The Society of Actuaries Issues Public Comment on IRS Notice

WASHINGTON, March 1 -- R. Dale Hall, managing director of research, and Timothy J. Geddes, Chair of the Retirement Plans Experience Committee at The Society of Actuaries, Schaumburg, Illinois, has issued a public comment on the Internal Revenue Service notice entitled "Mortality tables for 2021 Notice 2019-67". The comment was written and posted on Feb. 28, 2020:

* * *

The Society of Actuaries ("SOA") is the world's largest organization credentialing and serving the actuarial profession. We are an educational, research, and professional organization of more than 32,000 actuaries, dedicated to serving our members, students, the profession, and the public. We conduct a wide range of research to provide technical information resources for the profession, to advance the capabilities of the profession, to inform public policy development, and to promote public understanding and the public interest. This research includes many studies of historical experience and techniques for projections into the future. Experience studies have been at the core of SOA research activities since its formation in 1949 and were additionally a main activity of our predecessor organizations for many decades prior to that date.

Core principles for all SOA research projects are objectivity, quality, relevance and quantification. The SOA does not take advocacy positions on specific policy proposals.

This letter is in response to IRS Notice 2019-67, "Updated Mortality Improvement Rates and Static Mortality Tables for Defined Benefit Pension Plans for 2021," which requested comments in connection with the SOA release of the Pri-2012 Private Retirement Plans Mortality Tables Report in October 2019.

Specifically, the notice solicited comments regarding the following:

- Whether there are other studies of actual mortality experience of individuals covered by pension plans and projected trends in that experience that should be considered for use in developing mortality tables for future use under Sec. 430

- Which tables in the Pri-2012 Mortality Tables Report should be used to develop Sec. 430(h)(3)(A) mortality tables, if the Pri-2012 Mortality Tables were to be used for that purpose The SOA's Retirement Plans Experience Committee (RPEC or "the Committee"), a group of highly qualified actuaries working in the retirement field, performed the Pri-2012 study. The SOA has relied on the expertise of current RPEC members in preparing the below comments in response to the request in Notice 2019-67.

Use of Pri-2012 Mortality Tables to develop Sec. 430(h)(3)(A) mortality tables Code Sec. 430(h)(3)(B) provides that the Secretary shall make revisions at least every 10 years to reflect the actual experience of pension plans based on available independent studies of the mortality of individuals covered by pension plans. These tables were updated for the 2018 plan year (Regulations 1.430(h)(3)-1 Mortality tables used to determine present value), to reflect the RP-2014 study rates. The rates were adjusted back to the central year of that study's experience period, 2006, (later published by the SOA as RP-2006 rates) and projected forward from 2006 using mortality improvement scale MP-2016. Since 2018, 2 the basis has been updated only for subsequent mortality improvement scales MP-2017, MP-2018 and MP2019 (applicable to years 2019, 2020 and 2021 respectively).

The Pri-2012 study was published by the SOA in 2019, first as an exposure draft in May, then in final form during October 2019. The central year of pension plan experience was 2012 and the tables produced were not adjusted for mortality improvement after 2012. The SOA believes it would be appropriate for the Department of the Treasury ("Treasury") to utilize the information contained in this recent study of pension plan mortality in developing Sec. 430(h)(3)(A) mortality tables.

It should be noted that the Pri-2012 amount-weighted rates for the total employee and retiree data sets generate very similar present value factors as the RP-2006 basis. For example, refer to Table 1.1 in the Pri2012 Report. This indicates some degree of consistency between the RP-2014 and Pri-2012 study results.

The Pri-2012 study also confirms the findings in both the RP-2014 and RP-2000 reports that the blue collar / white collar distinction is significantly correlated with higher / lower (respectively) mortality.

The Pri-2012 report and tables explicitly reflect the higher mortality exhibited by surviving beneficiaries ("contingent survivors") of deceased pension plan participants. This higher mortality was observed and noted in the RP-2014 study, though final tables reflected the combined data for retirees and contingent survivors as "healthy annuitants". RPEC's Pub-2010 study of public sector (government-sponsored) experience also showed this difference; it was the first RPEC study to provide separate tables for contingent survivors and exclude contingent survivor experience from retiree table development.

