Projection-driven plan design decisions

For plan sponsors with defined benefit (DB) plans in their retirement programs, the conventional wisdom in today’s economic environment has been that they should phase out these plans and replace them with defined contribution (DC) plans, either gradually (by closing the DB plan to new participants and possibly freezing benefit accruals for all current participants) or immediately (by terminating the DB plan). Typically the reason for this is that plan sponsors are no longer finding themselves willing or able to deal with the volatility and administrative complexities that are inherent in DB plans, and instead prefer the simplicity and predictable costs that DC plans can provide.

However, there may be reasons why a plan sponsor would want to go in the opposite direction and reopen an existing DB plan or even consider establishing a new one if it does not sponsor one already. Two corporate departments that could see appeal in this approach are:

• Finance: The plan sponsor could be reporting a large prepaid asset related to its overfunded DB plan on its financial statements, which it would be required to immediately recognize as an expense if it were to terminate the DB plan. The plan sponsor may not be willing to absorb such a large hit to its income immediately and would instead prefer to wear it away over time by having the DB plan generate some additional expense each year.
  Furthermore, the inherent structure of a DB plan gives it an advantage over a DC plan because, while actual cash contributions will always be necessary to fund promised benefits in a DC plan structure, a DB plan that is overfunded may have built up a credit balance over time that can be used to satisfy future statutorily required contributions instead of contributing actual cash. Leveraging the excess funding to provide additional benefits within the DB plan in place of providing benefits in the DC plan therefore provides two benefits to the plan sponsor—the increased benefits will generate additional expense over time, which wears away the prepaid asset, and at the same time the actual cash that the plan sponsor would have otherwise contributed into the DC plan is freed up for other purposes.
• Human resources (HR): The plan sponsor’s HR department may be having issues recruiting and retaining talent and offering a DB plan could be seen as a valuable perk by potential and existing employees. This would help differentiate the plan sponsor from its competitors given the likelihood that very few (if any) of them have open DB plans. If properly designed, it could be offered at little to no additional cost to the plan sponsor.

Identifying and mitigating potential risks of a DB plan

Even if the plan sponsor decides to reopen its existing DB plan or to establish a new one, other factors need to be considered before deciding on exactly how to set up the new plan design:

• Volatility: How much volatility (measured, for example, in pension expense or statutory cash requirements) is the plan sponsor willing to take on?
• Investment policy: How should the investment policy be adjusted? If the plan sponsor is reopening an existing DB plan, is the existing investment policy still appropriate given the need to fund future benefit accruals?
• Administrative expenses: Will the plan sponsor find the administrative expenses resulting from the new plan design acceptable, including Pension Benefit Guaranty Corporation (PBGC) premiums?
• Future outlook: Does the plan sponsor anticipate any future events that may be detrimental to its goals as it relates to the new plan design?

It is at this stage where a projection model can be useful because it can help provide valuable insight into these factors. While the remainder of this article focuses on how deterministic projection models can help plan sponsors, the same can be said for stochastic projection models (with the potential for more robust analyses—see our recently published article on deterministic vs stochastics models¹).

Using a deterministic projection model to quantify the impact of certain variables and risks on a DB plan

As plan sponsors prepare financial budgets each year, they often find it helpful to look out a few years into the future to see which direction certain plan measures (like pension expense or statutory cash requirements) are moving so that they can budget for the plan appropriately. Such results are exactly what a projection model can provide. With a projection model, given a plan’s participant data and design (whether it is the current plan design or an alternative design under consideration), the plan sponsor can review relevant projection results prepared by its actuary over a set number of years using the projection’s baseline assumptions and then, if necessary, have the actuary change certain variables to see how they impact those results.

For example, if the plan sponsor is solely focused on how its plan (or potential plan) would affect annual pension expense under Financial Accounting Standards Board (FASB) Accounting Standards Codification (ASC) 715, then it may wish to analyze the impact of the following variables:

• ASC 715 discount rate
• Changes to the plan’s investment policy that would affect the expected return on assets
• Actual asset performance each year, including how much in administrative expenses is paid from plan assets
• Inflation
• A one-year shock, either positive or negative, to the plan assets (for example, what happens if assets drop by 20% in a given year)
• Contributions made by the plan sponsor (both amount and timing)

A properly calibrated deterministic projection model will allow the actuary to provide results for metrics important to the plan sponsor while at the same time retaining the ability to vary assumptions like those listed above. When considering changes to a retirement program, it is important to have the ability to show varying results under different assumptions reflecting both the existing plan (baseline) and any alternative designs being considered. The reason for this is that plan sponsors will want to quantitatively measure whether the proposed changes have the intended effect on their retirement programs.

