Sunday, October 11, 2009

The real problem with public employee pensions

Cross-posted from The Enterprise blog.

The Washington Post
reports today on the sorry state of funding in state and local employee pensions, focusing on the impact of recent poor stock returns. While a poor investment climate certainly hasn't helped, it's not the biggest reason public employee funds are in bad shape. A bigger reason public pension plans are underfund is that, in effect, we told them they can be. State and local pension plans use different and far less demanding accounting rules than do corporate pensions, even though public employee benefits are guaranteed by law while corporate pension benefits are not.

The key issue is how to "discount" future benefit obligations to the present, which tells us how much plans must have on hand today to fund their future liabilities. A high discount rate lowers the present value of a future obligation, while a low discount rate implies a higher present value.

Corporate pension plans must discount their future benefit liabilities at the low interest rates earned by high-quality corporate bonds, while public pension plans are allowed to use the much higher expected return on their assets, which include a high proportion of stocks and, more recently, hedge funds and private equity. The effects can be startling.

For simplicity, imagine a pension plan that owed a lump sum of $1 million 15 years from now. Discounting at a 6.25 percent interest rate – which is typical for corporate bonds today, although higher than several years ago – the present value of that obligation would be approximately $403,000.

Using an 8 percent return, which is not uncommon for public pension funds, the present value of that $1 million future obligation would be only $315,000. Plans that have at investments worth at least $315,000 would consider themselves fully funded and, in some cases, use this status to justify increasing benefits.

Defenders of current actuarial practice argue that public pension funds are different, since governments can't go bankrupt – a proposition that may well be tested soon – and because they can always raise taxes to fund deficits. The latter may be true, but surely the point of pension accounting is to give taxpayers some idea of the contingent liabilities hanging over them – which current methods do not.

Moreover, there is a good case that public pension funds should use lower discount rates than corporate pensions because public pension benefits are a safer asset for the beneficiary and thus a more binding obligation on the pension plan. Corporate pension benefits are not fully guaranteed if the sponsor goes bankrupt, while in most states accrued public pension benefits are treated as a binding obligation. In many states these benefits are guaranteed in state constitutions.

If these pension obligations are as binding as state government bonds, it makes sense to discount them at the same rates. Nationally, the yield on a state government bond with a maturity of 15 years averages around 3 percent. Discounted at that rate, a $1 million future obligation requires $642,000 in assets today – over twice as much as the funds themselves would consider necessary.

Moreover, while public pensions discount their future obligations at the "expected return" on their investments, this doesn't mean we can actually expect those assets to meet their goals. The reason is that funds take as the expected return the average return on the asset classes they hold, and the average return is always higher than the median or typical return. Imagine that a public pension fund invested $315,000 in assets with an expected return of 8 percent and a standard deviation of returns of 13 percent. Using a Monte Carlo simulation we can check how often this portfolio is likely to exceed $1,000,000 in 15 years time. The answer is a little over 40 percent, meaning that there's an almost 60 percent likelihood that even a "fully funded" public pension plan won't be able to meet its obligations.

Allowing public pension funds to discount their benefit obligations at the expected return on their investments doesn't just lower the amount of funding they must undertake, it also encourages them to take more risk with their investments. Were a fund to hold only safe investments like Treasury bonds it could discount its benefits only at a low interest rate. But the riskier the investments they make the higher discount rate they can use. It's easy to see where this leads. For instance, the expected return on the Profunds "Ultrabull," which doubles the returns on the S&P 500 would be, well, double the expected return on the S&P 500 – or around 20 percent per year. This would solve plans' funding problems on paper, but it's hard to believe this is the most sensible investment strategy to take.

Accounting is a boring subject and so it's not surprising that it doesn't get much attention in the press or by lawmakers. But it's hugely important.


Buce said...

Seems to me the appropriate question to ask would be: what would it cost you to buy that kind of protection in the annuity market. I.e., suppose the public employee who retires at 65 and wants (a) to match his final salary; and (b) annuity (not-outlive-my-money) protection; and (c) a survivorship benefit; and (this is a biggy) inflation protection. Vanguard will sell you a package like that; I assume others will too. Cost will be a function of final salary but it won't be cheap. Seeing as how they are offering inflation protection, they'll use a real (non-nominal) interes rate inthe 2-3 pct range.

Andrew G. Biggs said...


Part of my comments are to show that it would cost a LOT more in private markets, but that's not due to any efficiency on the government's part but simply because they ignore part of the costs of their plans. In markets, risk comes with a price -- that's why stocks have to pay higher expected returns than bonds -- but public employee pensions basically act as if stocks come with guaranteed returns.
DB pensions may have some pricing advantages on annuities, though smaller than Social Security does (since it's a broader system). But these differences are tiny compared to the market risk issue.