Retirees allege plans used decades-old mortality tables, reducing joint and survivor annuity payments
Kellogg and FedEx must now defend class action claims that their pension plans shortchanged married retirees by relying on decades-old mortality data.
In a decision filed on March 16, 2026, the United States Court of Appeals for the Sixth Circuit revived two class action lawsuits that accuse the companies of using outdated actuarial assumptions to calculate pension benefits for married retirees, resulting in smaller monthly payments than the law requires.
The cases were brought by retired workers enrolled in defined benefit pension plans sponsored by Kellogg (now Kellanova and WK Kellogg Co.) and FedEx Corporation. Each of the plaintiffs is married and receives a joint and survivor annuity – a form of pension that continues paying benefits to a surviving spouse after the retiree's death. Under federal law, that joint annuity must be the actuarial equivalent of the single life annuity the retiree would otherwise receive.
The problem, according to the retirees, is how the plans arrived at that equivalence. Both the Kellogg and FedEx plans allegedly relied on mortality tables built from data collected in the 1960s and 1970s. Because people live significantly longer today than those tables assume, the calculations underestimate how long retirees will collect benefits. That assumption, in turn, drives down the monthly payments made to married participants.
The Kellogg plans used the Uninsured Pensioners 1984 Mortality Table. FedEx used both that table and the 1971 GAM Mortality Table, and also applied age setbacks for spouses – effectively treating them as younger than they are – which reduced payments further still.
Lower courts in Michigan and Tennessee had dismissed both lawsuits, finding that the statute does not tell pension plans which mortality tables or actuarial methods to use. The Sixth Circuit saw it differently. Writing for a two-to-one majority, Judge Jane Stranch held that the concept of actuarial equivalence, as understood by the profession at the time Congress enacted ERISA in 1974, inherently requires the use of mortality data that reasonably reflects how long today's retirees are expected to live. Using data that no longer reasonably reflects current life expectancy, the court concluded, cannot satisfy the requirement.
The court was careful to note that it is not prescribing any particular table or method. Plan actuaries still have room to exercise professional judgment. But that judgment must fall within a reasonable range – and assumptions drawn from mortality patterns observed more than fifty years ago may no longer qualify.
Judge Nalbandian dissented, arguing that Congress knew how to write a reasonableness requirement into the statute and chose not to do so in this provision. He pointed to the Internal Revenue Code, which separately requires reasonable actuarial assumptions for pension plans seeking tax-qualified status, as the more appropriate enforcement mechanism.
The cases now return to the trial courts for further proceedings. No determination has been made on whether the plans' specific assumptions are in fact unreasonable – only that the retirees' claims are plausible enough to move forward.
For HR teams still administering legacy defined benefit plans, the ruling raises a straightforward question: when was the last time anyone checked the actuarial assumptions baked into the plan documents? If the answer involves a mortality table older than some of the retirees collecting benefits under it, this decision suggests it may be time for a closer look.
