Thomas Reichert v. Kellogg Co.

• Court: Court of Appeals for the Sixth Circuit
• Decided: March 16, 2026
• Docket: 24-1442
• Status: Published

Opinion

RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 26a0081p.06

UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT

┐ 24-1442 │ THOMAS N. REICHERT; STUART R. BUCK; KENNETH A. │ HENRICH, on behalf of themselves and all others similarly │ situated, │ Plaintiffs-Appellants, > No. 24-1442 │ │ v. │ │ KELLOGG COMPANY, et al., │ Defendants, │ │ BAKERY, CONFECTIONARY, TOBACCO WORKERS AND GRAIN │ MILLERS PENSION COMMITTEE; KELLANOVA, fka Kellogg │ Company; WK KELLOGG COMPANY; THE ADMINISTRATIVE │ COMMITTEE OF KELLANOVA PENSION PLAN; JOHN DOES 1–20, │ Defendants - Appellees. │ ┘

Appeal from the United States District Court for the Eastern District of Michigan at Detroit. No. 2:23-cv-12343—Stephen J. Murphy III, District Judge.

24-5945 ┐ │ ROBERT A. WATT; GARY J. FRIESEN; MICHAEL H. MCKENNA; │ GEOFFREY B. COE; CRAIG A. COVIC, │ Plaintiffs-Appellants, │ > No. 24-5945 │ v. │ │ FEDEX CORPORATION; FEDEX CORPORATION EMPLOYEES’ │ PENSION PLAN; RETIREMENT PLAN INVESTMENT BOARD OF │ FEDEX CORPORATION; JOHN/JANE DOES 1–10, │ Defendants-Appellees. │ ┘ Appeal from the United States District Court for the Western District of Tennessee at Memphis. Nos. 2:23-cv-02593; 2:23-cv-02516—John Phipps McCalla, District Judge. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 2 Watt, et al. v. FedEx Corp., et al.

Argued: May 8, 2025

Decided and Filed: March 16, 2026

Before: STRANCH, BUSH, and NALBANDIAN, Circuit Judges.

_________________

COUNSEL

No. 24-1442: ARGUED: Rachana Pathak, STRIS & MAHER LLP, Cerritos, California, for Appellants. Joseph J. Torres, JENNER & BLOCK LLP, Chicago, Illinois, for Appellees. Jeremy P. Blumenfeld, MORGAN, LEWIS & BOCKIUS LLP, Philadelphia, Pennsylvania, for Appellees. ON BRIEF: Rachana Pathak, Peter K. Stris, Douglas D. Geyser, STRIS & MAHER LLP, Cerritos, California, Oren Faircloth, Lisa R. Considine, David J. DiSabato, SIRI & GLIMSTAD LLP, New York, New York, for Appellants. Joseph J. Torres, Alexis E. Bates, Emma O’Connor, JENNER & BLOCK LLP, Chicago, Illinois, for Appellees. No. 24-5945: ARGUED: Louis M. Bograd, MOTLEY RICE, LLC, Washington, D.C., for Appellants. Jeremy P. Blumenfeld, MORGAN, LEWIS & BOCKIUS LLP, Philadelphia, Pennsylvania, for Appellees. ON BRIEF: Louis M. Bograd, MOTLEY RICE, LLC, Washington, D.C., Douglas P. Needham, Mathew P. Jasinski, M. Zane Johnson, MOTLEY RICE LLC, Hartford, Connecticut, Oren Faircloth, Lisa R. Considine, SIRI & GLIMSTAD LLP, New York, New York, Robert A. Izard, IZARD, KINDALL & RAABE LLP, West Hartford, Connecticut, for Appellants. Jeremy P. Blumenfeld, Jared R. Killeen, MORGAN, LEWIS & BOCKIUS LLP, Philadelphia, Pennsylvania, Matthew A. Russell, MORGAN, LEWIS & BOCKIUS LLP, Chicago, Illinois, Terrence O. Reed, Joseph B. Reafsnyder, FEDERAL EXPRESS CORPORATION, Memphis, Tennessee, Michael E. Kenneally, MORGAN, LEWIS & BOCKIUS LLP, Washington, D.C., for Appellees. Nos. 24-1442/5945: ON AMICUS BRIEF : Nancy J. Ross, Michael S. Scodro, Joshua D. Yount, MAYER BROWN LLP, Chicago, Illinois, for Amici Curiae.

STRANCH, J., delivered the opinion of the court in which BUSH, J., concurred. NALBANDIAN, J. (pp. 21–35), delivered a separate dissenting opinion. _________________

OPINION _________________

JANE B. STRANCH, Circuit Judge. Retired employees of the Kellogg Company and FedEx Corporation sued their respective employers and pension plans for violating the Employee Retirement Income Security Act (ERISA). Plaintiffs, each of whom are married participants in employee pension plans, alleged that their plans used outdated mortality data to calculate their Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 3 Watt, et al. v. FedEx Corp., et al.

benefits, which improperly decreased those benefits in violation of ERISA’s actuarial equivalence requirement. The district courts dismissed the actions on the ground that ERISA does not require pension plans to use particular mortality tables or actuarial assumptions when calculating benefits for married participants. We REVERSE and REMAND for further proceedings.

I. BACKGROUND

A. Statutory and Regulatory Background

Congress enacted ERISA in 1974 to “protect . . . the interests of participants in employee benefit plans and their beneficiaries” by establishing substantive requirements for employee benefit plans and “by providing for appropriate remedies” in federal court. 29 U.S.C. § 1001(b); see Pub. L. No. 93-406, 88 Stat. 829. Although ERISA does not “require[] employers to establish” benefit plans or “mandate what kind of benefits employers must provide,” it ensures that “employees will not be left empty-handed once employers have guaranteed them certain benefits.” Lockheed Corp. v. Spink, 517 U.S. 882, 887 (1996). ERISA’s requirements arise, in large part, from the common law of trusts—a body of law that courts have long utilized to impose and enforce obligations on fiduciaries of private welfare and pension plans. Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 110–11 (1989).

Under ERISA, as in trust law, fiduciaries are charged with protecting the interests of, and managing the money owed to, plan participants and their beneficiaries. See Akers v. Palmer, 71 F.3d 226, 229 (6th Cir. 1995). Such fiduciaries are named by the plan instrument, 29 U.S.C. § 1102(a)(2), and they must “discharge [their] duties” with “the care, skill, prudence, and diligence under the circumstances then prevailing that a prudent man acting in a like capacity and familiar with such matters would use,” id. § 1104(a)(1).

A key aspect of ERISA is its regulation of traditional, private pension plans. Under ERISA, a pension plan may be either a defined contribution plan, such as a 401(k), or a defined benefit plan, in which the plan guarantees a certain pension benefit upon retirement. See West v. AK Steel Corp., 484 F.3d 395, 398–99 (6th Cir. 2007); Smith v. CommonSpirit Health, 37 F.4th 1160, 1162 (6th Cir. 2022). This case deals with the latter type, defined benefit plans. Once an Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 4 Watt, et al. v. FedEx Corp., et al.

employer has promised a defined pension benefit to an employee, ERISA requires, among other things, that the employer adhere to specific vesting requirements and timely pay the promised, vested benefit to the employee. See, e.g., 29 U.S.C. § 1053 (imposing vesting requirements on private pension plans).

Relevant to this case, ERISA provides for different types of defined pension benefits depending on whether an employee is single or married. The Act requires that pension plans offer unmarried participants a single life annuity (SLA), in which a participant receives a defined benefit payment for the duration of his or her own life. See Spirt v. Tchrs. Ins. & Annuity Ass’n, 691 F.2d 1054, 1058 n.1 (2d Cir. 1982); Esden v. Bank of Boston, 229 F.3d 154, 159 (2d Cir. 2000). For married participants, however, ERISA requires plans to offer as an option a joint and survivor annuity (JSA). A JSA ensures that, should a married participant predecease his or her spouse, the spouse will continue to receive benefits for the remainder of his or her life. See Shields v. Reader’s Digest Ass’n, Inc., 331 F.3d 536, 539 (6th Cir. 2003). A participant receiving a JSA is likely to receive a lower monthly payment than a participant receiving an SLA because, under a JSA, “the payments are likely to continue for a longer period of time because of the survivor benefits component.” Id. at 539 n.2.

ERISA also places requirements on the JSAs offered to plan participants. Employers providing employee pension plans must offer married participants an “accrued benefit” in the form of a “qualified joint and survivor annuity” (QJSA). 29 U.S.C. § 1055(a)(1). Section 1055(d) defines a QJSA as:

an annuity . . . for the life of the participant with a survivor annuity for the life of the spouse which is not less than 50 percent of (and is not greater than 100 percent of) the amount of the annuity which is payable during the joint lives of the participant and the spouse.

Id. § 1055(d)(1)(A). In other words, a QJSA is a statutorily prescribed form of JSA that, like all JSAs, consists of an annuity (i.e., a fixed monthly payment) starting at the participant’s retirement and continuing through the “joint” lives of the participant and his or her spouse, followed by a “survivor annuity” (i.e., another fixed monthly payment) to the participant’s spouse for the remainder of the spouse’s life. That survivor annuity must be no less than 50% Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 5 Watt, et al. v. FedEx Corp., et al.

and no greater than 100% of the annuity paid during the participant’s lifetime.1 Critically, § 1055(d) specifies that a QJSA must be the “actuarial equivalent of a single annuity for the life of the participant.” Id. § 1055(d)(1)(B).

Though the QJSA is the default benefit option for a married participant, see id. § 1055(c)(1)(A), § 1055 requires employers to permit a participant, with the consent of his or her spouse, to waive the QJSA, id. The plan must then offer, as a minimum alternative, a “qualified optional survivor annuity” (QOSA). Id. § 1055(c)(1)(A). The statute defines a QOSA as:

an annuity . . . for the life of the participant with a survivor annuity for the life of the spouse which is equal to the applicable percentage of the amount of the annuity which is payable during the joint lives of the participant and the spouse.

