Let’s talk pension accounting!

Yeah, I know, exciting topic — but really: it’s actually timely and relevant to provide a bit of background on pensions.

Public pensions are quite a bit in the news — in particular, how underfunded they are, and the fact that their liability valuations, under accounting principles for government, uses a discount rate based on the expected return on plan assets, so, typically about 7 – 8%. As an actuary accustomed to private-sector pension plans, this just feels bizarre. Of course, in the case of a fully-funded plan, if the plan assets reached this return expectation over the long run, there’s a certain logic in saying that this is a measurement of the funding status and a tool to determine required contributions, and, in fact, for many years the standard funding calculations for private-sector plans used such an approach.

But we’re not talking about funding. We’re talking about how to measure liabilities in a more absolute sense, irrespective of whether or to what extent these plans are funded. In such a case, for private-sector plans, FASB presecribes the use of “high quality corporate bonds” to set the discount rate, which at the moment produces rates of about 5%. It’s been higher — when I first started as an actuary I think rates were more like 7% — and it’s been lower, dipping down to 4% over the past couple of years.

The original FASB statement on how to account for pensions, FAS 87, issued in 1985, prescribed the use of a rate that represented the discount rate at which the liabilities could be settled — in other words, the idea was not to use a coporate discount rate, but to use the rate at which a company could purchase annuities to offload their liabilities. Only later did interpretations specify that the rate to be used is the bond rate, which, in general, should reasonably approximate the cost of purchasing group annuities (note that some large companies, such as GM, have made the news, at least in the pension consulting world, for buying annuities for their pension liabilities, concluding that the cost of annuities isn’t really that much more than the liabilities on their books, and a good deal when considering the risk and administrative cost of the pensions).

Internationally, the funny thing is that discount rates for different countries differ based on the bond rates in that country — which can lead to quite goofy results, when a company’s liabilities are valued at 15% because that’s what government bond rates look like (and there are no corporate bonds to speak of in the country), or (in one case) we’ve thrown in the towel completely and used economic growth rates because there are no reasonable government bond rates to use!

In any event, permitting asset-return-based discount rates produces any number of unpleasant real-world consequences.  In Illinois, back in the days of Rod Blagojevich, the state failed to fund its pension obligations — but in 2003, decided to issue $10 billion in bonds, paying the low rates for state-issued bonds, and fund the pension plan with the aim of obtaining more aggressive returns.  Magic, eh?  Practically free money — or, at least, it genuinely did improve the balance sheet because of this accounting method, even though no sensible person would take on debt in order to invest in stocks. 

But I can also understand, to a certain degree, why government accounting differs:  a company, at any time, can go out of business, or shrink substantially (again, think General Motors:  until the settlement, its pension obligations were massive because its workforce had shrunk so massively).  A government entity is assumed to be a more “permanent” entity, with its population, and state workers, growing or remaining stable, more or less  — so that its obligations can be dealt with on a pay-as-you-go basis and the state will always be around to meet its obligations.

Of course, we’ve now learned that this isn’t true — in the case of cities, at least, population, and sustainable levels of government workers, can shrink substantially, down to half, or even a third of their peak size (Detroit dropped from 1.85 million in 1950 to 0.7 million in 2010).   And even if it were, it violates basic principles of pension accounting, that costs be recognized in the period in which they occur.

What’s more, it’s simply wrong for a government entity of any kind to promise benefits to be paid in the future, as a debt that a successor mayor/governor/legislature has to figure out how to pay.  I’d very much like for states to get out of the pensions business entirely, and leave that to unions via Taft-Hartley multiemployer plans.  But for existing liabilities, what’s done is done, and those liabilities need to be paid off just as much as any other.