As in prior RPEC studies (RP-2000, RP-2014 and Pub-2010), the Pri-2012 report provides separate tables for disabled retirees, who also have higher mortality rates compared to general retirees. All tables have distinct versions for male and female participants, and current employees have tables separate from retirees, disabled retirees and contingent survivors.

In general, the SOA believes Treasury should consider applying distinctions that RPEC studies of pension plan experience have shown to be significantly correlated with mortality, including: - A plan's "collar" designation -

o RPEC's work has identified a significant difference in mortality between employees who are either hourly-paid or union ("blue collar") and those who are salaried and non-union ("white collar"). The Committee has observed these differences in the past three reports, beginning with the RP-2000 study.

o For many plans, participants' blue collar or white collar classification could be identified from past working history. The SOA thus believes that it would be reasonable for pension plans to value liabilities based upon those participants' collar status.

Where sponsors could individually identify participants' collar status it would be reasonable to selectively apply blue collar and white collar mortality at the participant level, as noted in subsection 12.2.2 of the Pri-2012 report.

Where sponsors are unable to individually identify collar status, the plan could be evaluated based on the thresholds used in the Pri-2012 study. RPEC used a threshold of 70% when identifying a plan as blue or white collar for the Pri-2012 study./1

The SOA acknowledges that when setting minimum funding standards, consideration must be given to sponsors' incentives when selecting assumptions. With that in mind, precautions against adverse selection in the choice of assumptions similar to those in the substitute mortality guidance are likely warranted if the IRS were to consider permitting collar-based mortality tables for minimum funding purposes. Another potential control against anti-selection the IRS could consider would be to presume participants are white collar unless the sponsor knows they are blue collar (or can demonstrate that the participants have other indicators suggesting mortality consistent with blue collar employment). Such an assumption would be consistent with RPEC's observation that unknown collar participants exhibited mortality similar to that of white collar participants (see subsection 4.3.2 of the Pri-2012 report).

An annuitant's status as disabled

o RPEC has also consistently identified distinct mortality patterns associated with participants who retire by means of long-term disability in each of its last three studies.

The current minimum funding rules permit the reflection of differentiated mortality tables for certain participants who are disabled. The subject was most recently addressed in IRS Notice 2008-29 which reaffirmed the use of the mortality tables provided in Revenue Ruling 1996-7. The tables referenced in Revenue Ruling 1996-7 were created based upon data which significantly precedes that date and thus are more than 25 years old.

o The IRS could consider applying mortality improvement to the disability tables. RPEC reviewed the progression of disabled annuitant mortality with the release of the MP2014 improvement scale. The Committee encouraged application of Scale MP-2014 to RP-2014 Disabled Retiree mortality rates based on information provided in the 2012 OASDI Trustees' Report and analysis of the relationship between the RP-2014 and RP2000 disabled annuitant rates (MP-2014 Report, subsection 6.2). The Committee continues to encourage projection of disabled annuitant mortality rates using appropriate mortality improvement scales.

o The existing disabled annuitant mortality tables under Sec. 430 are bifurcated between pre-1995 and post-1994. For those disabled after 1994, the law allows for separate tables only for disabilities that meet the Social Security definition of disability. Thus, the Pri-2012 tables may be deemed inappropriate for minimum funding purposes without a statutory change.

o If enabled by a statutory change, using Pri-2012 tables for minimum funding purposes would enable plan sponsors to avoid the gap that exists in the current rules where healthy mortality must be used for participants who are disabled under the plan's definition but not verifiably disabled under the Social Security Act. The Pri-2012 tables were created by splitting the retiree and disabled retiree subgroups using the contributing plans' disability definitions.

- An annuitant's status as a contingent survivor -

o Consideration should be given to reflecting contingent survivor mortality for any annuitant in the plan census who is the surviving beneficiary of a deceased plan participant.

o As discussed in the Pri-2012 (and Pub-2010) reports, the reflection of contingent survivor mortality for joint-and-survivor benefits (where the original participant is still alive) has a relatively smaller impact. Typically, the majority of the present value for a joint-and-survivor annuity is associated with the period when the original participant is still alive.