Id. § 1055(d)(2)(A). A QOSA functions in the same way as the QJSA, except that it is required to provide a survivor annuity with a different percentage than that of the default QJSA. 2 Id. § 1055(d)(2)(B)(i). Section 1055(d) requires that a QOSA, like a QJSA, be the “actuarial equivalent” of an SLA “for the life of the participant.” Id. § 1055(d)(2)(A)(ii).

ERISA does not define or elaborate on the term “actuarial equivalent.” Congress, however, has conferred on the Department of the Treasury the authority to interpret certain provisions of ERISA, including § 1055. See § 1055 (authorizing the Secretary of the Treasury to prescribe regulations under § 1055). The Department also has interpretive authority over the parallel provision to § 1055 in the Internal Revenue Code, 26 U.S.C. § 401, which establishes the requirements for defined benefit plans to maintain their tax-qualified status. The Treasury Department has issued regulations interpreting § 401, 26 C.F.R. § 1.401(a)-11, which provide, among other things:

A qualified joint and survivor annuity must be at least the actuarial equivalent of the normal form of life annuity or, if greater, of any optional form of life annuity offered under the plan. Equivalence may be determined, on the basis of consistently applied reasonable actuarial factors, for each participant or for all

1 For example, a 75% QJSA will provide the surviving spouse with a fixed monthly payment that is equal to 75% of the monthly payment received during the participant’s life. See 29 U.S.C. § 1055(d)(1)(A). 2 Specifically, if the default QJSA pays less than 75% to the surviving spouse, the QOSA must pay 75%, and if the default QJSA pays 75% or more, the QOSA must pay 50%. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 6 Watt, et al. v. FedEx Corp., et al.

participants or reasonable groupings of participants, if such determination does not result in discrimination in favor of employees who are officers, shareholders, or highly compensated.

Id. at § 1.401(a)-11(b)(2).

B. Factual Background and Procedural History

1. Kellogg Plaintiffs

Plaintiffs Thomas N. Reichert, Stuart R. Buck, and Kenneth A. Henrich are retired employees of the Kellogg Company. Each is a married participant of one of two defined benefit pension plans, (1) the Kellanova Pension Plan and (2) the Bakery, Confectionary, Tobacco Workers and Grain Millers (BCTGM) Pension Plan (together, the “Kellogg Plans”).3 Both Plans offer married participants three JSAs that purportedly meet the criteria for QJSAs—a 50% default JSA, a 75% JSA, and a 100% JSA. Reichert and Buck are participants in the BCTGM Plan and elected to receive a 50% JSA. Henrich is a participant in the Kellanova Plan, through which he opted into a 100% JSA.

To convert an SLA to a QJSA, the Kellogg Plans use an interest rate and mortality table. Interest rates account for the time value of money by “discount[ing] the value of expected future payments to the date of the calculation.” R. 22, Am. Compl., PageID 122. Mortality tables estimate expected longevity of the participant by “predict[ing] the rate at which retirees will die at any given age.” Id. Both variables are combined to generate a “conversion factor,” which is used to determine the benefit that would be actuarially equivalent to the SLA for a given participant. Assume, for example, a married participant with a 50% QJSA, who, if single, would receive a monthly SLA benefit of $1,000. Applying a conversion factor of .90, the joint annuity would be $900 per month, and after the participant’s death, his spouse would receive $450.

To determine the applicable conversion factor, the Kellogg Plans use the Uninsured Pensioners (UP) 1984 Mortality Table, which is based on mortality data from the 1960s and

3 In 2023, Kellogg split into two entities: Kellanova and WK Kellogg Co. The Kellanova and BCTGM Pension Plans, which were previously both sponsored by Kellogg, are now sponsored by Kellanova and WK Kellogg, respectively. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 7 Watt, et al. v. FedEx Corp., et al.

1970s. Because mortality rates have improved over the last several decades, the use of this decades-old mortality table produces higher mortality rates that assume fewer total payments for the single-life comparator SLA, resulting in a lower conversion factor for the JSAs. The lower conversion factor decreases the monthly payment that Plaintiffs receive. If the Kellogg Plans were to use contemporary mortality data, each Plaintiff would receive a higher monthly payment.

On September 14, 2023, Reichert, Buck, and Henrich, on behalf of themselves and others similarly situated, filed a class action suit in federal court. On December 15, 2023, Plaintiffs filed an amended class action complaint against the Kellogg Plans and their administrators (collectively, “Kellogg”). Plaintiffs alleged violation of 29 U.S.C. § 1055 on the ground that the Kellogg Plans’ use of outdated mortality tables and actuarial assumptions failed to provide them with QJSAs that were actuarially equivalent to SLAs as required by § 1055(d), which resulted in the systematic underpayment of benefits. Plaintiffs also alleged that the Kellogg Plans’ failure to comply with § 1055(d) constituted a breach of fiduciary duty under 29 U.S.C. § 1104. Kellogg moved to dismiss Plaintiffs’ complaint for failure to state a claim. The district court dismissed the complaint in full, holding that Plaintiffs’ claims failed as a matter of law because § 1055 does not require the use of particular mortality tables or actuarial assumptions.

2. FedEx Plaintiffs

Plaintiffs Robert A. Watt, Gary J. Friesen, Michael H. McKenna, Geoffrey B. Coe, and Craig A. Covic are former employees of the FedEx Corporation and married participants in the FedEx Corporation Employees’ Pension Plan (FedEx Plan), a defined benefit pension plan. The FedEx Plan offers married participants three JSAs that, it asserts, meet the criteria for QJSAs: a 50% default JSA, a 75% JSA, and a 100% JSA. Pursuant to the FedEx Plan, Watt, Friesen, and Coe receive the Plan’s 100% JSA, while Covic and McKenna receive the 50% JSA.

Like the Kellogg Plans, the FedEx Plan uses an interest rate and a mortality table to generate a conversion factor, which is then used to calculate the QJSA that would be actuarially equivalent to a participant’s payments under an SLA. The FedEx Plan uses the UP 1984 Mortality Table (the same one Kellogg uses) and the 1971 GAM Mortality Table, which is based Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 8 Watt, et al. v. FedEx Corp., et al.

on mortality data from the 1960s. As with the Kellogg Plans, the FedEx Plan’s use of outdated mortality tables results in a lower monthly payment of benefits to each Plaintiff.4

On September 19, 2023, Watt, Friesen, McKenna, Coe, and Covic, on behalf of themselves and others similarly situated, filed a class action suit in federal court. On November 6, 2023, Plaintiffs filed an amended complaint against the FedEx Plan and its administrators (collectively, “FedEx”). The amended complaint alleged violation of § 1055 based on FedEx’s use of outdated mortality data and consequent failure to offer Plaintiffs QJSAs within the meaning of § 1055(d). And it alleged breach of fiduciary duty under § 1104 based on the violation of § 1055(d). FedEx moved to dismiss the complaint for failure to state a claim. The district court concluded that § 1055 does not require the use of particular actuarial assumptions and dismissed Plaintiffs’ complaint.

II. ANALYSIS

Both district courts had federal question jurisdiction under 28 U.S.C. § 1331. Plaintiffs, in turn, timely appeal the district courts’ final judgments, conferring this court with jurisdiction to hear the appeals under 28 U.S.C. § 1291.

A district court’s order of dismissal for failure to state a claim under Federal Rule of Civil Procedure 12(b)(6) is reviewed de novo. Jones v. City of Cincinnati, 521 F.3d 555, 559 (6th Cir. 2008). We must “construe the complaint in the light most favorable to the plaintiff, accept all well-pleaded factual allegations as true, and examine whether the complaint contains ‘sufficient factual matter, accepted as true, to state a claim to relief that is plausible on its face.’” Ohio Pub. Emps. Ret. Sys. v. Fed. Home Loan Mortg. Corp., 830 F.3d 376, 382–83 (6th Cir. 2016) (quoting Ashcroft v. Iqbal, 556 U.S. 662, 678 (2009)).

4 The FedEx Plan also applies “setbacks” that further diminish the amount of benefits each Plaintiff receives. Specifically, the Plan applies a three-year “setback” for beneficiary spouses under the UP 84 Mortality Table and a six-year “setback” for beneficiary spouses under the 1971 GAM Mortality Table. A “setback” “subtracts a specified number of years from a standard mortality table for purposes of calculating benefits”—for example, “if there is a 65-year old retiree who has a spouse that is also 65, but the plan states that there is a [three]- year setback for beneficiaries, then, for purposes of calculating benefits, the plan uses age 62 for the beneficiary’s age.” Using a younger age for the spouse overestimates the likelihood that the spouse will outlive the participant, which further reduces the conversion factor and, in turn, the participant’s monthly payments. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 9 Watt, et al. v. FedEx Corp., et al.

The core issue in these appeals is whether Plaintiffs sufficiently pleaded claims that Defendants’ use of outdated mortality tables violated § 1055(d)’s actuarial equivalence requirement. FedEx also raises two alternative bases for dismissal of the claims against it, neither of which were decided by the district court. We address each issue in turn.

A. Section 1055(d)’s Actuarial Equivalence Requirement

Plaintiffs assert that their employers’ use of unreasonably outdated mortality data to calculate their benefits violates ERISA. Specifically, they allege that the use of such data fails to produce JSAs that are actuarially equivalent to the SLAs they would otherwise receive, thereby denying them the QJSAs to which they are entitled under § 1055. Defendants argue that Plaintiffs’ claim fails as a matter of law because § 1055(d)(1)(B)’s actuarial equivalence requirement places no restrictions on the actuarial assumptions that a pension plan may use when calculating QJSAs. The viability of Plaintiffs’ claims, therefore, hinges on whether § 1055(d)(1)(B) prohibits the use of unreasonably outdated actuarial assumptions.

As always, our analysis begins with the “statutory text.” United States ex rel. Felten v. William Beaumont Hosp., 993 F.3d 428, 431 (6th Cir. 2021). We “interpret[] a statute in accord with the ordinary public meaning of its terms at the time of its enactment.” Bostock v. Clayton Cnty., 590 U.S. 644, 654 (2020). To do so, we must look to “the language itself” and “the specific context in which that language is used.” Felten, 993 F.3d at 431 (quoting Robinson v. Shell Oil Co., 519 U.S. 337, 340–41 (1997)). Should the text prove “unclear,” we may look to the statute’s broader context, structure, and purpose to resolve the ambiguity. Felten, 993 F.3d at 431.

