Public Pensions: Actuarial "expected-return" discounting violates intergenerational equity
Investments are a separate but related issue
Taxes are paid in exchange for government services. Those taxes are used to pay the government employees who provide the services. The employees’ compensation includes cash and benefits, including pensions.
What is intergenerational equity?
Suppose today’s taxpayers (Gen0) promise a pension payment in twenty years of $10,000 as part of an employee’s compensation package.
How much must Gen0 set aside now, and what should it be invested in, to be fair to Gen20, the taxpayers who will have to make the payment in twenty years?
“Fairness” in this context is called “intergenerational equity.” It is a principle of public finance applicable to public sector pensions. (It does not apply to corporate-sponsored pensions.1)
The 2014 Society of Actuaries Report of the Blue Ribbon Panel on Public Pension Plan Funding described intergenerational equity as “the desire for the full cost of public services, including pensions earned by public employees, to be paid by those receiving the benefits of those services.” The 2013 booklet called “Pension Funding: A Guide for Elected Officials” issued by representatives from a consortium of public pension advocacy groups described its attainment as when ”the cost of employee benefits is paid by the generation of taxpayers who receives services.” The Governmental Accounting Standards Board (GASB) version associated with expense recognition2 is embodied in the concept of “interperiod equity”:
…the burden of the cost of services is borne by present-year taxpayers… This burden is not shifted to future-year taxpayers … through an increase in the level of borrowing, …
Getting back to the example, consider two investments, the only ones in our hypothetical universe:
Investment A. A zero-coupon U.S. Treasury bond that pays off in twenty years with absolute certainty, i.e., no credit risk. Buying a payment of $10,000 in 20 years costs $4,146.
The implicit discount rate that reconciles the future payment and the price is 4.5%
Investment B. A risky investment pool that pays an unknown amount in twenty years based on a normal probability distribution. Buying enough that the statistical expected value (mean) and median are equal to $10,000 costs $2,584.
The implicit “expected” discount rate is 7.0%
At time zero, the members of Gen20 don’t yet know they will be Gen20, yet Gen0 has legally obligated them to pay $10,000. Gen0 could buy $4,146 of Investment A and place it in a trust, thus imposing no obligation or risk on Gen20 and fully paying for the services it received. Seems equitable, right?
However, what happens in current public pension practice is that Gen0 invests in Investment B, and Gen0’s actuary reports that the contribution needed for “full funding” is $2,584. The actuary claims that if $2,584 is contributed, intergenerational equity is satisfied because it is “expected” to be worth $10,000 in 20 years, based on a long-term “expected” (50% probable) annual return of 7%.
If you somehow knew you would end up in Gen20, this probably wouldn’t feel equitable. You’ve been saddled with the obligation to make a future payment of $10,000 but provided an asset with only a 50% chance of being sufficient to pay it.
Public pension funding based on current actuarial practices using “expected-return” discount rates inherently violates intergenerational equity. In the example, at time zero, Gen20 is effectively short (it owes) $4,146 of Investment A and long (it owns) $2,584 of Investment B.
Gen0 has appropriated the $1,562 difference from Gen20.
Gen0 need not worry that its financial statements will reflect this shortfall. GASB Statement No. 68, which governs “accounting and financial reporting by state and local governments for pensions,” requires expected-return discounting because (paragraphs 227-228):
…discounting using the long-term expected rate of return on pension plan investments, when pension plan assets are … invested using a strategy to achieve that return, reflects the long-term nature of the employer’s pension liability…
…[I]f the potentially significant effect of pension plan investment earnings is not considered in the measurement of the pension liability, … amounts recognized by the employer, including the employer’s cost of services associated with pensions as they are earned, potentially would be misstated and would fail to provide information appropriate for use in assessing the degree to which interperiod equity is achieved.
This reasoning is wrong. Knowing that the actuary used expected-return discounting is sufficient to say with certainty that “interperiod equity” is violated.3
Even the members of GASB who wrote this would pay more for an investment that pays $10,000 in 20 years with certainty than one with an “expected” payoff of $10,000 but with a wide distribution of possible outcomes.
