For decades, public pension plans in this country have estimated future benefit liabilities using a discount rate that is based on estimated future investment returns of fund assets. This approach, an “assumed rate of return” method, has come under mounting criticism by financial economists and public policy groups. Using work developed by these economists, many argue that a rate based on investment return assumptions vastly understates pension liabilities. In their view, the rate should be based on low risk, or even risk-free, bond rates. Today, these bond rates are substantially lower than the assumed rates of return used by most public pension plans.
The result? A huge disparity in the estimated liabilities of public plans. In one analysis, two financial economists, Robert Novy-Marx and Joshua Rauh, argue in “The Liabilities and Risks of State-Sponsored Pension Plans” that the collective liabilities of public plans in this country using a risk-free bond rate are approximately $5.17 trillion. In contrast, the authors state, the aggregate liabilities estimated by public plans using assumed rates of return is $2.87 trillion — more than $2 trillion less than the “true” liabilities. The conclusion of many political and economic commentators is that public pensions and their public employer plan sponsors are facing a “tsunami” of unrecognized debt that will eventually swamp state and local governments.
The Rockefeller Institute of Government recently adopted this view in a new report and concluded that public pension plans follow a “deeply flawed” funding approach. According to this report, “Strengthening the Security of Public Sector Defined Benefit Plans,” the problem stems in substantial part from mismeasurement of the costs of pension funding by the pension plans themselves. The “proper way to value future cash flows such as pension benefit payments is with discount rates that reflect the risk of the payments. This is separate from the question of the rate pension funds will earn on their investments.” Although the Rockefeller Institute is careful to state that a risk-free bond rate should not be used as a funding requirement, the bond rate should nevertheless be used to disclose the “full cost” of pension benefits.
Last fall, the Government Accountability Office stepped into the fray and issued its own report summarizing expert views. Its conclusion is that opinions are deeply divided between the “assumed-return approach” and the “bond based approach.” Noting that the different approaches can, and typically do, produce vastly different estimates of liabilities, the GAO nevertheless made no recommendations.
These disparate views can be disconcerting, at best, for public pension plan trustees.
Looking more closely at the views of financial economists, it becomes clear that using an estimated rate of return to discount future pension liabilities is not misleading. This discount rate actually reflects the costs of funding pension benefits far better than using a risk-free or low-risk bond rate of return. If appropriately set, the former will reflect an estimate that is much closer to the actual cost of pension benefits and therefore the liabilities of the system. In contrast, discounting these liabilities using a hypothetical bond rate reflects an estimate of the future value of these benefits to plan members. The latter valuation is important: Employers and taxpayers should know the value of pension benefits received by public employees. But estimating this benefit amount does not reflect the actual costs of funding public pensions.
Defined benefit plans are designed to be pre-funded. Contributions are held in trust, and invested, in the present to pay for future benefits. This pre-funding of future liabilities is almost unique in public finance. Most liabilities of state and local governments are funded on a “pay as you go” basis. The costs of policing the streets, teaching students and administering the courts, other than associated pension costs, are paid by tax revenues at the time they are incurred. Public works projects are often funded through the issuance of municipal bonds. Thus, these costs (including the interest costs of borrowing money) are funded on a dollar-for-dollar basis: Every dollar spent is a tax dollar collected.
In contrast, the costs of pension benefits are not solely funded by tax revenue. Tax dollars in the form of employer contributions may be invested, and earn returns, to help pay for these benefits. On average, well over 70% of the total costs of pension benefits are paid by investment earnings.
Nearly all public plans invest their assets in a diversified portfolio. This asset mix typically includes stocks, corporate bonds, real assets and private equity funds. It is beyond dispute that the returns actually generated by these investments will actually fund future benefit payments, or liabilities, of the plan. Thus, the estimated future cost of these liabilities should be discounted using an estimated rate of actual returns.
Financial economists such as Messrs. Novy-Marx and Rauh argue that using an assumed investment rate of return is inappropriate because it does not reflect the risk or, more accurately, the lack of risk that plan participants will not receive the stream of promised benefit payments.
As the Rockefeller Institute explains: “Pension liabilities … must be estimated based on the benefits expected to be paid — i.e., future cash flows. Financial economists and analysts ordinarily value future cash flows using a discount rate that reflects the riskiness of the payments. … This is a tenet of modern finance. Amounts that are extremely likely to be paid will have lower risk, and therefore a lower discount rate, than amounts that are less likely to be paid.”
