Pension discussions typically involve a language all their own: unfunded liability, net pension liability, actuarial liabilities, fiduciary net position, and fiduciary net position as a percent of total plan liability. I could list several more, but I want to focus on two terms: discount rate and investment rate (the latter is better known as assumed investment rate of return).
Perhaps more than any other decision, the rate used for those two items will have a significant impact when determining the amount of money New Jersey should appropriate each year to properly fund pension commitments.
Many people mistakenly use the two terms interchangeably — and use the same rate for both. But they are not the same. Used properly, the assumed investment rate of return refers to the assumption employed to project current and future investment income. For example, New Jersey assumes that its investment portfolio will earn 7.5 percent. (This has varied over the past few years and is projected to be reduced to 7 percent in 2023.) The discount rate, in comparison, is used to estimate today’s value of future pension liabilities.
A critical job of the actuary is to project the liability of the pension system. That requires projecting benefits paid in the future and “discounting” those projected benefits to the present. Using a proper discount rate is critical.
Using the assumed rate of return for the discount rate is a deeply flawed approach. It violates finance theory, which posits that a proper discount rate should reflect the riskiness of the liability and not the riskiness of the asset. (It also contrasts significantly with how private firms and other countries value pensions.)
Because pensions are generally protected by law and are likely to be paid even if poorly funded, the discount rate should reflect bond-market rates for low-risk assets, such as Treasury bills. Ordinarily, these rates will be far lower than the assumed investment return on a portfolio that includes stocks and bonds, because the latter assumes the portfolio will earn more, but at some risk.
Why would a higher discount rate be used for projecting liabilities? According to a report issued in 2012 by the, using a higher than appropriate discount rate has at least three results: 1) It creates pressure to invest in riskier assets; 2) pension plans appear healthier thus potentially creating incentives to reduce contributions or enhance benefits; and 3) it keeps employer contributions, and therefor appropriations, artificially low.
But using a lower discount rate properly reflects the liabilities of the pension system. There is no agreement among finance experts as to what specific discount rate reflects the riskiness of pension liabilities, and no disagreement that reflecting pension riskiness using the lower rate is the correct approach.
No federal agency or board issues rules for how state and local governments must fund their systems. However, the Government Accounting Standards Board (GASB) issued accounting and actuarial guidance effective 2015 that still permits an assumed investment rate of return to be used for projecting both liabilities and investments.
But there are other rules to guide actuaries. For instance, a state or local government that has a very poorly funded pension plan — based on defined criteria — must use a discount rate reflecting bond market rates, currently between 2.5 percent (treasury) and 4.5 percent (Moody’s corporate). These rates are far lower than a 7.5 percent investment rate and lead to much higher and more accurate estimates of liabilities.
And for those pension plans that are partially but not properly funded — again, based on defined criteria can use a “weighted average” of the two rates. New Jersey, not surprisingly, is a perfect case in point.
Clear? I doubt it. Could anything be more confusing? All state and local governments should be required to use a risk-free discount rate to determine liabilities — a logical conclusion, but not the rule. Rather we have a mixture of rules, with the lower risk rate only required for states that have very poorly funded pension systems. The impact: estimated liabilities for poorly funded plans have increased, while estimated liabilities for some pension plans have not — or only partially.
One of the most significant consequences of this mix-and-match approach, is that reports issued by most organizations assessing pensions work with different criteria when comparing the level of unfunded liabilities — in effect, comparing apples and oranges. Granted, under any criteria the states with poorly-funded pension systems, such as Illinois, Kentucky, Connecticut, and New Jersey will still be at the bottom of the list but it’s not possible to accurately assess how far they are from other states.
Moody’s Investor Service, which keeps close tabs on pension data, does prepare reports using a proper discount rate to ensure comparability among states. This “adjusted” data developed by Moody’s suggests the real total unfunded liability for all states is $3.9 trillion rather than the reported $1.6 trillion that results when using more favorable discount rates. GASB should issue new guidance that requires a low discount rate for all, as I detail below.
Until 2015, New Jersey used the assumed interest rate of return for its discount rate; for example, 7.9 percent in 2005, thus significantly understating its projected liabilities. Today, per GASB guidance, it uses a weighted average. For fiscal year 2017 the blended rate was based on the expected rate of return for investments (7.65 percent) and a municipal bond rate of 2.85 percent. There is a complicated computation made to determine how the rates were blended, but suffice to say the discount rate for the Public Employees Retirement System was 3.98 percent. Other systems used different rates. This blended rate is significantly lower than the 7.65 percent that would have been used prior to 2015, but still higher than the 2.85 percent that finance experts suggest should be used.
A report titledprepared by the Nelson Rockefeller Institute of Government at the State University of New York, reinforces my observations and draws several important conclusions:
Pension accounting and funding standards encourage investing in risky assets to reach projected yields.
Workers, retirees, and taxpayers ultimately bear the risk.
Those risks are poorly disclosed and understood.
Governments are not sufficiently disciplined to make adequate contributions — and consequently push significant costs onto future generations.
Based on these observations, the report suggests:
Pension funds of all state and local government should value liabilities and expenses with a low risk rate — not the assumed investment rate or the weighted-average approach.
Pension funds must disclose more fully the consequences of investment risk.
There needs to be an external downward pressure on investment risk.
Governments must keep their bargain and make realistic actuarially determined contributions.
New Jersey made many questionable decisions over the past 21 years creating the current massive shortfall in unfunded pension liabilities. In the current year, $3.2 billion has been appropriated for pensions — that’s good — but still approximately 50 percent short of the required amount. The shortfall would be even larger if the state used a discount rate recommended by Moody’s Investor Service and every public finance expert.
Unfortunately, because of poor policy choices New Jersey as well as other governments might very well need to rethink the size and scope of their pension systems and make significant changes.