I agree, with two caveats.

1. Political problem: people leap from saying the standard assumption is too high to saying the proper valuation is treasury rates or some other "riskless" method. That is a political statement designed to stoke a crisis mentality, one that demonizes public employees and of course any semblance of a union - as if a union were needed to increase public pensions shared by legislators! Actual discussion of a range of valuation disappears into the rhetoric of horror.

2. Times change. While it is survivorship, we also don't know the future and the people who do the valuations understand two additional points: that shortfalls are more a product of intentional underfunding - e.g., Illinois - and that adjustments which take effect over an extended period of time, meaning decades, will suffice unless underfunding is truly hideous or they handed all the money to a Bernie Madoff.

You, too, are an optimist. The current forward-looking yield of the S&P 500, according to Standard & Poor's, was 1.83% as of the close of business on last Thursday.

America is a country where investors care far, far more about market values than anything else. Few people care about dividends. Few people even think about forward-looking prospective total returns.

So at a most optimistic set of variables, most pensions still need to nearly double their return on equities to get close to the valuations they carry on their books? Lovely. I think this qualifies as a good definition of "crisis", regardless of jomiku's optimism.

In reality, pension funds' future liabilities are being discounted far too much, understating what they need as reserves by a huge amount. Thus, the assumption of jomiku that the underfunding can be made up for by small adjustments over a long period of time is untenable -- if the rate of return is truly this dismal, then small increases in pension reserves will not keep pace with the ever-increasing shortfall (remember, pension outflows increase over time as the overall population increases, the number of government workers increases, people live longer and the population of pensioners increases). Unless you properly account for an accurate future liability amount, you cannot calculate the correct amount of the underfunding, and cannot accurately budget for eliminating the shortfall.

My guess is that the true value of the shortfalls are so enormous, that pensions will not be able to get to a proper level of funding without drastic increases in state budgets (and most states cannot afford this). Small changes over time are not going to cut it.

an underfunded pension(by a public union) is a political decision to hide the costs from the taxpayers only to slam them with the constitutionally protected bill later.

Equity returns vary a lot depending on the starting year, of course.

Some nice charts for considering this are here: http://www.crestmontresearch.com/stock-matrix-options

The particular chart looking at real stock market returns with reinvested dividends is http://www.crestmontresearch.com/pdfs/Stock-Matrix-Taxpayer-Real1-11x17.pdf

Long term returns of 3 or 4% are very common. Returns of > 9% are rarer than negative returns. Given that the 20th was a good century, I don't think expecting a real return higher than about 4% is reasonable.

Buttonwood, you were using nominal return--- aren't these liabilities inflation indexed? If not there is one ray of hope.

@Joe In Morgantown

Even those charts will overestimate the returns you'll get. Portfolios need to be rebalanced periodically, incurring additional transaction costs. And more importantly, using the arithmetic mean instead of the geometric one makes a big difference, especially for shorter time-periods. Even in the U.S. the equity risk premium is tiny, and certainly nowhere near the 6% "rule of thumb" that many still use.

Given how high Price:Trailing 10 years earnings are now, any positive real return expectation for the next 20 years is optimistic.

What confuses me the most is that if equities return around 8% (or even 6%) every year on average, while the GDP increases by only 2-3%, then we see a long-term redistribution of all asset values into equities. This can't possibly last forever, and the long-term return on equities will by necessity be constrained by the risk-free rate. Although, yes, in the long run, we're all dead, so it's perfectly possible that such a value redistribution could take place throughout a person's lifetime but not on longer time scales, I would be leery of any investment strategy that guarantees (or, even worse, operates on the assumption of) a return greater than the long-term NGDP growth rate, discounted by population growth. If you can, it had better be arbitrage. Otherwise, it's likely just plain false.

Trying to learn more - A dividend yield that exceeds GDP growth is no problem--distributed earnings are a component within GDP. Of course GDP is a flawed measure full of double-counting, but that's for another day.

It would be--as you say impossible in the longer term--for the growth in dividend yields to outpace GDP expansion. In thory, we might see the dividend growth rate exceed that of GDP for awhile, if profits from offshore activities were substantial for the index concerned and rising more rapidly than the home economy (or alternatively falling more slowly).

How could this be a truism? "It is a truism to say that equity returns come from three sources; the dividend yield, dividend growth and the change in the price/dividend ratio. "

What about equities that have no dividend? Does one assume earnings as the surrogate dividend? Or is there a humongous dividend paid when the company winds up 20 years out?

First, I'd like to thank Buttonwood for a very engaging column in the January 29 paper. A few thoughts:

One major goal of funding pensions is to tie the obligation to pay for a pension to the same cohort of ratepayers, customers, etc, who enjoy the services of the employee at the time the employee's pension service is being credited. So on a "micro" level, it makes sense to fund pensions as a matter of proper cost allocation.

