Since 1974, ERISA has dictated that the normal retirement date (NRD) for single employer defined benefit (DB) plans be no later than age 65. (I don’t want to say this rule is old, but the original draft defined it as “Age LXV”.) Back then, much more focus was placed on early retirement rules; not much thought was given to exactly what happens to people who work beyond NRD.
Today we know the situation is different. People live longer (often reaching the ripe old age of C and beyond!) With all this extra time on their hands, they are choosing to work past age 65 more and more frequently. And while there are benefits to employees and employers from working longer, the potential cost increase within DB plans attributable to late retirement should be fully understood.
Actuarial equivalence adjustments
Many DB plan documents are written to pay late retirees no less than the “actuarial equivalent” of their pension earned as of NRD. Specific mortality and interest rates assumptions used to adjust benefits are usually defined within the plan document.
Most plan administrators are already familiar with the concept of actuarial equivalence through the early retirement calculation. If a person retires before NRD, their benefit is reduced using the plan’s actuarial equivalence definition to account for the fact that it will be paid out over a longer period of time. (So for example, a $1,000 per month pension payable at age 65 may be reduced to $800 if payments commence at age 62).
Actuarial equivalence mathematics are flipped for late retirements, and the pensions are increased to reflect the expectation that they will be paid out over a shorter period of time. (So the $1,000 age 65 pension may become $1,400 at age 68.)
Generally, the value of the pension payable at NRD is adjusted to make the benefit worth the same amount regardless of when payments actually start if the actuarial equivalence assumptions are met.
Plan actuarial equivalence definitions used for late retirement adjustments are often quite old, many dating back to the 1980s or even earlier. As such, they understandably reflect the period in which they were drafted, generally characterized as having significantly shorter life expectancy and significantly higher interest rates than today. It is not unusual to see the 1984 Universal Pensioners mortality table (“UP-84”) and an interest rate of 8%, or similarly outdated assumptions, being used to adjust benefits from NRD.
Pension liabilities, on the other hand, use more contemporary assumptions. Accounting liabilities and lump sum conversions are based on very recent mortality tables and market value corporate bond rates (currently in the neighborhood of 4%).
When the life expectancy assumption is longer, the reward for surviving beyond NRD is less since each additional year represents a smaller portion of the total life expectancy. (Life expectancy from age 65 of the IRS 2016 Unisex mortality table is approaching 21 years, almost five years longer than UP-84). When rates are lower, the interest paid for past deferred payments is also less. Combined, the effect of using current actuarial equivalence assumptions for late retirement would be significantly smaller increases.
DB anti-cutback rules, however, prohibit changing a plan’s actuarial equivalence definition if it results in a loss of early (or late) retirement subsidies. So while the actuarial basis for the determination of liabilities evolves with market conditions (longer lives and lower rates), the basis for adjusting an individual’s benefit is frozen in time — like a woolly mammoth in a glacier.
The disparity in these assumptions from different eras can result in significant, possibly shocking, late retirement subsidies as seen in the example below. The table compares lump sum values from a $1,000 monthly benefit at NRD adjusted for late retirement using two different actuarial equivalence assumptions: UP-84 and 8% against IRS 2016 mortality and 4%.
Example is for illustrative purposes only.
While the amount of the benefit is being adjusted for a relatively short life expectancy in the UP-84 column, that adjusted payment is still assumed to be paid out over a longer life when liabilities or lump sums are calculated. The result is an echo chamber of inadvertent subsidy that exponentially rewards individuals for survival and interest far more than current assumptions would dictate. The closer an individual gets to the end of the older actuarial equivalence mortality table (UP-84), the greater the subsidy becomes.
Not so frozen
As surprising as these results may be, they are potentially jaw-dropping to those who think they have already frozen their plans. Excess late retirement subsidies have the same effect as unfreezing accruals beyond NRD for all intents and purposes. If not handled correctly, the glacial woolly mammoth can thaw out and trample funding ratios and balance sheets.
Suspension or disbelief?
Unless regulators and legislators unexpectedly move to address late retirement issues directly, DB sponsors are left with a limited number of options to address unintended subsidies beyond NRD. Consultation with legal counsel and the plan’s administrator and actuary are highly recommended before taking any action.
One consideration is for sponsors to begin issuing “suspension of benefits” notices to current employees as they approach age NRD (usually age 65). If allowed by the plan document, this practice eliminates the need to perform the actuarial equivalence adjustment and simply pay out whatever benefit the employee has earned through their late retirement date. For frozen plans, this accomplishes the stated objective of actually halting new benefit accruals. (Note that adjustments from age 70½ are still required for employees working beyond that age due to minimum distribution regulations.)
Terminated vested participants (those no longer working but not yet collecting benefits) are trickier as suspension of benefits notices hold no power over former employees. Lump sum windows offer a potential solution by paying these participants out before they reach NRD [see previous post on “Pension Defenestration”] assuming you can find the participants to pay them!
After a certain age, however, a cynical person would suggest it’s better to not find them at all.*** Obsolete actuarial equivalence rules can produce handsome windfalls to tardy retirees. But from a plan sponsor cost perspective, the ideal retirement age may truly be better never than late!
*** The Department of Labor has just recently begun investigating plan practices regarding the location of terminated vested participants, so this probably isn’t a great idea.
Mike Clark, along with many of his actuarial equivalents, is a fellow of the Society of Actuaries (SOA) and a member of the American Academy of Actuaries (AAA).
The subject matter in this communication is provided with the understanding that Principal® is not rendering legal, accounting, or tax advice. You should consult with appropriate counsel or other advisors on all matters pertaining to legal, tax, or accounting obligations and requirements.
Insurance products and plan administrative services are provided by Principal Life Insurance Company, a member of the Principal Financial Group® (Principal®), Des Moines, IA 50392.