Public Pension Plans
ed bartholomew, an independent pension consultant, is the former chief financial officer of the Inter-American Development Bank.
Published April 28, 2017
According to the 2013 annual report for Detroit’s general retirement system, the city’s pension plan “is stable and secure and expects to meet all future retirement obligations to its members.”
Wait. Isn’t Detroit bankrupt?! Well, I fudged here just a bit. The annual report quoted preceded the city’s bankruptcy by some months. In fact, though promised pension payments were cut, the retirees didn’t do too badly. The bankruptcy judge blessed a settlement (at the expense of other creditors) that preserves the lion’s share of their benefits. At least for now.
But the mess in Detroit, alas, is only the beginning of a great unraveling of retirement plans for state and local employees in much of the United States. And even relatively happy endings can hardly be taken for granted.
The first question that occurs is who is going to be stuck with the bill. Taxpayers? Pensioners? Residents who will have to make do with lower-quality public services? Going forward, there’s also the question of how the process of funding promised benefits should be changed so the funds are sure to deliver. The answer requires a dip into financial theory. But don’t let the minor technical hurdles prevent you from understanding a pretty straightforward idea.
First, Some Facts
Not many of us work for employers who still offer “defined benefit” pension plans, which promise monthly income for life in retirement. In the private sector, such plans have largely been replaced by defined contribution plans — typically 401(k) plans that are less costly to employers and shift most of the risk of saving enough for retirement to employees. But if you’re a state or local government employee, you probably still have a defined benefit plan. What’s more, the plan may well have been a big factor in convincing you to work for the government because government jobs often make up mediocre pay with generous benefits.
That puts future retirees in the uncomfortable position of depending on contractual rights that may be contingent on courts willing to enforce them — a position made more uncomfortable by the fact that millions of them don’t have Social Security as a backstop. This, for example, is the case for the failing Dallas police and fire pension system, which actively resisted attempts to include pensioners in the Social Security system.
Consider, too, that, unlike private pension benefits, state and local pension benefits are not insured by the Pension Benefit Guarantee Corporation, which was set up by Congress in 1974 to insure private pension funds. This exclusion was based partly on a Constitutional objection. But it was also assumed that state and local taxing power would be sufficient to ensure that all promises would be kept. Now, in 2017, that assumption doesn’t seem quite so solid.
When the Unthinkable Must be Thought About
Detroit pensioners learned the hard way that, despite soothing messages repeated over and over in the annual reports, a guaranteed pension annuity is only as good as the guarantor. While outright defaults have been rare, they may or may not be so rare going forward. It depends on whom you ask.
Keith Brainard, the research director of the National Association of State Retirement Administrators, notes that unfunded pension liabilities vary widely among plans but believes that “for the vast majority of states, cities and plans,” these liabilities are manageable. He attributes the cases of the largest unfunded liabilities, like those in New Jersey, to employers paying less than their actuaries calculate they should — and, presumably, hoping for a wizard investment adviser to save them.
Financial economists, as opposed to the professional actuaries who are paid to tell pension funds what they need to contribute, would agree that is a big problem and a distinguishing characteristic of the worst-funded cases. But they do not agree with Brainard’s assessment that there’s no systemic underfunding problem with the “vast majority” of public pension plans.
Joshua Rauh, professor of finance at Stanford University, notes in a recent research report that public pension funds still commonly estimate that they’ll earn returns on investments of 7.5 to 8 percent annually. Those numbers, based on past earnings on pension portfolios, are probably unrealistic going forward. But, like virtually all financial economists who have studied the issue, he has a more fundamental beef with the actuaries’ approach. He argues that “this practice obscures the true extent of public-sector liabilities” because it treats an uncertain expected return as though it were a sure thing.
Under rules set by the Government Accounting Standards Board, pension liabilities are valued by discounting promised benefits at the expected rate of return on plan assets. Public pension actuaries employ a similar approach for estimating plan cost and funding requirements. Under this standard, if taking more investment risk allows a pension plan to expect a higher return on those investments, employer contributions can be smaller. The same goes for the value of accrued liabilities — what’s owed for benefits earned, but not yet paid.
It’s not hard to understand why this approach is popular. Investing in low-yielding but ultra-safe government bonds, which could absolutely ensure the payouts on guaranteed pensions, would vastly increase the cost of plans to employers and, arguably, make pension plans unaffordable.
