HEADNOTE: As with all farreaching changes, the proposal to accrue these costs is certain to continue as a controversial matter, particularly with public hearings to be held soon. The author, who has been involved with this topic from its beginning, describes more significant portions of the proposed accrual process and offers practical alternatives. Participants in the discussions of this project will want to consider these items carefully before the hearings begin. There are some things you just don't do. With apologies to the late singer Jim Croce, you don't tug on Superman's cape, you don't spit into the wind, and you don't mess around with accrual accounting. That explains, at least in part, why business executives and accountants seem reluctant to oppose the FASB's proposal to require accrual accounting for nonpension retirement benefits. They are prepared to argue the details of how to accrue the cost. But they are not prepared to argue the merits of cost accrual. Most think it is right; and most of the rest think it is inevitable.

Lynch the FASB

Unless you have spent the past six months Spelunking beneath Kentucky, you know the broad outlines of the proposals. Starting in 1992, companies would accrue the cost of all nonpension retirement benefits, principally health care benefits, over some part of the working lives of employees expected to receive the benefits. Starting in 1997, if the accrued liability has not reached a minimum amount, companies would record the shortfall as an added liability.

A field test of the proposals has produced some scary numbers. For a mature company, with many workers already retired, the change to accrual accounting would increase the cost from three to six times. For a young company, the increase would be more like 30 times.

Even worse, current tax laws allow companies to deduct only the pay- as-you-go cost. While the larger accrual cost will some day be deductible, SFAS 96 on income tax accounting sharply limits the future deductions that a company can count in figuring its current tax expense. So for many companies--probably most--any cost increase would drop straight to the bottom line with little or no immediate tax benefit. Balance sheets, too, would take a beating. For some, the new liability would be their biggest. For all U.S. companies, the total liability would be hundreds of billions of dollars. Little wonder a Business Week story on the proposals began with the headline, "First Thing We Do Is Kill All The Accountants."

But while business executives might cheerfully lynch the FASB, most people are not prepared to argue against accrual accounting. They wouldn't succeed if they tried. Accrual accounting is the tribal god of the accounting profession, it's the equivalent of motherhood, apple pie, and the American flag. The FASB's vote on cost accrual was unanimous. The two members who dissented to the whole statement objected to details, not to accrual accounting. So, despite Damon Runyon's advice never to offer more than three to one on events involving human beings, the odds on retaining pay-as-you-go accounting have to be at least 100 to one against. On this issue, the FASB believes it is on the side of the angels. And few in business or the accounting profession seem inclined to argue the point.

That does not mean the October and November public hearings on the proposals will be an exchange of pleasantries. To the contrary, the hearings are likely to be as fractious as any the FASB has held. The Board will get an earful of objections and suggestions for change. And there is a good chance the Board will agree to some of them and issue a standard that differs in important ways from the proposals.

"You're Gonna Love It"

Of all the objections, the one most likely to dominate the hearings focuses on how the cost is computed. Many of the scary numbers coming from the field test resulted not so much from accruing the cost as they did from the particular way the FASB wants the accrual computed. If for no other reason than to ease the financial statement impact of the proposals, companies are sure to seek changes in the computation routine.

As with many of the proposals, those for computing cost were brought forward largely intact from SFAS 87 on pension accounting. If you like pension accounting, the word is, "you're gonna love" the accounting for other retirement benefits.

Specifically, companies would project the amount and timing of benefit payments, discount the projected payments, and spread the discounted amount over years of employee service. The amount assigned to a year would then be adjusted for interest on the benefit obligation, for earnings on plan assets (if any), and for amortization of prior service cost, gains and losses, and a transition asset or obligation. The result would be a benefit cost arrived at by adding six separately computed components, a routine familiar to anyone who has computed pension cost under SFAS 87.

For most companies, the big dollars, and the big headaches, are in retiree health care benefits. Here, the FASB says, the projection of benefit payments is to include an assumption as to the future trend of health care cost, considering four things:

1. Future health care cost inflation. How much longer will growth in the cost of health care continue to outpace general inflation?

2. Technological advances. What new machines and procedures will scientists devise? When, if ever, will technological advances reduce the cost of health care?

3. Changes in utilization and delivery patterns. How many more people will elect triple bypass surgery? How many more people will decide to use HMOs?

4. Changes in the health status of plan participants. How much sicker will employees be 30 years from now?

While this list looks comprehensive, not everything that will affect future health care costs is to be considered. In particular, companies are not to consider future changes in the terms of their health plans or in government cost-sharing arrangements, such as Medicare. After benefit payments are projected, they are to be discounted using a current market "settlement" rate, that is, the rate at which a company could settle its benefit obligation currently. Recognizing that health care benefits can rarely be settled, the Board would let the discount rate be based on current interest rates for long-term, fixed income securities.

