute of limitations clock. In Young, we stated, Verizon argues that Young’s claim accrued ... when she received her lump-sum benefit.... At that time, however, the parties’ dispute over the correct interpretation of the Plan had not developed. And nothing suggests that the $286,095 payment that Young received should have been a red flag that she was underpaid.... The 1998 payment that Young received was not so inconsistent with her current claim for additional benefits as to serve as a clear repudiation. Young, 615 F.3d at 816. But Young does not control this case for two reasons. First, the right that the lump-sum distribution needed to “clearly repudiate” was very different in Young. In that case, the trustees were ignoring a scrivener’s error in the Plan document and distributing lump sums that were smaller than the Plan literally prescribed. Id. at 814. Thus, the lump-sum distribution did not place Young on notice that the Plan was ignoring one factor in a complex formula in the plan document. Here, in contrast, the lump-sum distribution merely needed to show that participants would receive their account balance and no more. That simple fact is what made the Plans unlawful. The second reason Young is not controlling is that unlike the present plaintiffs, the plaintiff in Young exhausted the plan’s internal remedies. Id. at 814. She thereby furnished an alternative accrual date: the date the plan finally denied her claim. Young persuaded us that her lump-sum distribution did not alert her to her injury (for the reasons detailed above), so we gave her the benefit of the later accrual date. Id. at 816. In view of those facts, the present plaintiffs are not really asking to be treated like the Young plaintiff at all. They have been given a pass on exhausting their internal remedies, and they now invite us to extend Young by allowing them to slip by with no accrual date. 9 We will not thereby approve nullification of the statute of limitations. Hence, we reject both the plaintiffs’ and the defendants’ objections to the district court’s conclusions on the limitations issue. B. Who Chooses the Method of Calculating Subclass A’s Recovery The Plan defendants argue that they are entitled to select the means of calculating subclass A’s recovery. They point to a host of cases supporting the notion that' ERISA fiduciaries are entitled to deference in their administration of the plan. In particular, they cite Conkright v. Frommert, — U.S.-, 130 S.Ct. 1640, 176 L.Ed.2d 469 (2010), for the proposition that their interpretations of the Plan are entitled to a deference that survives despite an initial impermissible interpretation. Briefly, in Conkright the Supreme Court reiterated the policy, most prominently articulated in Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989), of deferring to Plan fiduciaries when they are interpreting Plan terms. The Court clarified that ERISA fiduciaries are not stripped of deference because of an initial *608 improper interpretation; they do not labor under a “one-strike-and-you’re-out” regime. Id. at 1646. Accordingly, the Plan defendants argue, the district court should have selected their first proposal (the “spread method”), and alternatively, should not have modified their second method (the “five-year average”) before adopting it. We do not credit the defendants’ argument that they are owed deference here, because they did not make a discretionary decision entitled to deference, and indeed they could not have. The reliance on Conkright is inapt because the issue here is not interpretation, and “Firestone is limited to questions of plan interpretation. ...” Fletcher v. Kroger Co., 942 F.2d 1137, 1139 (7th Cir.1991). The doctrine “does not bring design decisions within ERISA.” Belade v. ITT Corp., 909 F.2d 736, 738 (2d Cir.1990); see also Rasenack v. AIG Life Ins. Co., 585 F.3d 1311, 1315 (10th Cir.2009) (“Under trust principles, a deferential standard of review is appropriate when trustees actually exercise a discretionary power ‘vested in them by the instrument under which they act.’ ”) (citation omitted). Here, the Plan documents made a general grant of discretion to the Plan administrators in § 11.3, but did not give them discretion to amend the Plan terms; the power to amend was reserved by the company in § 14.1. Cf. Brumm v. Bert Bell NFL Retirement Plan, 995 F.2d 1433, 1437 (8th Cir.1993) (describing a plan that gave the administrators “discretionary power ‘to define and amend the terms of the Plan and Trust....’”). Moreover, the Plans’ generalized grant of interpretive discretion did not authorize the administrators to controvert the clear terms of the Plan. See Marrs v. Motorola, Inc., 577 F.3d 783, 786 (7th Cir.2009) (“The administrator is not by virtue of such a grant of authority free to disregard unambiguous language in the plan....”); Call v. Ameritech Mgmt. Pension Plan, 475 F.3d 816, 822-23 (7th Cir.2007) (“[Unambiguous terms of a pension plan'leave no room for the exercise of interpretive discretion by the plan’s administrator.... ”); Cozzie v. Metropolitan Life Ins. Co., 140 F.3d 1104, 1108 (7th Cir.1998) (“[E]ven [given a broad grant of discretion], the [administrator] is bound by the terms of the document. Interpretation and modification are different; the power to do the first does not imply the power to do the second.”). These Plans did not give the administrators any discretion in how to calculate future interest for lump-sum distributees because the unlawful “wash” calculation was effectively codified in the Plans. 10 Section 5.2 declared (incorrectly) that the cash balance account value equaled the “actuarial equivalent” of the value at age 65. And under § 5.5(e), that “actuarial equivalent” was to be the amount of any lump-sum distribution. To further drive the point home, the Plan “deemed” the 30-year treasury rate to be “the reasonable rate of return ... for purposes of [the actuarial equivalency] projection.” §-5.2 (emphasis added). And the definitions section, § 2.1, also defined the “actuarial equivalent” by reference to the 30-year treasury, and linked this value to the lump-sum distribution amount. Thus, projecting future interest credits with a differ *609 ent rate would have been an abandonment, not an interpretation, of the Plans’ terms. Such a projection would presumably have violated § 11.2 of the Plan, which forbade the plan administrator from adopting rules “contrary to the specific provisions of the Plan.” 11 Accordingly, Plan defendants applied the wash calculation ministerially. The Plan’s prescriptive approach to calculating lump-sum future interest credits, although now likely regretted, was consistent with a controlling IRS regulation. IRS Notice 96 — 8, the authority of which we have recognized, 12 specifically requires (1) that a cash balance plan dictate a projection method; and (2) that the prescribed method must not allow discretion. The Notice states in relevant part, A frontloaded interest credit plan[ 13 ] that specifies a variable outside index for use in determining the amount of interest credits must prescribe the method for reflecting future interest credits in the calculation of an employee’s accrued benefit. In order to comply with [I.R.C.] section 401(a)(25), the method, including actuarial assumptions, if applicable, must preclude■ employer discretion. IRS Notice 96-8, “Cash Balance Pension Plans,” 1996-1 C.B. 359 (Feb. 5, 1996) (emphasis added). These restrictions allow a cash balance plan to comply with the requirement in I.R.C. § 401(a)(25) 14 that plan benefits be “definitely determinable.” Esden, 229 F.3d at 166. In sum, the Plan defendants did not exercise interpretive discretion over the projection rate for calculating future interest credits. Nor did the Plan terms permit such interpretation. Therefore, this is not case about the fiduciaries’ construal of the Plan, and the Supreme Court’s Firestone and Conkright decisions have little authoritative to say. Especially given the IRS Notice, we are loath to convert this into a matter of Plan discretion for the first time in connection with calculating damages for participants who have long since left the Plans. 15 In hindsight, it seems that the defendants had no way to escape being accountable for the unlawful wash calculation once it was codified in the Plans (other than perhaps procuring an *610 amendment of the Plan terms from the employer). This was an unfortunate predicament for the Plan defendants, but that does not mean they are now entitled to deference in calculating the plaintiffs’ posthoc recovery. Someone, however, must choose a method for making the inherently uncertain estimate this case requires. As no ERISA-specific exception applies, the district court should assume its accustomed responsibility for calculating the plaintiffs’ recovery. That has been the prevailing practice in comparable cases. See Esden, 229 F.3d at 177 (“It shall be for the district court in the first instance to determine the proper projection rate for the calculation of damages----”); West v. AK Steel Corp. Ret. Accumulation Pension Plan, No. l:02-cv-0001, 2005 WL 3465637, at *2, 2005 U.S. Dist. LEXIS 37863, at *5-6 (S.D.Ohio Dec. 19, 2005) (“While the financial impact is evident and not trivial, the Court again rejects Defendants’ argument that