Late retirement - variable rates

[lees]

• I have a plan that currently uses PPA rates for  late retirement actuarial increases.  No SOB notices, and the retirement benefit is the greater of the Ab with actuarial increases vs continued accruals.  The actuarial increase factors are based on rates as of BCD, rather than each 12/31 (it's debatable whether this is OK, but that's not where my questions lies).

Assume interest rates at BCD = 4%, 5% and 6% for the 3 segments.  Assume age 65 at NRD, and a.e. updates will occur at integral ages in the future (66, 67, etc.).   Also assume that the AB as of NRD, with actuarial increases,  will ultimately be the final winning benefit.  My question is how to determine the segment periods.  Depending on how you typically set up your late retirement comparison, you would get different results.

The 2 common ways to handle this type of calculation would be:

1. Do the comparison each 12/31 between the prior benefit with actuarial increase from prior date to current 12/31, and the AB determined as of current 12/31.  This comparison is repeated until BCD (format of the examples in the 1988 Proposed Regulation). 

2.  Determine the ABs (based on service, earnings, etc.) as of NRD and each 12/31, and then actuarially increase each of these values from the AB determination date all the way to BCD.  Then the greatest of all the Abs with applicable actuarial increases is determined as the final benefit at BCD.  This format is what is illustrated in Gray book examples 200-34 and 2007-17.

Either way, you typically (should) get the same answer.  Even when rates vary, you get the same answer as long as you vary the rates for the appropriate periods in Method 2.

However, with the PPA segment rates, there can be a difference (due to interpretation)

1.  In Method 1:  For the actuarial increase from 65 to 66, benefit payments for the NRD benefit are discounted from age 65 to 70 at 4%, 70 to 85 at 5%, and greater than age 85 at 6%.  However, when the next actuarial increase is applied from 66 to 67, the segment periods are being reset to 5, next 15, 20+.  So now the benefit payments based on the NRD AB from age 70 to 71 are discounted at 5% instead of the original 4%.  This anomaly repeats itself each time there is an update/comparison made, so that effectively the 1^st segment rate ends up applying to the AB determined at NRD for more than just the 1^st 5 years after NRD. 

2. In Method 2:  Since the actuarial increase is applied from AB Determination Date to BCD, the segment periods never change over time.  So the discount rate for benefits payable from age 70 to age 71, no matter what

Does anybody have an opinion or reference with regards to this issue?  Personally I feel that the result in #2 is correct (Method #1 can still be used as long as the segment periods are determined as of NRD).  Logically it doesn’t make sense to vary the segment periods.  This could result in a termination prior to age 65 (who defers benefit beyond NRD) getting a different benefit than someone who works past NRD, but doesn’t accrue a significant benefit after NRD, just because a comparison needs to be done to ensure that continued accruals don’t exceed the NR benefit.

If you feel the segment rates shouldn’t get reset for the Abs based on prior date (NRD in example above), should they still get reset for Abs determined at later dates? That is, for the AB determined at age 66, should the segment periods start at age 66, or should they start at age NRD?

 

 

 

[Calavera]

Just want to add more complexity to your way of thinking. What about mortality tables? Are you using the applicable mortality and therefore the table is different from year to year?