1. Section 1055(d) Prohibits the Use of Unreasonably Outdated Actuarial Assumptions

Section 1055(d) specifies that a QJSA must be the “actuarial equivalent of a single annuity for the life of the participant.” 29 U.S.C. § 1055(d)(1)(B). As all parties recognize, the term “single annuity” refers to the SLA benefits that a participant would receive in lieu of a QJSA. Section 1055 also makes clear that the relevant SLA must be “for the life of the participant.” Id. In other words, the QJSA must be actuarially equivalent to the amount of SLA Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 10 Watt, et al. v. FedEx Corp., et al.

benefits that a participant—here, each Plaintiff—would have received over the course of his or her life, from retirement to death.

The critical interpretive issue is the meaning of the term “actuarial equivalent” at the time of ERISA’s enactment in 1974. See Bostock, 590 U.S. at 655. The term “actuarial equivalent” has long been utilized in actuarial practice. See Walter J. Couper & Roger Vaughan, Pension Planning: Experience and Trends 189 (1954). We therefore evaluate this “term of art” as it was understood by actuarial scientists when ERISA was enacted. See Corning Glass Works v. Brennan, 417 U.S. 188, 201 (1974); see also Stephens v. U.S. Airways Grp., 644 F.3d 437, 440 (D.C. Cir. 2011) (treating “actuarial equivalence,” as used in 29 U.S.C. § 1054(c), as a “term of art”); United States v. Hansen, 599 U.S. 762, 774 (2023) (“[W]hen Congress ‘borrows terms of art in which are accumulated the legal tradition and meaning of centuries of practice, it presumably knows and adopts the cluster of ideas that were attached to each borrowed word.’”).5

Before ERISA’s passage in 1974, actuarial scientists understood the term “actuarial equivalent” to describe a situation in which one benefit has an equal present value to another benefit after accounting for certain actuarial assumptions, such as mortality rates. See Couper & Vaughan, supra, at 189; accord Michael Puchek, Pension Plan Policies and Practices 30 (1952) (explaining that two benefits are the “actuarial equivalent” of one another when one benefit is “equal to the value” of the other “discounted for death and interest”). That “established meaning” persists in the modern era—as the D.C. Circuit explained in 2011, “[t]wo modes of payment are actuarially equivalent when their present values are equal under a given set of actuarial assumptions.” Stephens, 644 F.3d at 440 (citing Jeff L. Schwartzmann & Ralph Garfield, Educ. & Exam. Comm. of the Soc’y of Actuaries, Actuarially Equivalent Benefits 1, EA1-24-91, https://perma.cc/XKD8-FDLW).

5 This is, at its core, the primary problem with the dissent’s theory of this case. The dissent considers all of the terms in § 1055(d) in the sense that a layperson might. But technical terms carry their technical meanings, so we need to consider how an actuary in 1974 would have understood these terms, not how a layperson might have understood them. See NLRB v. Coca-Cola Bottling Co., 350 U.S. 264, 269 (1956) (“But if the word be deemed to have a peculiar connotation for those intimate with [actuarial sciences], it is incumbent upon us to give the word its technical meaning[.]”). Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 11 Watt, et al. v. FedEx Corp., et al.

Well before 1974, actuarial scientists recognized that ensuring equal present value between or among benefits requires mortality data that appropriately reflects the expected lives of the recipients of the benefits. For example, Plaintiffs point to a 1964 book on the actuarial science of pensions, which explained that, to convert single life annuities into “reduced pensions” such as “joint and survivor annuit[ies],” “[i]t is general practice to determine the reduced pensions on the basis of actuarial equivalence calculated on appropriate mortality tables.” William F. Marples, Actuarial Aspects of Pension Security 40 (1964) (emphasis added). Similarly, in a 1964 article, a prominent actuarial scientist explained that one of the key starting points for determining actuarial equivalence is a “suitable” mortality table. Hilary L. Seal, “Actuarial Equivalence”: An Exposition, 101 Tr. & Est. 836, 836 (1962).

In the pension context, a mortality table was understood to be “appropriate” or “suitable” when it aligned with the life expectancy of the relevant participant population. See Dorrance C. Bronson, Concepts of Actuarial Soundness in Pension Plans 36–37 (1957). As mortality rates change, so too must the mortality data that actuaries select. See Wilmer A. Jenkins & Edward A. Lew, A New Mortality Basis for Annuities, 1 Transactions of Soc’y of Actuaries 369, 369 (1949). That understanding has persisted in the decades after ERISA’s enactment. See, e.g., Schwartzmann & Garfield, supra, at 11 (explaining that “interest and mortality assumptions play a key role” in determining actuarial equivalence, and that the assumptions used must be “reviewed and modified” so that they “continue to fairly assess the cost of the optional basis of payment”).

Pre-1974 judicial decisions evinced the same understanding of actuarial equivalence. For example, in 1954, a Pennsylvania court held that “the term ‘actuarial equivalent’ . . . means calculations made at retirement according to the best actuarial information and tables then available to the actuary.” O’Dea v. Pub. Sch. Emp. Ret. Bd., 88 Pa. D. & C. 593, 604 (Pa. Ct. C.P. 1954). A Washington court similarly held in 1959 that actuarial equivalence “reasonably requires the most accurate tables available.” King Cnty. Emp. Ass’n v. State Ret. Bd., 336 P.2d 387, 391 (Wash. 1959) (en banc); see also id. (explaining that a pension fund could only be “self-sufficient,” as was required by the relevant statute, if “the mortality tables in use . . . accurately reflect the life expectancy of members when they retire”). A California court echoed Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 12 Watt, et al. v. FedEx Corp., et al.

this understanding two years before ERISA’s passage, explaining that an “actuarial[ly] equivalent” retirement benefit is one that is “reasonably expected to be paid to” the participant “based upon the actuarial expectancy of the lives of” the participant and his spouse. Berry v. Bd. of Ret., 23 Cal. App. 3d 757, 759 (1972). The actuarial literature and historical caselaw point in the same direction—actuarial equivalence requires mortality assumptions that reasonably reflect the lives of the applicable benefit recipients. That means using up-to-date mortality data, not data from half a century ago.

This makes good sense. Actuaries are professionals, and a core part of their job is to perform calculations and make predictions based on real-world conditions. See Dan M. McGill, Fundamentals of Private Pensions 218–19 (2d ed. 1964). Competent performance of this professional obligation demands that actuaries “base [their] forecasts on assumptions which at the time of the valuation seem to be the most appropriate for the case at hand.” Id. Recognition of these basic concepts is part of our duty to evaluate the term “actuarial equivalent” “by reference to the art or science to which [it is] appropriate.” Corning Glass Works, 417 U.S. at 201 (quotation omitted). Defendants’ position that pension plans can use whatever assumptions they wish, including outdated and inapplicable mortality data, does not accord with these core concepts. Actuarial equivalence, in the context of lifetime pension benefits, cannot be attained if the mortality assumptions used to calculate such benefits fail to align with the life expectancy of the benefit recipients.

The remaining text of § 1055(d)(1)(B) illustrates why reasonably accurate mortality data is necessary to achieve actuarial equivalence. Consider, again, the requirement that pension plans offer each married participant a QJSA that is the “actuarial equivalent of a single annuity for the life of the participant.” 29 U.S.C. § 1055(d)(1)(B) (emphasis added). This means that the present value of the QJSA must be equal to the present value of the SLA that such participant would otherwise receive over the course of his or her lifetime. See, e.g., Couper & Vaughan, supra, at 189. Equivalence in present value can only be achieved if the mortality assumptions used to convert the payment under the SLA to the payment under the QJSA reasonably reflect the “life of the participant” who would otherwise receive the SLA. See, e.g., Berry, 23 Cal. App. 3d at 759; Marples, supra, at 40; Seal, supra, at 836; Schwartzmann & Garfield, supra, at 11. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 13 Watt, et al. v. FedEx Corp., et al.

In this context, “the participant” is not some hypothetical, abstract individual; it is a modern-day retiree. Accurately estimating the lifespan of the relevant participant, therefore, necessarily requires the use of mortality data reasonably reflecting the life expectancy of a retiree living in the present day. Accordingly, to provide a QJSA within the meaning of § 1055(d), a plan must provide a joint and survivor annuity with an equivalent value to the single life annuity the participant would otherwise receive, and that equivalence requires reasonably accurate mortality data.

This conclusion aligns with the longstanding view of the Treasury Department, the federal agency with interpretive authority over both § 1055(d) and its parallel provision in the Internal Revenue Code, 26 U.S.C. § 401(a)(13)(D)(ii), which delineates tax-qualification requirements for defined benefit plans and contains the same actuarial equivalence requirement. In 1977, shortly after ERISA’s enactment, the Treasury Department promulgated a regulation interpreting § 401 of the IRC, including the actuarial equivalence requirement. See 26 C.F.R. § 1.401(a)-11(b)(2). That regulation, which remains operative, provides that actuarial equivalence “may be determined, on the basis of consistently applied reasonable actuarial factors, for each participant or for all participants or reasonable groupings of participants, if such determination does not result in discrimination in favor of” certain employees.6 Id. (emphasis added). Although this interpretation of the actuarial equivalence requirement arises out of the IRC, rather than § 1055(d), it remains instructive. The Treasury Department, as the agency charged with interpreting and enforcing much of ERISA, has extensive experience with and expertise over its substantive requirements, and its “roughly contemporaneous” and “consistent” understanding of the actuarial equivalence requirement is entitled to great respect. Loper Bright Enters. v. Raimondo, 603 U.S. 369, 386 (2024). This regulatory backdrop reaffirms what is clear from the statutory text—§ 1055(d)’s actuarial equivalence requirement prohibits the use of

6 FedEx argues that the use of the term “may” in the regulation renders it “permissive” as to the use of reasonable actuarial assumptions. But that misreads the text of the regulation. The term “may,” as used in the regulation, refers not to the “reasonable actuarial factors” requirement, but to the subsequent clause permitting employers to determine actuarial equivalence with respect to either each participant individually or reasonable groupings of participants. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 14 Watt, et al. v. FedEx Corp., et al.

unreasonably outdated mortality tables or other unreasonable, inappropriate actuarial assumptions.