People pay for greater certainty. One can’t buy “certain” for the price of “expected.” Yet public pension actuaries will attest that an obligation to pay $10,000 in 20 years with certainty can be fully funded for the price of an asset with only a 50% probability of being worth $10,000 when the payment comes due.
One can’t buy “certain” for the price of “expected”
This economic nonsense is why economist Allison Schrager, in Bloomberg, described expected-return discounting as “a convention that makes financial economists’ heads explode.”
What about when things work out? Does that make everything ok?
Now, imagine we’ve made it to time 20 and it’s time to pay the $10,000. Back at time zero, Gen0 took the actuary’s advice and invested $2,584 in Investment B, and, lo and behold, the annual return was 8% instead of the “expected” 7%. Gen20 now has $12,044 and only needs $10,000. Gen20 has a $2,044 windfall to spend on other things that they did nothing to earn.
Was Gen20 still cheated? Unquestionably.
Gen20 used none of the services that required this payment at time 20, yet, at time zero, Gen20 was unknowingly saddled with a debt of $4,146. That amount was the economically meaningful pension “cost” at time zero that should have been invested on Gen20’s behalf. It can also be thought of as:
The value of the debt that taxpayers issued to the employee in lieu of other forms of compensation;
The value of the pension that the employer should have included as part of total compensation in its financial statements;
The cash compensation that the economically rational government employee would have forfeited in exchange for the future pension payment, or the amount of additional cash compensation that she would have required to give up the pension payment;
The amount at time zero that the employee should have considered part of her compensation if she had another job offer with a different employer and was comparing total pay packages; and
The market value of the promised future payment that the bond holders of the plan sponsor should have taken into account to assess the sponsor’s total debt burden.
For Gen20, $4,146 is the payment that should have been made to the pension trust by Gen0 at time zero to fairly compensate Gen20 for the debt burden that Gen0 imposed on them. There is no alternative. It doesn’t matter how the assets were invested or how things turned out.
If the $4,146 pension cost had been invested in Investment B at time zero (instead of the $2,546), Gen20 would have $19,324 instead of $12,044. In other words, for the obligation and risk foisted on them, Gen20 could have received $7,280 more from market transactions than what they got from Gen0.
Even though Gen20 looks to be ahead with $12,044, they have been cheated out of $7,280. Gen0 benefited at Gen20’s expense in violation of the principle of intergenerational equity.
If the full cost is contributed, what should it be invested in?
Now, suppose that everyone finds the above argument convincing and that Gen0 contributes the full pension cost of $4,146 to the trust on behalf of Gen20. Is there anything wrong with investing the $4,146 in Investment B?
A 2004 paper by Jeremy Gold and Larry Bader addressed this issue in depth.4 I recommend reading it.
The authors find no advantage to investing in risky assets.
Their conclusion derives largely from the observation that the pension plan is a financial pass-through. Taxpayers bear all of its experience. Whatever the pension plan invests in is an investment exposure for taxpayers. Likewise, the pension liability is effectively a taxpayer debt.
Therefore, one must include a taxpayer’s financial exposure from the public pension plan to her exposure from her own personal portfolio of assets and liabilities to determine her total financial exposure.
In the example, contributing $4,146 and investing it all in Investment B is not a violation of intergenerational equity in the same sense as contributing $2,584 would be (regardless of how it’s invested). There is no appropriation by Gen0 as long as the full cost of $4,146 is contributed.
But investing the $4,146 cost in risky assets means that:
Taxpayers are exposed to more asset risk and asset-liability mismatch risk than they realize.
