Reasonableness of Mortality Assumptions

[Guest mbklein]

In this still pre-GATT era, does anyone have any idea whether a 1963 George Buck mortality table is "reasonable" in the context of a standard termination? Would you clearly get a notice of sufficiency from the PBGC if you employed that table? It is the table employed to calculate lump sum amounts under the current plan document.

[david rigby]

I don't know that particular table, but the test of "reasonableness" is the result. That is, are the results you get from using it (such as normal cost, 412 contribution, 404 contribution, etc.) significantly different from the results you would get by using a mortality table that is generally considered "reasonable"? (Yes "significant" is also a word subject to interpretation.)

The answer you get to that question depends on several factors, such as how well funded a plan is, the funding method, the other actuarial assumptions, the plan design, and (especially) the demographics of the situation.

A good way to start this analysis is to look at the ratio of annuity values at two different ages: immediate annuity at retirement age, and a deferred annuity at the average current age. The ratio is defined as the annuity of the table in question, divided by the annuity of the 1983 GAM, of course using the same interest rate. Look at this ratio for males and females. Then use common sense to evaluate how significant the difference is.

Example, the ratio of the annuity value at age 65 between the 1983 table and the 1994 table is 1.062 for males and 1.008 for females. Thus, the significance of the 1994 table is much less for a female population than for a male population.

A warning: if you are asking the question because you are trying to "manage" the resulting contribution, you should also use common sense as to the appropriateness of that action, especially if the ratios discussed above are more than 10% off (that is, less than 90% or more than 110%). That 10% differential is my personal threshold, although I prefer a 5% or less; not a magic number.

Another important point is that the entire package of actuarial assumptions is being evaluated for reasonableness, not necessarily only one. If you have a set of assumptions which is "individually reasonable" except for the mortality assumption, then look at the output not the input.

Barnet Berin wrote a pretty good book on actuarial funding methods a few years ago, "The Fundamentals of Pension Mathematics", published by the Society of Actuaries in 1989. The first question at the end of the first chapter is extremely interesting, where he ignores all assumptions (such as interest rate, turnover, mortality, etc.) and proves that ALL funding methods produce the same contribution for a given set of data. Thus, it is the application of the various actuarial assumptions that produce variations on contribution levels. The funding methods merely modify the incidence of the annual costs.

This message has been edited by pax (edited 04-06-99).]

[This message has been edited by pax (edited 04-07-99).]