Opinion
Civil Action No. 21-928 (RDM) Consolidated with: No. 21-00931 (RDM) No. 21-02132 (RDM)
2023-02-06
TRUSTEES OF the IAM NATIONAL PENSION FUND, Plaintiffs, v. OHIO MAGNETICS, INC., et al., Defendants.
Anthony Cacace, Pro Hac Vice, Neil V. Shah, Pro Hac Vice, Proskauer Rose LLP, New York, NY, John E. Roberts, Pro Hac Vice, Proskauer Rose LLP, Boston, MA, Guy G. Brenner, Proskauer Rose LLP, Washington, DC, for Plaintiff Trustees of the Iam National Pension Fund. Guy G. Brenner, Proskauer Rose LLP, Washington, DC, Anthony Cacace, Pro Hac Vice, Neil V. Shah, Pro Hac Vice, Proskauer Rose LLP, New York, NY, for Plaintiff Trustees of the Iam National Pension Fund. Stephen K. Dixon, Morgan, Lewis & Bockius LLP, Washington, DC, Stacey Recht Eisenstein, Akin Gump Strauss Hauer & Feld LLP, Washington, DC, for Defendant Ohio Magnetics, Inc. David J. Laurent, Buchanan Ingersoll & Rooney, PC, Pittsburgh, PA, for Defendant Phillips Liquidating Trust. Stephen K. Dixon, Benjamin T. Kelly, Pro Hac Vice, Morgan, Lewis & Bockius LLP, Washington, DC, Deborah Davidson, Pro Hac Vice, Morgan, Lewis & Bockius LLP, Chicago, IL, Randall McGeorge, Pro Hac Vice, Morgan Lewis Bockius LLP, Pittsburgh, PA, for Defendant Toyota Logistics Services, Inc. Guy G. Brenner, Proskauer Rose LLP, Washington, DC, for Plaintiff Trustees of the Iam National Pension Fund. David J. Laurent, William P. Lewis, Buchanan Ingersoll & Rooney, PC, Pittsburgh, PA, for Defendant Phillips Liquidating Trust.
Anthony Cacace, Pro Hac Vice, Neil V. Shah, Pro Hac Vice, Proskauer Rose LLP, New York, NY, John E. Roberts, Pro Hac Vice, Proskauer Rose LLP, Boston, MA, Guy G. Brenner, Proskauer Rose LLP, Washington, DC, for Plaintiff Trustees of the Iam National Pension Fund. Guy G. Brenner, Proskauer Rose LLP, Washington, DC, Anthony Cacace, Pro Hac Vice, Neil V. Shah, Pro Hac Vice, Proskauer Rose LLP, New York, NY, for Plaintiff Trustees of the Iam National Pension Fund. Stephen K. Dixon, Morgan, Lewis & Bockius LLP, Washington, DC, Stacey Recht Eisenstein, Akin Gump Strauss Hauer & Feld LLP, Washington, DC, for Defendant Ohio Magnetics, Inc. David J. Laurent, Buchanan Ingersoll & Rooney, PC, Pittsburgh, PA, for Defendant Phillips Liquidating Trust. Stephen K. Dixon, Benjamin T. Kelly, Pro Hac Vice, Morgan, Lewis & Bockius LLP, Washington, DC, Deborah Davidson, Pro Hac Vice, Morgan, Lewis & Bockius LLP, Chicago, IL, Randall McGeorge, Pro Hac Vice, Morgan Lewis Bockius LLP, Pittsburgh, PA, for Defendant Toyota Logistics Services, Inc. Guy G. Brenner, Proskauer Rose LLP, Washington, DC, for Plaintiff Trustees of the Iam National Pension Fund. David J. Laurent, William P. Lewis, Buchanan Ingersoll & Rooney, PC, Pittsburgh, PA, for Defendant Phillips Liquidating Trust. MEMORANDUM OPINION RANDOLPH D. MOSS, United States District Judge
These consolidated cases require the Court to interpret various provisions of the Employee Retirement Income Security Act ("ERISA"), 29 U.S.C. § 1001 et seq., a prospect that can send chills down the judicial spine. But Judge Richard Posner, invoking a quote from Oliver Wendell Holmes, offers words of comfort to judges required to give meaning to ERISA's arcane terms. See Chicago Truck Drivers Union v. CPC Logistics, Inc., 698 F.3d 346, 353 (7th Cir. 2012). As Justice Holmes wrote to the English jurist Sir Frederick Pollock in a very different time and context but in words that, as Judge Posner notes, are apt to judicial efforts to interpret ERISA: "I am frightened weekly [by seemingly hard cases,] but . . . when you walk up to the lion and lay hold[,] the hide comes off and the same old donkey of a question of law is underneath." Id. (quoting 1 Holmes-Pollock Letters: The Correspondence of Mr. Justice Holmes and Sir Frederick Pollock, 1874-1932, at 156 (Mark De Wolfe Howe ed. 1941)). This is such a case. The complexity of the statutory regime and the intricate calculations required to implement it are, at first, daunting. But once that "hide comes off," both the question presented and the answer to it are readily discernible. Id.
The present dispute began in 2018, when three companies—Ohio Magnetics, Inc., Toyota Logistics Services, Inc., and Phillips Liquidating Trust (successor in interest to the Phillips Corporation) (collectively the "Defendants" or "Companies")—withdrew from the IAM National Pension Fund (the "Fund"). Dkt. 34-2 at 5 (Pl.'s SUMF ¶ 20); see 29 U.S.C. § 1383(a). The Companies' withdrawal entitled the Fund to assess withdrawal liability against each of them. 29 U.S.C. § 1381. Simplifying somewhat, withdrawal liability is a charge equal to each company's proportionate share of the unfunded pension benefits to which workers participating in the pension plan associated with the Fund have a vested interest. Id. § 1381(b)(1). Pension funds retain actuaries to calculate withdrawal liability, and the Fund's actuary did so here, basing its calculations on certain actuarial assumptions that it adopted on January 24, 2018. Dkt. 34-2 at 6 (Pl.'s SUMF ¶ 22).
In their briefing, the parties typically refer to the pension fund, rather than the pension plan associated with the fund. The statute and the bulk of the case law, however, speak primarily in terms of pension plans. See, e.g., 29 U.S.C. §§ 1381, 1383, 1393. There is no practical difference in most contexts, and, accordingly, the Court will refer to plans and funds interchangeably except where specifically noted.
The question presented here is whether the Fund's actuary was permitted to use the assumptions that it did in calculating the Companies' withdrawal liability. The Companies say that the actuary was not permitted to do so. They contend that because they withdrew in 2018, the actuary was required to use the assumptions that were "in effect" on December 31, 2017 to calculate their liability. Dkt. 37 at 5. For support, they point to provisions of ERISA that require liability for a withdrawing employer to be assessed based on a fund's unfunded vested benefits "as of" the end of the year prior to that in which the employer withdraws. Plaintiffs, the Fund's trustees, disagree. They maintain that, under the very same statutory provisions, the Fund's actuary was free to set its assumptions at any time and, in so doing, to consider any and all events occurring up to the time it made its withdrawal liability calculation, including events occurring after December 31, 2017. Dkt. 34-1 at 7.
The Court disagrees with both positions. It concludes that ERISA provides actuaries more flexibility than the Companies posit but less than they would have under the Trustees' theory. Because the Court's reading of the statute is at odds with those of the arbitrators whose awards are under review—all of whom sided with the Companies—the Court will GRANT Plaintiffs' motion for summary judgment, Dkt. 34, DENY Defendants' cross-motion for summary judgment, Dkt. 38, VACATE the arbitration awards, and REMAND the cases to their respective arbitrators for further proceedings consistent with this opinion and to resolve any further challenges to the withdrawal liability assessments.
I. BACKGROUND
A.
In a multiemployer pension plan, multiple employers make financial contributions to the same general trust fund, and the money in that fund is used to provide for the pensions of the various employers' employees. 29 U.S.C. § 1002(37); see Concrete Pipe & Prods. of Cal., Inc. v. Constr. Laborers Pension Tr. for S. Cal., 508 U.S. 602, 605-06, 113 S.Ct. 2264, 124 L.Ed.2d 539 (1993). These plans are maintained in accordance with collective bargaining agreements between the employers and a union and are governed by the provisions of ERISA. United Mine Workers of Am. 1974 Pension Plan v. Energy West Mining Co., 39 F.4th 730, 734 (D.C. Cir. 2022). Among other things, ERISA requires employers participating in multiemployer plans to "contribute annually to the plan whatever is needed to ensure it has enough assets to pay for the employees' vested pension benefits when they retire." Id.
