Understanding the difference in how liabilities are calculated in cash balance plans—and why it matters
Assumptions are made and most assumptions are wrong.
-Albert Einstein
As the funded status of their defined benefit (DB) pensions has improved over the last several years, many of our pension clients have systematically reduced their allocation to equity and other return-seeking assets in favor of bonds, which hedge the interest-rate risk of the pension liability. For accounting/reporting purposes, that liability is estimated as the present value of the future benefits paid to retirees, discounted using the current yields on corporate bonds. Generally, the estimate of benefit cashflows themselves is independent of current market conditions, so as yields rise the present value (liability) drops, and as yields fall, the present value increases.
In order to reduce the sensitivity of the funded status to changes in yields, plans invest in bonds which match the interest-rate sensitivity (duration) of the benefit payments. In an idealized case, a fully funded plan can eliminate investment risk by matching contributions to annual accruals (service/normal cost) and by investing all assets in a default-free bond portfolio whose cashflows (principal and coupon payments) match the future benefit payments. Of course, the real world is complicated by mortality experience, bond defaults, incomplete bond markets and a preference by sponsors to substitute higher market returns for contributions.
As mentioned above, the benefit cashflows in a traditional DB plan are independent of current market conditions. Those cashflows are re-estimated annually by the actuary in the annual valuation exercise, and generally reflect gains and losses associated with the previous year’s deviations between assumptions and experience (mortality, workforce profile, salary growth, etc.) as well as any changes to the assumptions themselves.
What is the interest crediting rate? How is it determined?
However, since the 1990s, cash balance (CB) plans have become the more common form of a defined-benefit plan. For these plans, the benefit payments at retirement (typically lump sums for CB plans) are estimated as the current balance projected forward to retirement using an interest crediting rate (ICR), then discounted back to the present using the discount rate.
Actuaries have, in our experience, usually used a long-term assumption for the ICR, so the retirement-benefit cashflows are still independent of the current market yield, thus the CB liability continues to have sensitivity to changes in yields (via the discounting process).
Understanding rate-sensitivity in cash balance plans
As an example, suppose the cash balance for an employee is now $100, the ICR is 3.00%, the employee is expected retire in 20 years and the discount rate is 3.5%. The future cash balance is projected to be $181 (growth at 3% for 20 years), and the present value is $91. Now, suppose the market yield drops by 100 basis points (bps) from 3.5% to 2.5%. The projected future balance doesn’t change, but the discounted present value increases to $110, reflecting an interest rate sensitivity of about 20%. This sensitivity could be reduced by investing in a 20-year zero-coupon bond.
What happens when a variable ICR assumption is used?
In recent months, however, we’ve seen cases where a client or actuary has considered using a variable ICR assumption which moves with market yields. For the example above, if yields drop by 100 bps, the ICR drops to 2.00% and the projected future balance also drops from $181 to $149, while the discounted present value remains at $91. In this case there is NO sensitivity of the CB liability to changes in yields. For a plan currently using long bonds, this shift in assumptions might suggest a shift in asset allocation, since long bonds are increasing rather reducing interest risk relative to the liability.
Results from our analysis
In conducting asset / liability analysis for these cases, we’ve observed some interesting issues. Most importantly, there is no reliable hedge for the growth (and variability) of the liability if the ICR is based on a medium- to long-duration bond yield. Securities whose coupon payments adjust with rates—such as floating rate notes or bank loans—might be helpful, but introduce other risks to the portfolio.
Let’s unpack this a bit. Suppose, for example, the ICR is linked to the yield on long credit bonds plus a spread. Even if that spread is zero, one cannot reliably achieve that yield by investing in long credit bonds without also being exposed to the interest-rate sensitivity. If yields rise, the value of the long credit portfolio will drop, but the value of the liability will actually increase. As yields continue to rise, the return requirement for the portfolio also rises with the ICR. The asset allocation process and the notion of de-risking as funded status improves become more complicated because there is no effective hedge. The estimation of a hedge ratio becomes a distraction in managing risk.
Another interesting observation in evaluating short-term risk for a portfolio which currently uses long Treasury bonds is that if those Treasuries are replaced (with the change in ICR assumption-setting), one loses the flight-to-quality advantage that Treasuries typically provide in extremely negative equity markets. While dropping the long bonds provides better outcomes in scenarios where yields rise, the overall reduction in risk (considering the entire range of scenarios) is muted by the lack of ballast in extreme equity scenarios—which generally comprise the worst scenarios.
The bottom line
Amid the ongoing volatility in markets, it’s more important than ever for cash balance plan sponsors to understand the differences in how liabilities are determined—and how a variance in actuarial assumptions can impact the liability-driven investing (LDI) allocation of a portfolio. We stand ready to assist with any questions you may have.