But this way of justifying business-as-usual rubs financial economists the wrong way because it implicitly treats investment risk as costless. A universal principal of financial valuation is that systematic risk — what’s left after diversifying as much as possible — has a cost because risk-averse investors would not otherwise be willing to bear it. Treating the risk premium as earned before the risk has been borne is fundamentally flawed. But that’s what the official accounting rules permit.
So why should you care about this in-the-weeds technical difference over the cost of risk and what that implies for the right way to measure pension costs? Because what you don’t know can hurt you — whether you’re a taxpayer who may be asked to pay more later, a municipal bond investor who may discover she’s last in line trying to collect behind a large hidden debt or a pension plan participant who may be at risk for what bankruptcy specialists euphemistically call a haircut.
For fiscal 2014, the total unfunded pension liability for U.S. state and local pensions, reported under the official Government Accounting Standards Board accounting rule, was $1.2 trillion. This $1.2 trillion gap is the difference between the board-estimated liability of $4.8 trillion and plan assets of $3.6 trillion. That means public plans have only 75 percent of the assets needed to cover future obligations, leaving a 25 percent gap to be covered by future taxpayers. And that 25 percent estimate assumes the pension funds will earn the high returns hoped for on a risky investment portfolio — returns of 7.6 percent on average into the distant future.
Financial economists, however, say the underfunding problem is much worse. Rauh, using a much lower interest rate to discount liabilities, put the total fiscal 2014 gap at $3.4 trillion, nearly three times the $1.2 trillion reported under Government Accounting Standards Board rules. If this larger estimate is the true funding gap, state and local plans are barely 50 percent funded. Moreover, adjusting for weak investment performance along with even lower interest rates on riskless investments in the two years since Rauh made his estimate, financial economists guesstimate that the fiscal gap in 2016 is $5 trillion to $6 trillion. That implies the plans have only 40 percent of what they need to meet their obligations.
The main cause for this large difference is, as noted above, the discount rate, which determines the present value of a dollar owed in the future. Higher discount rates make that future dollar owed less costly, and lower discount rates more costly. According to financial economics, the appropriate discount rate for a future cash flow is the rate on a default-free bond (like a U.S. Treasury) due at the same point in time, plus a spread for any risk of nonpayment. Returns expected on assets used to fund the liability simply aren’t relevant for the purpose of valuing the liability.
So, if a pension promise is thought to be free of the risk of default, a financial economist would discount it at a default-free rate. And this is what Rauh did.
But wait, you say. We just looked at the failing pension plans in Detroit and Dallas. If benefits are at risk, they’re not default-free. Right, but note that’s not what Keith Brainard and others advocating for expected-return discounting are basing their argument on. To put it another way, saying that pension debts are smaller because they might not be paid should give no one comfort.
Advocates for the Government Accounting Standards Board’s approach (a group that includes most public-pension actuaries) think they are answering a different question. They aren’t asking what the market value of the liability is — something they consider mostly irrelevant — but how much needs to be set aside today to pay the promise when due. And if they can reasonably expect a higher return by taking more risk, why can’t they assume that’s what the plan will get?
Resolving the discount rate debate hinges, in the minds of many, on whether risk stays risky in the long run. If investment risk vanishes — or at least significantly diminishes — over the very long time horizons in which pensions operate, advocates for the standards board’s approach claim that allows using the expected return on risky assets to discount liabilities, which would remain secure.
In the Long Run
If we think of risk only as the bumpiness of the ride, not as uncertainty about where the ride will end, then the so-called equity premium — the extra return expected for stocks — is merely compensation for fortitude and patience. And if earning an equity premium is guaranteed if we only wait long enough, there’s no risk in the long run.
Can this be right? You’ve probably been told by some financial advisor that a long-term investor — say one investing for retirement 20 years away — should invest mostly in stocks, because over long periods stocks outperform bonds. Framed more carefully, the rationale is not because stocks outperform over long periods, but because stocks will likely outperform — or because they have in the past.
Framing is important: saying simply “stocks outperform” leaves no room for uncertainty, while inserting “will likely” acknowledges they may not. Consider an analogy to another type of risk, the risk your house will burn down.
You would not claim that insuring your house against fire is a waste of money because houses don’t burn down — that’s clearly false. But you’d also not justify forgoing insurance because your house will likely not burn down, or because, historically, no houses in your neighborhood have burned down.
Forgoing insurance is unwise, even though doing so would almost certainly save you money. You insure not because you expect your house to burn down, but because it might. And you prefer a sure loss you can afford — the insurance premium — to the small chance of a huge loss you can’t afford. Only if you could afford the loss — say, because you’re rich — might forgoing house insurance make sense.
