Announcement: SOA congratulates the new FSAs for June 2019.

Issues in selecting the discount rate for GASB 45 Valuations

Issues in selecting the discount rate for GASB 45 Valuations

On what do actuaries base interest assumption? It is this author's hope that most will encourage public sector employers to determine the GASB 45 expense using a discount rate based on the funding policy for that fiscal year.
By Kevin Binder

Public employers now find themselves in a similar situation to private sector employers in the early 1990s. In the early 1990s the FAS106 accounting standard required private sector employers to measure post-retirement medical liabilities using accrual accounting. Similarly, the GASB 45 accounting standard is now requiring public sector employers to measure and disclose their liabilities on an accrual basis for the first time.

Since it is a new standard, there are areas where the intent of the standard could be interpreted in more than one way and the prevalent interpretation of the standard by the actuarial community has not been made. I would like to focus on what is to me the most interesting aspect of the new standard, the discount rate selection. In doing so, I would also like to particularly focus on what I believe will be a common situation for the next few years. Finally, I am concerned that the discount rate that will tend to be selected in this situation may be overly aggressive and ultimately reflect badly on the profession.

In the good old days, when actuaries were responsible for the interest rate used to determine minimum funding for pension plans, many actuaries based their interest rate assumption on the building block approach. Using this approach, the interest rate would depend upon the plan's investment allocation. For example, suppose that equities have historically earned 6 percent more than inflation, and bonds have historically earned 2 percent more than inflation. If we assume 4 percent inflation, and an investment allocation of 60 percent equities, 30 percent bonds and 10 percent cash, then the expected rate of return is 4 percent plus 6 percent times 60 percent plus 2 percent times 30 percent, or 8.2 percent. If we assume investment expenses reduce the investment return by 50 basis points, then using this approach, we would assume a 7.7 percent interest rate. If we increased our equity allocation to 70 percent, then the interest rate assumption would be 8.1 percent. This method of determining the investment return assumption pre-dates recent articles comparing actuarial practice and financial economics.

The authors of the GASB 45 standard embraced this concept. Except that like the private sector (and unlike pensions) many governments have not established a trust to pre-fund post-retirement medical benefits. So, if there is no OPEB trust, what is the source of assets used to pay for these benefits? The GASB 45 standard answer is whatever financial instrument the government uses to invest its general funds (typically money market like accounts). So, if there is no OPEB trust, the discount rate selected should be something like the anticipated long–term rate of return assumption for money market accounts, typically about 4 percent is being assumed. However, supposing the local government creates a trust to pay for post–retirement medical benefits, then GASB 45 allows for a discount rate that is similar to the interest rate assumed for a pension plan. The median interest rate assumption for public sector pension plans is currently 8 percent.1

Here is how paragraph 13 of the GASB 45 standard describes the discount rate selection (italics added).

"The investment return assumption (discount rate) should be the estimated long term investment yield on the investments that are expected to be used to finance the payment of benefits. ... For this purpose, the investments expected to be used to finance the payment of benefits are (1) plan assets for plans which the employer's funding policy is to contribute an amount at least equal to the ARC, (2) assets of the employer for plans that have no plan assets or (3) a combination of the two for plans that are being partially funded. The discount rate for a partially funded plan should be a blended rate that reflects the proportionate amounts of plan and employer assets that are expected to be used."

For the readers who are not familiar with public sector accounting terms, the ARC or Annual Required Contribution is the GASB term for the amount that needs to be reported and is equivalent to the net periodic benefit cost under FAS 106.

SSo, GASB 45 has created an incentive to pre–fund retiree medical benefits by establishing a two–tiered system for determining discount rates. Governments that are funding the ARC will typically use a discount rate similar to that assumed for public sector pension plans. Governments that are only making pay–as–you–go medical payments will have to assume a much smaller discount rate based on the anticipated long–term yields that money market type funds are expected to earn.Before I continue, a few points need to be made.

