By Stephen A. Alpert
One of the more difficult judgments an actuary must make is being “reasonable.” In addition to applying our training, experience, and professionalism, we also need to navigate the definitions and expectations that others—principals, regulators, accountants, attorneys, or the public at large—bring to any given situation.
While the tensions between multiple views and definitions likely apply across all actuarial practice areas, recent activity related to benefit suspension applications by multiemployer pension plans have highlighted these tensions across different stakeholders and brought them into sharper focus. The interactions among actuaries, government officials, plan management, and individual plan participants provide fertile ground for exploring difficult questions such as:
- Is “reasonableness” something that can be objectively measured?
- How does an actuary reconcile different judgments about reasonableness—whether between different actuaries or with someone outside the profession?
- What role does popularity (what “everyone else” is doing) play?
To set the stage for this discussion, a little bit of background about multiemployer benefit suspensions is needed. Multiemployer pension plans are collectively bargained and are governed by a joint board (union and employers) of trustees. The Multiemployer Pension Reform Act of 2014 (MPRA) permits certain deeply distressed multiemployer pension plans to reduce benefits to plan participants as a means of preventing plan insolvency, but only after the plan sponsor has exhausted all other options to remain solvent, and provided the U.S. Department of Treasury approves the application.1 One of the requirements is that the suspensions be “reasonably estimated to achieve, but not materially exceed, the level that is necessary to avoid insolvency.”2 The regulations implementing this provision clarify that “avoiding insolvency” means that an actuary must project the available assets relative to benefits due each year in the future, typically for 30 years or more.
The regulations3 provide a three-part test for whether any plan is “reasonably estimated to achieve” avoiding insolvency:
- No insolvency in any year in a deterministic projection;
- More than 50% chance that the plan will avoid insolvency in any year in a stochastic projection; and
- No decrease in solvency ratio in the last five years of the (deterministic) projection.
The regulations go on to define “reasonably estimated to not materially exceed…” as meaning that a lesser reduction (that is, a greater benefit payable to the participant) would cause a failure of the reasonably estimated to avoid insolvency test. The materiality threshold for this determination is the greater of 5% of the initial reduction or 2% of the initial benefit.
For example, suppose that an actuary determines—on the basis of numerous assumptions (more on those later)—that paying participants 60%4 of their originally scheduled benefits (that is, a reduction of 40%) meets all parts of the “avoiding insolvency” test each year for the next 30 years. To meet the materiality test, the actuary would have to show that paying participants 62% of their originally scheduled benefits (2% = the greater of 2% of the initial benefit or 5% of the 40% reduction) would cause the plan to fail the solvency test outlined above.
Most actuaries would quickly recognize that a long-term stochastic projection of investment return, time of benefit commencement, mortality, and other assumptions includes a large amount of inherent uncertainty. A small change in any of these assumptions could easily change the median (50th percentile) solvency ratio in the nth projection by much more than could be explained by a 3.3% change in the initial underlying benefits (2/60).
In essence, then, the regulation defines a reasonable range of results much more narrowly, and with greater mathematical precision, than we might expect from two actuaries independently applying their own professional judgment, or even what one actuary might identify as a reasonable range of results. A corollary point is that the regulation implicitly assumes that there is virtually no measurement uncertainty or model risk over a very long projection period; which is different from the view typically taken by those involved in financial forecasting, whether or not they are actuaries. Treasury has suggested that this approach is due in part because of a legal structure that makes the benefit reductions a one-time yes/no binary decision, with minimal opportunity for adjustment as future experience unfolds.5
In addition to requiring reasonable overall results, the Treasury Department, mainly through the IRS (the main federal agency for these and most other regulations around pension actuarial work), has a long history—going back to the regular periodic pension valuation requirements of the Employee Retirement Income Security Act of 1974 (ERISA)—of requiring that non-statutory assumptions also be individually reasonable. For these multiemployer projections, for example, the regulations require:
- Each of the assumptions must individually, and in combination, be reasonable, “taking into account the experience of the plan and reasonable expectations.”6
- Measurement-specific factors must be taken into account.
- Analysis and explanation justifying differences from assumptions used for other purposes must be provided.
- Description and “supporting evidence for the selection of” the major assumptions must be supplied. The list of assumptions and the required supporting detail is quite extensive.7
- Ten years of experience must be provided for “critical” assumptions: contribution amounts (including base units8 and negotiated rates), withdrawal liability9 payments, and the actual rate of return on plan assets.
