GASB Improves Public Pension Plan Accounting and Disclosures

Yesterday, the Governmental Accounting Standards Board (GASB) approved  two new standards  related to accounting and disclosure for public retirement plans. GASB 67 will change the plan level reporting and disclosure while GASB 68 will change the employer level reporting. This comes after much debate and public controversy as the media has had many a field day trumpeting the cumulative underfunding of public pension plans.

Statement 67, according to the GASB press release, enhances note disclosures and required supplementary information for public defined benefit and defined contribution plans. Among the new requirements are annual money-weighted rates of return and 10-year supplemental information schedules.

Here I warn the reader, particularly the unknowing one, about the value of the new requirement as compared to the cost of providing it. Any of these 10-year projections will be supplied by actuaries. Actuaries, as a group, are smart people. I am proud to say that I am an actuary. So, to be sure, I am not denigrating my profession here.

However, when an actuary presents various measures of plan liabilities and forecasts of future costs to a plan sponsor, be it a public sponsor or a private one, one thing we are sure of is that the numbers are not perfect. What they are is a best estimate based upon a set of actuarial assumptions and methods. Based upon our experience and training, our estimates will likely be better than yours, but the likelihood that the plan liability that we calculate will be a precise measure of the present value of benefits accrued to date is essentially nil.

So, consider what happens when, even with our level of training and experience, instead of looking one year into the future, we look ten years into the future. I don't know anyone who knows what will happen to the economy over the next ten years. The last 10 years or so illustrate this well. Suppose I had asked you on June 26, 2002 whether interest rates would increase, decrease, or stay relatively steady over the next 10 years. What would you have said? Well, in July 2002, I did ask that question to a group of roughly 100 professional asset managers. As a group, these people were pretty savvy about the economy. About half a dozen of them thought that rates would stay relatively steady over the next ten years; the remainder thought they would increase and most thought they would increase significantly.

Statement 68 makes three significant changes to accounting for public pensions:

• Projections of benefit payments will now include assumed projections of pay increases, projection of service credits, projections of automatic cost-of-living adjustments (COLAs) and projections of ad hoc COLAs if those ad hoc COLAs are nearly automatic. Over time, a meaningful number of public employers have kept their reported liabilities down by providing annual ad hoc COLAs rather than automatic ones. While they were essentially the same, the reporting for them was very different, but no more.
• Perhaps the single most key actuarial assumption used in calculating a pension liability is the discount rate. Statement 68 will allow generally well-funded public pensions (there aren't too many of them right now, but they do exist) to use an expected long-term rate of return on assets to discount the liabilities so long as the assets are invested in such a way as to reasonably expect that rate of return to be achieved. Plans that do not meet the criteria to use a long-term rate of return must use what is essentially a risk-free rate for governmental entities -- a yield or index rate on 20-year AA or higher-rated municipal bonds.
• Finally, plans are to be valued using an entry age [normal] actuarial cost method. To the extent that a plan is pay-related, this method should produce roughly a level percentage of pay annual cost. If the plan is not pay-related, the method should produce roughly a level dollar amount annual cost. To my mind, this is a significant improvement and a step that other rule-setting bodies including the United States Congress should [have taken] take a lesson from.

What will this ultimately do for public pensions? I suspect that we will see a better picture of the levels of underfunding of public pension plans, but will not unfairly punish those that have been funded responsibly. The new standards are not perfect, but in my opinion, this is a step in the right direction.

Accounting for Pension Buy-Ins

Over the last several years, so called pension buy-ins have become fairly popular in Europe. Just recently, the first one in a US qualified defined benefit plan was completed. The questions beg:

• What is a pension buy-in?
• How does the plan sponsor account for such a transaction? [NOTE: the author is not an accountant and does not provide accounting advice, but has drawn upon writings from and conversations with several large accounting firms.]

Briefly, here is how a pension buy-in works (from the standpoint of the plan sponsor). The plan purchases a contract from an insurer. When a benefit payment comes due for a participant covered by the buy-in, the plan pays that amount to the participant and the insurer pays an equal amount to the plan. Economically, this feels a lot like a buy-out contract where the plan purchases annuities from an insurer and the insurer pays all future benefits. The two contracts are probably priced the same as the insurer bears the same risk in each case. In the typical buy-in arrangement, the plan sponsor has the right to convert from a buy-in to a buy-out at no additional cost.

Buy-outs tend to generate what is know as settlement accounting under ASC 715 (previously FAS 88). In order for a transaction to be a settlement, it must satisfy a three-prong test:

1. The action must be irrevocable.
2. The action must relieve the plan sponsor of primary responsibility for the obligation being settled.
3. With respect to the plan sponsor, the event must eliminate significant risks related to the obligations and assets used to effect the transaction.

