CMC Response to Proposed ASOP 4 Revisions
July 24, 2018, 9:31am
To the Members of the Actuarial Standards Board:
Cavanaugh Macdonald Consulting, LLC thanks the Actuarial Standards Board for this opportunity to comment on the proposed revisions to ASOP 4. We recognize that these revisions are rooted, as noted in the background sections of these three documents, in the context of a perceived need for additional guidance for public retirement plans. As a leader in providing actuarial consulting services to state and local government pension plans, we have expended tremendous effort over the years working toward educating boards, staff, and sponsors about appropriately funding these retirement plans, and from this experience we have some observations that we believe may be of value in helping to shape ASOPs that can best advance professional practice.
Prescriptive or Principle-Based?
Before dealing with some specific items in the proposed revisions, we believe there are two broad issues that should be addressed. The first of these is to note that many of the proposed changes for ASOP 4 are somewhat prescriptive, rather than being primarily principle-based. Generally, we actuaries have prided ourselves in being professionals and being able to exercise appropriate judgment in applying the core principles of our profession. We have made allowances for complying with applicable laws and regulations, or in assisting the accounting profession, but we have not wavered from the ideal of conducting our work as professionals following these core principals.
The challenge in moving to a more prescriptive model in ASOP 4 is that there are now significantly more details that must be provided. No longer can the ASOP language call on the actuary to consider a course of action – now it must describe that course of action in a way that can fit all situations. We note that for corporate pension plans guided by ERISA, there are hundreds of pages of specific requirements so that the actuary can carry out the required tasks. The proposed ASOPs are much shorter, of course, but that comes at the expense of being sufficiently prescriptive for all possible cases. We have identified a number of issues arising from real public retirement plans in which the proposed prescriptive ASOP language may be inadequate to provide complete guidance. If the underlying prescriptive requirement is not based on a principle, it becomes very difficult, and perhaps impossible, to simply use professional judgment to fill in the missing steps. Our point is that in order to be prescriptive and handle the wide range of real-world practice, it may well take hundreds of pages rather than just a few paragraphs to provide adequate guidance.
Further, ERISA does not require an actuary to be a professional, but instead calls for an actuary to carefully follow required steps. The various actuarial organizations in the United States have certainly encouraged practicing pension actuaries to join with them in acting as professionals following principles, and significant numbers have. Nonetheless, an Enrolled Actuary can legally practice on ERISA plans without needing to join those of us who seek to act as professionals – because the emphasis has been shifted from principle to prescription. Most public retirement systems currently require that their actuary be a member of one of the actuarial organizations that subscribe to the Code of Professional Conduct and the Actuarial Standards of Practice, ensuring that they have someone who acts in a professional capacity. Some, however, accept the Enrolled Actuary designation alone as sufficient. If the ASOPs for public plans become more prescriptive, at some point there may be less perceived need for professionals and systems may opt to engage Enrolled Actuaries without any other designation – or even non-actuaries – who can calculate the numbers they need without including the additional ASOP-compliance information they don’t need or want, reducing the influence the profession has in this area. Principles require a professional, prescriptions require a technician.
A second broad observation is that much of the new material in the proposed ASOP 4 revision will not apply to the majority of pension plans. Section 3.20 does not apply when there is a prescribed assumption or method set by law, thereby excluding ERISA plans. Likewise, sections 3.14, 3.16, and 3.17 will not be applicable for ERISA plans because these methods are selected via federal laws and regulations. Of course, ERISA valuations performed by individuals who are only Enrolled Actuaries are already exempt from complying with any ASOPs. Another category of plans to which many of these new provisions would not seem to apply are unfunded plans. Because there is no intent to fund them formally, there would not technically be a funding valuation or the need for cost allocation procedures. Thus, many of these new sections really only apply to funded church and public plans.
