American Academy of Actuaries Issues Public Comment on IRS Notice

WASHINGTON, March 17 -- Elena Black, vice chairperson of Pension Committee at the American Academy of Actuaries, has issued a public comment on the Internal Revenue Service's notice entitled "Mortality tables for 2021 Notice". The comment was written on March 9, 2020, and posted on Feb. 24, 2020:

* * *

On behalf of the American Academy of Actuaries/1 Pension Committee, I write to provide the Department of the Treasury ("the Treasury") and Internal Revenue Service ("the IRS") comments on topics described in the Request for Comments in the Notice 2019-67.

In what follows, "RPEC" means the Society of Actuaries' Retirement Plan Experience Committee. "RP-2014" refers to RPEC's mortality experience study published in October 2014, and "Pri-2012" refers to RPEC's mortality experience study published in October 2019.

Baseline mortality tables for purpose of Internal Revenue Code (IRC) Section 430

The standard base mortality tables under Sec. 430(h)(3)(A) are required to be revised at least every 10 years/2 to reflect the actual mortality experience of pension plans and projected trends in that experience. These mortality tables are intended for measurement of pension liabilities of private pension plans in the United States. As such, the most relevant experience is that of the U.S. private-sector pension plans. Currently, these standard funding tables are based on the RP-2014 study (baseline underlying table RP-2006, which was separately published by RPEC in 2018).

These tables were derived from the U.S. private pension plan data during calendar years 2004- 2008.

The most recent RPEC mortality study--Pri-2012, published on Oct. 23, 2019--contains mortality tables reflecting experience of U.S. private pension plans during calendar years 2010- 2014. As such, it is logical to consider this new information to update relevant experience reflected in baseline mortality tables. We note that the volume of experience data for this study was almost 50% higher than in the RP-2014 study in terms of exposures (over 16 million versus 10.5 million exposures).

Some have noted that for some populations, the Pub-2010 study based on U.S. public-sector data produced mortality tables similar to certain tables in Pri-2012 study. We note, however, that for the total population covered by the Pub-2010 study, there are differences, including in the timing of exposures. Pub-2010 consists of families of tables separated by certain occupational categories. Reflecting this information in any manner for purposes of creating a table for IRC Section 430 would require some decision on what blend of these populations would be most appropriate for private-sector pension system as a whole.

In addition to covering a more relevant population (the total data set appears to reflect a reasonably representative blend of private-sector pension plans), Pri-2012 also spans a more recent experience period than Pub-2010. We believe that all of these considerations make it unnecessary to reflect additional experience captured in Pub-2010 for purposes of standard IRC Section 430 funding tables.

Because the current tables were revised to reflect RP-2006 and the related mortality improvement scales in 2018, any near-term action to reflect the new Pri-2012 tables should be well within the required 10-year time frame.

Contingent annuitants' mortality rates

For the first time, RPEC derived separate mortality tables for contingent annuitants. Data for contingent beneficiaries was tracked in previous studies (RP-2000 and RP-2014) but, at the time RPEC decided not to publish separate mortality rates for this group. Rather, it instead combined this experience with that of retirees to develop blended rates--similar to what was done in developing the PRI-2012 non-disabled annuitant tables. Below we offer a few comments related to these new tables and their potential use.

* It is not clear why, after declining to publish separate contingent survivor rates in previous studies, RPEC decided this time to place such a heavy emphasis on the contingent survivor rates. As discussed below, we are not convinced that the resulting liabilities will be sufficiently different from the results that would be obtained by using the blended non-disabled annuitant tables.

* As discussed in RPEC's report, the exposures were tracked only after the primary annuitant's death. This makes the observed mortality rates (assuming statistically credible data) contingent on another death event that could be highly correlated with the beneficiary's mortality. Exposures for the contingent survivor population are disproportionately represented by those with spouses who died at younger ages (because those with spouses who die at older ages don't make it into the population until later in life). As the couple likely shared similar habits and lifestyle, one might expect that the surviving spouse would have higher mortality than the general population (even after allowing for any temporary "grieving widow/widower" effect). Thus, tracking exposures only after primary annuitant deaths introduces selection bias to this population.

Therefore, although these tables might be appropriate for the period following the death of the primary participant, there is no support for applying these tables prior to that point.

* With regard to measuring plan liabilities for participants currently in receipt of Joint and Survivor (J&S) annuities, as well as for those assumed to elect a J&S in the future, or otherwise receive a future annuity upon the death of the participant, it is possible to apply the contingent annuitants' mortality rates after (future) primary annuitant death. This corresponds to approach 2 in Section 12.4 of the Pri-2012 Report. This methodology calls for different mortality rates for the contingent annuitant depending on whether the primary annuitant is alive or not. While this approach might have rationale and appeal on a theoretical basis, few valuation systems are currently equipped to handle this additional complexity. While systems could likely be adapted if this approach were required, it would represent a fundamental change in how liabilities are traditionally valued, with annuity factors representing the sum of a two-dimensional array of future outcomes (based on age and age of death of the primary annuitant) rather than the current approach, which sums a one-dimensional array (age only). As discussed below, we question whether the potential effect on costs would warrant the additional complexity.

* In prior studies, RPEC measured healthy annuitant rates based on the entire annuitant dataset (including contingent survivors), and these rates were utilized for all healthy annuitants. In Pri-2012 study, RPEC also published "healthy annuitant" rates by blending retiree and contingent annuitant rates weighted by exposures. This suggests an approach, not described in Pri-2012 Report, of using these blended rates to value J&S recipients' liability, an approach that would be consistent with prior and current practices.

* We evaluated the impact on pension liabilities for J&S recipients by comparing J&S annuities measured under two approaches. In both approaches, we use retiree mortality for the primary annuitant. For the contingent annuitant, the first approach uses retiree mortality prior to the death of the primary annuitant and contingent survivor mortality following the death of the primary annuitant, while the second approach uses the blended non-disabled annuitant mortality both prior to and following the death of the primary annuitant. For male/female (primary/contingent) couple receiving 100% J&S, the difference between these two approaches (assessed by comparing annuity factors derived under each approach at representative ages) is small--particularly considering that the weighted average J&S percentage for a given pension population is going to be much lower than 100%. For a female (primary)/male (secondary) couple, the impact is slightly higher, but still very modest. Considering that males are more likely to elect J&S annuities and weighting for typical demographic profile of J&S recipients' group, the overall impact on liabilities between these two approaches is a fraction of a percent. We would be happy to share additional supporting detail if that would be relevant to your deliberations.

The Pension Committee appreciates the opportunity to comment on this matter. Please contact Philip Maguire, the Academy's pension policy analyst ([email protected], 202-223-7868), if you have any questions or would like to arrange a convenient time to discuss these comments further.

Sincerely,

Elena Black, MAAA, FSA, EA, FCA

Vice Chairperson, Pension Committee

American Academy of Actuaries

* * *

Footnotes:

1/ The American Academy of Actuaries is a 19,500-member professional association whose mission is to serve the public and the U.S. actuarial profession. The Academy assists public policymakers on all levels by providing leadership, objective expertise, and actuarial advice on risk and financial security issues. The Academy also sets qualification, practice, and professionalism standards for actuaries in the United States.

2/ Sec. 431(c)(6)(D)(iv) requires a review at least every five years for the tables used to calculate current liability for multiemployer plans, but only requires an update of that table to the extent that the secretary determines it is necessary. Given that current liability does not have a significant effect on multiemployer funding, we have assumed that the 10-year time frame specified in Sec. 430 is more critical to driving updates in IRS-prescribed tables.

* * *

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