Subsection 12.4 and Appendix D.6 (and Table D.19) of the Pri-2012 report outline three approaches RPEC explored to calculate joint-and-survivor present values:

Approach 1: Use the same mortality basis as the original retiree for the life of the beneficiary. This approach disregards the contingent survivor tables for valuing joint-and-survivor annuities.

Approach 2: Use the same mortality basis as the original retiree while the original retiree is still alive, and use contingent survivor tables after the death of the original retiree

Approach 3: Use contingent survivor mortality for the full lifetime of the beneficiary

The SOA believes any of these three approaches could be reasonable. As mortality for beneficiaries prior to the death of the original retiree is generally not reliably tracked, it is difficult to ascertain which of these three approaches most accurately models beneficiary mortality in the period prior to the original retiree's death. It should be emphasized that these three approaches are solely for valuing joint-and-survivor annuities, and that the SOA believes that contingent survivor mortality tables should be considered for surviving beneficiaries of deceased plan participants.

5 Other Studies of Actual Mortality Experience of Individuals Covered by Pension Plans The Pri-2012 tables are based on mortality experience of private-sector defined benefit pension plans in the U.S., from 2010 through 2014. These tables are the most current published mortality tables drawn from experience of U.S. private-sector pension plans, including multiemployer plans. As such, the SOA believes that these are appropriate tables to use if updates to the Sec. 430 mortality tables are to be made.

The SOA is not aware of any other recently published tables covering this same population, to which Sec. 430 generally applies.

IRS Notice 2019-67 specifically asks whether the Sec. 430(h)(3)(A) mortality tables should take into account mortality experience of individuals covered by large public-sector plans, such as those included in the SOA's Pub-2010 Public Retirement Plans Mortality Tables (the "Pub-2010 tables"). The SOA does not believe that the Pub-2010 tables should be considered to update the Sec. 430 mortality tables. The Pub-2010 tables were created based on the experience of public sector plans which are not subject to Sec. 430. The resulting mortality rates were significantly different from the results for private-sector plans, likely due to the different nature of the work and industry for the public sector covered groups (see the comparison of annuity factors in Appendix D.5 of the Pri-2012 report). As such, the SOA does not generally believe that these tables are representative of the mortality experience of the private-sector plans, to which Sec. 430 applies. An exception might be to allow the Pub-2010 tables to be used for development of substitute mortality tables under Sec. 430(h)(3)(C), if the IRS were to allow such alternative standard tables, for plans for which the pattern and level of mortality experience is closer to one of those tables.

One other potential source of published mortality tables in the U.S. is the Social Security Administration.

However, because the Social Security mortality tables are based on general population mortality experience, these tend to show higher mortality levels (i.e., lower life expectancies) than for working populations, such as included in the Pri-2012 and the prior RP-2014 study. Using Social Security tables likely would understate liabilities for most plans covered by Sec. 430. Because of this, the SOA does not believe that the Social Security tables should be considered in updating the Sec. 430 mortality tables for individuals covered by private-sector pension plans.

Projected Trends in Mortality Experience of Pension Plans

As mortality varies by age and by calendar year, generational projection is best able to capture the two-dimensional nature of this variation. In the past, some actuarial software has not been able to accommodate a two-dimensional model, and it was desirable to approximate the impact of a two-dimensional scale with a one-dimensional static scale. However, at this point, generational projection has become the norm, and our understanding is that actuarial software can now generally handle generational projection. Thus, we believe that generational projection should be considered for determining minimum pension funding rather than a static approximation.

The SOA published the Mortality Improvement Scale MP-2019 Report in October 2019. The MP scales have been updated annually since 2014. Scale MP-2019 is based on the RPEC mortality improvement model, denoted RPEC_2014_v2019, reflecting the historical mortality data through 2017. Similar to prior MP scales, the scale was developed by following the principles outlined in the Scale MP-2014 Report and subsequent MP reports.