At first glance, this may appear reasonable, under the presumption that if there are not already requirements for determining funding requirements, an actuary should follow reasonable procedures in setting these. Of course, we believe that the current ASOP 4 actually does contain such guidance already. Moreover, many public plans have controlling legislation that dictates much of what an actuary may do (including specifying in statute the service-based salary increase assumption in one case we are familiar with), but because these rules are set by the states that established the retirement systems, the ASOP’s prescribed assumptions and methods exception does not apply.
Effectively, the ASOPs have set up a system whereby federally-set standards, some of which may have been legislated as a matter of delaying pension contributions to increase current tax revenues, are being presumed to be sufficient, while similar standards passed by state legislatures are deemed otherwise. It is not clear why the ASB should anticipate that officials elected to national offices will write legislation that is superior to that written by officials elected to state offices, especially since it was the same voters who elected both. Nonetheless, ERISA plans – which make up the majority of pension plans - are being exempted from significant portions of these new proposed ASOP changes, while public plans are not. We recognize there are some high-profile public plans where elected officials have chosen to delay funding, perhaps even against the advice of the systems’ actuaries, but we also are aware that the PBGC has concerns about future finances, suggesting that the funding methods that are mandated for ERISA plans may have weaknesses as well. Excluding plans from the ASOP proposals solely because of the type of legislative body writing the rules does not make sense.
(As an aside, we recognize that some may object that state legislators cannot objectively write funding rules for plans they establish, control, and fund. However, the plan provisions, the funding requirements, and significant information about the operations and funding progress of these systems are readily available to the public. If the legislators are not behaving responsibly in the eyes of the public, the public will ultimately vote them out of office and select new legislators. If we feel that we need these actuarial standards because publicly-elected officials cannot be trusted, we are effectively claiming that we as actuaries know public policy better than the public. This seems to be arrogant and well outside our scope of expertise.)
We next wish to address some specific issues where we believe the proposed revisions should be adjusted. First, we note that Section 3.11 requires that an actuary “calculate and disclose an obligation measure of effectively defeasing the investment risk” when performing a funding valuation. There are then some prescriptive elements of this calculation listed, including the valuing of benefits accrued as of the measurement date, the cost method, and a discount rate to be selected from one of two specific sources. As already discussed, this degree of specificity in this ASOP is in stark contrast with other ASOP language that frequently has terms such as “should consider” or “must consider”. As we discuss in the following paragraph, the prescriptive, inflexible nature of this requirement creates some situations that are not clear, but without the apparent ability of the actuary to apply judgment.
Many public plans have adopted benefit provisions that mitigate the employer risk by adjusting the member benefits based on such measures as asset performance or plan funded ratio. A common example is a COLA or 13th check that is payable when the plan reaches a certain funding threshold. Another example is the use of a variable interest crediting rate in cash balance plans which is tied to recent investment returns. In both of these examples, the benefits are based, to some degree, on actual asset performance. If investment risk is eliminated by purchasing Treasuries, the expected return will be significantly lower, thereby reducing benefits. The proposed language for the investment risk defeasement cost is not clear as to how to proceed. Should the actuary assume that the benefits will not be reduced (simply changing the discount rate in a computer program) leading to a model which has inconsistent assumptions? Or should the actuary reduce the projected benefit payments, thereby leading to a calculated risk defeasement measure that actually contains a combination of investment risk reduction and benefit cuts? Professional judgment does not work well in this situation since the calculation is being performed by prescription rather than by principle. Either approach seems to run afoul of other actuarial principles – we either build an inconsistent model or we improperly identify a measure. This is not simply a technical issue to be addressed – there will be other variants and odd situations that will arise, requiring more and more technical modifications. It is rarely possible to be partly prescriptive.