Buy-ins, as currently structured should not generate settlement accounting, despite the visually identical economics. Here is why.

• Typically, the action is not irrevocable as the contract often can be undone.
• Because the insurer could become insolvent and because the plan will own the contract, neither the plan nor the plan sponsor has eliminated significant risks with respect to the obligations and assets.

So, for accounting purposes, the plan is left with the same obligations it had before and a sizable asset of a type not previously considered in the accounting literature. We then ask how should this new-fangled asset be valued. After consideration of the thoughts of accounting professionals, we conclude that there are two options. First, the contract can be valued (essentially as an obligation), but using the rate at which such asset could be sold. The accounting profession seems to me to think that the current (at the measurement date) PBGC rates for single-employer pension annuities may be an appropriate discount rate. Second, and perhaps more intuitive to me, both the asset and the associated obligation would be valued at the discount rate at which such obligation could effectively be settled.

Under the first approach, the asset will likely exceed its associated obligation. Under the second approach, they will be identical. 

"I see", said the blind man.

So, there you have it. If my reading of the tea leaves is correct, it goes like this. A buy-out generates settlement accounting. A buy-in with the same economics does not, but it does give the plan sponsor a choice of two methods of accounting one of which will produce a smaller annual pension expense than the other.

Pension Accounting Changes Yet Again

Yesterday, IFRS announced changes to IAS 19 -- Employee Benefits. For those of you who are not familiar, IFRS is the major international accounting board and through IASB, the International Accounting Standards Board, they issue international accounting standards and related material.

According to the IFRS press release, the "IASB introduces improvements [emphasis added] to the accounting for post-employment benefits." In my opinion, this is very much a matter of opinion. Let's examine.

The most significant change that was made was eliminating deferred recognition of gains and losses. What exactly does this mean?

Most companies will measure their pension assets for IAS 19 purposes once each year, that occurring on the last day of the fiscal year. Under the new IAS 19, effective for fiscal years ending January 1, 2013 or later, actuarial gains and losses that occur during the year will be recognized in income immediately. Among the rationale for this is that this change will inhibit imprudent risk-taking.

I digress. When determining the obligations of a pension plan (IAS 19 calls this a defined benefit obligation or DBO), an actuary makes a myriad of assumptions (technically, the assumptions belong to the plan sponsor, but the sponsor typically defers to the actuary who is trained to make such assumptions) including, but not limited to, these:

• Discount rate: the rate at which those obligations could effectively be settled on the measurement date
• Salary increase rate: the rate(s) at which salaries are expected to increase
• Mortality: the likelihood of an individual dying at a particular age assuming that they live through the immediately preceding age
• Retirement: the likelihood of an individual retiring at particular age assuming that they remained in employment through the immediately preceding age
• Termination: the likelihood of an individual terminating employment prior to retirement at a particular age assuming that they remained in employment through the immediately preceding age

Those are some tricky assumptions to make. And, while actuaries are highly trained, there are a few things that I can tell you with certainty about those assumptions. First, they are reasonable; in fact, in many countries, they are required to be the actuary's best estimate. However, most actuaries would tell you that with regard to any assumption, there is probably a range within which any assumption might be considered a best estimate by some qualified and ethical actuary. Stated differently, it is very reasonable that on a particular date, I think the best estimate discount rate for a particular obligation is 5.80% and another actuary thinks the best estimate is 5.90%. This happens all the time and it is fine. Second, the assumption will not be perfect. In all my years of practice, only once can I remember a single defined benefit plan assumption being perfect for a single year. Or, as a colleague of mine once put it to a client when presenting actuarial valuation results, "We know these answers are wrong, but they are as close to being correct as any result you would get from anyone else."

Why did I tell you all this? It was to point out that these actuarial gains and losses that must be reflected in income are estimates. Sometimes, the estimates are very good, and sometimes, despite the best honest and ethical efforts of all parties involved, they turn out to be pretty bad.

Now, let's switch to the asset side. As I noted earlier, our clients usually measure the assets and obligations of a plan once each year, on the measurement date. Suppose a plan's fair value of assets on December 30 is $1 billion. And, further suppose that due to the breakout of civil war in Grand Fenwick (Grand Fenwick was a mythical country in a book called "The Mouse That Roared"), the plan's fair value of assets drops to $975 million on December 31. And, then when the markets realize over the New Year's Day break that the Grand Fenwick civil war has already ended, the plan's assets rebound to $999 million on January 2. The plan sponsor still reports assets of $975 million and the company's income statement will have suffered a $25 million hit because of this silly event.