6 Scale MP-2019 was constructed as follows: Historical rates were calculated using a two-dimensional Whittaker-Henderson graduation of the natural logarithm of U.S. population mortality rates with smoothness components based on the sum of the squares of third finite differences ("order-3 graduation"). Scale MP-2019 rates were developed from this RPEC_2014_v2019 model, applying the same committee-selected assumption set used to develop Scales MP-2016 through MP-2018; specifically: - Ultimate long-term rate of mortality improvement: flat 1.0% rate to age 85, decreasing linearly to 0.85% at age 95, and then decreasing linearly to 0.0% at age 115

- Until the ultimate rate is reached, improvement based on third-degree polynomial interpolation with an initial slope of zero, so that short-term rates of improvement reflect recent experience - Interpolation reflecting a 50%/50% blend of third-degree polynomials along fixed ages and fixed year-of-birth cohorts

- Convergence period for fixed ages of 10 years

- Convergence period for fixed year-of-birth cohorts of 20 years

- Applying a two-year step-back from the most recent year of historical data; in the case of Scale MP-2019, the most recent year of historical data is 2017, which along with the above convergence period results in ultimate improvement rates that are fully attained in 2035.

As noted in the Scale MP-2019 report, the SOA believes that Scale MP-2019 produces a reasonable mortality improvement assumption for measuring obligations for most retirement programs in the United States. In addition to Scale MP-2019, other assumption sets could be used with the RPEC_2014_v2019 model to produce a reasonable assumption for projecting future mortality improvement.

Since 2010, US mortality improvement rates have seen a significant decline, resulting in year-over-year volatility in the measurement of pension obligations. In 2018, RPEC began producing an additional model, denoted "RPEC_O2", that uses a similar methodology but with the goal of decreasing this volatility by reflecting a smoother graduation based on second finite differences ("order-2 graduation"). Scale O2-2019 is the corresponding scale using the same RPEC-selected assumptions described above for MP-2019. Scale O2-2019 and the RPEC_O2_v2019 model could also be reasonable for projecting future mortality improvement. A comparison of Scale MP-2019 and Scale O2-2019 is shown in subsection 3.3 of the Mortality Improvement Scale MP-2019 Report.

The Social Security Administration reflects its own mortality improvement scale for the 2019 OASDI Trustees Report. The current SSA scale generally assumes lower long-term improvement rates than Scale MP-2019, but higher short-term improvement rates. A comparison of Scale MP-2019 and the SSA scale is shown in subsection 4.2.2 of the Mortality Improvement Scale MP-2019 Report.

RPEC's MP scales rely upon the same historical data as that used by the SSA. One principal methodology difference between the two mortality improvement assumptions is that the MP scales reflect mortality trends by year-of-birth cohort. The Committee's review of historical experience shows significant mortality patterns by year-of-birth cohort, and RPEC incorporates these patterns in projecting mortality improvement rates.

7 Conclusion

The SOA recognizes there are other factors Treasury may consider in deciding if, when and how to update the applicable mortality basis. Some of the considerations above may require more time to implement than is feasible for updating Sec. 430 mortality tables for 2022 and, if the Treasury wishes to incorporate them, could be reflected over a period of multiple years.

If you think it would be helpful, representatives from the SOA and RPEC would be available to discuss any of the items discussed in this letter - or any related topics - with you in more detail.

* * *

1/ RPEC classified a plan as blue collar if 70% or more of its participants are union or paid hourly. The Committee classified a plan as white collar if 70% or more of its participants are both salaried and non-union. Plans which satisfied neither of the above conditions were considered mixed collar. These thresholds have remained unchanged since the RP-2000 study.

The notice can be viewed at: https://www.regulations.gov/document?D=IRS-2019-0053-0001

TARGETED NEWS SERVICE (founded 2004) features non-partisan 'edited journalism' news briefs and information for news organizations, public policy groups and individuals; as well as 'gathered' public policy information, including news releases, reports, speeches. For more information contact MYRON STRUCK, editor, [email protected], Springfield, Virginia; 703/304-1897; https://targetednews.com