In concluding that § 1055(d) prohibits the use of unreasonable assumptions, we note that reasonableness is a range, not a precise prescription. Actuarial science requires reasonable estimates, and different actuaries can come to different conclusions on what the exact best estimates are in each situation. See McGill, supra, at 218–19. Recognizing this reality, courts have long afforded deference to the determinations of plan actuaries when assessing reasonableness under other ERISA provisions, upholding actuaries’ chosen assumptions if they are “within the scope of professional acceptability.” Concrete Pipe & Prods. of Cali., Inc. v. Constr. Laborers Pension Tr. for S. Cali., 508 U.S. 602, 635 (1993) (discussing the express reasonableness requirement in § 1393’s withdrawal provision); accord Sofco Erectors, Inc. v. Trs. of Ohio Operating Eng’rs Pension Fund, 15 F.4th 407, 421 (6th Cir. 2021) (applying § 1393(a) and noting that an actuary’s estimates “deserve deference”). The same principle applies here. An employer would not violate § 1055 were it to use actuarial assumptions that are “within the scope of professional acceptability” in the field of actuarial science, i.e., reasonable. Concrete Pipe & Prods., 15 F.4th at 421. But if the assumptions selected by an employer do not reasonably reflect the life expectancy of its plan participants, that employer has exceeded the scope of actuarial acceptability and failed to adhere to § 1055(d)’s actuarial equivalence requirement.

With this statutory backdrop in mind, we turn back to Plaintiffs’ complaints. Plaintiffs allege that Defendants used mortality data from the 1960s and 1970s to calculate their benefits. This decades-old mortality data, they assert, is seriously outdated, and therefore does not reasonably accord with the life expectancy of a modern-day plan participant, in contravention of basic actuarial standards. As a result, the JSAs offered by Defendants to married participants have lower present values than the SLAs that these participants would otherwise receive, meaning none of the JSAs offered by Defendants meet § 1055(d)’s actuarial equivalence requirement. These allegations, accepted as true, plausibly allege that Plaintiffs were denied a Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 15 Watt, et al. v. FedEx Corp., et al.

QJSA within the meaning of § 1055(d)(1).7 Plaintiffs have thus stated plausible claims for violation of § 1055 and breach of fiduciary duty under ERISA.

The dissent’s proposed dictionary definitions of “actuarial equivalent” support our interpretation. It says that actuarial equivalence requires looking at “‘the expected remaining lifetime of the recipient’” and “‘giving adequate consideration to actuarial factors such as . . . incidence of mortality[.]’” Dissent at 22 & n.1 (citations omitted). The dissent fails to explain how a plan administrator can adequately consider the mortality rates of, and the remaining lifetime of, the recipient population using data from over half a century ago. The conclusion that plan administrators must use mortality data that reasonably reflects the mortality rates of the current recipient population is the only interpretation that makes sense given those definitions. Plaintiffs have adequately stated a claim that the plan administrators violated ERISA by using the unreasonably outdated mortality data.

2. Defendants’ Arguments to the Contrary Are Unavailing

Defendants resist this conclusion, largely on the ground that § 1055(d) does not expressly mandate that plans use “reasonable” actuarial assumptions or delineate particular actuarial factors that plans must follow. Defendants, like both courts below, contend that the absence of such language is notable, given that other sections of ERISA either expressly impose a “reasonableness” requirement or identify particular actuarial factors that plans should use when calculating other forms of payment. See 29 U.S.C. §§ 1393(a)(1), 1432(e), 1055(g), 1083(h)(1), 1085a(c)(3)(A). Because § 1055(d) lacks any such language, they assert, it cannot be read to impose any substantive limits on the actuarial assumptions selected by employer plans. As the district court in Reichert summarized, under this construction, an employer may “use[] any mortality table” it chooses without running afoul of § 1055(d). Docket No. 2:23-CV-12343, R. 36, Op. & Order, PageID 883.

7 Whether the mortality assumptions selected by Defendants indeed fail to produce actuarially equivalent benefits is a question of fact that cannot be conclusively determined on a motion to dismiss. At this stage, we simply determine that Plaintiffs have stated a plausible claim for relief under ERISA. See Ohio Pub. Emps. Ret. Sys., 830 F.3d at 382–83. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 16 Watt, et al. v. FedEx Corp., et al.

Adopting this view of the statute would render the text of § 1055(d)(1)(B) meaningless. When a court interprets a statute, it “must give effect to each clause and word.” Tiger Lily, LLC v. U.S. Dep’t of Hous. & Urban Dev., 5 F.4th 666, 669 (6th Cir. 2021). Under Defendants’ interpretation of § 1055(d), employers would have full discretion in determining the actuarial assumptions used to calculate each QJSA and, by extension, full discretion in determining the amount of benefits provided by each QJSA. That would authorize employers to select data that allows the conversion between an SLA and a QJSA to work any way the employer wants, without regard to whether the two forms of benefits have a present value that is in fact equivalent. The section’s substantive requirements would thus remain the same regardless of whether the “actuarial equivalent” statutory language is present, depriving § 1055(d)(1)(B) of meaning and significance.

Indeed, as the district court in Reichert acknowledged, Defendants’ construction would allow employers to use dramatically outdated mortality data, including a mortality table from the sixteenth century, without running afoul of ERISA. Lacking any real response to this argument, Defendants simply note that they do not currently use such outlandish and outdated data. But that does not address the core interpretive problem, which is that their construction would allow an employer to use mortality rates that are completely out of step with the mortality rate of modern-day plan participants and would, therefore, result in a QJSA significantly lower than the SLA that a participant would otherwise receive. Generally, courts should avoid interpreting statutes in a manner that could result in “absurd” outcomes. Dodd v. United States, 545 U.S. 353, 359 (2005) (quoting Hartford Underwriters Ins. Co. v. Union Planters Bank, N.A., 530 U.S. 1, 6 (2000)). Defendants’ position does precisely that by impermissibly reading § 1055(d)(1)(B) out of the statute.

Perhaps recognizing this quandary, Defendants attempt to imbue substantive meaning into § 1055(d)(1)(B) by arguing that it effectively acts as a disclosure mandate, requiring employers to disclose their actuarial assumptions in their plan documents and to adhere to those disclosed assumptions when calculating the QJSA. But that interpretation makes little sense. ERISA already requires employers to disclose plan terms in writing. See 29 U.S.C. § 1102(a)(1) (providing that employee benefit plans must be “established and maintained pursuant to a written Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 17 Watt, et al. v. FedEx Corp., et al.

instrument”); id. § 1104(a)(1)(D) (requiring that fiduciaries discharge their duties “in accordance with the documents and instruments governing the plan”). And the Internal Revenue Code, which determines whether pension plans receive favorable tax treatment, provides that “whenever the amount of any benefit is to be determined on the basis of actuarial assumptions, such assumptions” must be “specified in the plan in a way which precludes employer discretion.” 26 U.S.C. § 401(a)(25). Given this preexisting disclosure regime, Defendants’ attempt to construe § 1055(d)(1)(B) as a mere disclosure requirement would impermissibly render the provision entirely superfluous. See Hibbs v. Winn, 542 U.S. 88, 101 (2004) (noting that courts should construe a statute “so that no part will be inoperative or superfluous”).8

Defendants’ alternative theory of § 1055(d)(1)(B), moreover, is incongruous with the provision’s basic text and function. Section 1055(d)(1) is a definition provision that sets forth the substantive criteria a JSA must meet to constitute a QJSA, and those criteria include the actuarial equivalence requirement in § 1055(d)(1)(B). Interpreting § 1055(d)(1)(B) as an implicit mandate to disclose assumptions and to adhere to those disclosed assumptions imposes no substantive limitations on the nature of the benefit itself. Defendants’ construction, in short, fails to add any “mean[ing]” to the term QJSA in § 1055(d)(1).

Defendants’ reliance on other ERISA sections, which expressly include a reasonableness requirement or delineate particular actuarial factors, is similarly unavailing. It is true that “[w]here 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.” Russello v. United States, 464 U.S. 16, 23 (1983) (quoting United States v. Wong Kim Bo, 472 F.2d 720, 722 (5th Cir. 1972)). The problem for Defendants, however, is that this presumption does not apply when the sections are dissimilar,

8 FedEx, attempting to take this disclosure argument one step further, argues that reading § 1055(d)(1)(B) to require reasonable actuarial assumptions would contravene ERISA’s disclosure requirements by permitting employers to depart from the specified assumptions “whenever, in their judgment, alternative assumptions are more reasonable.” FedEx Appellee Br. 18. That is simply wrong. Requiring employers to use reasonable actuarial assumptions does not permit them to depart from the requirement that actuarial assumptions be specified in a written plan document. It simply ensures that, when employers set forth their actuarial assumptions in a written plan, they must select assumptions that produce a QJSA that is in fact actuarially equivalent to the SLA the participant would otherwise receive. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 18 Watt, et al. v. FedEx Corp., et al.

with different language and different formulations addressing different circumstances. See City of Columbus v. Ours Garage & Wreckage Serv., Inc., 536 U.S. 424, 435–36 (2002) (“The Russello presumption that the presence of a phrase in one provision and its absence in another reveals Congress’ design . . . grows weaker with each difference in the formulation of the provisions under inspection.”).

Here, each section relied upon by the district courts and Defendants is dissimilar to § 1055(d). Defendants point specifically to §§ 1393(a)(1), 1432(e), 1083(h)(1), 1085a(c)(3)(A), and 1055(g), each of which express some sort of reasonableness requirement or require the use of certain actuarial factors. But §§ 1393(a)(1), 1083(h)(1), and 1085a(c)(3) appear in different provisions with different formulations. Rather than governing benefit conversion, these sections seek to ensure that plans remain funded throughout the lives of the plan participants by setting minimum funding standards for employer plans and imposing liability on employers who seek to withdraw from multi-employer pension plans. Section 1432(e) also falls under a different provision and focuses on determining eligibility for special financial assistance. Most notably, none of these five sections, including § 1055(g), contain the term “actuarial equivalent” or the clause “for the life of the participant.” Given the absence of such language, it makes sense that Congress found it necessary to impose a reasonableness requirement in other ERISA provisions, but not § 1055(d), the text of which inherently requires mortality assumptions that reasonably reflect the life of a modern-day plan participant. And, as discussed, that actuarial equivalence requirement confers on plan actuaries reasonable discretion in their selection of mortality data, which differentiates § 1055(d) from provisions like § 1055(g) that mandate the use of particular interest rates and mortality tables.