An employer who participates in a multiemployer plan is free to withdraw from the plan and to terminate its obligation to make annual contributions. 29 U.S.C. § 1383. But an employer's withdrawal does not divest any worker enrolled in the plan of the pension benefits he or she has earned; the fund and its remaining contributors must still provide for the vested pension benefits of all its participants. See Energy West, 39 F.4th at 734-35 & n.2. This structure can create perverse incentives: if a plan's funding begins to lag—say, because a market downturn decreases the value of its assets—participating employers will be required to make larger annual contributions in order to comply with ERISA. Milwaukee Brewery Workers' Pension Plan v. Jos. Schlitz Brewing Co., 513 U.S. 414, 416-17, 115 S.Ct. 981, 130 L.Ed.2d 932 (1995). And as required annual contributions grow, so too does the incentive for participating employers to withdraw. Id. Withdrawals further exacerbate funding shortfalls, and a shortfall-withdrawal-shortfall cascade can send a plan into a "death spiral." Energy West, 39 F.4th at 734. ERISA created a federally chartered insurance corporation, the Pension Benefit Guaranty Corporation ("PBGC"), to backstop troubled pension plans and to head off death spirals. 29 U.S.C. § 1302. But, in practice, the existence of this safety net only further encouraged withdrawals and threatened to stretch the PBGC's obligations beyond its means. See Connolly v. Pension Benefit Guar. Corp., 475 U.S. 211, 214-15, 106 S.Ct. 1018, 89 L.Ed.2d 166 (1986).
Congress enacted the Multiemployer Pension Plan Amendments Act of 1980 (the "MPPAA"), Pub. L. 96-364, 94 Stat. 1208, to address this problem. To ensure that employers pay their fair share (and to discourage strategic withdrawals) the MPPAA requires withdrawing employers to pay for the privilege. 29 U.S.C. § 1381. Under the MPPAA, an employer that withdraws from a multiemployer plan must pay "its pro rata share of the pension plan's funding shortfall," also known as its withdrawal liability. CPC Logistics, 698 F.3d at 347; 29 U.S.C. § 1381(a), (b). More specifically, withdrawal liability is imposed based on "the employer's proportionate share of the plan's 'unfunded vested benefits,' calculated as the difference between the present value of vested benefits and the current value of the plan's assets." Pension Benefit Guar. Corp. v. R.A. Gray & Co., 467 U.S. 717, 725, 104 S.Ct. 2709, 81 L.Ed.2d 601 (1984) (quoting 29 U.S.C. §§ 1381, 1391); see 29 U.S.C. § 1393(c).
Withdrawal liability is a function of both known variables and indeterminate assumptions. For instance, when calculating withdrawal liability, a plan's actuary knows how many employees are enrolled in the plan and what benefits their pensions provide. But the actuary must estimate, among other things, how long these employees will work and how long they will live. Energy West, 39 F.4th at 735. The assumption with the greatest effect on the withdrawal liability bottom line is the rate at which the plan's assets will grow "by the miracle of compound interest"—that is, the discount rate. CPC Logistics, 698 F.3d at 348. The higher the discount rate, the faster the fund's assets are projected to grow on their own, and thus the smaller the present value of the plan's liabilities, the lower the funding shortfall, and the less a withdrawing employer's withdrawal liability. See id. And, conversely, the lower the discount rate, the slower the assets are assumed to grow, and thus the greater the present value of the plan's liabilities, and the more a withdrawing employer must pony up. See id. The MPPAA requires plans calculating withdrawal liability to use "actuarial assumptions and methods which, in the aggregate, are reasonable (taking into account the experience of the plan and reasonable expectations) and which, in combination, offer the actuary's best estimate of anticipated experience under the plan." 29 U.S.C. § 1393(a)(1).
Of particular importance here, the statute also requires that withdrawal liability be calculated "not as of the day of withdrawal, but as of the last day of the plan year preceding the year during which the employer withdrew," also known as the "measurement date." Jos. Schlitz Brewing Co., 513 U.S. at 418, 115 S.Ct. 981 (citing 29 U.S.C. §§ 1391(b)(2)(A)(ii), (b)(2)(E)(i), (c)(2)(C)(i), (c)(3)(A), and (c)(4)(A)). So, for a plan operating on a calendar year, withdrawal liability is based on the plan's unfunded vested benefits "as of" December 31 of the year before the year in which the employer withdraws. Id. This is true regardless of when in the year the withdrawal takes place. Employers who withdrew from a calendar year plan on January 1, 2022, June 30, 2022, and December 31, 2022, would all be liable for their share of the plan's unfunded vested benefits "as of" December 31, 2021. 29 U.S.C. § 1391(b)(2)(E)(i).
Congress apparently adopted "this calculation date" to foster "administrative convenience." Jos. Schlitz Brewing Co., 513 U.S. at 418, 115 S.Ct. 981. Irrespective of any withdrawals, ERISA and the internal revenue code both require plans to prepare "a valuation of the plan's liability" every year. 29 U.S.C. § 1084(c)(7)(A); 26 U.S.C. § 431(c)(7)(A). Setting a plan-year-end measurement date allows an actuary to use the calculation "it must prepare in any event," thereby "avoiding the need to generate new figures tied to the date of actual withdrawal." Jos. Schlitz Brewing Co., 513 U.S. at 418, 115 S.Ct. 981. The MPPAA makes this option explicit, stating that an actuary "may rely on the most recent complete actuarial valuation" and "reasonable estimates for the interim years of the unfunded vested benefits" when determining an employer's withdrawal liability. 29 U.S.C. § 1393(b)(1).
Withdrawal liability can be substantial, and, not surprisingly, plans and withdrawing employers disagree about which assumptions to use. A withdrawing employer that wants to dispute the calculations made by a plan's actuary must do so through arbitration in the first instance. 29 U.S.C. § 1401(a)(1). A plan's determination of withdrawal liability receives considerable deference in the arbitration process and is "presumed correct" by the arbitrator unless the withdrawing employer "shows by a preponderance of evidence" that either the actuarial "assumptions and methods" used were unreasonable "in the aggregate," "taking into account the experience of the plan and reasonable expectations," or that the plan's actuary "made a significant error" in applying those assumptions or methods. Id. § 1401(a)(3)(B); see also id. § 1401(a)(3)(A). After arbitration, "any party can seek 'to enforce, vacate, or modify the arbitrator's award' in district court." Energy West, 39 F.4th at 736 (quoting 29 U.S.C. § 1401(b)(2)).
B.
1.
IAM National Pension Plan is a multiemployer pension plan. It provides retirement benefits to certain employees who performed work for employers maintaining collective bargaining agreements with the International Association of Machinists and Aerospace Workers, AFL-CIO (or with affiliated local and district lodges). Dkt. 34-2 at 2 (Pl.'s SUMF ¶¶ 1-2). The plan's assets are held in the Fund, which is governed by a trust agreement. Id. (Pl.'s SUMF ¶¶ 3-4). Under that agreement, the Fund's fiscal year and ERISA plan year correspond to the calendar year. Id. (Pl.'s SUMF ¶ 4); Dkt. 4-1 at 44 (Pl.'s Ex. A). The agreement also provides that withdrawal liability shall be calculated using the methodology set forth in 29 U.S.C. § 1391(b). Dkt. 34-2 at 2 (Pl.'s SUMF ¶ 5); Dkt. 4-1 at 32 (Pl.'s Ex. A).
Cheiron is the Fund's actuary and performs the Fund's annual valuations and withdrawal liability calculations. See Dkt. 34-2 at 3 (Pl.'s SUMF ¶¶ 7-8). In November 2017, Cheiron issued the Fund's 2016 Plan Year valuation, which concluded that as of January 1, 2017, the Fund had $448,099,164 in unfunded vested benefits. Id. (Pl.'s SUMF ¶¶ 10-11). To reach this result, Cheiron employed a discount rate of 7.5%. Id. (Pl.'s SUMF ¶ 12). In making this calculation, Cheiron did not include any assumption for the Fund's future administrative expenses, even though these expenses are paid out of the Fund's assets and therefore contribute to the Plan's unfunded vested benefits. See Dkt. 4-1 at 79 (Pl.'s Ex. B); Dkt. 38-12 at 2 (Phillips Stip. ¶ 4).
On January 24, 2018, the Fund's trustees held a meeting at which Cheiron recommended, and the Trustees unanimously approved, certain new withdrawal liability assumptions. Dkt. 34-2 at 4 (Pl.'s SUMF ¶¶ 13-14); Dkt. 37-8 at 4 (Def.'s Resp. Pl.'s SUMF ¶ 14); Dkt. 4-1 at 126 (Pl.'s Ex. D). Two such assumptions are relevant here: First, the withdrawal liability discount rate was set at 6.5%, versus the 7.5% rate that had been used for the 2016 Plan Year valuation. Dkt. 34-2 at 4 (Pl.'s SUMF ¶ 14). Second, an administrative expense load assumption equal to 4.0% of the present value of vested benefits was put in place, with the proviso that this assumption would be "[r]edetermine[d] annually upon completion of the actuarial valuation." Dkt. 4-1 at 126 (Pl.'s Ex. D); Dkt. 34-2 at 4 (Pl.'s SUMF ¶ 14); see id. at 5 (Pl.'s SUMF ¶ 18). All else equal, both of these assumptions would result in greater withdrawal liability than would the assumptions used in the 2016 Plan Year valuation.