First, in order to use the higher discount rate assumption, an employer is not initially required to attain a substantial level of funding compared to plan liabilities. The table above shows a five–year GASB 45 projection for an employer that is fully funding the ARC. After four years the plan is 25 percent funded.

Thus, public employers are not required to issue an OPEB bond to fund the entire (or close to) accrued liability to use the higher interest rate.

Second, the discount rate to be used depends upon the government's funding policy, not the actual level of funding.

Third, GASB 45 also has substantial disclosure requirements, including disclosure of the liabilities, the assumptions, the funding progress and the Net OPEB Obligation (NOO) which is analogous to the accrued expense under FAS 106.

Here is a common situation:

  • Jurisdictions are seeing ARC amounts that are two to four times their current pay–go costs.
  • The unfunded expense may be close to double the funded expense.
  • Jurisdictions are unable to raise the cash to fund the ARC within their current budget, and are unwilling (or unable) to slash benefits to a level that could be funded, and also unable to immediately raise taxes to pay for funding the OPEB cost on an accrual basis.
  • Since governments are not–for–profit organizations, the only real impact for most jurisdictions are possible rating cuts with an increase in the cost of borrowing. That said, many elected officials will be nervous about possible credit rating downgrades occurring during their term of office.
  • The rating agencies have informally said that they will not slash bond ratings as long as governments have a plan to eventually reach (five years for AAA jurisdictions, more for jurisdictions with lower bond ratings) the funded expense.

Given these constraints, many public sector employers are looking at a multi–year plan to gradually reach the funded expense. For example, the State of Maine recently announced that it is adopting a funding policy to gradually increase funding to the ARC over 10 years. These employers can manage to find the additional funds (or make some modest benefit adjustments) to get through the first year of the GASB 45 standard. They will worry about the more substantial future cash requirements or more substantial benefit reductions later; in the meantime, the commitment to a multi–year plan will probably keep the jurisdictions bond rating unchanged.

So here is the gray area, supposing the government plans on increasing its OPEB funding over a phase–in period to the funded GASB 45 expense, what discount rate should be assumed in the phase–in period? Note that the GASB 45 standard is very clear that the discount rate is based on the funding policy. Let's say the unfunded discount rate is 4 percent, the funded discount rate is 8 percent, and the government elects to gradually increase funding over a 10–year period to the funded ARC. One way of determining the blended discount rate is to look at just the current year's funding policy. Here is an example of how the blended discount rate might be determined during the first year of the funding policy.

  1. ARC based on 4% discount rate $180.00
  2. Pay as you go cost (PAYG) $ 40.00
  3. Arc based on 8% discount rate $100.00
  4. Gap between PAYG and ARC on 8.0% $ 60.00
  5. Proposed funding $ 46.00
  6. Additional funding (above PAYG) $ 6.00
  7. Percent of GAP funded (6. / 4.) 10.00%
  8. Difference in discount rates (8%–4%) 4.00%
  9. Increase in discount rate (7. x 8.) 0.40%
  10. Blended discount rate (4% + 9.) 4.40%

Using this approach, if the government kept the 10–year phase–in schedule, in the second year the discount rate used would be 4.8 percent and so on.

However, an alternative view is to look at the long–term funding policy. The government is planning on gradually increased funding to the trust over the phase–in period, but after the phase–in period the plan is to contribute the funded expense. If the plan is realized, then perhaps we could consider all of the foreseeable projected contributions under the funding policy, and compare them to the funded expense over the same time period. How many years should we consider?