We can (and should) think about investment risk the same way. It’s not correct to say stocks outperform as a timeless truth, like “the sun rises in the east.” While acknowledging that equities have historically outperformed bonds and likely will do so in the future, there’s still a nontrivial risk they won’t — and not just over short periods, but over long ones as well. It only makes sense to hold equities, then, to the extent the investor can afford the potential loss.
For financial economists, the above paragraph seems a no-brainer. To see why, let’s assume the opposite and see that it implies something that makes no sense.
Suppose that over a sufficiently long holding period — say 30 years — there is no chance that a diversified portfolio of stocks will return less than a 30-year default-proof U.S. Treasury bond. If that were true, one could obtain what financial economists call riskless arbitrage — a sure profit with no money down and no risk, the financial equivalent of a perpetual-motion machine.
Imagine a bank that borrows $1 billion by issuing a 30-year bond, and invests the proceeds (but none of its own money) in stocks, held in a legal trust with dividends reinvested to secure repayment of the $1 billion plus accumulated interest on the bond in 30 years. First, recognize the bond is free of default risk in this scenario since it’s guaranteed by the stock portfolio, which in 30 years is sure (by assumption) to be worth more than principal and interest owed on a default-free bond. Second, recognize there will be money left over after paying off the bond for the same reason. Having put no money down, the bank is certain to make a profit. Can this be possible?
Even if this were possible at some point in time, the very act of exploiting the opportunity — buying stocks and borrowing — would bid up stocks and interest rates, until it was no longer possible. Thus, the assumption that stocks can never underperform a default-free bond ultimately cannot be true. So, stocks are risky, even over the long run. In addition to the one-in-two chance they’ll return less than expected, there is a nontrivial chance they’ll return less than a default-free bond.
To see why this is not a small problem for pension funds, consider how badly underperformance compounds over the long periods that pensions must operate. Suppose you’re expecting to earn 7.5 percent per annum on a risky asset portfolio over 30 years, but instead earn 4.5 percent. If you were looking out just one year, you’d have 3 percent less than expected — a manageable problem. But what if the difference in returns is compounded over 30 years?
Investing $1,000 and earning the 7.5 percent expected every year for 30 years would result in an end balance of $8,755. If instead you earned 4.5 percent, the end balance would be only $3,745 — 57 percent less than expected. Now, suppose your average annual return over 30 years is an even-worse 1.5 percent. In this case, you would end up with only $1,563, or 82 percent less than expected. While such a dismal return is not likely, it is certainly possible. Indeed, Japanese equities have performed much worse than that over the nearly 30-year period since 1989.
Suppose you participate in a defined benefit plan offered by your employer. During your working years, you earn, in addition to wages, the right to an annuity when you retire — a monthly check for life. For example, suppose your employer’s plan promises a monthly pension equal to 2 percent of final pay for every year worked. So, if you have worked for 40 years, your pension will equal 80 percent of final pay.
But what if your employer doesn’t have the money when the time comes? To mitigate that risk, pension plans are typically prefunded. During your working years, funds are set aside and invested, so there will be enough to pay the promised pension.
How much, then, does your employer need to contribute each year to fund the plan and secure the promise? In deciding that, professional pension actuaries make several assumptions — most critically on how much the plan will earn on its investments and on how long participants will live.
If the actuary knew precisely how much each participant would receive annually in retirement and for how long, as well as the return on the pension plan’s investments, the math would be straightforward. But since this information is not known, it must be estimated. And this is where risk creeps in on little cat feet.
The actuary’s calculation is based on expected values, where “expected” is the estimated statistical mean for a distribution of possible outcomes. If there’s an equal chance that the realized outcome for each assumption will fall on either side of what’s expected, then there will be a one-in-two chance that the calculated funding will be less than what’s required to pay the promise. Maybe a little less, maybe a lot less, depending on how much uncertainty there is around each assumption.
Now, while any individual’s life expectancy is highly uncertain, the average life expectancy for a group becomes increasingly less uncertain as its size grows. And so, defined benefit plans for large companies that have thousands of participants are effective vehicles for diversifying this type of risk.
Other risks, however, are linked to the discretionary behavior of the plan managers — for example, investment risk. Pension plans could hold low-risk bonds with cash flow matched closely to the profile of the benefit that will be owed. And indeed, insurance companies that sell fixed annuities, which function very much like defined benefit pensions, follow an investment strategy much like this.