The parties disagree about how best to characterize what occurred at this meeting. The Trustees assert that Cheiron adopted the assumptions after a discussion with the Trustees. Dkt. 34-2 at 4 (Pl.'s SUMF ¶ 14). The Companies, however, dispute whether there was any discussion. Dkt. 37-8 at 4-5 (Def.'s Resp. Pl.'s SUMF ¶ 14). The minutes of the meeting state that "following discussion," the Trustees "unanimously approved" the assumptions as "recommend[ed] [by] the Fund's Actuary, Cheiron." Dkt. 4-1 at 126 (Pl.'s Ex. D); Dkt. 38-13 at 128 (Phillips Ex. D). They further indicate that, at the meeting, Cheiron "confirmed that all of [the] changes to the withdrawal liability calculation and the actuarial assumptions" were "reasonable and defensible." Dkt. 34-2 at 4 (Pl.'s SUMF ¶ 15) (alteration in original). Although the parties' disputes about the meeting may at some point become significant, they are immaterial for present purposes. What matters for today is simply that these assumptions were adopted at the January 24, 2018 meeting.
On April 17, 2019, Cheiron issued the actuarial valuation for the 2017 Plan Year. Id. (Pl.'s SUMF ¶ 16). This valuation employed a 6.5% withdrawal liability discount rate and a 3.5% expense load assumption. Id. at 5 (Pl.'s SUMF ¶ 18). The resulting unfunded vested benefits figure was $3,043,369,928. Id. at 4 (Pl.'s SUMF ¶ 17).
2.
As of the beginning of 2018, the Companies were party to collective bargaining agreements that required them to make annual contributions to the Fund. Id. at 5 (Pl.'s SUMF ¶ 19). All withdrew from the Fund over the course of the year: Phillips as of April 7; Ohio Magnetics as of June 30; and Toyota Logistics as of December 29. Id. (Pl.'s SUMF ¶ 20). The Fund assessed withdrawal liability against them on the following days and in the following amounts: On April 2, 2019, the Fund assessed $2,013,028 and $477,475 against Phillips and Ohio Magnetics, respectively. Id. at 6 (Pl.'s SUMF ¶ 22); Dkt. 38-12 at 6 (Phillips Stip. ¶ 28); Dkt. 38-11 at 5 (Ohio Mag. Stip. ¶ 26). And on June 18, 2019, the Fund assessed $1,289,384 against Toyota Logistics. Dkt. 34-2 at 6 (Pl.'s SUMF ¶ 22); Dkt. 38-13 at 5 (Toyota Log. Stip. ¶ 24). Each calculation was prepared using essentially the same assumptions adopted in the January 24, 2018 meeting and contained in the 2017 Plan Year valuation: a 6.5% withdrawal liability discount rate and a 3.5% expense load. Dkt. 34-2 at 5, 6 (Pl.'s SUMF ¶¶ 18, 22); Dkt. 37-9 at 2 (Def.'s SUMF ¶ 8). The Companies requested review of their respective assessments, and each demanded arbitration after the Fund denied those requests. Dkt. 38-11 at 6 (Ohio Mag. Stip. ¶¶ 27, 29); Dkt 38-12 at 7 (Phillips Stip. ¶¶ 30-31); Dkt. 38-13 at 5-6 (Toyota Log. Stip. ¶¶ 25-26).
To be sure, the 3.5% expense load was slightly lower than the 4.0% adopted at the January 24, 2018 meeting. But for obvious reasons the Companies have not argued that 4.0% should apply, and, as a result, neither the arbitrations nor this case raise the question whether Cheiron erred in applying an expense load assumption that was not adopted until April 2019 and that was lower than the assumption approved in January 2018.
3.
In each of the three arbitrations, the parties agreed that the arbitrator would resolve certain issues at the outset based on stipulated facts before addressing any further challenges to the withdrawal liability assessments. Dkt. 34-2 at 6-7 (Pl.'s SUMF ¶ 24). These threshold issues differed to some extent between the arbitrations, but each arbitrator was asked to decide a variation of the following question: Was it permissible for the Fund to assess withdrawal liability for the Companies, which withdrew in 2018, based on actuarial assumptions adopted in January 2018, or was Cheiron required as a matter of law to use assumptions that had been adopted prior to December 31, 2017? Dkt. 37-1 at 6-7 (Ohio Mag. Award); Dkt. 37-2 at 1-2 (Phillips Award); Dkt. 37-3 at 1 (Toyota Log. Award).
All three arbitrators concluded that Cheiron erred in basing its withdrawal liability calculations on assumptions adopted after December 31, 2017. Dkt. 37-1 at 37 (Ohio Mag. Award); Dkt. 37-2 at 10-11 (Phillips Award); Dkt. 37-3 at 14 (Toyota Log. Award). Although each arbitrator employed slightly different reasoning, all relied on the Second Circuit's decision in National Retirement Fund On Behalf of Legacy Plan of National Retirement Fund v. Metz Culinary Management, Inc., 946 F.3d 146 (2d Cir. 2020), which had adopted the following bright-line rule:
[I]nterest rate assumptions for withdrawal liability purposes must be determined as of the last day of the year preceding the employer's withdrawal from a multiemployer pension plan. Absent any change to the previous plan year's assumption made by the Measurement Date, the interest rate assumption in place from the previous plan year will roll over automatically.Id. at 152; see Dkt. 37-1 at 29-33 (Ohio Mag. Award); Dkt. 37-2 at 12-14 (Phillips Award); Dkt. 37-3 at 15 (Toyota Log. Award). Because Cheiron's withdrawal liability calculations ran afoul of this holding, the arbitrators directed Cheiron to recalculate withdrawal liability using the actuarial assumptions that the Fund had most recently adopted before December 31, 2017: a 7.5% discount rate and no expense load. Dkt. 37-1 at 39 (Ohio Mag. Award); Dkt. 37-2 at 15 (Phillips Award); Dkt. 37-3 at 17 (Toyota Log. Award). Because this issue proved dispositive to the withdrawal liability assessments, the arbitrators did not reach any of the other questions before them, which questions did not turn on the same timing principle.
On April 4, 2021, the Trustees filed two separate lawsuits. In the first, they asked the Court to vacate the arbitration award entered in favor of Ohio Magnetics, Dkt. 1 at 1 (Compl. ¶ 1), and, in the second, they sought vacatur of the arbitration award entered in favor of Toyota Logistics, Trustees of the IAM Nat'l Pension Fund v. Toyota Logistics Servs., Inc., No. 21-931, at Dkt. 1 at 1 (Compl. ¶ 1). In both cases, the defendants counterclaimed to enforce the arbitration awards. Dkt. 10; Toyota Logistics Servs., Inc., No. 21-931, at Dkt. 8. On the Trustees' motion, the Court subsequently consolidated the Ohio Magnetics and Toyota Logistics cases. Dkt. 19. Then, on August 10, 2021, the Trustees brought a third lawsuit, this time challenging the arbitration award in favor of Phillips. Trustees of the IAM Nat'l Pension Fund v. Phillips Liquidating Trust, No. 21-2132, at Dkt. 1 at 1 (Compl. ¶ 1). As in the two earlier filed cases, the defendant counterclaimed to enforce the arbitration award. Id. at Dkt. 7. The Court consolidated the Phillips case with the two earlier filed actions and set a briefing schedule in the consolidated litigation for cross-motions for summary judgment. Min. Order (Aug. 16, 2021).
The parties' briefing is now complete, and the case is ripe for decision.
II. STANDARD OF REVIEW
This case implicates two distinct standards of review: the typical summary judgment standard under Federal Rule of Civil Procedure 56 and the specific standard by which the Court reviews an ERISA arbitration award.
Starting with the more familiar of the two, summary judgment is warranted if a party can "show[ ] that there is no genuine dispute as to any material fact and [that the party] is entitled to judgment as a matter of law." Fed. R. Civ. P. 56(a). "[I]n ruling on cross-motions for summary judgment, the court shall grant summary judgment only if one of the moving parties is entitled to judgment as a matter of law upon material facts that are not genuinely disputed." See Muslim Advocs. v. U.S. Dep't of Just., 833 F. Supp. 2d 92, 98 (D.D.C. 2011).
When reviewing an arbitration award under ERISA, however, the Court (1) presumes that the arbitrator's findings of fact are correct "unless they are rebutted 'by a clear preponderance of the evidence' " and (2) reviews the arbitrator's legal conclusions de novo. Energy West, 39 F.4th at 737 (quoting 29 U.S.C. § 1401(c)). The Fund contends, and the Companies do not dispute, that the only question raised by each of the arbitration awards at issue in this case is a pure question of law. See Dkt. 34-1 at 15. The Court agrees, so its review is de novo.