Defendants also point to the Supreme Court’s admonition that ERISA is a “comprehensive and reticulated statute,” and that its “carefully crafted and detailed enforcement scheme provides strong evidence that Congress did not intend to authorize other remedies that it simply forgot to incorporate expressly.” Mertens v. Hewitt Assocs., 508 U.S. 248, 251, 254 (1993) (quotations omitted). Plaintiffs are not asking this court to read unwritten requirements into the statutory text. Rather, they rely on § 1055’s express statutory mandate that defined benefit plans offer married participants actuarially equivalent QJSAs within the meaning of Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 19 Watt, et al. v. FedEx Corp., et al.

§ 1055(d)(1). ERISA, in accordance with its derivation from the common law of trusts and imposition of fiduciary duties to protect plan participants and their spouses, confers the right to sue to enforce that statutory mandate. 29 U.S.C. § 1132(a)(3). Indeed, “[t]he statutory object of the qualified joint and survivor annuity provisions, along with the rest of § 1055, is to ensure a stream of income to surviving spouses.” Boggs v. Boggs, 520 U.S. 833, 843 (1997). Recognizing that § 1055(d)(1)(B) imposes substantive obligations regarding the actuarial assumptions used by employers and their pension plans helps satisfy the statutory obligation of the fiduciaries to protect the pension payments promised to married couples by providing “a stream of income to surviving spouses.” Id.

Finally, FedEx raises various policy concerns—namely, the ostensible difficulty of judicially determining what constitutes a “reasonable” set of actuarial assumptions and the claimed prospect of increased litigation that substantially interferes with employers’ pension plans. Pure policy considerations, of course, do not constitute a proper method of statutory interpretation. See Husted v. A. Philip Randolph Inst., 584 U.S. 756, 779 (2018). And they certainly cannot override a statute’s unambiguous text. See Helix Energy Sols. Grp., Inc v. Hewitt, 598 U.S. 39, 59 (2023). Even so, these concerns do not accord with core ERISA concepts or routine ERISA litigation. ERISA is heavily derived from the common law of trusts, a body of law that has long been used to impose responsibilities and obligations on fiduciaries of private pension plans. See Firestone, 489 U.S. at 110–11. Courts, particularly in the ERISA context, have frequently been, and continue to be, called upon to evaluate reasonableness, which often requires careful assessment of whether a fiduciary has acted with reasonable care in performing his or her obligations to protect the interests of, and manage the money owed to, plan participants and their beneficiaries. See id. In short, this type of litigation is well within the wheelhouse of ERISA. And, as discussed, courts have long afforded a degree of deference to the determinations of plan actuaries when assessing reasonableness. See, e.g., Concrete Pipe & Prods., 508 U.S. at 635; Sofco Erectors, 15 F.4th at 421. That reasonable deference should apply with equal force to § 1055(d).

We therefore hold that § 1055(d) prohibits employers from using unreasonable, inappropriate actuarial assumptions when calculating QJSAs. Because Plaintiffs have plausibly Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 20 Watt, et al. v. FedEx Corp., et al.

alleged that Defendants used unreasonable actuarial assumptions in calculating their benefits, thereby denying them QJSAs within the meaning of § 1055(d)(1)(B), their claims survive.

B. FedEx’s Alternative Grounds for Dismissal

On appeal, FedEx offers two additional grounds for dismissal of the claims against it, neither of which were addressed by the district court. Absent “exceptional circumstances,” this court “will not address issues on appeal that were not ruled upon below.” Maldonado v. Nat’l Acme Co., 73 F.3d 642, 648 (6th Cir. 1996); see also Stoudemire v. Mich. Dep’t of Corr., 705 F.3d 560, 576 (6th Cir. 2013) (same); Christian Healthcare Ctrs., Inc. v. Nessel, 117 F.4th 826, 856–57 (6th Cir. 2024) (noting that because this court is a “court of review, not first view,” it generally “cannot consider an issue not passed on below” (quotations omitted)). FedEx points to no exceptional circumstance, and we can think of none, that might warrant our consideration of either of these alternative grounds for dismissal. We thus leave it to the district court to address these issues in the first instance, and we express no view on the merits of either argument.

III. CONCLUSION

For the foregoing reasons, we REVERSE the judgments of the district courts and REMAND for further proceedings consistent with this opinion. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 21 Watt, et al. v. FedEx Corp., et al.

DISSENT _________________

NALBANDIAN, Circuit Judge, dissenting. The majority and I agree that, under 29 U.S.C § 1055(d), “[t]wo modes of payment are actuarially equivalent when the present values are equal under a given set of actuarial assumptions.” Maj. Op. p.10 (quoting Stephens v. U.S. Airways Grp., 644 F.3d 437, 440 (D.C. Cir. 2011). But that’s the extent of our agreement. The majority relies on a collection of academic sources and state-court caselaw to read into the statute an atextual requirement that the given set of actuarial assumptions be “reasonable.” Nothing in Article III permits us to blue-pencil a federal statute to match our policy preferences, so I respectfully dissent. And I write separately to explain why the statute’s text can’t support the majority’s holding.

I.

This inquiry should begin, and end, with the ordinary meaning of the statute’s text. Recall that § 1055(d) requires pension plans to offer qualified joint and survivor annuities that are “the actuarial equivalent of a single annuity for the life of the participant.” 29 U.S.C. § 1055(d)(1)(B), (d)(2)(A)(ii). The word “reasonable” doesn’t appear in that command. And yet the plaintiffs advocate for a standard that reads “actuarial equivalent” to mean “reasonable actuarial equivalent.” That doesn’t persuade.

Start with first principles. We look initially to Congress’s definitions to determine a statutory term’s meaning. If Congress supplied its own definition, it’s “virtually conclusive.” Dep’t of Agriculture Rural Dev. Rural Hous. Serv. v. Kirtz, 601 U.S. 42, 59 (2024) (quotation omitted). But when it hasn’t troubled itself to define its words, we generally give statutory terms their ordinary meaning. United States v. Santos, 553 U.S. 507, 511 (2008); Taniguchi v. Kan Pac. Saipan, Ltd., 566 U.S. 560, 566 (2012). And the “best place to start” the ordinary-meaning inquiry is with dictionaries in effect at the time of enactment. See J. B-K. ex rel. E.B. v. Sec’y of Ky. Cabinet for Health & Fam. Servs., 48 F.4th 721, 726 (6th Cir. 2022) (quoting Keen v. Helson, 930 F.3d 799, 802 (6th Cir. 2019)). Still, that’s just the start. We don’t lift words out of Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 22 Watt, et al. v. FedEx Corp., et al.

place and muse upon them in a sterile vacuum. See Salazar v. Paramount Glob., 133 F.4th 642, 650 (6th Cir. 2025). Rather, “it remains ‘a fundamental canon of statutory construction that the words of a statute must be read in their context and with a view to their place in the overall statutory scheme.’” Id. (quoting West Virginia v. EPA, 597 U.S. 697, 721 (2022)).

Those principles lead to only one conclusion: Congress used the term “actuarial equivalent” unencumbered by a latent connotation of “reasonableness,” as appellants and the majority would have it.

A.

Right off the bat, “reasonableness” is conspicuously absent from contemporary dictionary definitions of actuarial equivalence. Black’s Law Dictionary defines an “actuarial equivalent” as “[t]he amount of accrued pension benefits to be paid monthly or at some other interval so that the total amount of benefits will be paid over the expected remaining lifetime of the recipient.” Actuarial Equivalent, Black’s Law Dictionary (12th ed. 2024).1 So for benefit stream A to be actuarially equivalent to benefit stream B, both must pay out the same “total amount” over the “expected remaining lifetime of the recipient.” Id. Here’s how that definition plays out in translating an SLA into an actuarially equivalent QJSA. Under the appellants’ benefit plans, the SLAs are regular streams of pre-defined, periodic payments that the employee receives for the rest of his life upon retirement. And a QJSA is the same thing, with a modification: it’s a stream of periodic payments that the employee and his spouse receive for the rest of both their lives. So a QJSA that’s actuarially equivalent to an SLA pays out the same “total amount,” but distributed over a different “expected remaining lifetime”—that of both spouses, instead of just the one. Id.

Failing to find any hints of “reasonableness” in the first-order definition, let’s dig deeper into some downstream terminology. As the majority recognizes, calculating actuarial

1 To be sure, Black’s Law Dictionary’s 1968 edition—which most closely predates Congress’s enactment of ERISA—doesn’t include “actuarial equivalence” at all. That dictionary didn’t include the term until 1999. But each edition since 2014 notes that the term’s earliest usage was 1920 and gives no indication that the definition has changed over time. In addition, the Black’s definition is consistent with other, contemporaneous dictionary definitions. See, e.g., Actuarial Equivalent, Dictionary of Labor-Management Relations 9 (1957) (“A pension of equivalent value from an actuarial standpoint—i.e., giving adequate consideration to actuarial factors such as interest, form of pension, incidence of mortality, etc.”). Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 23 Watt, et al. v. FedEx Corp., et al.

equivalence requires a process. See Maj. Op. p.12 (explaining that “to convert the payment under the SLA to the payment under the QJSA,” “the present value of the QJSA must be equal to the present value of the SLA”). That process brings more terminology. And those terms carry their own meanings. But none of those meanings involve reasonableness.

Take the first step, which involves calculating the SLA’s “present value,” or “[t]he expected value of a contingent cash-flow stream, such as payments from an annuity.” Actuarial Value, Black’s Law Dictionary (12th ed. 2024). More specifically, the first step calculates a single figure, in today’s dollars, that encapsulates the value of all the fixed monthly payments the retiree would receive over his or her remaining lifetime. See Actuarial Present Value, Black’s Law Dictionary (12th ed. 2024); see Present Worth, Ballentine’s Law Dictionary (3d ed. 1969). How does a pension plan accomplish that task without knowing (1) the precise value of the dollars received at each point in the retiree’s life compared to today’s money and (2) how long the retiree will live? It uses two actuarial tools: interest rates and mortality tables, respectively. Examine both in turn.