After the projected payments are discounted, the discounted amount would be spread over years of employee service according to the terms of the plan. If the plan is silent (most health care plans are) an equal amount would be assigned for each year of service, starting with the date an employee is hired and ending with the date he or she is fully eligible for benefits. If the date of full eligibility precedes the expected retirement date, this proposal would spread the cost of benefits over a period shorter than an employee's full service time. Each proposal offers much that will be criticized. Let's start with the trend of health care cost.

Unreliable Numbers, Inconsistent Accounting

Actuaries, whose happy job it would be to make the projections, have frankly expressed doubts about their ability to do it with reasonable reliability. Two years ago, a panel of actuaries warned that "the measurement of retiree medical benefit obligations is a young art striving to be a mature science...Study and research in this area have just begun...It has only been in the last five years that the subject of retiree medical programs has received a lot of attention from the actuarial community." The actuaries expressed particular concern that "existing data...used as the basis of extrapolating health care trends may or may not be appropriate for the longer term. Existing statistical data has almost never been produced with the long-term future in mind-- most frequently the data would be used for the short-term pricing of one-year health care coverage."

Beyond that, the actuaries said, "the rate of increase in health care cost is still a matter of some controversy." Actuaries do not agree about "how the trend of claim costs in the future will proceed," and they know little about the "sensitivities of assumptions when projecting health care cash flows-over many decades." As a result, the panel said, actuaries are forced to rely on "a substantial amount of subjective judgement...Two actuaries valuing the same plan...are likely to differ substantially" in their projections.

Yet, even a small difference in the cost trend rate can make a big difference. Using a typical plan, one study found that a one-percentage- point change in the trend rate changed the cost by 10% to 40% depending on the portion of the covered group that is currently retired. Differences of that size reduce reliability to levels below acceptability.

Compounding the problem of unreliability is that of inconsistency. The FASB would include changes in health care costs in the projection, but would exclude changes in plan terms and in government cost-sharing arrangements. The changes are interrelated. Changes in health care costs trigger changes in plan terms and in government cost-sharing arrangements. And, as the history of Medicare illustrates, changes in government cost-sharing arrangements trigger changes in health care costs. Projections, the FASB is sure to be told, should include all of the relevant future, or none of it. A cost computation based on a projection that includes future changes in health care costs but excludes future changes in plan terms and in government cost-sharing arrangements would be inconsistent and unbalanced.

Few are likely to suggest that companies project changes in retiree health care plans and in government cost-sharing arrangements. The unreliability problem is bad enough already. Nor is there any reason for companies to project such changes. They are events of the future; they can be recognized when they occur.

This is equally true of future changes in health care costs due to health care cost inflation, technological advances, changes in utilization and delivery patterns, and changes in participants' health status. These, too, are future events that can be recognized when they occur.

Thus, many will suggest an alternative computation, one that does not give up the advantages of accrual accounting, but at the same time, does not require unreliable projections or inconsistent treatment of related events. The alternative would simply exclude projections of the health care cost trend rate from the computation, or at least project only general price level inflation.

Arbitrary Exercise

The proposal to compute the cost as the sum of six separate components is also certain to draw fire. Critics will object that this is an inappropriate extension of pension accounting and far too complicated.

As applied to pension plans, the computation makes sense. The typical defined benefit pension plan promises benefits in the form of fixed or determinable dollar payments, which vary with employee service--the longer the service, the greater the benefits. Since each year of service earns additional benefits, it is not illogical to begin the cost computation by computing each year's service cost as the present value of the additional benefits earned during the year. Such a cost can be fairly described as a measure of the financial effect of an event, the earning of benefits during the year. The cost could then be adjusted for interest on the benefit obligation, for earnings on plan assets, etc. While the computation is complicated, at least it flows logically from an attempt to measure the financial effects of an actual event.

But the logic breaks down when the computation is applied to the typical nonpension retirement benefit plan, particularly retiree health care plans. Such a plan promises benefits in the form of services to be purchased when needed at prevailing prices. The length of employee service has no effect on the benefits the employee will receive. That being the case, computing each year's service cost as the present value of the added benefits earned by the year's service has no basis in fact. Service during the year does not earn added benefits.