III. ANALYSIS
The issue before the Court is a narrow one. The Court need not decide whether Cheiron correctly calculated the Companies' withdrawal liability or whether it applied substantively reasonable actuarial assumptions. Instead, the Court must simply decide—as a matter of statutory interpretation—whether Cheiron erred when it applied actuarial assumptions that were adopted after the measurement date. The parties offer markedly different answers to that question.
A.
An employer's withdrawal liability, as explained above, is equal to its proportionate share of the plan's unfunded vested benefits—that is, the present value of its liabilities less the current value of its assets. See 29 U.S.C. § 1391(b). So that an employer is not held responsible for unfunded liability accrued before it participated in the plan—a potential disincentive to join a multiemployer plan in the first place—the MPPAA creates "default rules (that is, rules that govern unless the plan provides otherwise)" for apportioning to a withdrawing employer "a share of only so much of the plan's funding shortfall as occurred while the employer was participating in the plan." CPC Logistics, 698 F.3d at 348. These rules are complex and involve calculating, allocating, and ultimately combining the change in a plan's unfunded vested benefits during each year in which a withdrawing employer participated. See id. Mercifully, the intricacies of this methodology are not at issue in this case. The critical thing, and common ground among the parties, is that the employer's withdrawal liability is calculated based on the plan's unfunded vested benefits "as of" the measurement date.
The statutory provision that most clearly sets forth the "measurement date" requirement reads as follows:
An employer's proportional share of the unamortized amount of a change in unfunded vested benefits is the product of . . . the unamortized amount of such change (as of the end of the plan year preceding the plan year in which the employer withdraws); multiplied by [the fraction of that amount attributable to the employer].29 U.S.C. § 1391(b)(2)(E)(i) (emphasis added). So, simplifying somewhat, when an actuary calculates withdrawal liability, it must determine the withdrawing employer's share of what the plan's unfunded vested benefits were "as of the end of the plan year preceding the plan year in which the employer withdraws." Id.
In addressing the temporal aspect of what an actuary must calculate, the parties cite more frequently to § 1391(b)(2)(A). That paragraph provides that "[a]n employer's proportional share of the unamortized amount of the change in the plan's unfunded vested benefits" is "the sum of the employer's proportional shares of the unamortized amount of the change in unfunded vested benefits for each plan year in which the employer has an obligation to contribute under the plan ending . . . before the plan year in which the withdrawal of the employer occurs." 29 U.S.C. § 1391(b)(2)(A) (emphasis added). Perhaps the measurement date requirement can be gleaned from this provision too, but, in the Court's view, § 1391(b)(2)(E)(i) provides a much clearer articulation of the requirement. The Court's reliance on § 1391(b)(2)(E)(i), moreover, is consistent with the approaches taken in the handful of judicial decisions that have addressed the measurement date requirement, all of which rely on the "as of" language found in § 1391(b)(2)(E)(i). See, e.g., Jos. Schlitz Brewing Co., 513 U.S. at 417-18, 115 S.Ct. 981; Metz, 946 F.3d at 148; Nat'l Ret. Fund ex rel. Legacy Plan of Nat'l Ret. Fund v. Metz Culinary Mgmt., Inc., No. 16-2408, 2017 WL 1157156, at *7 (S.D.N.Y. Mar. 27, 2017); Trustees of IAM Nat'l Pension Fund v. M & K Emp. Sols., LLC, No. 21-02152, 2022 WL 4534998, at *2 (D.D.C. Sept. 28, 2022). It is also consistent with the test the parties invoke in their briefs, which, despite citing to § 1391(b)(2)(A), repeatedly employ § 1391(b)(2)(E)(i)'s "as of" terminology. E.g., Dkt. 34-1 at 10, 12, 22, 32; Dkt. 37 at 5, 7, 10, 11, 13, 14, 15, 16, 18, 19, 20.
Although the parties agree that ERISA requires actuaries to calculate the withdrawing employer's share of unfunded vested benefits "as of" the measurement date, they disagree about what this means. In the Companies' view, plan actuaries must "apply the actuarial assumptions in effect as of the . . . [m]easurement [d]ate" for purposes of calculating an employer's withdrawal liability. Dkt. 37 at 9. What they evidently mean by this is that when, as here, a plan has adopted actuarial assumptions at any time prior to the measurement date, those assumptions remain "in effect"—and are thus binding—for purposes of calculating withdrawal liability, unless the plan's trustees affirmatively approve an alternative set of actuarial assumptions prior to the measurement date. Id. at 9-18. They contend that this rule is "implicit" in ERISA's requirement that actuaries employ "reasonable assumptions" and that it is necessary to avoid "widespread manipulation" of assumptions by actuaries seeking to punish withdrawing employers. Id. at 14, 17, 22; see also 29 U.S.C. § 1393(a)(1).
An example may make the Companies' argument concrete. Under the rule that they propose, if a plan adopted a discount rate assumption in June 2020 and did not adopt a different assumption before December 31, 2022, then withdrawal liability for employers that withdraw in 2023 would have to be calculated using the assumption adopted in June 2020. This would be true even if in January 2023, after reviewing the data for the year 2022, the actuary concluded that the assumptions made in June 2020 no longer made sense. It would be true, for instance, even if the economy sustained a shock akin to 1929's Black Thursday in December 2022, before the measurement date. The actuary could change the assumptions in January 2023 to reflect this development, but the new assumptions could not be used to calculate withdrawal liability for employers who withdraw in 2023; the old assumptions would govern for withdrawals prior to January 1, 2024.
The Trustees, in contrast, argue that actuarial assumptions are unconstrained by the measurement date. Dkt. 34-1 at 7, 17-19. They posit that the only limitations on actuarial discretion in setting assumptions are found in 29 U.S.C. § 1393, which, as relevant here, provides that "[w]ithdrawal liability . . . shall be determined by each plan on the basis of . . . actuarial assumptions and methods which, in the aggregate, are reasonable (taking into account the experience of the plan and reasonable expectations) and which, in combination, offer the actuary's best estimate of anticipated experience under the plan." 29 U.S.C. § 1393(a)(1) (emphasis added). So, if the actuary's assumptions are reasonable and if they reflect the actuary's best estimate of the plan's anticipated experience, nothing more is required. The measurement date exists, the Trustees say, for the fixed, knowable components of the withdrawal liability calculation: the then-existing value of the plan's assets and liabilities (before discounting and the application of other assumptions). Dkt. 34-1 at 21. But, they continue, actuarial assumptions can be refined at any time prior to the plan's issuance of a notice and demand for payment of withdrawal liability, id. at 7, 17-19, which can occur long after the measurement date, see, e.g., Dkt. 38-11 at 174-79 (Def.'s Ex. F) (Notice Issued to Ohio Magnetics). Not only that, when adopting assumptions, an actuary can (and should) consider the latest information available at the time it makes its calculation. Dkt. 34-1 at 19. So, for example, if an employer withdrew from a plan on December 1, 2022 (making the measurement date December 31, 2021), an actuary could consider data generated (and events occurring) up to the time at which it calculates withdrawal liability—i.e. long after both December 31, 2021 and December 1, 2022. See id. A market downturn or jump in interest rates in March 2023, months after the measurement date, could be fair game. See id.
Neither position is correct. As Judge Lamberth recognized in Trustees of IAM National Pension Fund v. M & K Employee Solutions, whatever unfunded vested benefits were "as of" the measurement date constitutes a snapshot of the information available "as of" that date. 2022 WL 4534998, at *14. Because an actuary must take time to consider what that information was (presumably it is not setting its assumptions at 11:59 p.m. on New Year's Eve), and because withdrawal liability is necessarily calculated after the measurement date, this means that the actuary need not determine before close of business on the measurement date what assumptions to use in generating its "best estimate" of the plan's "anticipated experience" as of the measurement date. See 29 U.S.C. § 1393(a)(1). Otherwise, the actuary could not reasonably account for developments up to the measurement date. By the same token, an actuary may only consider information that was available by the measurement date. It may not set assumptions reflecting developments that occur in the following plan year. What matters is the informational landscape on the measurement date—not whatever (possibly outdated) assumptions had been set prior to that date and not anything (including a significant economic shock) that happens after.
The Court notes that "information available" is not necessarily the same thing as "facts in existence." Many facts "exist" in some theoretical sense long before they are ever known, understood, or reasonably available to actuaries; a person may have cancer, for example, years before she is diagnosed—or even could be diagnosed with existing medical technology. The Court's holding limits actuaries to consideration of information available on the measurement date and does not permit an actuary to base its assumptions on facts that, although extant on the measurement date in some abstract sense, were not actually available for practical use on that date.
