Return to The Consulting Actuary Index

Session 102 - The Generational Gap: Mortality Assumptions in the 21st Century
Presenters: Christopher M. Bone and Emily K. Kessler
Recorder: Michelle Koontz

Actuaries are coming to grips with a consistent pattern of mortality improvements and moving to get ahead of this trend.  Panelists discuss the appropriateness of building mortality improvements into actuarial valuations.  In addition, the mechanics of PPA mortality and custom tables for large plans are covered.

Recent Tables
Before PPA, GAM-83 was the most popular table with 75% of plans using it (according to a 2003 Watson Wyatt study).  Following GAM-83, three new tables reflected pensioner experience, the UP-94, GAM-94, and RP-2000.  UP-94 reflected uninsured pensioner experience projected to 1994 while the GAM-94 was group annuitant experience projected to 1994.  RP-2000 reflects retirement plan experience projected to 2000 and was developed specifically for current liability.

Regulatory Update
The final IRS regulations for the new current liability mortality basis for 2007 funding are based on RP-2000.  The rates were projected on a static basis from 2000 to valuation date and then further projected 7 years for annuitants and 15 years for actives.  A blended table is available for all plans. 

Proposed 2008 mortality assumptions will take on plan specific mortality tables.  These tables must be recent in that the experience study end date must be within 3 years of the proposed date of the first use.  They also must be credible with at least 1,000 deaths over a 2-4 year experience study period.  There are also limits on the ability to “cherry pick” rates.  There would have to be significant evidence to say mortality for one group is different than the general group.   These tables must also be generational and gender specific, but the base rates can be uniquely determined or be a fraction of the standard table.  Mortality table creation has changed since GAM-83 in that now technology permits changes and projections easily.  In the assumption setting process, the final step for the practicing actuary in setting the mortality assumptions is to decide whether to project or not project.  With regards to PPA, the actuary can continue to use plan specific tables only if they continue to accurately predict future mortality.

An SOA study (Kays 2005) found that consistently using mortality tables that were not projected, at least to current effective date, would accumulate assets that are less than ideal.  Ideal was defined to be assets sufficient enough to cover actual mortality improvement.  The results led to the recommendation that mortality assumptions be updated periodically and at least to the valuation date by the appropriate mortality scales.  In other words, if the mortality adjustment is not consistently improved then rates will slide and eventually result in an underfunded plan.

Projections
The most commonly used projection scale is AA.  This scale was constructed with GAM-94 and UP-94 tables.  It was based on mortality trends from CSRS data (1977-1993) and SSA 107 (1977-1988).  Rates were graduated and smoothed.  This scale also reflects how mortality improved for a specific group of individuals which can be attributed to improved health treatments and technology.

Two ways to project mortality can be seen by using a static table versus a generational table.  A static table projects each value by X years to build in X years of mortality improvement at each age.  A generational table is built with unique tables for each birth cohort based on number of years between birth cohort and “base group.”  The effect of using a generational projection is improved mortality, especially among males. 

There still remains a debate on how to treat future mortality trends.  One side suggests that we push survival out to older and older ages (stretching the curve).  The other suggests improving survival at older ages but still set an upper age limit (squaring the curve).  Generally, people are getting extra years of “healthy life” rather than, for example, living in a nursing home.  See the “Living to 100” monograph on the SOA website for more detail.