Further, the described goal of this measure is to find the cost to “effectively” defease the investment risk. However, the discount rates described in 3.11.c are related to methods that are not capable of actually defeasing the investment risk. A simple illustration will suffice to demonstrate this: Suppose that we as very wise actuaries had decided to implement this strategy in 1980. We would have worked with a plan sponsor to purchase bonds that very nicely lined up with projected cash flows for the next 30 years, and then purchased some 30-year Treasuries with the coupons removed that we intended to reinvest in 2010 to cover the remaining 40-60 years of payments that would still be due. However, in 2010, interest rates were much lower and so we would have found that we did not have enough money to make the benefit payments. The risk was not “effectively” defeased because some (substantial) risk remained. Of course, we all knew in 1980 that interest rates were high and that such levels were not likely to continue. We feel much more confident in 2018 buying bonds to defease the risk, because rates are not likely to be much lower in 2048 than they are now. But, if rates are higher in 2048 (even if still below historic averages), we would have more money than was needed. Thus, we still have risk – if we set aside too much money, we gave up some other use of that money in the meantime. If it were possible to buy options to purchase Treasuries 30 years and 60 years from now, then it would be possible to say investment risk can be “effectively” defeased. Otherwise, we are not accomplishing the stated goal of this section, but are instead feeling foolishly smug about our capabilities to predict interest rates well past our lifetimes.
We also note that while the goal of this disclosure calls for defeasing investment risk, one of the options (3.11.c.2) is to determine the cost of settling pension obligation, thereby eliminating all risks. While investment risk is usually the largest, there is an inconsistency between the stated intent and the method to measure it. If this method is allowed, we suggest that the actuary be further required to disclose which risks are being eliminated so that users will not be misled. Quantifying the cost of eliminating risks may indeed be valuable (and certainly is something actuaries will be considering under ASOP 51), so it is not clear why only one of those risks be identified in ASOP 4.
We do appreciate the efforts of the ASB to avoid requiring a market value of a settlement cost be determined. Because many of the funds we work with could purchase a few insurance companies themselves, there really is not a market for settling the plan obligations for large public retirement systems. Even for those systems where it is legally possible to do such things as purchase annuities for retirees, the state governments are still frequently the back-up for failed insurance companies – and so the sponsoring government would most likely be responsible for benefit payments if the insurer failed, meaning the risk is not really completely transferred. This, however, is a reminder that a settlement cost in 3.11.c.2 does not really fully settle liabilities.
Section 3.11 is stated to be applicable to funding valuations. As a practical matter for public plans, any study of proposed changes to benefits or funding policies will require an analysis be performed on this basis as well as the funding basis, since the decision makers will need to know the implications of any proposed changes for the risk defeasement cost to be disclosed. Later we discuss some of the issues relating to multiple measures of a liability, but it should be emphasized that this measure will be guiding decisions and will be higher profile than just a simple disclosure in a report.
While the cost to defease investment risk may be of interest to some, we anticipate that it will not be a useful measure for many public retirement systems. After all, it reflects an action that cannot be implemented by most plans. Most public plans are open to new members and many are prohibited from reducing future benefits for anyone who is in the plan. Attempting to defease the risk on a hypothetical benefit for a mid-career employee is purely an academic construct – the accounting effort to actually carry out such a task would be incredibly challenging. Of course, as noted earlier, there are not financial instruments available that can actually defease the risk over the 80-100 years of remaining payouts expected to be made to current members. Further, these plans are investing in diverse portfolios designed to take an appropriate level of risk so as to provide benefits at a lower cost than could be managed in the absence of risk. In fact, the trustees of the plan are required by law to be prudent, and it is doubtful that a “no risk” investment portfolio would be deemed prudent. Thus, actuaries will be required to disclose a cost of an action that is neither legal, possible nor desirable. We realize that some other disclosures may not be useful or appreciated by the clients we serve, but such information tends to either be needed for other actuaries to opine on the reasonableness of the work, or to alert the plan sponsors of some potential problem that they need to know about. This disclosure, however, does not assist other actuaries in reviewing the work (and, in fact, adds another item to review) and does not in most situations give the sponsor any useful information since there is no opportunity to actually defease risk. On the contrary, our clients will have to pay to have this work done, causing a small amount of actual harm to their funding situation. We question how such a requirement benefits anyone beyond answering some academic curiosities.