This rule best reflects what the statute says. See Sebelius v. Cloer, 569 U.S. 369, 376, 133 S.Ct. 1886, 185 L.Ed.2d 1003 (2013). Section 1391(b)(2)(E)(i) provides that an employer's withdrawal liability is based, in relevant respects, on its share of the fund's unfunded vested benefits "as of the end of the plan year preceding the plan year in which the employer withdraws," 29 U.S.C. § 1391(b)(2)(E)(i), that is, "as of" the measurement date. Although much of the language used in ERISA is technical in nature, the phrase "as of" is not. It simply means "at or on (a specific time or date)," As Of, Webster's Third New International Dictionary 129 (1976), or "as things stood on (a date)," As Of, Oxford English Dictionary (3d ed., June 2011), https://www.oed.com/view/Entry/11307 (last visited February 6, 2023). Accordingly, an employer's withdrawal liability is based on a snapshot of its "allocable amount of unfunded vested benefits" as things stood on the measurement date, here December 31, 2017. 29 U.S.C. §§ 1381(b), 1391(b)(2)(E)(i).
Neither party's argument fares well under the statute's plain language. Consider first the Trustees' position that actuaries may consider events occurring after the measurement date when deciding what the appropriate discount rate was "as of" the measurement date. That argument fails, because as a matter of ordinary meaning and common sense, the condition or valuation of something "at," "on," or "as things stood" on a given date is necessarily independent of events that occur subsequent to that date. Imagine, for example, that a benchmark interest rate with which a fund's assets are highly correlated stood at 5% on the measurement date but then climbed to 12% a year-and-half later when the plan's actuary got around to calculating the employer's withdrawal liability. As the Trustees would have it, the actuary would be permitted (and perhaps required) to consider the 12% interest rate when setting its assumptions to calculate unfunded vested benefits. But that argument ignores the fact that the concept of "unfunded vested benefits" turns on, among other things, the actuary's estimate of the anticipated returns of the fund's assets. Hence, an actuary following the Trustees' approach would not be following the statute, because he could not fairly be said to be looking at a snapshot of the world as "things stood" on the measurement date with respect to an essential component of unfunded vested benefits.
To take another example, imagine that the statute required actuaries to determine the average life expectancy of their plans' participants "as of" December 1, 2019. In making that assessment at some later date, an actuary might reasonably consider the age of the average plan participant on December 1, 2019, other demographic information, and data that existed on December 1, 2019 regarding projected life expectancies. The actuary could not consider, however, any impact that the COVID-19 pandemic—or any other intervening public health event—has had on average life expectancies. To do so would violate the mandate to make the calculation "as of" the specified date, because "as things stood" on December 1, 2019, there was no basis for taking account of a then-quiescent virus.
If anything, the Companies' position fares even worse on the face of the statutory text. In their view, if a plan has adopted certain actuarial assumptions prior to the measurement date—say, for example, for purposes of conducting the required valuation for the preceding year—and if it does not adopt new assumptions before the measurement date, it is bound to apply the previously adopted assumptions. This is true, moreover, even in cases in which significant events occur before the measurement date. The difficulty for the Companies is that it is unreasonable as a matter of both language and common sense to read the statutory requirement that an actuary determine what unfunded benefits were—and thus what the appropriate discount rate was—"as of" the measurement date to require an actuary to ignore the information available on that date.
To see the problem, assume that a plan applied a discount rate of 5% for purposes of its 2016 valuation and did not "adopt" a different actuarial assumption before a measurement date of December 31, 2022. The Companies would require the plan's actuary to apply that rate—even if that assumption was proved wildly wrong before the measurement date. Indeed, even if the fund was invested principally in stocks, and the stock market suffered a cataclysmic crash after the 5% discount rate was adopted but before the measurement date, the Companies would say that the 5% discount controls. That cannot be right, however, because no one would say that a discount rate based on a 5% expected return on the plan's assets reflects how "things stood" on December 31, 2022. Nor can the Court find in § 1391, or any other provision of ERISA, an "implicit" requirement that an actuary use a discount rate that is disconnected from reality. Cf. Energy West, 39 F.4th at 740 ("[T]he discount rate assumption cannot be divorced from the plan's anticipated investment returns."). Although Congress could have required actuaries to do so, the Court will not strain to reach such a result in the face of a much more obvious reading of the statute.
In one sense, the very different arguments the Trustees and the Companies press suffer from a common flaw: both ask the Court to decouple the various components of the concept of "unfunded vested benefits." As explained above, a plan's unfunded vested benefits are a function of certain knowable values—a plan's assets, the number of its beneficiaries, the generosity of the plan's benefits, and the schedule by which those benefits vest, among other things—and certain indeterminate assumptions, like the discount rate and the life expectancy of the beneficiaries. Both parties agree that the knowable values are set based on what they were on the measurement date. Dkt. 34-1 at 21; Dkt. 39 at 9; Dkt. 37 at 13-14. And both parties recognize that the data used to determine these values can be and as a practical matter must be compiled and analyzed after the measurement date. Dkt. 34-1 at 21; Dkt. 39 at 9; Dkt. 37 at 13-14. Yet both of them want to treat actuarial assumptions differently. The Trustees would allow actuarial assumptions to take account of developments subsequent to the measurement date, despite recognizing that this practice would not be permissible for the value of the plan's assets or its number of beneficiaries. Dkt. 34-1 at 19; Dkt. 39 at 9. And the Companies would forbid actuaries from engaging in the exact same practices they insist are required for assets and liabilities—producing values after the end of a plan year based on information available as of the measurement date—to formulate the appropriate assumptions. See Dkt. 37 at 13-14.
The statute draws no such lines and, instead, treats unfunded vested benefits as a single, unified concept, at least as a temporal matter. Section 1391 is the only provision that says anything about timing, and it speaks only of what "unfunded vested benefits" were "as of" the measurement date. 29 U.S.C. §§ 1391(b)(2)(E)(i), 1391(b)(2)(A). It does not distinguish, for example, between the pre-discounted value of a plan's future liabilities and the discount rate that is used to generate the present value of those liabilities. And it does not include so much as a stray clause hinting that plan actuaries must determine the relevant actuarial assumptions based on events occurring before some cutoff prior to the measurement date or based on the previous issuance of other plan reports (including annual valuations). Read in this manner, the statutory text is clear, and it requires actuaries to consider the concept of unfunded vested benefits as a whole and to calculate that amount "as of" the measurement date.
For simplicity, the Court disregards the separate set of actuarial assumptions, not at issue here, used to determine future benefit obligations, such as assumptions regarding the life expectancy of participants.
The Companies at times seem to recognize as much. They candidly acknowledge that the statute "does not expressly prohibit" the application of actuarial assumptions determined after the measurement date that are based on information available or events occurring before the measurement date. Dkt. 37 at 22. As explained above, that is, if anything, an understatement, given the statute's complete absence of any such restriction. But more importantly, although the Companies recognize that the statute is silent as to the limitation they support, they ignore the clear takeaway from that silence: Congress did not impose any such limitation. The same principle applies to the Trustees' argument. Congress treated "unfunded vested benefits" as a single concept with respect to the measurement date, and the Court has no warrant to separate, for example, the current value of a plan's assets and future value of its liabilities from the assumptions used to project future returns on those assets and the present value of those liabilities.
Similarly, the statutory section devoted to actuarial assumptions does not even suggest that separate timing rules exist for these assumptions than for other components of the withdrawal liability calculation. 29 U.S.C. § 1393. Not only that, the language of that section cuts against the Companies' argument that the Court should read such rules into the statute. Section 1393(a)(1) directs plan actuaries to use assumptions that "are reasonable (taking into account the experience of the plan and reasonable expectations) and which, in combination, offer the actuary's best estimate of anticipated experience under the plan." Id. § 1393(a)(1). At least at times, an actuary could not possibly comply with these directives while also following the Companies' reading of the measurement date requirement. If a major economic shock altered the return profile of the plan's assets before the measurement date but after the plan had issued its annual valuation for the prior plan year, the Companies would prohibit the plan's actuary from considering those events when calculating withdrawal liability (unless it revised its assumptions prior to the measurement date). But if the actuary failed to take account of these developments, it could not make a "reasonable" assumption based on the plan's actual experience.
Nor could the actuary give its "best estimate of anticipated experience" for the plan, if it were required to ignore such significant events. Something is "best" when it "excel[s] all others." Best, Webster's Third New International Dictionary, supra, at 208. And an estimate excels all others when it is the most accurate. The statute, accordingly, should be read to permit an actuary to make its most accurate estimate so far as is possible from the perspective of the measurement date. See Energy West, 39 F.4th at 738 (explaining that the best estimate requirement "lay[s] down . . . a substantive rule that the assumptions reflect the characteristics of the plan"). Under the Companies' contrasting view, the actuary would be reduced to admitting that its actuarial assumptions were wrong and could say only that its hands were tied by a restriction that appears nowhere in the statute. That presents not only an untenable position for the actuary but, more importantly for present purposes, an untenable reading of the statute.