- Sensitivity analysis, including investment return that is 1% and 2% less than the baseline assumption and growth in “contribution base units” that is 1 percentage point less than the baseline assumption, must be provided.
Some of these requirements, like measurement-specific factors or disclosure of the rationale for assumption selection, echo similar requirements in the profession’s actuarial standards of practice (ASOPs).10 Others, such as the requirement for both individual and aggregate reasonability or precise sensitivity analysis thresholds, go much further and are much more prescriptive.
So far, we have a regulatory formula that appears to provide a measurable definition of “reasonable” (for purposes of the regulation), and regulatory framework that leaves open the question of who gets to make the judgment about reasonability, although there is a strong implication that through the application and approval process Treasury claims at least oversight, if not the outright ability to substitute its own judgment for that of the actuary in determining reasonability.
Before we get to how Treasury evaluates “reasonableness,” which will add the question of popularity to the questions of measurement and conflicting judgment already noted, it is worthwhile to review the broader context of public expectations and standards in which actuaries operate.
In everyday usage, we see a wide range of meanings, in part depending on context. From the dictionary11 we have up to a half-dozen possible definitions:
- Being in accordance with reason (“a reasonable theory”)
- Not extreme or excessive (“reasonable requests”)
- Moderate or fair (“a reasonable chance” or “a reasonable price”)
- Having the faculty of reason
- Possessing sound judgment
ASOP No. 1 describes the idea further for actuaries:
The intent is to call upon the actuary to exercise the level of care and diligence that, in the actuary’s professional judgment, is necessary to complete the assignment in an appropriate manner.
Because actuarial practice commonly involves the estimation of uncertain events, there will often be a range of reasonable methods and assumptions, and two actuaries could follow a particular ASOP, both using reasonable methods and assumptions, and reach different but reasonable results.
Both the ASOPs and the Code of Professional Conduct note that any conflicting or more restrictive “applicable law (statutes, regulations, and other legally binding authority)”—as spelled out in each ASOP—takes precedence over the ASOPs.
Across practice areas, “reasonable” is also used in a more nuanced way that may encompass a broader range of the dictionary meanings throughout the definitions in several ASOPs.12
- “Reasonable, appropriate, and attainable” costs are a key element of actuarially sound Medicaid capitation rates. (ASOP No. 49)
- “Reasonable dividend expectations” of policyholders are “[t]he expectations that the current dividend scale will be maintained if the experience underlying the current scale continues, and that the dividend scale will be adjusted appropriately if the experience changes.” (ASOP No. 33)
- “A reasonable estimate of unpaid claims liabilities” is a component of “incurred claims.” [ASOP No. 5]
- Moderately adverse conditions include “one or more unfavorable, but not extreme, events that have a reasonable probability of occurring during the testing period.” (ASOP Nos. 22 and 28)
- A “reserve evaluation” is the “process of evaluating the reasonableness of a reserve.” (ASOP No. 36)
- A data review is done to “determine if [the] data appear reasonable and consistent for purposes of the assignment.” (ASOP No. 23)
“Reasonable” or “reasonableness” also appears numerous times within the body of the ASOPs, perhaps most illuminatingly within ASOP No. 41, Actuarial Communications, which requires that the actuary
“identify the methods, procedures, assumptions, and data used by the actuary with sufficient clarity that another actuary qualified in the same practice area could make an objective appraisal of the reasonableness of the actuary’s work as presented in the actuarial report.”
Taken together, all these uses and definitions make it clear that, for actuaries acting solely within the requirements of the ASOPs, “reasonableness” depends heavily on the context in which it is used and how an actuary applies judgment to arrive at a conclusion of reasonability. There is no single, absolute definition of reasonable—indeed, such a hard-and-fast rule would seem to be the antithesis of what “reasonable” represents.
However, as ASOP No. 1 notes, “reasonableness” is just one element of professional judgment. Assignments must also be completed “in an appropriate manner.” In many situations, one might assume that the input of a principal or outside regulator would be needed to determine what this means.
Through regulation, rulings, and informal discussions, Treasury and the IRS have provided significant guidance on what they consider to be “an appropriate manner” for multiemployer benefit suspensions. This guidance appears to have significantly restricted the range of the actuary’s reasonable judgment.