In theory, many taxpayers could adjust their portfolios (as described in the Gold-Bader paper) to bring their total exposures to those desired. In reality, things are too opaque for that to be possible. And people without a personal asset portfolio to adjust probably shouldn’t be involuntarily exposed to a lot of risk, as their employment prospects may be correlated with those for risky assets. That is, in a recession, one may become unemployed at the same time the bill to reimburse the plan for investment losses comes due;
The plan sponsor’s bondholders are subject to a greater risk of default. The volatility of the net pension debt that results from investments that don’t track market pension liabilities increases the likelihood that the pension assets will be insufficient to make the payment. If that happens, bondholders will be competing with pensioners for available resources.
Taxpayers will (and currently do) have significant exposure to risky and opaque private assets and other so-called alts (alternative assets) that have become popular investments for public pension plans.
Most taxpayers do not have the income or wealth to qualify as an “accredited investor,” which would be required to invest in these assets outside the pension plan. Yet, they have significant exposure thanks to public pension trustees.
If public pensions have fully funded properly-measured liabilities, there is little reason for either earlier generations of taxpayers or later ones to invest in non-liability-matching risky assets. If fixed-payment liabilities backed by risky assets worked as advertised in pension plans, state and local governments could generate free money for general expenditures by issuing (non-pension) debt and investing the proceeds in risky assets.
Conclusion: properly measured costs should be fully funded and hedged
It seems clear that costs should be fully funded and invested in liability matching (hedging) assets.5 Annual costs will still be volatile because market interest rates are volatile, and governments are said to prize stable and predictable annual pension contributions. But regardless of any short-term “smoothing” of actual contributions relative to volatile actual costs that may be considered expedient, accurate liability and cost measurements are always the proper benchmarks against which to judge funding and investment practices, the net debt owed by taxpayers, the value of the total employee compensation package, and the risk to a plan sponsor’s bondholders.
I’ll gladly post responses in this space that disagree with anything in this essay (or any others).
If a pension is underfunded, then the corporation’s owners should, all else equal, receive less on the sale of their shares, in an amount sufficient to account for the underfunding.
As distinct from funding, the rules for which are often prescribed by statute.
Unless the assets happen to be invested in assets that “match” the financial characteristics of the liability, but no one does that.
Bader, Lawrence N. and Gold, Jeremy, "The Case Against Stock in Public Pension Funds" (2004). Wharton Pension Research Council Working Papers. 411. https://repository.upenn.edu/prc_papers/411
Another excellent critique of actuarial discounting and the resulting generational inequities in public plans:
Gold, Jeremy. “Risk Transfer in Public Pension Plans” in The Pension Challenge: Risk Transfers and Retirement Income Security, ed. Olivia S. Mitchell and Kent Smetters (Oxford: Oxford University Press, 2003); available at https://pensionresearchcouncil.wharton.upenn.edu/wp-content/uploads/2015/09/WP18_Gold1002.pdf
This paper includes: “Actuaries and sponsors of public sector defined benefit pension plans agree that each generation of taxpayers should bear its fair share of the long term plan cost. Actuarial methods and assumptions are designed to equate expected costs across generations. This paper uses arbitrage principles to show that equating expected costs unfairly lowers risk-adjusted costs for early generations and raises them for later generations. The use of expected rather than risk-adjusted returns on risky assets leads to sub-optimal asset allocations, granting of valuable options (skim funds), and costly financing strategies such as Pension Obligation Bonds.”
The assets should match liabilities as closely as possible and be adjusted to keep the match close over time.
Larry, your analysis, as always, is rock solid. The source of the problem goes beyond these calculations. When the employer makes a new promise to employees of a defined benefit in the future, it immediately sets up an intergenerational problem, as it will not fund that promise immediately but only over time. The extreme case, as when DB pension plans first began, involved no requirement for funding but a windfall for the older workers in the firm.
I also struggle with how to let current employees benefit from the higher returns in the stock market. The risk to traditional bonds held for long periods of time, largely due to inflation, is greater than the risk to stocks. DB plans, particularly for a permanent employer like a state government, do offer the possibility of allowing an intergenerational sharing of risk both because of the implicit requirement to hold onto the stocks for long periods but also because the plans can avoid some of the wins and losses that derive from the timing of the deposit and the withdrawal.