6th Cir.

Reichert v. Kellogg Co.

March 16, 2026 ·24-1442 ·Published ·Jane B. Stranch · By James Taylor

The Sixth Circuit reversed the dismissal of ERISA claims alleging that pension plans used outdated mortality data to calculate benefits for married participants. The court held that the statutory requirement for actuarial equivalence prohibits the use of unreasonable, inapplicable actuarial assumptions that fail to reflect modern life expectancies.

Retired employees of Kellogg Company and FedEx Corporation sued their respective employers and pension plans, alleging violations of the Employee Retirement Income Security Act, or ERISA. The plaintiffs, all married participants in defined benefit pension plans, claimed that the plans used mortality tables based on data from the 1960s and 1970s to calculate their benefits. Specifically, the plans used the Uninsured Pensioners 1984 Mortality Table and the 1971 GAM Mortality Table to determine the conversion factor between a single life annuity and a joint and survivor annuity. Because these tables assume higher mortality rates than modern data, the resulting monthly payments to the participants were lower than they would have been if current life expectancy data were used. The plaintiffs argued this violated ERISA's requirement that a qualified joint and survivor annuity be the actuarial equivalent of a single life annuity. The district courts dismissed the complaints, ruling that ERISA does not mandate the use of specific mortality tables or actuarial assumptions, leaving the selection of data to the discretion of the plan administrators.

The Sixth Circuit, in an opinion by Judge Stranch, focused on the statutory text of ERISA Section 1055(d), which mandates that a qualified joint and survivor annuity must be the actuarial equivalent of a single life annuity for the life of the participant. The court determined that 'actuarial equivalent' is a term of art with a specific meaning in the field of actuarial science at the time of ERISA's enactment in 1974. The court explained that actuarial equivalence requires that the present value of the joint and survivor annuity equal the present value of the single life annuity. To achieve this equality, the mortality assumptions used must reasonably reflect the life expectancy of the actual participants. The court reasoned that using mortality data from the 1960s or 1970s to calculate benefits for modern retirees fails to meet this standard because it does not accurately reflect the expected remaining lifetime of the current population. The court rejected the defendants' argument that the statute allows any assumptions to be used, noting that such an interpretation would render the 'actuarial equivalent' requirement meaningless and lead to absurd results. The court also addressed the dissent's reliance on dictionary definitions, clarifying that technical terms must be understood within their professional context. While acknowledging that other ERISA sections explicitly use the word 'reasonable,' the court found that the concept of reasonableness is inherent in the definition of actuarial equivalence itself. The court concluded that the plaintiffs plausibly alleged that the use of decades-old mortality data resulted in benefits that were not actuarially equivalent to the single life annuities, thereby stating a valid claim for relief.

The decision reverses the dismissal of the cases, allowing the litigation to proceed to the merits. Pension plan administrators can no longer rely on the argument that they have unfettered discretion to use any mortality table they choose. Moving forward, plans must ensure that their actuarial assumptions, particularly mortality tables, are reasonable and reflect current life expectancy data to satisfy the actuarial equivalence requirement. The cases are remanded to the district courts to determine whether the specific mortality tables used by Kellogg and FedEx were indeed unreasonable and to calculate any potential benefit shortfalls. The decision leaves open the question of what specific timeframe constitutes 'outdated' data, though it establishes that data from the 1960s and 1970s is likely insufficient for modern participants.