We have already seen that, within the regulations, the range of “reasonable” long-term projection results is considerably narrower than what most actuaries would assume in the absence of other information. In our example of a 30-year stochastic solvency projection, assets need to be projected to be greater than liabilities 50% of the time, with no more than a 3.3% difference in initial benefit amounts. In non-actuarial terms, that may be the equivalent of landing multiple consecutive darts in the same hole in the bull’s-eye.
The IRS has imposed additional restrictions on these multiemployer benefit suspensions through application rulings and informal discussions. For example:
Treasury indicated that a single value for long-term average deterministic investment returns was not appropriate, and should instead be replaced by select and ultimate investment return assumptions that take into account the timing of the plan’s cash flows.13 The implicit assumption here is that forecasting short-term investment returns (for example, annual returns over a 1- to 10-year period) is “more refined” and supposedly “better” (as opposed to just being different) than forecasting an average return over a long period of time. This contrasts to what statistics courses teach: even if there is an expectation that near term average returns might differ from average long-term returns, it is much more difficult to predict the sequence of random events like flipping a coin or the rate of return in a particular period than it is to estimate the proportion of heads over a large number of flips or the rate of return over a large number of periods.
- Several times in the same letter, Treasury concludes that an assumption is not reasonable if it produces a materially different result than a more “refined” calculation. These conclusions would appear to conflate complexity with accuracy. While that may be true in some cases, in others, differences—even material differences—are just that, and it is impossible to say that one is better than another, especially for very-long term projections. It is entirely possible that reasonable differences in judgment or reasonable differences in assumptions could produce a range of results that would be materially different from one another. These could be particularly difficult in the context of the very narrow definition of materiality embedded in the multiemployer benefit suspension regulations.
- As a starting point for evaluating the reasonable range of the investment return assumption, Treasury uses a survey of (non-actuary) investment managers’ expected returns, focusing on the expectations of the second- and third-quartile managers.14 Actuaries have many open questions about this survey, including questions about methodology, measurement-specific factors, suitability to purpose, or why first- and fourth-quartile survey results did not represent a reasonable difference of opinion.
- With regard to demographic assumptions, Treasury has placed a lot of weight on a specific, clear, and direct connection to the experience of the plan as a basis for assessing reasonability.15 Although “taking into account the experience of the plan” has long been a part of the pension actuary’s world, the level of required detail and documentation around this requirement is new.
It took several years after enactment of the law for this guidance to develop, including several applications that were turned down by Treasury as being “unreasonable.” More recently, however, plan sponsors have learned from the earlier denials, and benefit suspension applications have had more success at demonstrating reasonableness and gaining approval from Treasury.
In the absence of other information, “reasonable” may be as simple as a Goldilocks test—neither “too hot” nor “too cold”—based on an actuary’s training, experience, and sound judgment. However, even that simple formulation raises the question of how much is “too hot” or “too cold.” There also may be a substantial albeit indeterminate and gradually changing gray area between definitely reasonable and definitely unreasonable.
However, actuaries seldom if ever get to make this determination in a vacuum. Reasonableness will ultimately be evaluated by principals or regulators, based on their specific needs and requirements—what is “appropriate to the situation.” For example, because of the considerable public and political effects of reducing retirees’ pensions, multiemployer benefit suspension applications received significantly heightened scrutiny from Treasury.
So, while it is important to understand the needs and requirements of principals and regulators, and to clearly communicate why the actuarial calculations are appropriate for those needs and requirements, the actuary’s interest in reasonableness doesn’t stop there. The public interest demands that we also apply our own skills, training, and judgment. To continue with our benefit suspension example, an actuary might have to wrestle with questions such as:
- How does an actuary reconcile regulatory constraints on a reasonable range of individual assumptions with a reasonable end result?
- Does an actuary need to form an opinion about the reasonability of the process, even if the result of that process is a result that falls within the actuary’s reasonable range determined by other means?
- How much detail should the actuarial report include to educate others about (and possibly reconcile) different viewpoints of reasonableness?
Fortunately, ASOP No. 4116 provides some guidance to help an actuary sort through these issues:
3.2 (as noted above): … [the actuary should] identify the methods, procedures, assumptions, and data used by the actuary with sufficient clarity that another actuary qualified in the same practice area could make an objective appraisal of the reasonableness of the actuary’s work as presented in the actuarial report.