As an example, we analyze the impact of the discount rate (variable) on a plan’s annual pension expense (metric). In this example, the plan sponsor’s finance and HR departments have differing goals. Finance wants to terminate the plan eventually and HR is concerned about attracting and retaining talent. Consider the following baseline scenario results, shown in the table in Figure 1, produced by a deterministic model developed for a plan sponsor’s existing overfunded DB plan with frozen benefits and a large prepaid asset. The baseline projection uses a discount rate (DR) of 2.96%, which was determined as of the beginning of the 2022 fiscal year.

As can be seen in Figure 1, the current plan is projected to produce income of over $2 million each year, with that amount increasing over time. Remember that a plan sponsor maintaining a plan with a large prepaid asset will take a significant hit to its pension expense upon termination, and pension income (as shown in Figure 1) will only serve to increase this prepaid asset over time. Given how much interest rates have shifted in 2022, the plan sponsor can (and should) consider how these results would change if a more recent discount rate was used. Assuming a 5.50% discount rate is applicable for all future years, we now have the following results shown in the table in Figure 2.

With this scenario, the plan sponsor can see that even in the current interest rate environment its DB plan is still producing income each year at a rate that will never allow it to terminate the plan without incurring a large hit to its financial statements. Similar analysis can be done with other assumptions (such as changing the expected return on asset assumption to illustrate changes in the investment policy or introducing an asset shock scenario whereby significant asset losses are incurred in a given year) to illustrate the plan’s sensitivity to various market conditions.

Using a deterministic projection model to decide whether a change in plan design is desired, and if the plan design chosen is sufficient

Now suppose the Finance and HR departments of the plan sponsor discussed these results and decided to request that the actuary produce another set of deterministic projections reflecting an alternative plan design. The plan sponsor wishes to explore reopening the existing DB plan as of the beginning of the 2023 fiscal year to provide all participants with cash balance accruals each year. Like with the baseline results shown above, it is helpful to see the results of this new design under both the beginning-of-year discount rate of 2.96% and the more recent rate of 5.50% so that the impact of the discount rate can be isolated. The results of a specific cash balance² design are shown in the table in Figure 3.

Now the plan sponsor can see that, even though income is still being generated each year, the overfunding in the plan is being put to good use by providing meaningful benefits to participants, and the growth in the prepaid asset each year has been slowed dramatically as evidenced by a much lower pension income. Unlike with the current design, which shows an increasing pension income amount, this design is on course to generate expense in the future. This analysis can be repeated by changing different variables or creating various scenarios to help the plan sponsor decide whether it is willing to accept the volatility presented within the results. For example, the plan sponsor may want to run a pessimistic scenario, where asset performance is well below expectations for a period of time, to see if the resulting pension expense exceeds set tolerance levels.

In this example, the new plan design meets the Finance department’s goal of getting to a point in the future where terminating the plan is an option (by utilizing the plan’s excess funding to wear down the prepaid asset over time), while also meeting the HR department’s goal of retaining and attracting talent (by providing meaningful benefits to employees). But the use of deterministic projections does not end here. The plan sponsor will want to periodically revisit these projections to monitor trends and ensure the design continues to meet intended goals.

Conclusion

When a plan sponsor is faced with the challenge of what to do with an existing DB plan that is not performing in a manner that meets its long-term goals, it is often confronted with questions that cannot be answered by the data available from the latest valuation. With access to a projection model that gives the plan sponsor the ability to customize assumptions and plan designs, it has the tools necessary to plan ahead and effect changes that realign the plan with its long-term goals. This article explored, as an example, the way the plan sponsor of an overfunded plan used deterministic projections to decide to reopen its frozen plan in a way that met its long-term goals. Being able to customize assumptions and plan designs allows plan sponsors to see the impact of relevant assumptions (such as discount rate) on various metrics (such as pension expense) while at the same time letting the plan sponsor zero in on a plan design it can be comfortable with, potentially helping the sponsor resolve and condense complex and difficult questions into a few relatively simple decisions.

¹ Townsend, L. (September 28, 2022). Deterministic vs. stochastic models: A guide to forecasting for pension plan sponsors. Milliman Insight. Retrieved December 16, 2022, from https://www.milliman.com/en/insight/deterministic-vs-stochastic-models-forecasting-for-pension-plan-sponsors.

² The results in Figure 3 reflect a cash balance plan design that grants annual pay credits of 4.5% per year and interest credits of 5.0% per year.