For example, suppose the phase–in period is five years, the fully funded discount rate is 8 percent and the unfunded discount rate is 4 percent. While it is true that in the phase–in period the funding policy will lead to additional contributions to the trust that are approximately half of the additional contributions needed to equal the funded expense, if the next 15 years are also considered (20 years in total), then for 75 percent of the years considered, government will be contributing the funded expense. So, for the first 20 years the government will be contributing about 87.5 percent of the additional contributions needed to reach the funded expense level which would result in a blended discount rate of about 7.50 percent being selected. If we look at the next 45 years (50 years in total), for 90 percent of the years considered, the government will be contributing the funded expense. So, for the first 50 years of the standard the government will be contributing about 95 percent of the additional contributions needed to reach the funded expense level, which would result in a blended discount rate of about 7.80 percent. Or if the government commits to repaying the shortfall after the phase–in period, over a long time–period (e.g., years six to 20), can we assume the funded discount rate of 8.00 percent for the entire 20–year period?

There are many ways to present the second approach. But they all end up with a discount rate close to the funded discount rate by averaging a long time–period using the funded discount rate with a shorter time–period using the unfunded discount rate.

I am very uncomfortable with the long–term approach for the following reasons:

  1. The approach is back–loaded; it is relying on future behavior over a long period of time that will require a high degree of sacrifice, either from taxpayers in the form of higher taxes, public sector retirees in the form of less generous benefits or users of local government services in the form of less government services.
  2. What ever plans are made now could be reversed by a future administration.
  3. Future costs could increase above and beyond the current costs. Most public sector employers are assuming medical trend increases similar to those used for FAS 106 valuations. The private sector experience is that these assumptions have been overly optimistic. If that experience is repeated in the public sector, costs at the end of the phase–in period could well be considerably higher than currently anticipated.

All of my reservations stem from the same concern. Let's see a track record of contributing the funded expense into a trust before we rely on this assumption for setting the discount rate. This is especially so if the policy is only verbal. I have slightly less reservations if the policy is in writing and publicly available. But I even have some reservations if the policy to gradually increase funding to the funded expense is turned into a statute. Statutes can be repealed or amended!

What will happen if the actuary used a high discount rate close to the funded rate, based on a long–term funding plan and some time during the phase–in period the government finds it is unable to contribute the funded expense. Presumably the discount rate will have to be reduced. When the GASB 45 expense rises due to the lower discount rate what will happen?

I have two concerns:

First, the actuary may be blamed for the incorrect discount rate selection, therefore reflecting badly on the profession. In its basis for its conclusions of the GASB 45 standard, paragraph 123 had this to say about the blended discount rate. "Additional research indicated that actuaries would be able to develop an estimate of a blended discount rate based on the expected long–term rates of return on plan and employer investments, as required for partially funded plan." The point being that clearly the accounting profession believes that actuaries will be heavily involved in setting the partially funded discount rate. We certainly could be blamed if they do not work out as planned.

My second concern should be more important. What about the plan participants? By using overly optimistic assumptions we may delay the date that plan sponsors realize that their current programs are unaffordable in the long term. Typically, benefit reductions for those already retired are less substantial than the benefit reductions for current employees. So when the plan sponsors realize that benefit reductions must be made, the reductions to current employee benefits could be more substantial than would have been necessary if the problem had been faced earlier. This is basically, what I have observed happened after the FAS 106 standard was adopted. The early valuations proved overly optimistic, expense rapidly increased, the result was often a series of benefit reductions as the true cost of these benefits became known. The result for many companies is that there is a group of grandfathered participants with generous OPEB benefits and many, if not all, employees with no OPEB benefits at all.

As actuaries we are frequently pressured to use more aggressive assumptions than we might initially use. However, my experience is that over the long run we are better off as a profession if we use realistic assumptions that are borne out over time than if we yield to these pressures. Even if in the short term the organizations we consult for would prefer a different answer.

My hope is that most actuaries will encourage public sector employers to determine the GASB 45 expense using a discount rate based on the current funding policy. If we are pressured to base our interest assumption on a planned long–term policy, we should at a minimum document our objections very clearly and make it known that the actuary did not select the discount rate.

The author would like to thank Adam Reese who kindly agreed to review the article and who made many helpful suggestions and comments.

Kevin Binder, FSA, MAAA, EA, is an actuary with Bolton Partners.

Footnotes:
1 2006 Wilshire Report of City and County Retirement Systems