The Trustees, for their part, take this theme even further and argue that the "best estimate" provision means that an actuary who is calculating withdrawal liability months or years after the measurement date may, and should, consider any and all intervening events that might materially affect the relevant, actuarial assumptions. Dkt. 34-1 at 19-20. But their argument misunderstands the relevance of actuarial assumptions in the statutory scheme. The actuary is required to make its "best estimate of anticipated experience under the plan" for purposes of determining the "unfunded vested benefits" of a plan, which, in turn, is used to calculate the employer's withdrawal liability. 29 U.S.C. § 1393(a). As explained above, unfunded vested benefits is a concept premised on both current assets and liabilities and actuarial assumptions regarding future returns, and § 1391(b)(2)(E)(i) specifies that this amount must be valued "as of" the measurement date. Id. § 1391(b)(2)(E)(i). Accordingly, the "best estimate" requirement must be understood as limited by the measurement date requirement.
The Court pauses to note that although the Companies and the arbitrators repeatedly discuss the "retroactive" application of actuarial assumptions, see, e.g., Dkt. 37 at 9, 13, the rule the Court adopts today does not allow for any more retroactivity than that which all agree the MPPAA requires. As the Court reads the statute, a withdrawing employer cannot be held liable for events that take place after the measurement date. As a result, the relevant informational inputs are set before the measurement date. It is only the actuary's application of judgment and analytical methods to that pre-existing information that occurs after the measurement date. There is nothing unusual or troubling about applying judgment to past events; that is, after all, what courts typically do. And in any event, retroactivity of this type—if that is, in fact, the right word—is inherent in the statutory scheme, which sets a measurement date that invariably antedates an employer's withdrawal and does not require an actuary to determine an employer's withdrawal liability until after the employer withdraws. See 29 U.S.C. §§ 1391(b)(2)(E)(i), 1391(b)(2)(A). The Court does not understand the Companies to take issue with this basic process, and the Court's decision does not create an extra layer of "retroactivity."
Here, moreover, the plan's actuary had already adopted the assumptions used to calculate the Companies' withdrawal liability at the time the Companies withdrew. So any elevated retroactivity concern that might arise when an actuary sets an assumption for the first time after an employer withdraws is not present, and the Court expresses no view on whether such a scenario would pose a problem. It also seems unlikely that when the Fund's trustees approved the use of a 6.5% discount rate (and use of an administrative expense load) on January 24, 2018 it relied on data generated or events occurring after the measurement date, that is December 31, 2017. But because the Court's decision precludes the reliance on information that first becomes available after the measurement date, that is an issue that can be addressed, if necessary, on remand. For present purposes, however, the Court concludes that the Companies' concerns about retroactivity are overblown and bear little, if any, relation to the facts of this case.
B.
Although the D.C. Circuit has yet to address the specific question presented in this case, it did address withdrawal liability in Combs v. Classic Coal Corp., 931 F.2d 96 (D.C. Cir. 1991). Combs considered an employer's (Classic Coal's) challenge to a plan's withdrawal liability assessment. 931 F.2d at 97. The plan's actuary had revised its discount rate assumption upward after Classic had withdrawn from the plan (and thus also after Classic's measurement date), and Classic sought the benefit of that revised assumption. Id. at 98-99. The D.C. Circuit was unpersuaded. It rejected Classic's contention that the plan's actuary was required to "consider[ ] evidence gathered after Classic's withdrawal"—i.e., the evidence that had supported the discount rate change—in setting its assumptions for Classic's withdrawal liability. Id. at 102. The court explained that a withdrawal liability calculation "is like a snapshot, in that it represents the actuary's best estimate given the evidence then available." Id. (internal quotation marks and citation omitted). As such, "[o]nce a withdrawn employer's liability is fixed, changes in [unfunded vested benefits] are irrelevant to the inquiry regarding withdrawal liability," and there is no basis for "requir[ing] the [fund's] Trustees to base their assumption on information gathered after the fiscal year-end of the Plan[ ]." Id. "Just as an employer's liability is not increased if the plan suffers losses in the withdrawal year," the court continued, "the employer is not entitled to benefit from actuarial changes subsequent to its withdrawal." Id. This narrow holding has little bearing on the question presented here, which has nothing to do with whether a plan may—or must—consider evidence gathered after the employer withdraws.
The parties' respective attempts to spin Combs in their favor are unconvincing. The Trustees observe that at one point Combs appears to approve of the plan's reliance on the information "available at the time of [Classic's] withdrawal" to calculate withdrawal liability. Dkt. 39 at 23 (quoting 931 F.2d at 102). Read in isolation, this language might be construed to support the Trustees' view that an actuary can consider developments after the measurement date, as long as those developments occur before an employer's withdrawal. But the court followed that sentence by explaining that the relevant cutoff is "the fiscal year-end of the Plan[ ]" and that events "in the withdrawal year"—that is, at any time after the measurement date—cannot be considered. Combs, 931 F.2d at 102. So, if anything, Combs supports this Court's conclusion that events occurring after the measurement date are off limits. Nor is this decision the first to read Combs this way. Judge Lamberth took the same view in M & K Employee Solutions, treating withdrawal liability as a "snapshot," contra the Trustees, and basing that snapshot on "the evidence 'then available' by 'the fiscal year-end of the [p]lan[ ].' " M & K Emp. Sols., 2022 WL 4534998, at *16 (first alteration in original) (quoting Combs, 931 F.2d at 102).
The Companies' reading of Combs is even less convincing. They first suggest that when Combs refers to "evidence" available and "information gathered" as of the measurement date, it means the actuary's discount rate assumption. Dkt. 37 at 19. That is implausible. An assumption is not itself "evidence" or "information," and one cannot "gather" an assumption. An assumption is something that an actuary makes based on its judgment and experience and the evidence and information it has gathered. Combs is crystal clear on this point, observing that "[t]o require the Trustees to base their assumptions on information gathered after the fiscal year-end of the Plans would discourage actuarial updating." 931 F.2d at 102 (emphasis both added and omitted).
The Companies' next move cuts less ice still. They argue that when Combs said an employer cannot "benefit from actuarial changes subsequent to its withdrawal," 931 F.2d at 102 (emphasis added), what it "should have" said is that an employer cannot "benefit from actuarial changes in the withdrawal year," Dkt. 37 at 19 (quoting Dkt. 37-3 at 16 (Toyota Logistics Award)). Tempting though this liberated approach to reading precedent may be, the Court is constrained to consider only what the D.C. Circuit actually said. And Combs said nothing about actuarial changes made in the withdrawal year, at least so long as they are made before day on which the employer withdrew.
It is possible that there is more to Combs' statement that an employer cannot "benefit from actuarial changes subsequent to its withdrawal" than first meets the eye, but it is not what the Companies think. 931 F.2d at 102. Read on its own, the passage suggests that an actuary may not set assumptions after an employer withdraws, even if the actuary relies only on information available as of the measurement date. But in Combs itself, the plan did not formally adopt the discount rate that the D.C. Circuit ultimately upheld until after Classic had withdrawn from the plan. Combs v. Classic Coal Corp., No. 84-1562, 1990 WL 66583, at *1-2 (D.D.C. Apr. 6, 1990) (noting that Classic withdrew from the plan in March 1981 and that the plan set its relevant discount rate assumption in May 1981). Strictly speaking, this was not an actuarial change, because it was the first withdrawal liability discount rate assumption the plan adopted after the MPPAA's passage in late 1980. Nevertheless, this factual context, combined with the fact that the statutory text does not tie actuarial assumptions to the withdrawal date, could be reason to think that adopting assumptions after an employer has withdrawn—but considering only information available and developments that had occurred by the measurement date—is permissible. The Court need not resolve this question, however, since the actuarial assumptions at issue here were adopted before the Companies withdrew.
Even putting these flaws in the parties' arguments aside, the Court would place little weight on the relevant paragraph from Combs. The question presented in that case was narrowly focused on whether the plan was required to base its actuarial assumptions on evidence gathered after the withdrawal date. Here, in contrast, the question is whether a plan is permitted to consider evidence that likely was likely available before the measurement date for purposes of making actuarial assumptions. There is no reason to believe that the D.C. Circuit gave the question presented here any consideration, and the parties' attempts to divine an answer to that question from competing phrases—or snippets of phrases—appearing in Combs almost certainly ascribes greater meaning to the court's words than the court intended.
C.
The Companies also place considerable weight on the Second Circuit's decision in Metz, as did the arbitrators. Metz, however, is neither controlling in this jurisdiction nor persuasive.
Metz involved an employer that withdrew from a multiemployer pension plan in May 2014. 946 F.3d at 147. As of that time, the plan's actuary had for several years used a 7.25% discount rate assumption to calculate withdrawal liability. Id. at 148. But the plan retained a new actuary, and in June 2014 the new actuary set a withdrawal liability discount rate assumption of approximately 3.25%. Id. The actuary then used that assumption to calculate the employer's withdrawal liability, which was tied to a December 31, 2013 measurement date. Id. at 149. The employer challenged the withdrawal liability assessment, arguing that the use of a discount rate adopted after the measurement date was impermissible. Id. The employer prevailed in arbitration, but the district court vacated the award, concluding that the statute imposed no constraint on when an actuary can set its assumptions. Id. The Second Circuit then reversed, holding that "the assumptions and methods used to calculate the interest rate assumption for purposes of withdrawal liability must be those in effect as of the [m]easurement [d]ate." Id. at 151. "Absent a change by a Fund's actuary before the [m]easurement [d]ate," it concluded, "the existing assumptions and methods remain in effect." Id.
