• Global Home
• Subscriptions/Research Downloads
• Contact Us
• Investor Relations
• Careers

How Would the Proposed Changes to Mortality Standards Affect Plan Sponsors?

Pending pension reform would require defined benefit plan sponsors to change the mortality assumptions they use to value their plan liabilities, so Watson Wyatt analyzed the impact of the proposed changes. We find that the new mortality assumptions would increase liabilities somewhat for most plan sponsors, especially for plans in which men outnumber women. But as auditors demand greater justification of mortality assumptions, these more realistic mortality standards would likely enhance credibility with analysts and auditors.

Both the Pension Security and Transparency Act (PSTA), recently passed by the Senate, and the Pension Protection Act (PPA), currently moving through the House, would impose a new standard for mortality assumptions used to value pension liabilities for funding purposes.

Actuaries currently must use the 1983 Group Annuity Mortality table, otherwise known as GAM83. The Society of Actuaries developed GAM83 based on mortality data from the 1960s that were then projected to 1983. A Watson Wyatt survey found that in 2004, 70 percent of firms surveyed used GAM83 to determine both their projected and current liability.¹

The pending legislation would require pension sponsors to use a more current mortality table called the Retired Pensioners Mortality Table² (RP-2000), also developed by the Society of Actuaries. RP-2000 is based on mortality experience from 1990 to 1994, which is then projected to 2000. Although the bill's wording is not perfectly clear, it appears that plan sponsors would use RP-2000 projected forward by a mortality improvement factor, in this case Scale AA, to the year 2006, with no anticipated mortality improvements thereafter.

The Treasury secretary would have to revise the table every 10 years. The legislation would also give firms the flexibility to adjust the assumptions if they could demonstrate that their mortality experience differed from RP-2000.

Getting mortality right is an important issue facing the actuarial profession. Pension sponsors rely on their actuaries to be the experts on mortality and are under increasing pressure from auditors to justify their mortality assumptions for pension accounting purposes. The proposed changes to mortality standards reflect today's longer life expectancies. In addition, actuaries need to consider the impact of further longevity improvements, and indeed, actuarial standards of practice governing mortality assumptions state: "The Actuary should consider factors such as…the likelihood and extent of mortality improvements into the future."³

The GAM83 tables were developed when a committee of the Society of Actuaries found that mortality had decreased significantly from the 1971 group annuity mortality table — the previous standard. The Society of Actuaries developed the RP-2000 tables to keep up with life expectancy trends that have evolved since then. The 1983 and 1971 tables are "static," meaning they do not assume future mortality improvements.

Mortality tables that take projected future mortality improvements into account are called generational mortality tables. There are two standard methods for accommodating evolving longevity trends. One is projecting a mortality table forward to a specific year by applying a mortality improvement factor — the current legislation specifies Scale AA. The other is using a generational mortality table that employs more dynamic projections of mortality improvements for each cohort. The Society of Actuaries recommends basing all projections of mortality improvement trends on Social Security and Civil Service participants' experience from 1977 to 1993.⁴ Any mortality table, including GAM83 and RP-2000, can be applied in either a static or a dynamic fashion.

Figure 1 shows life expectancy at age 65 for different mortality tables. For men, life expectancy gets progressively higher between GAM83, RP-2000 with no projections, RP-2000 projected to 2006 and RP-2000 with generational projections. For women, however, GAM83 projects longer life expectancies than the three other measures. This is primarily because GAM83 has a 10 percent load factor, or reduction in mortality probabilities, to reflect mortality improvements, while RP-2000 has no such load factor. Nonetheless, the load factor applies to both sexes; hence the data indicate a narrowing mortality gap between men and women.

By raising the estimated cost of annuities, higher life expectancies create higher pension liabilities. Given that the new mortality table extends the life expectancies for men but shortens them for women, the net effect on pension liabilities will depend on the gender makeup of the plan's population.

To measure the effect of different mortality standards on pension sponsors, we applied the various standards to two actual plans. We selected plans with different ratios of men to women as well as different durations of plan liabilities. Plan One is smaller and has a notably higher ratio of active participants, thus a longer duration of liabilities, and a significantly higher percentage of women (Table 1).

Figure 2 shows what happens to liabilities as the two plans move from GAM83 to the RP-2000 mortality table, from GAM83 to RP-2000 projected to 2006, and from GAM83 to RP-2000 generational projection. For Plan One, changing from GAM83 to RP-2000 without projections actually lowers pension liabilities, owing to the lower female life expectancy that results from removing the 10 percent load factor included in GAM83.

Moving from GAM83 to RP-2000 projected to 2006 — as called for by pending legislation — raises Plan One's liabilities by .5 percent and Plan Two's liabilities by 1.75 percent.

Figure 3 shows the percentage difference in life expectancies between GAM83 and the RP-2000 standards for men, and Figure 4 shows comparable results for women. Numbers greater than zero indicate higher life expectancy — and hence higher liabilities — under the RP-2000 standard as compared to the GAM83 standard. While RP-2000 projected to 2006 does project lower life expectancies for women at all ages through 93, it also predicts significantly longer lives for men through age 86. The resulting higher liabilities for men swamp the lower liabilities for women, even for Plan One. Moving to RP-2000 generational projections increases liabilities by over 2.5 percent for both plans.

The bottom line for plan sponsors is that changes to mortality standards would likely increase pension plan liabilities, particularly for plans with many more men than women. However, the higher costs of increased liabilities for some sponsors would be offset by the benefits of improved credibility with analysts and auditors.