This is part of the slow rush to mark-to-market accounting. Does it make sense? Sometimes? I can argue either side and I often have. But, in this case, for an ongoing plan, this is a long-term obligation which means that the asset pool should be used to mitigate risks associated with a long-term obligation. But, the amended IAS 19 will cause plan sponsors to treat this as a short-term obligation. This, in turn will cause more sponsors to freeze or terminate their plans which will cause the obligation to be, in fact, a short-term obligation.

Prophetic, isn't it ...

Pension and OPEB Accounting - Immediate Gain and Loss Recognition

I read this morning that Verizon is changing its accounting methods for recognition of actuarial gains and losses. Rather than amortizing gains and losses outside of a corridor over an extended period of time (I could get really technical here, but suffice it to say that since roughly 1985-1987, virtually every American company with a defined benefit pension plan has been using an amortization method similar to the one I described here), they will be expensing all of these gains and losses in the year they occur. Presumably, this is to prepare for global accounting convergence.

To me, this is part of mark-to-market hysteria, and it is just wrong. Let's make the assumption that Verizon's current intention is to keep these plans going (if they intend to terminate, that's a whole different discussion). What happens then is that changes in a hypothetical bond portfolio that Verizon doesn't even own will have a direct effect on the company's annual earnings. And, to make it worse, this earnings effect will be based on the yield on that hypothetical portfolio on a specific day. So, if the yields on that bond portfolio increase on December 31 after a decrease on December 30, and then followed by a big decrease on January 2, then Verizon gets to book extra income (or less expense). Similarly, if the assets that the plan holds rise on December 31, that rise will be reflected in that year's income.

This makes no sense to me. The obligations under the plan are long-term obligations. Short-term deviations in the valuation of those liabilities should be irrelevant for most companies. Similarly, short-term deviations in the valuation of plan assets should be irrelevant. If the plan has a short remaining life span, or if the company is in a very bad financial position, these deviations become far more relevant, but for a solid company with ongoing plans, this is, IMHO, just foolhardy.

I don't know how many people, if any, will comment on this, but I feel certain that at least some readers will disagree with me. That's ok. You are entitled to your opinion. And, you know, I wouldn't be so adamant about this if, for example, companies got to pick a discount rate for valuation of these obligations based on some sort of moving average of bond yields (even a weighted moving average), but people shouldn't have to hang on tenterhooks waiting to see what is happening with bond rates on December 31, not in my world anyway.

I guess I think it's time for the JLASB (if you can't figure it out, I just might not tell you).

Comparability in Financials, What Comparability?

A news release from SEI said that in its study of FAS 87 (now ASC 715) discount rates for 2009 fiscal years, 90% of them fell within the 161 basis point range from 5.27% to 6.88%. I wonder, where were the other 10% and how were they justified?

In the early days of FAS 87, in my experience, most companies were fairly cavalier in their choice of discount rates. Many used the same rate as they used for funding under Code Section 412. Others used their current liability interest rate. Still others seemed to use a dart board or a Ouija board.

Then the SEC intervened. They interpreted the FAS 87 guidance that companies look to the rate on high-quality fixed income instruments to mean that the rate should approximate that found on Moody's Aa bonds. A fairly typical approach was to look at Moody's Aa's a month or two before the measurement date, round up, usually to the next higher quarter of a point (but sometimes more than that) and use that rate unless there was significant change before the measurement date.

Accounting firms gradually gave this more and more scrutiny and then came Sarbanes-Oxley. Suddenly, the Big 4 (and other accounting firms generally followed) were asking companies to justify their discount rates. Discount rates became more and more tied to bonds of similar duration to the plan's obligations. Rounding up became taboo.And, all the while, the underlying discount rates on high-quality bonds were falling. Pension cost went up at the worst possible times.

So, what happened? Consulting firms came to the rescue. Since the accountants asked that discount rates be tied to specific bonds of similar duration to the obligations, actuaries developed tools to assist defined benefit plan sponsors in selecting and perhaps optimizing discount rates. As methods became more and more sophisticated, some plan sponsors went discount rate shopping. Think about it: if you were required to book the costs for a $billion pension plan and for a fee of, say, $10,000, you could buy yourself and additional 30 basis points in discount rate, would you do it?

So, while FAS 87 was supposed to improve the comparability of accounting for pension plans (and it probably did for a while), it appears that we have returned to days of less comparability. Come up with a rule and give smart people enough time and they will find the angle and find a way to optimize results.

It'snot as bad as it could  be, but comparability, we don't have no stinkin' comparability.