The Metz decision relied on four non-textual arguments. First, the court observed that, "[i]n the context of multiemployer pension plans, interest rate assumptions cannot be altered daily and must have a degree of stability." Id. at 150. For support, the court pointed to the fact that the very plan at issue there had used a "7.25% rate for several years and [the plan's] annual reports to the government reflect[ed] [that] ongoing rollover" of the same rate over time. Id. Second, the court looked to a different provision of the statute, § 1394, which provides that "[n]o plan rule or amendment . . . may be applied without the employer's consent with respect to liability for a withdrawal or partial withdrawal which occurred before the date on which the rule or amendment was adopted" and requires a plan to give notice to employers "of any plan rules or amendments" that the plan adopts. 29 U.S.C. § 1394. As the Metz court explained, the legislative history of this provision "demonstrates that it was designed to protect employers from the retroactive application of rules relating to the calculation of withdrawal liability." Metz, 946 F.3d at 150. The court recognized that § 1394 "does not define 'plan rules and amendments' " and that § 1393, which governs actuarial assumptions "does not specifically address retroactivity." Id. at 151. But it nevertheless reasoned that "the retroactive selection of interest rate assumptions for purposes of withdrawal liability [is] . . . inconsistent with Congress's legislative intent." Id. Third, the court invoked yet another provision of the statute, § 1021(l)(1), which requires a plan, at an employer's request, to provide the employer a "[n]otice of potential withdrawal liability." Id. (alteration in original) (quoting 29 U.S.C. § 1021(l)(1)). In the court's view, this notice would be of "no value" if "retroactive changes in interest rate[ ] assumptions may be made at any time." Id. Finally, the Metz court posited that "the selection of an interest rate assumption after the [m]easurement [d]ate would create significant opportunity for manipulation and bias," because "[n]othing would prevent trustees from attempting to pressure actuaries to assess greater withdrawal liability on recently withdrawn employers than would have been the case if the prior assumptions and methods actually in place on the [m]easurement [d]ate were used." Id.
This Court does not see things the same way. Starting with the Metz court's first argument, the Court takes no issue with the observation that interest rate assumptions cannot fluctuate daily nor remain open forever. But that sensible insight does not compel the result in Metz. Under the rule this Court adopts today, an actuary may set its assumptions after the measurement date, but it may do so based only on information that was available as of that date. This universe of relevant evidence does not fluctuate or remain open indefinitely, and thus there is no reason to expect that an actuary following this Court's approach would have cause to alter its assumptions daily or to revise them long after the fact. To the contrary, repeated revision of assumptions that must be based on a fixed universe of available facts and that must represent the actuary's "best estimate of anticipated experience under the plan" would be suspect and open to challenge under the best estimate requirement. See 29 U.S.C. § 1393(a)(1). Although it is, of course, conceivable that after setting an assumption "as of" the measurement date once, an actuary might reinterpret the pre-existing evidence and change its mind, there is no reason to believe that this practice would be commonplace, particularly given the burden of justification that such an actuary would face if its decision was contested. The prospect that an actuary might, on occasion, alter its assumptions based on evidence that comes to light between an earlier determination and the measurement date, moreover, is precisely what the statute contemplates.
The Court is also unpersuaded by the inference the Metz court drew from § 1394 and its legislative history. Metz concluded that because § 1394 expressly limits retroactivity for changes to plan rules and amendments, the statute should be read also to limit retroactivity for changes to actuarial assumptions—even though § 1393, the provision dealing with actuarial assumptions, contains no such limitation. But as Judge Lamberth explained in M & K Employee Solutions, "[t]he presence of an anti-retroactivity provision in the section dealing with plan rules and amendments, and the absence of one in the section dealing with actuarial assumptions, suggests that anti-retroactivity was purposefully omitted in the latter." 2022 WL 4534998, at *18; see also Barnhart v. Sigmon Coal Co., 534 U.S. 438, 452, 122 S.Ct. 941, 151 L.Ed.2d 908 (2002) ("[W]hen Congress includes particular language in one section of a statute but omits it in another section of the same Act, it is generally presumed that Congress acts intentionally and purposely in the disparate inclusion or exclusion." (internal quotation marks omitted)). But even if actuarial assumptions like the discount rate and expense load were plan "rules or amendments"—the Companies (wisely) do not argue that they are—and thus § 1394 was directly applicable to them, its retroactivity provision would make no difference in this case. Section 1394 limits the application of a rule or amendment to an employer's withdrawal liability only if the rule or amendment is adopted after the employer withdrew. 29 U.S.C. § 1394(a). The disputed assumptions here, however, were set before any of the Companies withdrew. Dkt. 34-2 at 4-5 (Pl.'s SUMF ¶¶ 13-14, 20).
More generally, Metz's § 1394-based appeal to legislative purpose fails to consider what was, according to Congress, its predominant purpose in enacting the MPPAA: preserving and protecting multiemployer pension plans and their beneficiaries. See 29 U.S.C. § 1001a(c) ("It is hereby declared to be the policy of this Act . . . to alleviate certain problems which tend to discourage the maintenance and growth of multiemployer pension plans, to provide reasonable protection for the interests of participants and beneficiaries of financially distressed multiemployer pension plans, and to provide a financially self-sufficient program for the guarantee of employee benefits under multiemployer plans."). That predominant purpose is served by permitting actuaries to make their best estimates of the "anticipated experience" of the plan on the measurement date, rather than tying their hands based on possibly outdated assumptions that no longer reflect expected performance. Id. § 1393(a)(1).
This is not to say that Congress was unconcerned with unfair withdrawal liability assessments. The problem with the Metz decision, however, is that it emphasizes this concern over the principal concern that the MPPAA was enacted to address and that it does so at the expense of the statutory text. See W. Va. Univ. Hosps., Inc. v. Casey, 499 U.S. 83, 98, 111 S.Ct. 1138, 113 L.Ed.2d 68 (1991) ("The best evidence of [statutory] purpose is the statutory text adopted by both Houses of Congress and submitted to the President."). Nor does Metz consider the statute's alternative means of ensuring that employers are treated fairly, most notably the requirement that actuaries employ "reasonable" actuarial assumptions based on "the experience of the plan and reasonable expectations" and that they "offer" their "best estimate[s]" of "anticipated experience." 29 U.S.C. § 1393(a)(1).
The Court must also part ways with Metz on the significance of the withdrawal liability notice provision. That provision states in relevant part that a "plan shall, upon written request, furnish to any employer who has an obligation to contribute to the plan a notice of . . . the estimated amount which would be the amount of such employer's withdrawal liability . . . if such employer withdrew on the last day of the plan year preceding the date of the request." 29 U.S.C. § 1021(l)(1)(A) (emphasis added). According to Metz, any such notice would be "of no value if retroactive changes in interest rate[ ] assumptions may be made at any time." 946 F.3d at 151.
The Metz court's argument, however, misunderstands how § 1021(l)(1)(A) operates. As § 1021(l)(1)(A) provides, if an employer makes a request in year X, the plan must provide it with an estimate of what its withdrawal liability would be had the employer withdrawn on the last day of year X - 1. That is, for purposes of the notice provision, the plan must act on the assumption that the relevant measurement date is the last day of year X - 2. But, if the employer actually withdrew in year X, its measurement date would be the last day of year X - 1. See 29 U.S.C. § 1391(b)(2)(E)(i). As a result, any estimate provided according to the terms of § 1021(l)(1)(A) will be based on information and assumptions that are a year out of date, as compared to the employer's actual measurement date for any given year.
The Metz court is correct that its holding removes an element of surprise. If plans are required to use actuarial assumptions that are adopted prior to the measurement date, employers will know with certainty what actuarial assumptions will apply if they elect to withdraw—the assumptions would be fixed and known on the measurement date, which, by definition, will precede the decision to withdraw. But that fact has nothing to do with the operation of § 1021(l)(1)(A), which looks back a year earlier and does nothing to guard against changes that may occur up to the measurement date. Indeed, if anything, § 1021(l)(1)(A) cuts against the reasoning in Metz and the Companies' argument here, because the procedure that Congress made available to employers to obtain estimates to employers of their withdrawal liability provides no assurances regarding—and takes no steps to protect employers from—potentially evolving actuarial assumptions. See M & K Emp. Sols., 2022 WL 4534998, at *18 ("Congress did not require that employers receive the applicable actuarial assumptions prior to a decision to withdraw, which is consistent with a statutory scheme that allows for formulation of those assumptions after the measurement date.").