Neither tool necessary to compute the total present value of an SLA—interest rates and mortality tables—need be “reasonable” as a definitional matter. To start, an interest rate is just a “fixed ratio” by which money grows over time. Interest Rate, Oxford English Dictionary (1st ed. 1933) (senses 10 and 11); see Interest Rate, Black’s Law Dictionary (12th ed. 2024). To be sure, prevailing interest rates fluctuate. See Monetary Policy: What Are Its Goals? How Does It Work?, Bd. of Govs. of the Fed. Reserve Sys. [https://perma.cc/YD7A-XECB] (last visited Nov. 17, 2025). But that doesn’t mean that an actuary’s election to use an outdated figure makes it any less of an “interest rate,” definitionally speaking.

Mortality tables follow the same principle—although different tables suit different populations, they aren’t inherently reasonable or unreasonable. Mortality tables set out an “organized chart” of data “indicating life expectancies for people in various categories (such as age, family history, and chemical exposure).” Mortality Table, Black’s Law Dictionary (12th ed. 2024); see Life Table, Black’s Law Dictionary (12th ed. 2024). So a mortality table is a purely statistical creature—it describes, based on an underlying dataset, how likely a given person is to Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 24 Watt, et al. v. FedEx Corp., et al.

die within a given timeframe. See Mortality Tables, Ballentine’s Law Dictionary (3d ed. 1969) (“[s]tatistical tables which show the probable expectancy . . . of life of a normal person according to sex and age”); Mortality Table, Webster’s Seventh New Collegiate Dictionary (7th ed. 1970) (“an actuarial table based on mortality statistics over a number of years”). There’s nothing inherently reasonable or unreasonable about it; sure, a mortality table using data from the 1620s will paint different life-expectancy probabilities than one derived from this year’s data. But applying the former table to people born in the latter era doesn’t make the table itself any less valid, as a definitional matter.

Now move to the second step of the SLA-to-QJSA calculation. It’s just the reverse of the first. It starts with the first step’s result—the SLA’s present value. Then, it uses the same actuarial methods to distribute that present value, but now over the expected remaining lifetimes of both the employee and his or her spouse. That yields—roughly speaking—a revised monthly- payment figure that’ll approximate the same total payout to both spouses as the SLA would pay out to the retiree alone.

So none of the concepts or terms that go into calculating an actuarial equivalent entail a reasonableness requirement, as defined by dictionaries of the time. The term “actuarial equivalent” doesn’t itself include that requirement. And neither do terms that apply downstream—that is, throughout the process of calculating an actuarially equivalent benefit— like “present value,” “interest rate,” or “mortality table.”

B.

With the dictionaries in place, we should also look to ERISA’s “overall statutory scheme,” since we “don’t scrutinize a statute atomistically.” Salazar, 133 F.4th at 650. The overall scheme points to the same conclusion: that actuarial equivalence doesn’t connote reasonableness. To start with, Congress used the word “reasonable” when it meant to. In ERISA, it specified that pension plans must select “reasonable” actuarial assumptions to calculate present values in a couple of other scenarios. See 29 U.S.C. §§ 1055(g), 1083(h)(1)(A), 1085a(c)(3)(A), 1393(a)(1), and 1432(e)(1). Rather than canvas them all, I’ll stick to one example: § 1083(h). That provision, enacted by Congress as part of the Pension Protection Act Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 25 Watt, et al. v. FedEx Corp., et al.

of 2006, Pub. L. No. 109-280, 120 Stat. 780,2 sets minimum funding standards for certain kinds of pension plans. Applying those standards entails actuarial computations that, much like the SLA-to-QJSA conversion, require calculating present values. To that end, Congress specified that any “actuarial assumptions” used to determine “any present value or other computation under” § 1083 must be “reasonable (taking into account the experience of the plan and reasonable expectations).” 29 U.S.C. § 1083(h)(1). So because Congress took pains to specify that plans’ “actuarial assumptions” in funding calculations must be “reasonable,” it must be the case that Congress doesn’t understand actuarial assumptions as inherently necessitating reasonableness—if it did, it would’ve had no reason to say so out loud. Indeed, if Congress “intended to restrict” pension plans to calculate QJSAs using reasonable actuarial assumptions under § 1055(d), “it presumably would have done so expressly,” as it did under § 1083(h). See Russello v. United States, 464 U.S. 16, 23 (1983).

So “actuarial equivalent” doesn’t connote reasonableness on its face or in context. Neither its direct definition, nor the definitions of any associated terminology, inject a fuzzy reasonableness requirement into upright math. And the rest of ERISA’s provisions show that Congress shared that understanding. When Congress wants to modify actuarial mathematics to include a reasonableness component, it knows how to do so. See 29 U.S.C. §§ 1055(g), 1083(h)(1)(A), 1085a(c)(3)(A), 1393(a)(1), and 1432(e)(1).

C.

The majority reaches a contrary conclusion by stepping twice off the beaten path. First, it concludes (in just two sentences) that actuarial equivalence is a term of art. Standing alone, I don’t think that conclusion is necessarily wrong—contrary to the majority’s charge that the “primary problem” with my view of this case is my failure to approach “actuarial equivalence” as a term of art. Maj. Op. p.10 n.5. Although I’ll explain why I wouldn’t rush to describe the phrase as a term of art as hastily, the crux of my problem with the majority’s technical-meaning

2 The Pension Protection Act thoroughly amended § 1083 as originally enacted. See Pub. L. No. 109-280, § 102, 120 Stat. 780, 789 (2006). But the previous version also included the reasonableness language quoted below. See Pub. L. No. 93-406, § 302, 88 Stat. 829, 871 (1974). So from the moment Congress first passed ERISA, it spoke explicitly when it wanted to specify reasonableness as part of actuarial calculations. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 26 Watt, et al. v. FedEx Corp., et al.

analysis is its second, more serious misstep: resorting exclusively to non-binding sources to establish that the term’s meaning definitively includes a reasonableness requirement. On both fronts, our precedents demand more.

1.

Start with error number one: the majority’s haste to deem “actuarial equivalent” a technical term of art. Typically, we hew to ordinary meaning “until and unless someone points to evidence suggesting” that Congress used a term in its technical sense. Niz-Chavez v. Garland, 593 U.S. 155, 163 (2021); see Feliciano v. Dep’t of Transp., 145 S.Ct. 1284, 1291 (2025). For example, when Congress “transplant[s]” a well-worn term from “another legal source, whether the common law or other legislation, it brings the old soil”—the term’s “accumulated” meaning—“with it.” Sekhar v. United States, 570 U.S. 729, 733 (2013) (quoting Morissette v. United States, 342 U.S. 246, 263 (1952), and Felix J. Frankfurter, Some Reflections on the Reading of Statutes, 47 Colum. L. Rev. 527, 537 (1947)). The Supreme Court has applied that standard to ERISA. See Nationwide Mut. Ins. Co. v. Darden, 503 U.S. 318, 322–23 (1992). And in doing so, it instructed us to assess the Act’s use of the term “employee” under that term’s “accumulated settled meaning under . . . the common law.” Id. (quoting Cmty. for Creative Non- Violence v. Reid, 490 U.S. 730, 739–40 (1980)). But it reached that conclusion only after looking first to the statutory definition and other ERISA provisions. Because the former was “completely circular and explain[ed] nothing,” and the latter failed to provide “specific guidance on the term’s meaning,” the common law controlled. Id. As I explained above, that’s not the case here. Congress left us generous material from which to conclude that actuarial assumptions mustn’t be inherently reasonable. See 29 U.S.C. §§ 1393(a)(1), 1432(e), 1083(h)(1), 1085a(c)(3)(A), and 1055(g). So it doesn’t seem entirely correct to conclude that the term is “technical” in any way that’d change how we search for its meaning.

In any case, and as I mentioned before, the majority’s sprint toward technical meaning doesn’t trouble me much because I “do not see much daylight” between actuarial equivalent’s technical meaning and its ordinary meaning. United States v. Tate, 999 F.3d 374, 389 (6th Cir. 2021) (Murphy, J., concurring in the judgment). Instead, the technical meaning aligns with the Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 27 Watt, et al. v. FedEx Corp., et al.

ordinary meaning derived from the sources I explored above. That’s because courts often resort to standard dictionaries and statutory context to determine the meaning of even “technical” terms—just as I did above, and as the majority fails to do. See, e.g., United States v. Hansen, 599 U.S. 762, 775 (2023) (explaining that “context differentiates” between competing meanings of a term, including “when the choice is between ordinary and specialized meanings”); Vill. of Hoffman Estates v. Flipside, Hoffman Estates, Inc., 455 U.S. 489, 501 n.18 (1982) (citing a standard dictionary for a “technical term . . . in the drug paraphernalia industry”); United States v. Wilkes, 78 F.4th 272, 283 n.4 (6th Cir. 2023) (citing three standard dictionaries for the meaning of “stereoisomer”); Tate, 999 F.3d at 389 (Murphy, J., concurring in the judgment) (quoting Black’s Law Dictionary for the meaning of “dangerous weapon” in the United States Sentencing Guidelines); United States v. Easter, 981 F.2d 1549, 1558 (10th Cir. 1992) (citing Webster’s Third New International Dictionary for the meaning of “base” in the federal drug- trafficking statutes); cf. United States v. Caseer, 399 F.3d 828, 838 (6th Cir. 2005) (surveying standard dictionaries to determine whether a “scientific or technical term” was sufficiently discernible to the public).

2.

Rather than survey the full field of sources available to us, the majority uses its “technical meaning” conclusion as a license to consult exclusively an eclectic collection of academic and legal sources. But this motley crew of citations looks a lot like “the equivalent of entering a crowded cocktail party and looking over the heads of the guests for one’s friends.” Cf. Conroy v. Aniskoff, 507 U.S. 511, 519 (1993) (Scalia, J., concurring in the judgment).