4.1.3 [the actuarial report should disclose]…
d. any cautions about risk and uncertainty (see section 3.4.1);
e. any limitations or constraints on the use or applicability of the actuarial findings contained within the actuarial communication including, if appropriate, a statement that the communication should not be relied upon for any other purpose; …
g. any information on which the actuary relied that has a material impact on the actuarial findings and for which the actuary does not assume responsibility (see section 3.4.3); …
4.3: … the actuary should disclose the following in the actuarial report …
a.the assumption or method that was set by another party;
b.the party who set the assumption or method;
c. the reason that this party, rather than the actuary, has set the assumption or method; and
- that the assumption or method significantly conflicts with what, in the actuary’s professional judgment, would be reasonable for the purpose of the assignment; or
- that the actuary was unable to judge the reasonableness of the assumption or method without performing a substantial amount of additional work beyond the scope of the assignment, and did not do so, or that the actuary was not qualified to judge the reasonableness of the assumption. …
4.4: If, in the actuary’s professional judgment, the actuary has deviated materially from the guidance set forth in an applicable ASOP … the actuary can still comply with that ASOP by providing an appropriate statement in the actuarial communication with respect to the nature, rationale, and effect of such deviation.
For any assignment, an actuary might need all these tools, especially when dealing with an issue as complex and nuanced as multiemployer benefit suspensions. Taking the ASOPs together—particularly ASOP No. 41 and the ASOP No. 1 requirement to “exercise the level of care and diligence that, in the actuary’s professional judgment, is necessary to complete the assignment in an appropriate manner”—leads me to a potential personal working definition for reasonable:
Can the actuary communicate the reasons for actuarial judgments and opinions in a way that is both educational and persuasive to the intended audience (both direct and indirect)?
Being able to answer this question affirmatively would likely mean that the actuary would have answered all the questions that we started with, as well as covered both dictionary and principal definitions of what it means to “be reasonable.”
STEPHEN A. ALPERT, MAAA, FSA, FCA, has been on the Pension Practice Council for 10 years and spent 30 years at Mercer. He is also the immediate past president of the Academy. The views in this piece are his and do not necessarily reflect those of the Academy.
- Benefit suspensions are an exception to the general rule that pension benefits cannot be reduced once earned. The conditions under which suspensions may be allowed are spelled out in section 432(e) of the Internal Revenue Code. Under MPRA, “Treasury, in consultation with the Pension Benefit Guaranty Corporation (PBGC) and the Secretary of Labor (DOL), must approve an application upon finding that the plan is eligible for benefit suspensions and has satisfied the applicable statutory requirements.” (May 6, 2016, letter to the Board of Trustees, Central States, Southeast and Southwest Areas Pension Plan)
- Code section 432(e)(9)(D)(iv).
- 1.432(9)-1(d)(5) and following paragraphs.
- A flat 60% is a simplified example; the MPRA rules are complex, and certain retired or disabled participants are protected from reductions.
- Unfortunately, a single decision now, with no automatic adjustment to reflect emerging experience, is a fairly common legislative or regulatory outcome across all areas of actuarial practice. Social Security, health care, long-term care insurance, flood insurance are all examples where the inability to make actuarial adjustments has created cost or solvency pressures. Benefit suspension decisions, like many of these other programs, ultimately reflect a political balancing act between the rights and benefits of the beneficiaries, and those of the taxpayers and others responsible for the program’s solvency.
- Reg. 1.432(9)-1(d)(5)(iv).
- See appendix B of Rev. Proc. 2017-43.
- Contribution base units are units of work such as hours, days, or weeks, on which employer contributions are based.
- When an employer withdraws (stops contributing) to a multiemployer plan, the plan will assess a withdrawing employer a portion of the underfunding. Both the assessment and amount of actual payment from the employer to the fund involve complex calculations that are beyond the scope of this discussion.
- For pension purposes, these include, but are not limited to, ASOP Nos. 4, 27, 35 and 41.
- The examples that follow are taken from the Definitions from Actuarial Standards of Practice and Actuarial Compliance Guidelines of the Actuarial Standards Board, which is available from the Other Documents page of the ASB website.
- Department of Treasury letter to Board of Trustees, Central States, Southeast and Southwest Areas Pension Plan, May 6, 2016.
- Meeting of the American Academy of Actuaries Multiemployer Plans Committee and representatives from the Department of the Treasury, PBGC, and Department of Labor, February 23, 2018.
- In addition to the preceding sources, see for example, Department of Treasury letter to Board of Trustees, Automotive Industries Pension Plan ATPA, May 9, 2017.
- Refer to ASOP No. 41 for full details and context.