the 30-year Treasury rate should be used.”), aff'd, 484 F.3d 395 (6th Cir.2007); Berger v. Xerox Ret. Income Guar. Plan, 231 F.Supp.2d 804, 820 (S.D.Ill.2002) (“As to Defendants’ contention that the Court should refer this case to the ... administrator to determine the proper rates for calculating benefits for ... participants, the Court rejects this position.”), aff'd as modified, 338 F.3d 755 (7th Cir.2003) 16 ; Lyons v. Georgia-Pacific Corp. Salaried Emples. Ret. Plan, 196 F.Supp.2d 1260, 1266 (N.D.Ga.2002) (“[Biased on the Eleventh Circuit’s opinion [in the same litigation] ... the court concludes that [a specified rate] is the appropriate interest crediting rate.”); but see Durand, 560 F.3d at 442 (stating that the district court on remand “could simply award injunctive relief that requires [the Plan defendant], in the first instance, to do what the law requires.”). The Plans are undoubtedly correct that determining how to measure future interest presents special challenges, but estimating damages is often a speculative, counterfactual inquiry. The district court is practiced in this discipline and is equal to the task in this case. C. Method of Calculating Future Interest Credits on Remand In view of the above conclusions, we believe the best procedure is to reverse the district court to the extent that it held that some deference was owed to the Plan defendants’ preferred calculation method. We cannot be certain that this belief did not inform the court’s selection of the “five-year average” approach. After all, this methodology originated with the Plan defendants. 17 For the foregoing reasons, we Affirm in part, Reverse in part and Remand for *611 proceedings 18 consistent with this opinion. Footnotes [1] I.R.C. § 417(e)(3)(A)(ii)(II) (2000). This statutory prescription was in effect from 1994 to 2006. Prior to 1994, the federal Pension Benefit Guaranty Corporation set the appropriate discount rate. See Tax Reform Act of 1986, Pub.L. No. 99-514, § 1139(b), 100 Stat. 2085, 2487 (1986). Effective in 2007, this entire calculation was no longer required because Congress sanctioned lump-sum distributions at no more than the cash balance account value. See Pension Protection Act of 2006, Pub.L. No. 109-280, § 701(a)(2), 120 Stat. 780, 920, 984 (2006). [2] There are actually two Plan defendants (and two Plans) here: The S.C. Johnson Plan, and the JohnsonDiversey Plan. The JohnsonDiversey Plan split from the S.C. Johnson Plan effective January 1, 1999 — seven months after the transition to a cash balance plan. The relevant provisions of the two plans are substantively identical. Dates relevant to the statute of limitations are different, but insignificantly so. To avoid repetition, we refer throughout to the dates that apply to the S.C. Johnson Plan. This opinion applies to all parties and resolves the entire appeal. [3] The parties have supplied detailed briefing on several methods of calculating future interest credits, but our holding makes it unnecessary to discuss them in depth. [4] Conkright v. Frommert, -U.S. -, 130 S.Ct. 1640, 176 L.Ed.2d 469 (2010). [5] Our accrual analysis differs from the district court's because the district court relied on language that derives from cases considering violations of ERISA § 510, "Interference with protected rights.” See Tolle v. Carroll Touch, Inc., 977 F.2d 1129, 1139 (7th Cir.1992). The reason we formulated an independent accrual framework for § 510 claims in Tolle was precisely because we believed that the considerations underlying accrual in cases brought under § 502(a) (like the present case) were distinguishable. We do not hold that the Tolle analysis is inapplicable, but we prefer to decide this case in reliance on accrual precedents specific to § 502(a). [6] In that case, the plan unequivocally repudiated a guarantee of no-cost medical benefits by a statement in the SPD that after a certain trust was depleted, "monthly premium contributions from retirees will be required.” Id. [7] See, e.g., Carey v. IBEW Local 363 Pension Plan, 201 F.3d 44, 46 (2d Cir.1999) (pension plan unequivocally repudiated Carey's benefits with letter stating "you lost all your pension services credits due to the fact that you incurred a break in service prior to being vested.”); Daill v. Sheet Metal Workers’ Local 73 Pension Fund, 100 F.3d 62, 66 (7th Cir.1996) (Union unequivocally repudiated Daill's § 502(a)(1) claim for a pension when, in "[a] three-page letter, the fund carefully and comprehensively explained the basis for its decision” that Daill was 'not entitled to a pension from Local 73.’ ”). [8] Here, we must pause to address an argument by the plaintiffs that surfaced for the first time at oral argument. We discuss it in the interest of getting the law right. The plaintiffs referred to Walker v. Monsanto Co. Pension Plan, 614 F.3d 415 (7th Cir.2010), for the proposition that the right to future interest credits derives from the ERISA plan, and not from the ERISA statute. Thus, the plaintiffs argue, they needed to see the Plan document to know that they had been injured. We will not recount the complex and entirely distinguishable facts of Walker, because it does not say what the plaintiffs contend. The present plaintiffs did not need to reference the Plan to understand their injury; they needed *607 to reference the ERISA statute and law interpreting it. Those sources may be obscure, but that will not be held against the defendants. [9] Or at least, the plaintiffs would like us to apply an accrual date that presupposes that the injury was somehow concealed from them. We have rejected the plaintiffs' theory that any such concealment has occurred. [10] In fact, the record contains several Plan documents, because there are two Plan defendants and because there were revisions over the course of the class period. We have reviewed the Plan documents thoroughly in connection with this discussion. The language of the relevant provisions evolved over the relevant period, but not in a way that injected an opportunity for interpretation of the appropriate lump-sum distribution value. Indeed, it seems that the unlawful policy actually became more explicit over time. [11] See also 29 U.S.C. § 1104(a)(1)(D) (requiring plan fiduciaries to manage the plan "in accordance with the documents and instruments governing the plan...."). [12] See Berger, 338 F.3d at 762; see also West v. AK Steel Corp. Ret. Accumulation Pension Plan, 484 F.3d 395, 410 (6th Cir.2007); Esden, 229 F.3d at 168-69. [13] A “frontloaded interest credit plan” is simply a cash balance plan that pays the participant interest for the period between leaving the plan and age 65. A cash balance plan must be frontloaded to be tax-qualified. See Berger, 338 F.3d at 762. [14] I.R.C. § 401 sets forth the basic prerequisites to enjoy the preferential tax treatment the Code provides for retirement plans. Thus, a plan that fails to "comply with Section 401(a)(25)” would, for all practical purposes, fail altogether. [15] In this connection, we must reject as too clever by half the Plan defendants' argument that they should be allowed the discretion to adopt a new method that does not permit discretion in its application. Formally, the interest crediting method must be prescribed in the plan, and the only persons who can amend these Plans are the employers — precisely the persons forbidden from exercising discretion in Notice 96-8. Technicalities aside, the defendants’ proposed distinction between discretion in formula selection and formula application is illusory in this scenario, because the recovery calculation method will only be used one time. The defendants are asking to control the only calculation that matters. We think that conferring on the Plan defendants the discretion to devise the entire formula ex post would miss the point of the IRS Notice, and I.R.C. § 401(a)(25). See also Treas. Reg. 1.411(d) — 4, Q & A-4, Q & A- 5 & Q & A-7. [16] The Plan defendants urge that this court’s Berger decision supports their belief that their calculation method is entitled to deference. The reality is quite the opposite, since the district court rejected that argument and was affirmed on appeal. Since the Berger defendants apparently did not raise the deference issue on appeal, we did not opine on it. So Berger either offers the defendants no support or undermines their position. [17] This is not to say that we perceive any problem with the five-year average methodology the district court adopted. It seems a great deal more administrable than the plaintiffs' probabilistic "stochastic” method, and more closely tied to the Plans’ actual interest crediting method (which is based on investment returns) than is the defendants' interest rate-based "spread" method. Although we do not decide the question, we note that Treasury "safe harbor” regulations and several precedents support the use of such an average. See Treas. Reg. 1.401(a)(4)-8(c)(3)(v)(B); Berger, 338 F.3d at 760; Esden, 229 F.3d at 170. [18] We assume that the proceedings on remand need not be lengthy or burdensome since the district court has already received extensive briefing on possible future interest crediting methods. The district court simply needs to select an interest crediting method with the understanding that it is in charge of the decision.