However, the GAO said it found one significant area where there is some, but not universal, room for agreement. Specifically, many experts supported providing multiple measures of liabilities for different purposes to provide a more complete picture of pension plan finances.

Though it made no formal recommendations, in a report to Tom Harkin (D-Iowa), the Chairman of the Committee on Health, Education, Labor, and Pensions in the United States Senate, the GAO said there may be value in providing multiple measures of liability and cost, using both assumed-return and bond-based discount rates—carefully labeled to describe their purpose (e.g., with some measures, such as funding targets, not even necessarily labeled “liabilities”)—and with explanations of what these measures do and do not represent.

“The measurements resulting from these different discount rate approaches can ultimately improve the understanding, management, and governance of the finances of pension plans. In short, there may be value in having multiple liability measures to arrive at funding, benefit, and investment policies that will better balance risks and rewards to plan participants and all other stakeholders,” the agency wrote in its letter.

The GAO said policy options to address DB plans’ challenges may be addressed by fostering the use of appropriate liability measurements and discount rate assumptions and increased transparency concerning their financial health. However, options should also be sensitive to the crucial need to ensure that benefits remain adequate.

In its report, the GAO explained that determining which discount rate to use to estimate plans’ future liabilities is important because different rates can cause great differences in calculations. For example, the present value of corresponding liabilities for a $1,000 benefit payable seven years from today is $760 at a 4% discount rate and $583 at an 8% discount rate. For a benefit payable seven years from today, the liability measured at 4% is 30% higher than the liability measured at 8%. Further, for a benefit payable 15 years from today, the liability measured at a 4% discount rate is 76% higher than the liability measured at an 8% discount rate.

The report said methods for determining a plan’s discount rate can be categorized into two primary approaches—the assumed-return and bond-based approaches. These primary approaches in turn can have different variations, such as the use of “smoothing” with bond-based approaches, as is used by private sector single-employer plan sponsors under the Employee Retirement Income Security Act (ERISA), where bond interest rates are averaged over multiple current and historical years.

The assumed-return approach bases the discount rate on a long-term assumed average rate of return on the pension plan’s assets. As employed by U.S. public pension plan sponsors, this approach often produces discount rates between 7% and 8%. The bond-based approach uses a discount rate based on market prices for bonds, annuities, or other alternatives that are deemed to have certain characteristics similar to pension promises. Under this approach, the discount rate is independent of the allocation of plan assets. The relevant bond quality (e.g., AAA-rated, AA-rated, etc.) can depend on the specific purpose of the liability measurement, which can result in rates that vary considerably.

Public plans and private sector multiemployer plans—which typically use an assumed return approach, and thus, higher discount rates—generally report higher funded ratios, and their liabilities generally appear lower, than those of comparable private sector single-employer plans. GAO noted that this approach generally produces lower liabilities than variations of bond-based approaches with little or no smoothing (which often produces lower discount rates), as used by private sector single-employer plan sponsors for financial reporting purposes. For example, the report said, Mercer estimated that at the end of 2013 an average private sector single-employer plan sponsor would have a discount rate of 4.88%, and according to the National Association of State Retirement Administrators, public plan sponsors assumed a return of 7.72% on average as of December 2013.

Some experts the GAO spoke with view differences between public sector and private sector single-employer discounting approaches as appropriate because they believe public plans can best estimate their pension costs using very long-term assumed returns as their discount rate. There were other experts, however, who disagree with this viewpoint or see value in both types of measures.

Some experts said that the assumed-return approach could incentivize public plan sponsors to invest in riskier assets because doing so can increase the assumed-return discount rate, thereby lowering reported liabilities and reducing funding requirements. Some experts told the GAO it is also possible that a plan’s discount rate approach could influence future benefit levels. At the most basic level, the cost of benefits typically will appear lower using an assumed-return discount rate than using a bond-based discount rate, perhaps leading to compensation packages that are weighted toward more retirement benefits or to larger overall compensation packages. 

Further, some experts expressed concern that sponsors of plans that have earned more than the assumed return, such as in a bull market, have given this extra return to participants as a benefit increase, but that benefits would not be cut at the same rate during periods of low returns. To the extent this occurs, it would mean that an assumed-return discount rate would need to be lowered, or the plan liability increased in some other manner, to reflect the fact that future bull-market gains would not be fully available to offset future bear market losses.