The final concern we have with Section 3.11 is that this measure may not be fully understood by those who see the number. While actuarial reports for corporate plans are presented to CFOs, human resource managers, and others in similar positions, public plan actuarial reports are placed on web sites, presented to legislative committees, and written about in newspapers. We believe that it is fully possible to explain to corporate senior management the nuances of unit credit versus entry age normal and funding discount rates versus risk-defeased yield curves, but we note that many in the general public may not be as familiar with these concepts as the typical CFO. While the general public is not a direct intended user of these actuarial reports, those of us who practice in this area are cognizant that the broad public is an indirect user.
As an illustration of this, the American Academy of Actuaries wrote the Missouri State Employees’ Retirement System Board a letter on October 23, 2017 in which they noted several times that the information provided to certain deferred vested employees might not be fully understood. In our experience, state employees have, on average, a higher level of formal education than the general population. If the Academy is correct in their concern that these individuals (with no particular actuarial background) would not understand certain actuarial concepts, why do we think the general public will understand these points?
Providing two sets of liability measurements in the same report that differ because of an alternate purpose makes sense to us as actuaries. It likely makes sense to those people we work for when we have had sufficient opportunity to provide background and education. For the general public, however, two sets of numbers may serve to create confusion. Because most public retirement systems must make their valuation reports available to the public, the presence of two sets of numbers can be easily misunderstood or intentionally misused. While under Precept 8 of the Code of Conduct actuaries must take reasonable steps to make sure that our work product is not used to mislead others, publicly presenting two sets of numbers will allow those with an agenda to eliminate public retirement plans the opportunity to say that the higher number is the “true” liability. They will be able to assert that actuaries are perhaps being dishonest in providing the calculation used for plan funding purposes. Apart from this intentional misuse, it would certainly be an easy argument that “actuaries make up all these numbers – look how different these two numbers are”. Actuaries are thus maligned as somewhere between being arbitrary to being outright liars – certainly not a step towards advancing the profession.
We strongly believe that this section 3.11 should include an “actuary shall consider” clause to provide for discretion when the issuance of additional disclosures will create confusion, especially since these disclosures are for an action that is neither possible, nor desirable for public retirement systems. In light of the ASOP 51 suggestion of a measure such as this as being an option, we actually believe 3.11 could be eliminated altogether.
The proposed language for this section should be modified to indicate that this applies only when amortizing a positive unfunded liability, and not when there is a negative unfunded liability (sometime called a surplus). As written, there could be some ambiguity in this regard.
We also believe, if this section will be prescriptive rather than principle-based, that there should be some clarification regarding the phrase “[i]f the actuary selects” to address the situation when legislation or the retirement system boards select the method with some degree of input from the actuary. In the public sector, the actuary rarely is able to solely select the actuarial methods or assumptions, but nonetheless often has significant influence on the process. If the language read “if the actuary has sole authority and selects” or “if an actuary recommends” it would be clear that the actuary was still allowed to use a method adopted by a board against the actuary’s recommendation.
The proposed language for this section should be modified to allow for a contribution rate to remain above the actuarially determined contribution for an indefinite time. The language in 3.16.a prevents the smoothed rate from being too much higher, which should provide protection against unreasonable intergenerational equity issues. We do not see any reason that a smoothing method should be compelled to be lowered, which 3.16.b currently requires. Some cash balance plans intentionally contribute well above the actuarial rate so as to provide the opportunity for “surplus” to be added to member’s accounts.
This section applies for funding valuations when there is not a prescribed method or assumption set by law. In the real world, this will essentially be applicable to funded church and public plans. As noted earlier, many, if not most, public plans have some additional legislative requirements for setting a method or assumption, but these requirements do not meet the definition of “prescribed assumption or method set by law” as defined in ASOP 4. Frequently, the legislative methods will nonetheless meet the requirements of this section, so that the actuary need only show one set of contribution numbers. In other cases, however, the actuary may be compelled to calculate an alternate contribution rate that may not be materially different from the legislated rate in order to satisfy the requirements of this section. There are real cases where the current method used does not technically meet the requirements of the proposed language, but our reasonable and compliant alternative would be identical. We can’t help but wonder if our work to provide a distinct “compliant” actuarial rate that is nearly or completely identical to the “non-compliant” rate may, in fact, serve to discredit the actuarial profession.