Finally, the potential for "manipulation and bias" does not justify the result in Metz. 946 F.3d at 151. For one thing, as Metz itself acknowledged, the Supreme Court in Concrete Pipe expressed a less suspicious view of actuaries than did the Second Circuit. Metz, 946 F.3d at 151-52. Concrete Pipe explained that actuaries are "apparently unbiased professional[s], whose obligations tend to moderate any claimed inclination to come down hard on withdrawing employers" and who are not "vulnerable to suggestions of bias or its appearance" because they are "trained professionals subject to regulatory standards." 508 U.S. at 632, 635, 113 S.Ct. 2264.
Moreover, even if one took a less rosy view of the actuarial profession, the statute has ways of protecting against bias and manipulation. As explained above, § 1393(a) requires that actuaries calculating withdrawal liability employ assumptions and methods that are "reasonable" and that "offer the actuary's best estimate of anticipated experience under the plan." An actuarial assumption adopted at the insistence of a plan's trustees that ignores professional standards or that is analytically unsound can be attacked under this provision twice over. First, unfounded actuarial assumptions or those that are changed simply to suit the trustees' preferences at any given time can be attacked as "unreasonable[,] both in arbitration and on judicial review." Concrete Pipe, 508 U.S. at 633, 113 S.Ct. 2264. The consequences of actuarial assumptions, moreover, can vary based on circumstances. "For example, the use of assumptions (such as low interest rates) that would tend to increase the fund's unfunded vested liability for withdrawal liability purposes would also make it more difficult for the plan to meet the minimum funding requirements of § 1082." Id. Yet, any effort by the plan's trustees (or, more accurately, their actuary) to manipulate the relevant assumptions to achieve favorable results on both counts, without intervening cause, would invite a compelling reasonableness challenge. See Energy West, 39 F.4th at 741 (holding that a plan's minimum funding rate and withdrawal liability discount rate assumptions "must be similar"). Second, in addition to its substantive requirements, the best estimate clause contains a "procedural rule" of actuarial independence—namely, that the actuary's assumptions be made by the actuary, not by a plan's trustees. Id. at 738; see also CPC Logistics, 698 F.3d at 357 (noting that the best estimate requirement "exists to maintain the actuary's independence"). Congress struck the balance between the need for actuarial flexibility—and thus accuracy—and outright manipulation by giving professional actuaries the necessary leeway, checked by the reasonableness requirement.
Nor does Metz solve the problem of actuarial bias. As Judge Lamberth pointed out, plans and their actuaries committed to manipulation have tools at their disposal no matter how one understands the measurement date requirement. The Metz rule, for example, does nothing to stop a plan from slow rolling an update to its assumptions until after the next measurement date in order to lock in for another year assumptions that might be more punitive to employers than is justified under current conditions. See M & K Emp. Sols., 2022 WL 4534998, at *18. Combs itself worried about such a scenario, cautioning against interpreting the measurement date requirement in a way that would "discourage actuarial updating." 931 F.2d at 102. In sum, the MPPAA's text reflects a balance struck by Congress between the competing considerations of actuarial flexibility and fairness to employers, and it is not for this Court to rewrite that legislative balance.
D.
The parties' other arguments are equally unavailing. The Companies invoke Roofers Local No. 30 Combined Pension Fund v. D.A. Nolt, Inc., 719 F. Supp. 2d 530 (E.D. Pa. 2010), aff'd, 444 F. App'x 571 (3d Cir. 2011), but that decision has little to do with this case. Dkt. 37 at 14-15; Dkt. 38-2 at 13. Nolt rejected a plan's reliance on an error in its unfunded vested benefits calculation discovered in 2003 to revise an employer's withdrawal liability in 2006, where the plan had previously calculated the employer's liability in 2002 based on a December 31, 2000 measurement date. 719 F. Supp. 2d at 539-40, 550-51. Nolt therefore did not address the only question presented here: whether an actuary can initially set its withdrawal liability assumptions after the measurement date. Nor does the Nolt court's reasoning help in resolving this case. Nolt relied almost entirely on two PBGC opinion letters that speak only to the propriety of a plan making "additional assessment[s]" of withdrawal liability if errors in the unfunded vested benefits calculation are uncovered; neither opinion letter addresses the constraints on an actuary making an assessment in the first instance. See PBGC Op. Letter 94-5 at 1; PBGC Op. Letter 90-2 at 2. No broadly applicable rule can be extracted from these letters and therefore from Nolt.
The Trustees, on the other hand, argue that the entire debate about the measurement date is beside the point, because actuarial assumptions can only be attacked as unreasonable or at odds with the actuary's "best estimate," not on the basis of a measurement date violation. Dkt. 34-1 at 6, 15-17 (citing 29 U.S.C. § 1393(a)(1)). But this contention begs the question of what makes an assumption unreasonable. And the Court has little doubt that an assumption that violates a provision of the MPPAA is unreasonable by the MPPAA's lights.
Finally, the parties joust over the significance of a MPPAA provision that permits an actuary "[i]n determining the unfunded vested benefits" to "rely on the most recent complete actuarial valuation used for purposes of section 412 of title 26 [a provision of the IRS code] and reasonable estimates for the interim years of the unfunded vested benefits." 29 U.S.C. § 1393(b)(1). The Trustees say that this provision shows than an actuary can rely on information learned after the measurement date. Dkt. 34-1 at 18; Dkt. 39 at 13-15. After all, a plan's "most recent" actuarial valuation may postdate the most recent measurement date and, for that matter, the relevant company's withdrawal date. Dkt. 39 at 13-15. And nothing in the statute purports to limit what information an actuary may consider when creating an actuarial valuation for reporting purposes. Id. The Companies, in contrast, insist that § 1393(b)(1) should be understood to refer to "the most recent complete actuarial valuation" that predates the measurement date. Dkt. 37 at 16-18. They point out that the statute's reference to "reasonable estimates for the interim years of the unfunded vested benefits" presumes that there will be "interim years" between an actuarial valuation used for this purpose and the measurement date to which unfunded vested benefits must be tied. Id.
Both arguments fall short. The Trustees' position that § 1393(b)(1) frees actuaries from any constraints imposed by the measurement date fails for several reasons. For starters, if the Trustees were right, there would be no point to the statute's imposition of a measurement date in the first place, and the Court must, to the extent possible, avoid reading one provision of a statute to nullify the significance of another. See TRW Inc. v. Andrews, 534 U.S. 19, 31, 122 S.Ct. 441, 151 L.Ed.2d 339 (2001) (rejecting an interpretation of a statute that would leave a provision "insignificant, if not wholly superfluous"). The Trustees' argument also proves too much. If § 1393(b)(1) was as freeing as they claim, an actuary could presumably use not only the actuarial assumptions from the most recent actuarial valuation, but also the asset and liability values it contains, even if these values postdate the measurement date. But the Trustees concede that an actuary must rely on what those figures were on the measurement date, Dkt. 34-1 at 21, and much like the provision that establishes the measurement date, § 1391(b)(2)(E)(i), § 1393(b)(1) provides no basis for treating assets and liabilities differently from actuarial assumptions. Finally, the Trustees' analysis, even if persuasive on a blank slate, cannot be squared with Combs' statement that "an employer's liability is not increased if the plan suffers losses in the withdrawal year." 931 F.2d at 102. If the inference the Trustees draw from § 1393(b)(1) were sound, an employer's liability could increase if the plan suffers losses in the withdrawal year, so long as the plan's actuary calculated the employer's withdrawal liability based on its actuarial valuation for that year, which the Trustees contend it would be free to do.
The Companies' position, in turn, would do little to help them, even if it was correct. Subsection 1393(b)(1) states than actuary "may rely on the most recent complete actuarial valuation." 29 U.S.C. § 1393(b)(1) (emphasis added). So even if the Court read into § 1393(b)(1) the words "the most recent complete actuarial valuation that predates the measurement date"—and it is far from clear that the reference to "interim years" provides sufficient basis to so fundamentally rewrite the provision—the Companies would be no better off, because use of that valuation would be an option, not a requirement. An actuary would remain free to conduct a fresh valuation, and, for the many reasons given, could rely on all information present on the measurement date in so doing.
CONCLUSION
The Court, accordingly, concludes that when setting actuarial assumptions applicable to a given measurement date, an actuary may look to the information that was available as of the measurement date but may look no further. So long as the actuary adheres to this rule, it may set its assumptions after the measurement date.
Because all three arbitrators applied a contrary rule, the Court will VACATE the arbitration awards. It will also REMAND the cases to their respective arbitrators to determine whether the Fund's actuary selected the disputed interest rate and expense load assumptions based only on information that was available as of the measurement date, and to resolve any further challenges the Companies have to their assessed withdrawal liability. The Court, accordingly, will GRANT the Trustees' motion for summary judgment, Dkt. 34, and DENY the Companies' cross-motion, Dkt. 38.
A separate order will issue.