In contrast to the investment incentives that public plan sponsors (and multiemployer plans) may face, the use of a bond-based discount rate for private sector single-employer plan sponsors can create an incentive to invest in bonds to make pension contributions more predictable or financial reporting results less volatile, GAO said. It explained there are various legal constraints on the ability of plan sponsors to reduce future or current benefit accruals, which vary further for public and private sector plans. For plans using bond-based discount rates (with little or no smoothing), liability values will fluctuate with changes in market interest rates. A bond-based investment policy can be used so that plan asset values will move in tandem with liability values as interest rates fluctuate. The greater the match between a plan’s investment assets and the amount and timing of its projected benefit payments, the more stable the plan’s funded status will be.

However, GAO notes that holding bonds means forgoing potentially higher returns from equities. Thus, the more that a plan matches assets to liabilities by purchasing similar-duration low-risk bonds, the more expensive the plan may become to fund, which may provide a disincentive to invest more in bonds.

Some experts GAO spoke with believe the appropriate discount rate to use depends on the purpose of the measurement. Regardless of whether they believed that, all experts interviewed pointed to at least one among six considerations that influenced their views about discount rate policy:

• Level and predictability of cost – Bond-based discount rates can lead to costs that are too high (depending on market conditions), or too volatile from year to year for sponsors to bear. On the other hand, an emphasis on ensuring predictable and level costs from year to year may mean plan sponsors are not contributing enough to adapt to changing market conditions.

• Benefit security and risks to stakeholders – Basing funding on assumed returns increases the risk that insufficient assets could be on hand when needed, or that contributions will have to be increased, or promised benefits reduced, in the future.

• Plan and sponsor characteristics – Key risk factors include the size of the plan relative to the size of the plan sponsor, the maturity of the plan, and the strength of the plan sponsor.

• Intergenerational equity – Experts agreed that each generation should pay its fair share for pension costs, but disagreed about what this meant in practice. One viewpoint is that using assumed returns passes uncompensated risk to future generations. The other viewpoint is that using bond rates charges current generations an amount greater than the expected long-term cost.

• System sustainability – One viewpoint is that the use of bond-based rates pushes plan sponsors to abandon sponsoring DB plans. The other viewpoint is that use of assumed returns leads to poor risk management practices (for both investment and benefit policy) and to crises of poorly funded or failing DB plans that cause sponsors, or create pressures, to abandon these plans.

• Transparency and comparability – Providing a bond-based measure in addition to an assumed-return measure may help outside parties get a transparent, comparable view of plan liabilities, based on market measures. However, some experts argued that multiple measures might not enhance transparency because such information could be confusing or misleading about the likely cost to fund a plan.

Nearly half of the experts GAO interviewed supported the use of multiple measures for valuing pension plan obligations. Many experts stated that reporting multiple measures of liabilities would be useful in providing transparency. Some said that reporting liabilities based on multiple discount rates would provide fuller transparency into a plan’s finances than using a single rate. Some experts also took the view that public plans providing liabilities at both a bond-based and assumed-return discount rate could provide a broader range of information to plans and employers to guide plan policies, and could potentially provide a useful check on the assumed-return measurement.

Nearly one-quarter of the experts GAO interviewed argued that only a bond-based approach should be used to value plan obligations, while nearly one-third favored use of only the assumed-return approach.

The GAO looked at practices in other countries—Canada, the Netherlands, and the UK—and found they apply a variety of approaches to discounting. Canada requires determination of multiple measures of plan obligations, based on both assumed returns and high-quality bond rates and annuity prices. The Netherlands requires that plan obligations be measured based on market interest rates, but allows the use of assumed returns for determining plan contributions or developing recovery plans. In the United Kingdom, discount rates are determined on a plan-specific basis and can include some allowance for assumed returns in excess of high-quality bond rates, depending on plan characteristics and the strength of the sponsor.

To the extent that plans in these countries use long-term assumed rates of return, they are generally lower than those used by many U.S. public plans under recent market conditions, the GAO found. Experts GAO interviewed in these countries described a greater degree of government oversight which the GAO said might help explain their use of lower assumed returns.