And the majority’s sources aren’t on-point. For example, it reads mid-twentieth century Washington state caselaw (among others) to hold that actuarial equivalence requires the use of reasonable or “accurate” actuarial assumptions. Maj. Op. p.11 (quoting King Cnty. Emp. Ass’n v. State Ret. Bd., 336 P.2d 387, 391 (Wash. 1959)). That approach has several problems. At the outset, the Washington Supreme Court’s decision in King County wasn’t about whether actuarial assumptions used in benefits calculations must be reasonable per se. Instead, as the same court later explained, it was about whether state employees had a “contractual right to . . . a benefit Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 28 Watt, et al. v. FedEx Corp., et al.

calculated based on any particular set of mortality tables adopted by the [state retirement board].” Lenander v. Wash. State Dep’t of Ret. Sys., 377 P.3d 199, 212 (Wash. 2016). King County didn’t hold that outdated actuarial assumptions “would be actuarially unsound” because of their outdatedness. Instead, the court said that it would be improper to apply “mortality tables rejected by” the state retirement board because “the statutes carve out and retain for [the state board] the authority to periodically update actuarial regulations.” Id. In other words, King County decided that it would be “actuarially unsound” for the state retirement system to use actuarial assumptions that it had abridged by subsequent regulation—not because the old actuarial assumptions were “unreasonable” in some technical sense.3

Aside from its inapplicability, the majority’s cited state caselaw doesn’t represent the universal view. Indeed, a quick survey yields other state caselaw that holds just the opposite. Take Alaska as an example. That state’s public-employee retirement system adopted a new actuarial table in 1982, “based on then-current mortality and interest earnings data, which superseded the table in effect since 1972.” Sheffield v. Alaska Public Employees Ass’n, Inc., 732 P.2d 1083, 1084 (Alaska 1987). The Alaska Supreme Court held that the state pension plan was obligated to use the outdated actuarial assumptions for retirees hired before the switch. See id. at 1089. To do otherwise would “vitiate” their contractual expectations (enshrined in a provision of the state’s constitution) to receive not just any pension, but a “particular amount” based on the actuarial factors specified by the plan at the time they signed up. Id. In so holding, the Alaska

3 The other state cases the majority cites falter for similar reasons. Start with the Pennsylvania trial court’s decision in O’Dea v. Public School Employes’ [sic] Retirement Board, 88 Pa. D. & C. 593 (Ct. C.P. Pa. 1954). There, the court held that the state’s retirement board could exercise its statutory authority to apply new actuarial assumptions retroactively, not that the state’s retirement board had to use reasonable actuarial assumptions. Id. at 602–03 (explaining the state’s statutory scheme, allowing the retirement board to adopt new mortality tables and apply them even to retirees who entered the system when older tables were in effect). Although the court opined that “actuarial equivalent . . . means calculations made at retirement according to the best actuarial information and tables . . . available to the actuary” at that point in time, it went on to uphold the state system’s use of a 1934 table to a woman who retired in 1947—not because it was the best or most reasonable table in a mathematical sense, but because it was the most up-to-date table issued by the board. See id. at 604 (internal quotation marks omitted). And the California case the majority cites falls even further afield. There, the intermediate appeals court resolved a question of family law—whether an ex-wife could claim her deceased ex-husband’s retirement benefits under their divorce decree despite the decedent’s remarriage—and only in passing mentioned the concept of actuarial equivalence. Berry v. Bd. of Ret., 23 Cal. App. 3d 757, 758–59 (Cal. Ct. App. 1972). Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 29 Watt, et al. v. FedEx Corp., et al.

Supreme Court expressly rejected its Washington-state counterpart’s reasoning in King County— the case the majority believes establishes that actuarial assumptions must be reasonable.

So I think the majority overemphasizes marginal sources.4 And at the same time, it discounts the best information we’ve got—the other ERISA provisions (discussed above) that do mention reasonableness in actuarial assumptions. The majority discards those provisions by reasoning that the “Russello presumption”—that is, the common-sense notion that Congress really means it when it omits a term from one provision while including it elsewhere—doesn’t apply here. In so doing, it reads § 1055(d) to say something that Congress itself opted not to express.

The majority’s reasoning doesn’t support such a departure from our customary role to interpret, not legislate. It claims the power to import reasonableness into § 1055(d) because that provision is sufficiently “dissimilar” from the other ERISA sections to overcome the Russello presumption. Maj. Op. pp.17–18. But the single case the majority cites for support doesn’t help. Although the majority is right that the Supreme Court discarded the Russello presumption in a case where “difference[s] in the formulation of the provisions under inspection” made it a poor fit, City of Columbus v. Ours Garage & Wrecker Serv., Inc., 536 U.S. 424, 436 (2002), we’ve explained elsewhere that the type of difference matters. Specifically, the presumption applies “when two provisions are sufficiently distinct that they do not—either explicitly or implicitly— incorporate language from the other provision.” Grand Trunk W. R.R. Co. v. United States Dep’t of Lab., 875 F.3d 821, 825 (6th Cir. 2017) (quoting Port Auth. Trans-Hudson Corp. v. Sec’y, U.S. Dep’t of Lab., 776 F.3d 157, 164 (3d Cir. 2015)). But it weakens “with each difference in the formulation of the provisions under inspection.” Ours Garage, 536 U.S. at 435–36. Said

4 Alongside the caselaw, the majority quotes a few academic sources that advise actuaries to use “appropriate mortality tables” and to “review[] and modify[]” their assumptions to “continue to fairly assess” prevailing conditions. Maj. Op. p.11 (first quoting William F. Marples, Actuarial Aspects of Pension Security 40 (1964); and then quoting Jeff L. Schwartzmann & Ralph Garfield, Educ. & Exam. Comm. of the Soc’y of Actuaries, Actuarially Equivalent Benefits 11, EA1-24-91, https://perma.cc/XKD8-FDLW). But professionals’ views about their responsibilities don’t give independent meaning to Congress’s words. And even crediting the cited works at face value, they don’t speak for all contemporaneous actuarial academics—just like King County didn’t speak for all state supreme courts—and no one has given us good reason to privilege one set of academics’ views over others. Indeed, as the appellees explain in their supplemental briefing, other academics “were agnostic about the selection of actuarial assumptions to calculate an ‘actuarial equivalent’ benefit.” Suppl. Appellee Br., pp.3–8 (collecting examples). Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 30 Watt, et al. v. FedEx Corp., et al.

differently, the provisions must be distinct enough that they don’t presumptively incorporate each other’s words, but they must not be so different that a comparison would be inapt.

Given that standard, §§ 1055(d) and 1083(h) should receive the Russello presumption. As the majority points out, the two provisions appear in different parts of ERISA that handle different problems. Maj. Op. p.18. So they wouldn’t presumptively incorporate each other’s commands. At the same time, both sections substantially involve the same issue: actuarial decisionmaking. Section 1083(h) concerns the actuarial assumptions used to calculate “any present value or other computation” (including actuarial equivalence) for plan-funding purposes while § 1055(d) homes in on one such computation—actuarial equivalence—for benefit- translation purposes. Cf. FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 133 (2000) (explaining that the “meaning of [a] statute may be affected by other Acts, particularly where Congress has spoken subsequently and more specifically to the topic at hand”). So for our purposes, any differences in formulation are merely superficial. Both sections involve the same mathematical concept—actuarial calculation. And although Congress opted in one section to specify that the assumptions used in those calculations should be reasonable, it omitted that requirement in another section about substantially the same thing. We should take Congress at its word.5

Ultimately, both the ordinary and technical meanings of actuarial equivalence require only that a QJSA calculated under § 1055(d) have “present values . . . equal” to that of the employee’s SLA “under a given set of actuarial assumptions.” Stevens, 644 F.3d at 440; see Actuarial Equivalent, Black’s Law Dictionary (12th ed. 2024). Our traditional tools of statutory interpretation show that those actuarial assumptions need not be “reasonable” under § 1055(d).

5 Another one of Russello’s underlying rationales also weighs in favor of applying the presumption here. As then-Judge Alito explained, “[t]he Russello canon is based upon a hypothesis of careful draftsmanship.” Kapral v. United States, 166 F.3d 565, 579 (3d Cir. 1999) (Alito, J., concurring). That makes sense—the more carefully drafted a document, the more likely an omission can’t be “ascribe[d] to a simple mistake in draftsmanship.” Id. (quoting Russello, 464 U.S. at 23). We’ve recognized that rationale in this circuit, too. Grand Trunk, 875 F.3d at 826 n.5 (quotation omitted). And if there’s any statute that can be described as “carefully drafted,” it’s ERISA. See Sprague v. General Motors Corp., 133 F.3d 388, 401–02 (6th Cir. 1998) (finding that an omission in one ERISA provision of a requirement stated in another provision “was no mistake” because “ERISA, after all, is a ‘comprehensive and reticulated statute’” (quoting Nachman Corp. v. Pension Benefit Guar. Corp., 446 U.S. 359, 361 (1980))). Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 31 Watt, et al. v. FedEx Corp., et al.

By concluding otherwise, the majority departs from our typical statutory analysis and, indeed, from the constitutional confines of our judicial role into the legislative.

II.

The majority perhaps bends the text because of its concern that “Defendants’ construction would allow employers to use dramatically outdated mortality data,” an allegedly “absurd” outcome. Maj. Op. p.16 (quoting Dodd v. United States, 545 U.S. 353, 359 (2005)). But “[t]he disposition required by the text here, though strict, is not absurd.” Dodd, 545 U.S. at 359. That’s because a distinct body of law—the Internal Revenue Code (IRC) and its implementing regulations—requires pension plans to use “reasonable” actuarial factors to calculate QJSAs or risk losing their tax-exempt status. So to the extent the majority worries that a faithful reading of ERISA would allow pension plans to toy with their beneficiaries’ benefits, the IRC’s much weightier penalties already address that concern.

