Aggregated 401a4 Test

[JBones]

If a plan sponsor adopts a new DB plan in 2010 that provides for 5 years past service credit and that the plan is being aggregated with a profit sharing plan for 401a4 using the annual accrual method, are the entire 6 years of DB accruals included in the first year's test or just the 2010 accrual? For example, would a participant benefiting under a 2% per year of credited service formula with 10+ years of service be included in the test with a 12% DB normal accrual rate?

If the previous 5 years are not included in the test, are they required to be tested in any other way?

Thanks in advance.

[AndyH]

My vote would be they all count in that one year. Just one vote, let's hear from others.

[SoCalActuary]

You should also consider the issues of testing on accrued-to-date method.

Discuss this with such experts as Kevin Donovan or Larry Deutsch, or Adrian LaBombarde if he is available.

[John Feldt ERPA CPC QPA]

On a design somewhat similar to this, according my memory of what Larry Deutsch said last month, the IRS declined to reply to some type of an information request (a request that Larry thought the IRS was actually required to provide a response). The request was to determine what can be used in the denominator for the number of years when testing a combo like this using the accrued-to-date method. My notes from Larry' symposium indicate that some firms have done this, applied for full-scope determination letters (showing how it was tested), and that some have received D letters. Larry indicated that the IRS field agents have been told that it is probably okay to do this (I think that means they were told it is okay to divide by a number that does not exceed the number of years in older of the two plan, even though the other plan is not as old). I have not seen a formal IRS response either way on this.

[AndyH]

Useful info but note the original question was about the annual method.

I am surprised to hear that there might be an issue with using a denominator greater than one for a first year accrued to date test that grants past service credit. I don't think the regulation or published guidance is clear enough to prevent that.

[John Feldt ERPA CPC QPA]

Correct. Sorry for throwing this off on a tangent.

The discussion at the symposium seemed to agree with you that the regulations do not appear to prevent that, but what the regulations allow and what the IRS thinks should be allowed sometimes do not align (go figure), which I think may have been the reason for the information request.

[AndyH]

To the original question, see 1.401(a)(4)-3(d)(2) "In the case of a measurement period that is the current plan year, testing service for the plan year equals one (1)."

I don't see any way around that.

To the accrued to date question, see 1.401(a)(4)-3(d)(1)(iv) "Testing service. (A) Testing service means an employee's years of service as defined in the plan for purposes of applying the benefit formula under the plan......."

I don't see any valid argument for denying past service credit to be used as testing service using the accrued to date measurement period unless the service credit is determined to be discriminatory or some other unusual situation covered in 1.401(a)(4)-11(d)(3) applies. Maybe such a situation, imputed service or past service exceeding 5 years for example, was implicit in the Larry Deutsh seminar discussion.

[John Feldt ERPA CPC QPA]

I think it was more about "the plan". If the DC plan is 20 years old and a new DB is added and now tested together, years of service as defined in "the plan" means what? 20 for the DC and 1 for the DB? No, you combine both and divide by the number of years. Some agents would argue that you could not use 20 as your divisor when testing the combo accrued-to-date because the DB only had 1 year. Larry's point was that he did not care either way, he just wanted to know (officially) their stance so he could design plans either way, without having to worry about one IRS office disagreeing with another.