I have a takeover plan which was amended 12/28/94 from a PIA offset formula to a unit benefit formula. The amendment was made effective 1/1/89. Do I have to grandfather the PIA offset AB as of 12/28/94?

Second question: what might one use as assumptions in calculating the PIA in this situation (ex: actual salary history and then a salary scale going back, level future salaries, 3.5% wage base, 3.5% CPI)? I will try to duplicate the prior administrators work, but some background info will help.

Thanks.

It makes sense that the AB would be grandfathered in 1994. Doesn't the prior plan provisions specify how the PIA offset is to be determined? I would follow the plan provisions.

I agree with Gary. In fact, because the plan was amended in any way that might affect the accrued benefit, then you *must* grandfather the amount as of the date of change (that is, the later of the effective date or the adoption date).

Gary is also correct in his reference to plan provisions. If the plan is silent on this (hard to believe it could omit a definition of PIA), then you should probably adopt whatever has been done in the past.

In more than 20 years, I have never seen a definition of PIA that included any increases in future salary of wage base. The likely definitions are to assume level future earnings or zero future earnings.

Absolutely the accrued benefit must be grandfathered. There is a specific prohibition on doing otherwise with a SS offset. It's been too long since I've done one of these to remember the cite, but clearly that answer is yes.

On the backwards salary issue, there a reg or something to the effect (around 1983 I think) that you could use either an index provided by Social Security or the IRS, or you could use a flat salary scale, not less than a certain percent (I believe it was 6%, but it's been a while, so you'll have to research it). Whichever is used, it has/had to be in the document and the participant has to be given the option of providing actual salary history instead. (This assumes a safe harbor plan-if it was general tested there were other things that could have been done).

I also don't think that any assumption of continued wages (or future indexing) is/was allowed.