First, a word on the law. Many of ERISA’s requirements have “nearly verbatim replication” in the Internal Revenue Code. Thornton v. Graphic Commcns. Conf. of Int’l Broth. of Teamsters, 566 F.3d 597, 602 n.4 (6th Cir. 2009) (quoting Cent. Laborers’ Pension Fund v. Heinz, 541 U.S. 739, 746 (2004)). So when a pension plan violates one of ERISA’s substantive requirements, it can also “lose[] its tax-exempt status” under the IRC. Id. Section 1055 follows that mold, finding its IRC parallel in 26 U.S.C. § 401. Both sections require defined-benefit pension plans to offer QJSAs. The ERISA version “impose[s] [the] substantive” requirement, while the IRC version “condition[s] the eligibility of pension plans for preferential tax treatment on compliance” with that substantive requirement. Cent. Laborers’, 541 U.S. at 746. And both provisions provide that a qualified QJSA must be actuarially equivalent to the participant’s SLA, without further defining what constitutes actuarial equivalence. 29 U.S.C. § 1055(d); 26 U.S.C. § 401(a)(13)(D).

Of course, we have no occasion to definitively interpret the IRC or its implementing regulations here. But those provisions counter the majority’s contention that ERISA’s most natural reading would “allow an employer to use mortality rates that are completely out of step with the mortality rate of modern-day plan participants.” Maj. Op. p.16. Sure, a plan wouldn’t Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 32 Watt, et al. v. FedEx Corp., et al.

violate ERISA (properly understood) by using outdated rates. But that’s not much to grieve. Using those rates could violate the IRC and its implementing regulations, which would bring a far worse fate for the pension plans. See Tax Consequences of Plan Disqualification, Internal Revenue Serv. [https://perma.cc/6PQB-T3ZA] (last visited Nov. 17, 2025).

So because the specter of a massive tax burden probably provides a greater incentive than ERISA’s private remedies, the majority needn’t rush in to do the work of the legislature. After all, ERISA’s penalties are standard fare: per-day fines, damages necessary to make up shortfalls in benefit distributions, and the like. See 29 U.S.C. § 1132. The IRC’s penalties differ in degree and in kind. Among other benefits, plans in compliance with § 401 qualify for tax-free growth on plan assets and immediate deductibility of employer contributions. See 26 U.S.C. §§ 404, 501. Upon disqualification, those benefits vanish. And that deprivation is retroactive—meaning that the plan doesn’t just lose its deductions and tax-exemptions going forward, but also stretching back to the time it breached § 401’s requirements. See Martin Fireproofing Profit Sharing Plan & Tr. v. Comm’r, 92 T.C. 1173, 1184–89 (T.C. 1989). For plans as large as the ones at issue here,6 the value of those tax exemptions and deductions likely dwarf any plausible liability under ERISA.

To put a finer point on it all: the planes won’t fall from the sky if we give § 1055(d) its most natural reading. The IRC imposes a reasonableness requirement on pension plans’ actuarial assumptions when calculating QJSAs. And avoiding the IRC’s penalties is probably a far greater impetus for the pension plans in this case than worrying about ERISA’s private remedies. So putting aside whether such considerations should ever weigh into our textual analysis (they shouldn’t), I struggle to credit the majority’s concern that pension plans will start using

6 According to forms filed with the federal government, the FedEx Corporation Employees’ Pension Plan’s assets had a market value of nearly $26 billion as of June 2024. FedEx Corporation, Form 5500 (filed Mar. 11, 2025) [https://perma.cc/F754-EUXV]. The Kellanova Pension Plan accounted for just shy of $1.25 billion as of January 2024. Kellanova, Form 5500 (filed Oct. 15, 2025) [https://perma.cc/Q5FQ-9Y5M]. And the BCTGM Plan had about $450 million by December 2024. WK Kellogg Co., Form 5500 (filed Oct. 21, 2025) [https://perma.cc/J5FE-S9YS]. All those figures are matters of public record, obtainable via the Department of Labor’s Form 5500 search tool available at https://www.efast.dol.gov/. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 33 Watt, et al. v. FedEx Corp., et al.

“dramatically outdated mortality data” like “a mortality table from the sixteenth century” as a consequence of our fidelity to the statutory text.7 Maj. Op. p.16.

In all fairness, the majority recognizes that § 401 and its implementing regulations have a role in this analysis—it just goes about it the wrong way. Rather than see those sources as an independent check on pension plans’ selection of actuarial assumptions, the majority takes the Treasury’s definition of an IRC term and uses it to overwrite Congress’s definition of the same term under ERISA.

That approach is problematic. At the outset, the fact that the Treasury deemed it necessary to specify a reasonableness requirement by regulation shows that the plain meaning of the statute doesn’t convey that requirement. In other words, and in keeping with the Russello presumption explained above, the Treasury’s inclusion of reasonableness in its regulation shows that Congress meant it when it left reasonableness out of its definitions.

More importantly, the majority misapplies the Supreme Court’s recent opinion in Loper Bright Enterprises v. Raimondo, 603 U.S. 369 (2024), to argue that the Treasury’s regulatory definition of QJSA is entitled to “great respect”—great enough to effectively amend Congress’s own definition of the same term. Maj. Op. p.13. Loper Bright did recognize the “due respect” courts can accord to “roughly contemporaneous[]” agency interpretations of statutory terms. Loper Bright, 603 U.S. at 386. At the same time, though, the Court emphasized that “[s]uch deferential review [is] cabined.” Id. at 389. It extends only when “a particular statute empowered an agency to decide how a broad statutory term applied to specific facts found by the agency.” Id. at 388. This isn’t such a case. As the majority recognizes, the Treasury promulgated its QJSA definition in 1977—one year before the Reorganization Plan of 1978 gave it interpretive authority over ERISA. See 43 Fed. Reg. 47713, 92 Stat. 3790 (1978). That fact alone undermines the majority’s focus on contemporaneity.

And even after it received authority over ERISA, the Treasury twice amended the relevant regulation without extending its interpretation of “QJSA” under the IRC to ERISA’s

7 In addition, as I note below, if companies play games with ERISA, Congress can address it, which it’s done before. Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 34 Watt, et al. v. FedEx Corp., et al.

shores. Both amendments mentioned the Treasury’s authority to interpret ERISA, but neither exercised it over § 1055(d) (the provision at issue in this case). See 53 Fed. Reg. 31837, 31838– 39 (Aug. 22, 1988) (providing that the updated regulation’s “requirements also apply to employee plans subject to Title I of ERISA,” but not that the regulation interprets ERISA’s text); 68 Fed. Reg. 70141, 70141–42 (Dec. 17, 2003) (providing that the updated regulation applies to ERISA § 205(c)(3), codified at 29 U.S.C. § 1055(c)(3), but no more broadly). That history shows that the Treasury established a reasonableness requirement only for QJSAs under the IRC, not ERISA. And so it wouldn’t make much sense to follow an agency’s interpretation of a statutory term—no matter how close in time to the statute’s enactment—if the interpretation isn’t oriented toward the statute at issue, but a statute that happens to contain the same language.

III.

Finally, I’d be remiss not to point out that the majority’s intrusion into the legislative role carries consequences of its own. ERISA, like all legislative enactments, represents a delicate balancing of competing interests hashed out by the people’s democratically elected representatives. And we disturb Congress’s judgment at our own peril.

In crafting ERISA, Congress took care to not make the statute “so complex that administrative costs, or litigation expenses, unduly discourage employers from offering [ERISA] plans in the first place.” Conkright v. Frommert, 559 U.S. 506, 517 (2010) (alteration in original) (quoting Varity Corp. v. Howe, 516 U.S. 489, 497 (1996)). It would be one thing if the text of the statute required such an outcome. But here we face—at most—congressional silence. And because the plain language of the statute doesn’t contemplate a reasonableness requirement for a plan’s actuarial assumptions, I see no reason to impose one by judicial fiat.

The most likely result of the majority’s decision is even more litigation and higher administrative costs. These costs will be passed on and will ultimately injure the continued viability of the plans. So it’s hard to see how this standard will benefit anyone besides the attorneys who litigate these cases.

Moreover, Congress has repeatedly shown that it’s willing and able to amend ERISA to respond to new developments in the world of retirement benefits security. For example, a House Nos. 24-1442/24-5945 Reichert, et al. v. Kellogg Co., et al. Page 35 Watt, et al. v. FedEx Corp., et al.

committee report on the Pension Protection Act of 2005 (which didn’t itself become law, but largely mirrored the act of the same name passed the following year) noted that Congress wanted to amend part of ERISA to address the use of “outdated” mortality tables in plan-funding calculations. H.R. Rep. No. 109-232, pt. 1, at 61 (2005). And Congress addressed that concern in the Pension Protection Act of 2006 by requiring pension plans to use mortality tables promulgated “at least every 10 years” by the Secretary of the Treasury to “reflect the actual experience of pension plans and projected trends in such experience.” Pub. L. No. 109-280, § 102, 120 Stat. 780, 800 (2006) (codified at 29 U.S.C. § 1083(h)). Similarly, a House committee report on the Retirement Protection Act of 1994 described why Congress adopted a cap on interest-rate assumptions in some actuarial calculations: “to prevent plans from using unreasonably high interest rates to determine the present value of participants’ benefits,” which “could lower the single sum paid to participants.” H.R. Rep. No. 103-632, pt. 2, at 57 (1994).

All this is not to say that we should make any decision about what the text means based on policy considerations or predictions of congressional priorities. Indeed, my point is just the opposite: we should leave policy considerations to Congress as the people’s elected representatives. We judges, on the other hand, should focus our energies only on interpreting Congress’s words as written. We pierce the veil of Congress’s underlying policy rationales at our own peril, limited as our sightlines are by the nature of litigation as opposed to legislation. So because “ERISA is a ‘comprehensive and reticulated statute,’” we should be “reluctant to tamper with [the] enforcement scheme’ embodied in the statute by extending remedies not specifically authorized by its text.” Great-W. Life & Annuity Ins. v. Knudson, 534 U.S. 204, 209 (2002) (alteration in original) (first quoting Mertens v. Hewitt Assocs., 508 U.S. 248, 251 (1993); and then quoting Mass. Mut. Life Ins. v. Russell, 473 U.S. 134, 147 (1985)).

* * *

Because the plain meaning of the statute doesn’t support the majority’s interpretation, I respectfully dissent.