How plan sponsors can use projection models to guide retirement program decision-making processes
As a Plan sponsor, you may wonder about the need for pension projections when you already receive an annual valuation from your actuary. In short, it’s like comparing a photograph to a movie. Annual valuations are a great snapshot of where a pension plan is currently at; projections are useful to see where a plan may be going.
Annual valuations are a picture of how your plan stands at that point in time. For example, a January 1, 2022, valuation report would have a plan’s funded status of January 1, 2022. For many pension plans, 2021 experience broadly was favorable, especially from an asset performance perspective. As a result, many corporate pension plans experienced an increase in their funded status between the valuations of January 1, 2021, and January 1, 2022. The aggregate funding percentage for the 100 largest U.S. pension plans increased from 88% to 96%.1 Corporate pension plans appeared to be gaining ground as the calendar year switched from 2021 to 2022.
But as 2022 has played out, many of those gains from last year are at risk of being lost. While projections couldn’t predict the roller-coaster ride 2022 has taken pension plans on, they could have helped plan sponsors weather the ups and downs. For example, a projection could have modeled the impact on a plan’s funding requirements of a large negative asset return during the year. And if that potential impact was meaningful to a plan sponsor, the sponsor could have budgeted and planned for such a hit to its cash requirements. Just as stress tests may be necessary for other aspects of a business, they should also be considered an integral tool for a plan sponsor’s pension plan.
As a simplistic example of a projection, let’s consider a frozen corporate pension plan that, as of January 1, 2022, had $100 million in liabilities and $85 million in assets. So, the plan’s funded status as on New Year’s Day was 85%. Now the shortfall, $15 million, must be funded over 15 years, meaning the plan sponsor would have a 2022 minimum required contribution of about $1 million per year ($15 million / 15 years).
Again, let’s keep things simple; assume there is no interest on accrued liabilities, no benefit payments, and no expenses are paid out of the plan. Then our liability as of January 1, 2023, is still $100 million.
For our first projection, let’s assume the plan sponsor contributes $1 million at the beginning of the year and 2022 investment earnings are 5%, a relatively modest earnings assumption. The plan would be projected to have $90 million ($86 million x 1.05) in assets on January 1, 2023, and thus the shortfall shrinks to $10 million. The minimum required contribution for 2023 would project to be about $0.7 million ($10 million / 15 years), or almost one-third lower than what the plan sponsor was required to contribute in 2022.
What if the investment return was not 5%, but instead the plan projects a loss of 20%? For this second projection, assets on January 1, 2023, would then be $69 million ($86 million x 0.80). The plan still has a liability of $100 million, so we now project a deficit of $31 million in 2023. Now the projected minimum required contribution for 2023 would jump up to about $2.1 million ($31 million / 15 years). Depending on how likely a 20% asset loss is to a plan sponsor, the results of this projection suggest the plan sponsor may want to ensure it has enough funding flexibility to cover a 2023 contribution that could be double what it paid in 2022. Additional projection scenarios can be explored by varying the asset return over a few years and seeing what the resulting contribution requirements would be over that time period.
Besides the asset return, the impact of changing other aspects of a pension plan can be studied with this type of projection model. Plan sponsors may wonder what rising interest rates could do to their accounting expense in future years. Or maybe a plan sponsor has enough to make extra contributions now; how might paying more up-front affect their funding requirements in the future? Perhaps inflation is a concern—how much might plan expenses or even benefit accruals themselves increase? A good basic pension projection would allow the levers to be adjusted and help the plan sponsor understand the impact of all the different components. Such a projection model is commonly called a deterministic projection.
Instead, what if you didn’t want to explicitly control the levers? After all, in the real world you can’t predetermine an asset return as a 20% loss.
Another type of projection, called a stochastic projection, runs many different randomized scenarios, and then reports both the impact and the likelihood of each scenario. In these projections, capital market assumptions are used as inputs. These projections generally run hundreds or even thousands of separate trials and report back the results of all of them.
So, the stochastic projection would report back the likelihood of that 5% gain or 20% loss happening, given the parameters set by the capital market assumptions. The likelihood of an outcome is a key characteristic of a stochastic projection and is what differentiates itself from a deterministic projection where the outcome is predetermined.
These differences between a deterministic projection model and a stochastic projection model are further explored in a recent article published by Milliman.2
Ideally, both types of projections would be used by plan sponsors to help them gauge the risks their pension plans face. The deterministic projection model is relatively simpler and more direct; the plan sponsor can see right away how a change in an assumption would affect its plan. The stochastic model adds to that by factoring in randomness.
Like any model, it should be noted that both types of projections are only as good as the inputs provided. A deterministic model where the lever is turned too far out will return a result that is out of whack. A stochastic model’s volatility input will need to be calibrated properly to provide meaningful results. A plan’s investment managers, administrators, and actuaries could be valuable resources to make sure the inputs that go into the model are suitable when projecting plan results.
A pension plan has too many variables surrounding it that are constantly changing to be checked upon only once a year. Good governance by a plan sponsor involves not just knowing where the plan stands today but knowing what risks are out there and how to mitigate or even take advantage of them. Projections reveal these risks and opportunities better than a snapshot valuation alone. The photograph pauses time, the movie starts it up again.
1 Wadia, Z., Perry, A.H., & Clark, C.J. (April 2022). 2022 Corporate Pension Funding Study. Milliman White Paper. Retrieved November 18, 2022, from https://us.milliman.com/en/insight/2022-corporate-pension-funding-study.
2 Townsend, L. (September 28, 2022). Deterministic vs. stochastic models: A guide to forecasting for pension plan sponsors. Milliman Insight. Retrieved November 18, 2022, from https://www.milliman.com/en/insight/deterministic-vs-stochastic-models-forecasting-for-pension-plan-sponsors.