The only reason pensions — especially public pensions — deviate from this approach is that it lets their actuaries assume a higher investment return, thereby reducing funding requirements. But this approach comes at a cost: a higher risk of falling short, perhaps by gigantic sums. A trillion here, a trillion there...
No Free Lunch
Return now to what’s wrong with the financial management of public pensions in the United States. It boils down to this: the plans, as currently managed, do not account for the cost of investment risk. As a result, the true cost of providing a secure pension benefit is more than what is reported, so funding is insufficient.
This means benefit payments down the road will depend to a significant degree on the ability of future governments to make up potentially significant shortfalls. The money incorrectly thought saved in smaller government contributions today is just the unaccounted-for market price of risk. That leaves the risk itself to be borne by someone in the future.
Consider, too, that since taking investment risk justifies assuming a higher asset return, it makes benefits appear cheaper. This creates a bias in favor of taking risk. And since the expected return is an assumption, not an observable parameter, there’s a bias to be optimistic. Note that, if contributions were based on an unbiased expected return — and sufficiently large to track actuarial estimates — there would still be a 50 percent chance of shortfall. However, since the average public plan is only about 75 percent funded — and this is based on an optimistic 7.6 percent expected return — the reality is a much greater than a 50 percent chance of shortfall. For the public plans in Illinois, New Jersey and Kentucky, which are just 40 percent funded, the situation is more grave.
Yet another complication of ignoring the economist’s view of risk: decisions about benefit increases are based on lowballed cost estimates. For example, given a choice between offering bumped-up pension benefits or higher salary in union contract negotiations, sponsors will be more inclined to improve benefits since future benefits are cheap. And if a plan seems to be more than 100 percent funded following several years of good asset returns, then benefit enhancements may seem covered, not requiring additional payment by the sponsor. Many plans did just this following the long bull market of 1982 to 1999.
Of course, once granted, benefit enhancements are owed and cannot be rescinded following market setbacks — unless, of course, the plan goes belly up and a judge is in charge.
There seems no limit to the ways in which ignoring risk undermines pensioners’ security. One example: the deferred retirement option plan, under which some pension plans allow retirees to keep working while investing their pension checks in accounts earning a guaranteed rate equal to what the plan expects to earn on its risky portfolio — say 7 or 8 percent.
The Dallas police and fire pension offered such a program. When the plan’s solvency was questioned, participants sought to withdraw their funds before the window shut — as it ultimately did. Indeed, thanks to the unfunded liabilities created by the plan, even basic benefits are now seriously at risk.
An income for life in retirement — the core feature of public defined—benefit pension plans — is a wonderful benefit, provided it is secure. It’s also an expensive benefit, becoming more so by longer life expectancies and lower interest rates. But reducing funding and taking on more investment risk does not make it more affordable, just less secure.
Following the path of least political resistance, pension plans are typically guilty of (a) assuming a discount rate that makes payouts less than certain, and (b) allowing pension funds to slip far below 100 percent funding, even when the estimate is based on ill-advised assumptions about the discount rate.
The commitment of a future government to make up any shortfall is like the commitment behind a bond. And like a bond, a pension obligation may not be fully met if the burden is too great, as in the case of Detroit. But public pensioners, unlike bondholders, cannot manage credit risk — the risk of not getting fully paid — by diversifying.
No bond investors would sensibly bet all their savings on the credit of a single issuer if there were any risk of default. But pensioners with underfunded pension plans are unwittingly doing exactly that, and in many cases the credit of the sponsor is deeply suspect.
To fix this large and growing problem, two things need to be done. First, government accounting standards and public pension actuarial practice need to recognize the cost of risk, thereby revealing how much a secure pension benefit really costs. Second, public pensions need to be governed by independent trustees who (a) are technically qualified to exercise oversight, (b) have no conflicts of interest and (c) have a clear mandate to represent the interests of the beneficiaries.
While taxpayers and other stakeholders (including the aforementioned professional pension consultants) may still find it less painful to accept the risks than to come up with the cash to secure pensions, the consequences of kicking the can down the road would at least be more transparent.
If the sponsors were committed to making public pension benefits secure, they could transfer risk currently borne by pensioners, who can’t manage that risk, to bondholders who can. If the government sponsor is not so committed — perhaps because the shortfall is already too large — it would be better to have the discussion about default sooner rather than later. Then all could consider how the obligations might be restructured to preserve as much value as possible, as fairly as possible.
It’s only human nature to delay difficult decisions — especially if the current generation of deciders can dump the decisions on the next. Figuring out how we got in this pickle, alas, is much easier than whacking a pathway out.