We note that section 3.19 requires the actuary to disclose if a funding policy will not accumulate sufficient assets to pay the benefits of the plan. Ultimately, this is the central issue at stake. If there is a policy in place that will accumulate sufficient assets, there is little apparent value in providing an alternate measurement of a contribution rate. We believe that the requirements of 3.20 should be applicable only if the actuary has reason to believe that the current funding policy is inconsistent with accumulating sufficient assets. In that situation, the additional information serves to illustrate the needed change in contributions to fund the promised benefits, information which would be needed by the plan sponsor. (Of course, the solution may be to reduce benefits rather than increasing contributions – this paragraph presumes that contributions are the only solution.)
Section 3.20.g calls for the reflection of timing between the measurement date and the contribution date. For public retirement systems with multiple tiers (where successive tiers usually have lower benefits and normal cost rates) where the contribution rate is determined as of a measurement date a year or two before the implementation of the rate, this may add a great deal of complexity for very little value. The requirement, as written, does not allow for the actuary to exercise judgment regarding the significance of the factors involved. May the actuary ignore the anticipated reduction in normal cost rate of a few basis points? Must the actuary reflect that school district payroll is not exactly uniform throughout the year? May materiality be considered? In on-going plans, such differences are frequently inconsequential when compared to the probable variation in population size, payroll experience, and the like. This provision should be restated to include the phrase “The actuary should consider...” so as to avoid prescribing a high level of detail which will only serve to increase the complexity of the calculation with little impact on the result.
The Background section of the exposure drafts indicates that these proposed changes were made in response to the Pension Task Force, which in turn was formed in response to concerns about public sector pension plans. The implication seems to be that the additional requirements of these ASOPs will somehow provide actuaries with the guidance that is needed to help them act in a way that leads to changes in how public pension systems are valued, funded, designed, or presented. As we reflect over the past decade or so, there have been major shifts that have occurred with public retirement plans:
the distribution of investment return assumption has shifted significantly lower
the mortality assumptions of many systems has moved to a generational mortality basis, including some systems that adopted these changes around 2005
contribution rates for employers have increased in most systems, and members have increased contributions in many cases, as well
benefit provisions for new hires (and for future accruals where legal) have been reduced or adjusted to share risk
contribution determination methods have been developed to help move toward fully funding on-going plans, with some systems targeting funded ratios in excess of 100%
the practice (inadvertently prompted by accounting standards) of an open 30-year level percentage of payroll amortization method has all but disappeared
- the level of detail provided to board members and the general public has increased significantly
Through all of this, we see actuaries playing a significant role in helping public policy-makers make these changes. This does not suggest that actuaries are unsure of what to do and needing additional guidance, but rather that actuaries are able to work from the existing principles of the ASOPs to help advance the practice and solve problems.
With these proposed prescriptive requirements, however, actuaries may be placed in a position of pointing out that a requirement of a state law is in conflict with actuarial standards, and supply the “correct” number that is nearly identical. (We realize “correct” is not what the ASOP says, but it is how the public will understand it.) Then the actuary will show two values for what would appear to many as the plan liability – proving to some that actuaries simply make up the numbers or lie.
What is not clear is that any of these proposed changes will cause law-makers to suddenly realize that they must make changes to public policy and choose pension funding over road improvements or school funding. They may believe that their role as elected officials is to serve the citizens who elected them by guiding public policy and making the challenging choices about how limited resources are to be allocated to best serve the public over time. And they may be right.
Cavanaugh Macdonald Consulting would strongly encourage more deliberation and discussion before implementing any changes.
Brent. A. Banister, PhD, FSA, EA, FCA, MAAA