Accrued to date general testing method- fixes in event of failure

[AndyH]

Assume a DB plan is being tested using the accrued to date method and a simple example of 30 NHCES with a Normal Accrual Rate of 1.1% who must be elevated to 1.3% to pass.

Assume the desired fix is to grant an increase in the accrued benefit of say $100. This would work if only the current year were being considered. But since I am testing on all past years, this gets divided among all years in the testing period and substantially diluted, right? Or is there some way around that?

The alternative might be to test using the annual method, in which case clearly I have only one year to fix. But there are data issues which make using the annnual method undesirable.

Is there any way around this need to fix all years in the testing period if the accrued to date method is used?

The usual tricks such as component testing are not available and banding, imputing, etc. have all been exhausted. I have limited options because my group is a small percentage of the total nonexcludables (controlled group situation with different plan year ends that cannot be permissively aggregated). I would pass average benefits easily but cannot get by the NCT.

I think I'm stuck but am grasping for straws.

[Mike Preston]

NCT?

I think you are stuck, given the parameters you identify.

Remember that you can test on any period, not just "all years of service". Does going back only 2 years work? How about 3?

Why isn't component testing available?

[AndyH]

Thanks for the comments, Mike. Hope you had a good vacation.

The situation I am faced with is extremely difficult and I'm about out of time. There are multiple formulas involved, mutliple early retirement provisions, different groups, etc. It is a spinnoff and new controlled group, and the period I'm dealing with is after the expiration of the disposition/grace period coverage rules.

Due to the complexities and because it is a takeover, I also am faced with limited data. And none of it is easy to recreate because there are combinations of offsets, minimums, excess formulas, etc. I could go back and recalculate lots of stuff, but I'm not sure that it would put me into a better spot. The one thing I know is that the accrued benefits I have are good and I've figured out the derivation of each.

So to avoid muddying the picture I'm trying to keep my questions simple. I already had to bring in employees of a related company to get to the NON safe harbor coverage 20%. So that is where I start, with a ratio percentage of 20%, now I'm testing under (a)(4) retroactively under 11-(g)(iv)(b) (year is 2002 and GUST FDL is pending).

Of course all controlled group members that have plans that could be aggregated have different plan year ends, so I cannot permissively aggregate anything.

I can't component test because I can't split my 20% and have each portion pass coverage.

I think I could pass if I could add a flat dollar add-on or a flat dollar minimum, combined with some other stuff, such as changing an early retirement provision.

But I don't want to go retro with a fix (it would be too expensive). In reading through the fresh start stuff, it looks like I may have some argument for fixing 2002 and not looking back even with the accrued to date method, but with all the cross references it's tough reading.

NCT=Nondiscriminatory Classification Test

p.s. I've heard speakers remark that in general testing it seems strange that the goal is the LESSER of the plan's ratio percent or the midpoint and question HOW the plan's R/P could ever be lower. Well, I found the quiz answer: When the plan's R/P is below the safe harbor %, (i.e. at the unsafe harbor), meaning that you are relying on facts and circumstances to pass 410(b). I got the circumstances.

[AndyH]

Mike, just a followup to your comments. How in a DB general test can you only use 2 or 3 years. Wouldn't the measurement period be limited to only the current year, the current year and all prior years, and the current year and all prior and future years?

It seems to me that these are your options unless you use a fresh start, in which case usually a plan amendment is required to "separate" the old from the new. . Am I mistaken about the measurement period options? Perhaps you were referring to comp averaging?

[Mike Preston]

I was speaking about average compensation. But I should have used numbers of 1, 3, 4, etc. 2 is the only one that isn't available. Of course, if there was an amendment in a prior year such that a fresh start could be claimed, that would work, too.

[AndyH]

Agreed. Thanks for your help.