The FASB attempts to overcome this awkward fact by assuming that each year's service earns a pro rata share of the total benefit. But that simply reduces the computation to an arbitrary exercise; it cannot be described as measuring the financial effects of an event. There is no event; there are no financial effects to measure. There is only the arbitrary allocation of a cost.

That, by itself, is not objectionable. Sometimes arbitrary allocation is the best that can be done. But when it is best, complexity cannot be justified. There is no reason to use a complex computation of benefit cost as the sum of six separately computed components instead of a simple computation. The complexity would add neither relevance nor reliability, only time and expense.

Worse, the accumulation of a cost by separate components, each with its own catalogue of detailed computational rules, would suggest a precision in the result that would be misleading. Keep in mind, computing the benefit cost as proposed would call for heroic assumptions about the level of health care costs in future decades, with the result discounted at a rate that could fall anywhere in a range of 400 basis points. And the discounted amount would be spread over a period that may or may not bear a reasonable relationship to the period benefited. Anything suggesting precision in such an exercise is to be shunned.

While the FASB might appropriately switch to any one of several simple methods, one that will be suggested is the so-called aggregate method. That method (see Exhibit 1) would spread the cost as a percentage of payroll. As with the FASB's proposed method, it would allocate the cost arbitrarily. But it would do so openly, avoiding any appearance of measuring the financial results of an event. More important, the method would be simple, and it would avoid the misleading appearance of precision.

The cost computation could be further simplified by requiring plan valuations only at three-year intervals. The small sacrifice of relevance would be more than offset by the reduced effort and expense.

Problems, Problems, Problems

Other criticism of the proposals will focus on the current market "settlement" rate for discounting purposes. Of what relevance is a settlement rate without a settlement vehicle? Why should the benefit cost and obligation bob up and down with every ripple in the money market? Even if the FASB insists on the greater relevance of a current market rate, why not at least let it be a company's own internal cost of capital rate? Wouldn't that rate be more relevant to the company's particular circumstances?

And why, critics will ask, should the cost be spread over less than an employee's full service life? Granted, at the date an employee is eligible for benefits, an obligations exists; the employee could quit tomorrow, and the company would owe the promised benefits. But not many eligible employees will quit tomorrow. Allocating the cost on the assumption that they will quit not only fails to charge all years worked with a share of the cost, but also misstates the company's economic obligation.

Critics are also certain to attack the minimum liability reporting proposal. However, this proposal sounds worse than it really is. A company would be required to report a liability at least equal to the discounted value of all benefit payments expected to be paid to employees who are already retired, and to active employees fully eligible for benefits. But the requirement would not take effect until five years after a company begins to accrue the cost of benefits for retirees and all active employees, including those not yet fully eligible. For many companies, five years' accrual would build up a liability nearly equal to--in some cases, greater than--the required minimum. For those companies, the proposed minimum liability reporting requirement would have little or no effect.

Nevertheless, it is sure to attract criticism, and not just from the companies that would be affected. Some critics will object in principle to a provision that treats one piece of the benefit obligation as more important than other pieces. Some will object to the added complexity of a proposal that would, in most cases, have only a temporary effect. And still others will object to the intangible asset offsetting the adjustment needed to raise the liability to the minimum. As with the similar intangible asset required by SFAS 87, they believe the only interest investors would have in it would be to make sure it is eliminated from their analyses.

As with many FASB statements, this one would contribute to the "Bloated Footnote Syndrome" (BFS), a malady characterized by more swelling than substance. Among the proposed footnote additions sure to be labeled BFS are these:

* A breakdown of benefit cost into its components. As noted, this would add an appearance of precision contrary to fact.

* A reconciliation of the plan's funded status to the balance sheet liability. The FASB enjoys complex schedules. It thinks others do, too, but it is wrong.

* A measure of what purports to be the vested benefit obligation--in this case, a "walkaway" liability based on the assumption that all employees entitled to benefits will quit tomorrow. Critics will suggest that there is no reason to doubt the continued existence of American business.

Another proposal certain to attract criticism would supersede Technical Bulletin 87-1, which permits companies to convert from pay-as- you-go accounting to accrual accounting for nonpension retirement benefits in either of two ways:

* By spreading the effect of the change over future years; or

* By charging the cumulative effect of the change to expense immediately.

Under the proposed standard, the second method would no longer be permitted. Instead, companies would compute a transition obligation (or, rarely, a transition asset) equal to the unfunded cost attributed to prior years, then amortize the obligation during future years. As a result, future years would bear not only their own cost, but prior years' costs as well.