Because public pension plan payments have very little risk of not being paid, under “tenets of modern finance” a risk-free bond rate should be used to discount public pension liabilities — in other words, a bond rate that reflects little or no risk to the buyer, the Rockefeller Institute report concludes.
But estimating the present value of a future stream of cash flows is not the same as estimating the future costs of paying for them. This distinction is illustrated by an example used in the Rockefeller Institute report. “If the government has a firm commitment to pay you $1,000 in fifteen years, you will use a lower discount rate to determine the value of the promise today than if your shiftless brother-in-law promises to pay you the same amount.” Of course this is true, if the discount rate is used to value the benefit to the recipient. On the other hand, this calculation will not, at least not necessarily, reflect the cost, in today's dollars, of paying out $1,000 in 15 years. The cost to the obligor, either the government or the brother-in-law in the example, will depend upon how much is earned on the $1,000 over the course of 15 years. If the brother or the government earns only a risk-free rate of return, then the present value should be calculated using a corresponding bond rate. If the government earns a higher rate of return, then a corresponding higher discount rate will be used. In either case, the rate of investment return is the appropriate discount rate.
The premise of the Rockefeller report is that the promise to make future payments of benefits is the same as the promise to pay a note or a bond. In other words, the price at which the pension liabilities would trade in the open market if they were packaged as bond or a note. But estimated liabilities of a pension plan are neither bank loans nor are they bonds.
Unlike banks, plan members do not lend money to their pension plan (or employer) at an agreed-upon interest rate. This interest rate is essentially the product of two factors: 1) the costs of capital to the lender bank and 2) the profit the bank expects to earn on the loan. Public pension plan members are not in the money-lending business and certainly do not earn a profit on their unpaid benefits.
Similarly, the financial marketplace does not price these future cash flows because they are non-transferrable. Unlike bonds or commercial notes, unpaid pension obligations cannot be bought or sold in the market. Because the sponsors of governmental pension plans — the employers — are clearly “going concerns,” liquidation and transfer of their liabilities is not realistically feasible. Instead, estimated pension liabilities are just that — estimates of future obligations that are primarily funded through investment returns.
Finally, using a discount rate that is based on the credit-worthiness of the pension system (or the plan sponsor) creates an anomalous result: Plans that are poorly funded will record lower liabilities than those that are well-funded. Because the risk of not receiving future benefit payments is higher with underfunded plans, a “bond rate” will also be higher. A higher discount rate results in lower estimates of future liabilities and, if the rate is used to calculate contributions, lower levels of contributions. The result: Poorly funded systems will report lower liabilities and will receive smaller contributions than well-funded plans.
Estimating an investment rate of return for plan assets is not an invitation to act imprudently. Unreasonably high discount rates can artificially depress expected costs and contribute to plan underfunding. The rate should be set using a diligent and thorough evaluation of the amount of investment risk the plan can prudently assume and reasonable estimates of future investment performance.
Investment experts generally agree that over the long term higher-risk investments (like common stocks) will yield higher levels of returns. They also agree that higher-risk portfolios generally have greater volatility in returns. For many plans, higher volatilities might threaten the long-term soundness and health of the plan. Diversification of assets can mitigate some volatility risk but trustees should also evaluate the composition of the current investment portfolio, the current funded status, and key plan and sponsor characteristics. These characteristics should include:
- Plan size. A large plan, especially one with larger unfunded liabilities, with a relatively small plan sponsor (in terms of tax revenues) might have great difficulty handling large investment losses in any particular year.
- Plan maturity. An older plan with fewer new members will have greater liquidity needs than a younger plan. Investment losses in any one year will therefore have a greater negative impact on the funding of future liabilities.
- Plan sponsor strength. A sponsor with strong revenues is better positioned to handle investment risks than a poorly funded one.
- Correlative risks. A financial market downturn might decrease both asset values of the plan and the financial health of the plan sponsor. Plans might thus raise contribution rates at a time when plan sponsors can least afford increases.
Each of these factors may be evaluated using stochastic and stress-testing models. Identifying these risks, designing an asset allocation that meets them and then setting a realistic expected rate of return will do more to meet the plan's obligations than using a hypothetical discount rate based on the value of benefits to plan members.
Peter Mixon is the former general counsel of the California Public Employees' Retirement System and is currently a partner at K&L Gates LLP. The views expressed in this article are his own and do not reflect those of his current or former employer.