On the other hand, if all financial assets are just claims on future incomes or revenues, as the Jan 29 column notes, wouldn't it be just as well on a "macro" level to say "why bother?" funding pensions, because the next generation is just going to have to pay for my retirement one way or another?

Perhaps the balancing point in this micro-macro conundrum is that saving now for future pensions increases the current stock of invested surplus. That in turn lowers the cost of capital and borrowing, which hopefully leads to higher rates of economic growth. It is however, easier to imagine this effect working if the investments support capital purchases (be they public or private) and not operating costs.

On the valuation of equities discussed in the blog above (which is not my area of expertise), it seems that retained earnings are lost in Buttonwood's dividend-driven valuation model. I won't say that retained earnings deserve the same respect as cash dividends, but a complete lack of consideration seems unwarranted. Anyone care to weigh in on this?

Here are some online sources. Not great, but not bad. Anyone care to share better links?
http://www.flatworldknowledge.com/node/29251#web-29255
http://www.capital-flow-analysis.com/investment-theory/earnings-per-shar...

Turning to public pension funds' discount rates, my sense of the field is that investment professionals' prospective expectations for a 70/30 fund are more in the range of 7.0% to 7.5%. This assumes inflation in the 2.5% range. So there is some professional pressure downwards on public pension plans using 8.0%.

Another piece of pressure is coming from the Governmental Standards Accounting Board, which is considering a change to discount rates used for financial accounting (as opposed to discount rates used for purposes of determining cash contributions to pension funds. The proposed change is a composite discount rate, using the expected return on investments (say 8.0%) to the extent the liabilities are funded by invested assets in a trust, and a risk-free rate (4.0%, for example) to the extent that the trust fund is expected to run out of money prior to satisfaction of all liabilities. An interesting compromise.

Mind you, the 8.0% discount rates are also typically applied to a definition of liabilities which factors in future salary and the effect of aging on pension costs: the "entry age normal" method, which is designed to address the cost allocation considerations described above. "Market value" liability measurements using risk-free rates typically consider only benefits earned to date. Not only the discount rate, but also the actuarial cost method matters.

One point of these comments being (I hope not to endlessly repeat earlier comments) that there is a bit more to the story than the headline grabbing (seeking?) claims of those who discount pension liabilities with "risk-free rates", or the "all is well" claims of those who stand to benefit from unrealistic actuarial assumptions. It is a frustrating experience to work on the inside of a profession which gathers occasional but heated attention. There is an awful lot of good academic and professional literature to be absorbed out there. And it is a lot easier to want a conclusion and then seek supportive facts than it is to become an expert and understand all sides of the issue.

Finally, it occurs to me that those who support the risk-free valuation basis for pensions might consider using market annuity purchase rates instead of gilts. After all, if an independent for-profit company will take it off your hands for a certain price, isn't that the very definition of a market value of liabilities? I'm surprised this approach hasn't gotten more play.

Incidentally, I believe insurers face similar challenges in valuing annuities - what mix of assets to invest in, and what discount rate to assume? This subject is not my area, but I would imagine that there is ample reason (regulatory valuation rules aside) to reflect the expected risk premium of invested assets when discounting liabilities. This leads to a similar issue as with pensions - investing in riskier assets leads to lower liability prices, and gaming the assumptions results in more ducats in the pocket today. Is this a flawed model, or a model which requires discipline and regulation of its practitioners? I will continue to argue for the latter.

One other bit of pessimism, which I didn't notice being covered: most public entities have the same "feature" in their pension future as other American organizations. Which is that, in the near future they will see a large cohort of Baby Boomers retiring and starting to draw pensions.

This has the unfortunate downside of substantially reducing the time in which small adjustments to funding can be made and still resolve the current shortfall. Actually, the common practice of allowing retirement at a younger age than non-public entities means that that big cohort is going to appear even sooner than elsewhere.

PensionActuaryPDX:
"On the valuation of equities discussed in the blog above (which is not my area of expertise), it seems that retained earnings are lost in Buttonwood's dividend-driven valuation model. I won't say that retained earnings deserve the same respect as cash dividends, but a complete lack of consideration seems unwarranted. Anyone care to weigh in on this?"

Modigliani and Miller's Dividend Irrelevance theorem?

When firms do their capital budgeting 'cost of equity(k)' is all that matters irrespective of how much retained earnings they have. Retained earnings is just another shade of equity capital and market will demand the same return on it.

Buttonwood sez: "Let's add in a further 1% for buy-backs."

Over longer periods, share dilution (new shares) have reduced returns by 1.5% to 2% per year. So shouldn't you be subtracting at least 1% rather than adding 1%?

http://www.efficientfrontier.com/ef/702/2percent.htm
http://www.efficientfrontier.com/ef/102/pie.htm

Why do stock and bond index predictions have anything do with actuarial return assumptions?
http://hedgefund.blogspot.com
Invest in skilled strategies not assets. Use 10% as the long term discount rate.
Don't take ANY beta risk. Long only alpha.