Critics are sure to point out that avoiding this dual cost was one of the reasons the APB Opinion 20 concluded that "most changes in accounting should be recognized by including the cumulative effect...of changing to a new accounting principle in net income of the period of the change." Admittedly, not every company would want to use the "cumulative-catch-up" method. It produces a one-time hit to earnings and equity that some companies simply could not stand. They will want the "prospective" method retained. But companies that can stand the one-time hit may urge the "cumulative catch-up" method, too. It would not only relieve future operations of prior cost, but would immediately convert both the balance sheet and the income statement to what they would have been had accrual accounting been followed in the past.

Other critics may suggest a more modest change: that the Board stretch out the amortization period for the transition obligation (asset). Here, too, the proposed standard follows SFAS 87; the amortization period would generally be the average remaining service life of active participants, but need not be less than 15 years. Since the transition obligation for nonpension retirement benefits is likely to be larger than it was for pension benefits, the FASB will be urged to lengthen the amortization period to, say, 20 or 30 years.

Odds on Change

At this point, a prudent person would stop. Not being noted for prudence, I will stick my neck out and make book on the likelihood that the FASB will adopt the recommended changes. (see Exhibit 2.)

While there is virtually no chance the FASB will back away from requiring accrual accounting, the odds are at least even that it can be persuaded to eliminate the requirement to project the rate of inflation in health care costs. While such a projection would produce a more relevant cost measure, at least in theory, the correspondingly greater measurement imprecision clearly troubles the Board. It may be willing to sacrifice theoretical relevance for measurement reliability.

The odds are equally good that the FASB can be persuaded to reduce the complexity in cost computation. Many complexities were carried over from pension accounting, and the Board may decide they have less relevance here. Since that would reduce the complexity of the footnote disclosure requirements, the odds of that are also good.

There is less chance, however, that the Board will permit discounting at something other than a current market rate. The FASB clearly believes that current measurements are more relevant than measurements based on long-term expectations. It may downplay the use of a "settlement" rate, since few settlement vehicles exist, but it is likely to require some form of current market rate.

The odds are much better that the Board will lengthen the period over which the benefit cost will be spread to full service life. The two dissenting members favor it, and the Board itself flip-flopped on the issue. Besides, this would be the kind of compromise the FASB likes to make in order to show that it listens.

The odds of eliminating minimum liability reporting are not good. While two Board members favor eliminating it, the others believe the five-year-delayed effective date takes care of most practical problems. So the odds (not coincidentally) are five to two against.

At least as poor is the likelihood that the FASB will permit both the "cumulative-catch-up" and the "prospective" method of converting to the new standard. The FASB does not like alternatives.

However, a longer period for amortizing the transition obligation (or asset) is a good bet. The choice of 15 years, while paralleling the requirement of SFAS 87, is arbitrary. Here, too, is a concession to practicality without a sacrifice of principle.

Practicality Without Sacrificing Principle

That, in fact, would be a good theme to adopt at the public hearings: concessions to practicality without sacrificing principle. Those who foresee problems if the proposals become a standard are more likely to be heard if they seek changes that do not violate principles the FASB holds sacred.

Accrual accounting for the benefit cost is one such principle. The choice of a technique for computing the accrual cost is not. Recording a liability and eliminating accounting alternatives are matters of principle. The length of the attribution period or the period for amortizing the transition obligation is not.

Yet changes that do not require the FASB to violate its principles can have important effects. Using the aggregate method, one company estimated, would cut its first year health care cost by more than 30%. That cut is probably bigger than most companies could expect. But reductions around 20% should be common. Eliminating the medical care inflation assumption could bring about even larger reductions. The effect of a longer attribution period would be smaller and harder to predict, but cost reductions of 5% have been estimated. By lengthening the period for amortizing, the transition obligation could cut cost by 10%. Cost reductions aside, there are sound accounting reasons for urging changes. Admittedly, accruing the cost of nonpension retirement benefits may be a good thing. But the Board's accrual proposals seem to have been designed according to Liberace's maxim: "too much of a good thing is simply wonderful." Exhibit 1 and 2 Omitted

(*)See Also: "Accruing the Cost of Other Postemployment Benefits--The Measurement Problem," by Dale L. Gerboth, The CPA Journal, November 1988.

Dale L. Gerboth, CPA, is a Partner in the National Office of Ernst & Young in New York. He is a member of the New York and Massachusetts Societies of CPAs. Mr. Gerboth was a member of the AICPA Task Force on Conceptual Framework for Accounting and Reporting, Committee on Employee Benefit Plans, Committee on Relations with Actuaries; the FASB Task Force on Accounting for Other Postemployment Benefits, and the Pension Implementation Group.