2008 lump sum

[Effen]

I'm curious what others are doing about partial ages under the new lump sum methodology.

Are most taking the "easy" way and interpolating between ages or are "right" way using the 1440 line approach mentioned by timesup in a prior post

[Mike Preston]

I am interpolating to months (either rounded or completed, depending on the history of the plan). I have found that the difference between the 1440 approach and the interpolation results are less than 0.07% (sometimes a lot less) and therefore an error factor (if you want to call it that) of less than $70 on a $100,000 distribution is something I can live with.

I also think the interpolation factors will always result in higher amounts, not lower amounts, so on anything other than 415 limit issues, it is something I would think should be acceptable. If at the 415 limit, I expect that some will be attached to plans that are being terminated and for those, it is expected (although I get not universal), that an IRS submission will be made and the methodology therefore made clear to the IRS before approval.

[Andy the Actuary]

I am using the "right" way but frankly do not believe when it comes to an estimate that the "right" way is any more right than the "wrong" way is wrong! There were actuaries employing alternate computational approaches long before PPA (Putrid Pension Act) was born.

[zimbo]

I am leaning to interpolating between the results determined by using whole ages. I have also found error rates less than 1%.

Curious what the software programs such as Datair or Relius will be doing with their benefit calculations.

[tymesup]

I once had a plan with a normal form of a J&S with a cash refund. The brute force technique used a two-dimensional spreadsheet for the J and the S with the different survival orders. I'd hate to do that one again with segmented rates.

[ak2ary]

When doing things the "right" way, are people accounting for the fact that mortality is not level throughout a year that, in fact mortality at age 62 and 11 months is greater than at 62 and 1 month? Because, it seems to me that, if mortality is not being somehow weighted toward the end of the year, the right way is no more right than interpolating. Seems to me that interpolating is both easier and just as accurate (even if less precise)

[SoCalActuary]

At your challenge, I went back to my "very old" 1970 textbook on finite differences to create an Lx table for monthly payments.

It does weigh the end of year payments to reflect a mortality curve as suggested.

For you old actuaries, I used 4th difference methods to create the curve on a look-forward basis, not a central point basis, where five Lx values are used.

Email me to request a copy.

[ak2ary]

and I would then agree that you would get closer to a "right" answer than simply applying the annual q prorata across the year...I just wondered how many of those doing it right, were making any adjustment to the annual q

[Andy the Actuary]

Yes, spreadsheet adjusts mortality discount by month

[tymesup]

The current syllabus for SOA Exam M lists uniform distribution, constant force of mortality and hyperbolic assumption. UDD was certainly on the May 2006 exam; I didn't pay any attention to hyperbolic and don't remember it being on the exam.

There won't be many lawyers or judges who will be interested in this. Heck, I bet nobody's written any comments to IRS asking for a methodology.