General Testing of Aggregated DB/DC Plans.

[flosfur]

Mutli-part question.

Non-safe Harbor DC plan.

1) Has anyone ever used an Accrued-To-Date method for a DC plan?

If so, how are the Accrual Rates determined?

(i) Each participant's accrued benefit is determined using the Current Account Balance Or the Total Employer Contributions allocated to date (ie. ignoring gains and losses to date)?

(ii) Testing Service is the Service from the date participation eligibility (age 21/1 yr service) was met - regs 1.401(a)(4)-11(d)(3)(i)(B).

(iii) If the "Years during which a participant benefited" is used as the Testing Service, does this include only the Years during which the participant actually received an allocation of Employer Contribution or All Participation Service whether or not an allocation was received during a year (eg. the employer skipped contribution to its PSP for some years)?

DB/DC Combo (not an Offset arrangement).

Situation: DB/DC plans are aggregated to pass the 410(b) coverage test and hence must be aggregated to test for 401(a)(4).

For 401(a)(4), must both plan use the same method to determine the accrual rates i.e. for both plans, use either the Annual method or the Accrued-To-Date method?

[Mike Preston]

1) Yes.

(i) Current Account Balances

(iii) Only those years where the participant benefitted within the meaning of 410(b). Even if no contribution was made, forfeitures might still have been allocated.

Yes, the aggregated DB/DC plan must be tested on a single basis, either rates or accruals.

[Guest merlin]

An additional question. How are the aggregate allocation/accrual rates cal'd when the plans have different definitions of compensation? Specifically, the dc plan uses the increased EGTRRA comp limit, but the db plan does not. Is each component of the aggregate rate calc'd using its own particular comp?

[AndyH]

Flosfur, just to clarify, the DB and DC aggregated rates must be done the same way, i.e. benfits or contributions basis. But, the plan can be broken into components, with different component plans tested differently.

For example, the plan could be broken into components with one tested on the annual method and one on the accrued to date method.

And, one component can be tested on benefits, and another on contributions basis. I think what Mike is saying is that the DB and DC aspects within these components must be tested the same way.

Merlin, I'm not sure of the answer to your question, but I would think you could use the higher limit for both, or different limits for each.

[Mike Preston]

Andy, you nailed my intent. Sorry if it wasn't clear.

Merlin, compensation used in testing is independent of compensation used in the plan. How you come up with your benefits in each plan is almost irrelevant. Once you have them, though, you test against any 414(s) definition of compensation you want to use. What the db plan does doesn't really matter because 414(s) won't allow you to eliminate the increased dollar limits.

[flosfur]

1) Yes.

.......

Yes, the aggregated DB/DC plan must be tested on a single basis, either rates or accruals.

But here is the problem (or may be it is not a problem?):

Under the Accrued-To-Date method the testing service for the DC plan must be the "Benefiting Service".

Whereas for the DB plan, the testing service can be the service from the date eligibility was met - regs 1.401(a)(4)-11(d)(3)(i)(B) (I know some on this Board don't agree with this). If this testing service was not permitted then the DB plan on its own or the aggregated DC/DB plan will not pass 401(a)(4) unless HCEs' benefits are reduced or NHCs' benefits are increased to the point which makes the DB plan unattractive to the employer.

So is this approach considered using the same method?

Needless to say, the plans will be submitted to the IRS for a DL but it would be better to know the answer before finalizing the design and implementation of the plans.

[Mike Preston]

I'm not sure I understand what the options are that you are choosing between.

I believe, without looking it up, that the benefits are aggregated and then you divide by the appropriate denominator. I don't think you aggregate the fractions.

Then again, my Gust deadline is tomorrow so I'm pretty much useless at the moment, as my head is into the plans on my desk.

If my memory is serving me, I would convert the DC to benefits and then add the benefits from the DB and then finally divide by service.

What this does is elongate the divisor with respect to the benefits accrued in the plan with less service.

I can't see how it is done any other way, especially once you consider imputation of social security.

[flosfur]

....But, the plan can be broken into components, with different component plans tested differently.

For example, the plan could be broken into components with one tested on the annual method and one on the accrued to date method.

......

Componing (my terminology) a plan is fine in theory and may be easily accomplished in large plans but for the small plans is not a practical proposition because of the sponsors' unwillingness to pay for the time and effort involved.

Unless one is doing this day in and day out, one can spend eternity on just one plan trying to figure out the right components and still come up with no solution.

[Mike Preston]

Very true. But there are some general guidelines. Componing (!) works best when you can eyeball the EBAR's/Allocation rates and see where an HCE can be paired with an NHCE group (what Larry Deutsch refers to as the "magic number") of NHCE's to build a foundation for that HCE.

From there it really isn't that tough.

There are so many options, though, that you are correct that one can spend an eternity if one chooses to do so.

[AndyH]

flosfur, I do it all the time with DC plans (but only when they would otherwise fail) and a couple of times a year with DB/DC plans (which have fees to sustain the testing), and at first I found it to be difficult.

But now I find it quite easy, having done quite a bit of it. And I find DB/DC combo testing to be much easier now with the 7.50% (less the PV of the average NHCE DB accrual) gateway to work with.

[flosfur]

I'm not sure I understand what the options are that you are choosing between.

I believe, without looking it up, that the benefits are aggregated and then you divide by the appropriate denominator.  I don't think you aggregate the fractions.

I'wll try again:

In summary Reg 1.401(a)-8(b)(2) says- after having computed the Equiv. annual benefit, the rate is computed by dividing the Equiv benefit by the number of years in which the employee benefited under the plan during the measurement period.

Let's take the simple case where no employee is participating in both the DC and the DB plan.

Per the regs, in applying the accrued-to-date method to the DC plan, one is using (as one must) the benifiting (participation) service to determine Equiv "DB" accrual rate.

Whereas in applying the accrued-to-date method to the DB plan, one can (and one would want to) use employment service from the date the employee met the participation requirement (i.e. from the date he could have entered the plan, had the plan been effect at that time).

Obviously, for a DB plan in which benefits are based on participation service, the accrual rates using the testing service which includes pre-participation service produces lower accrual rates Vs. using the participation (benefiting) service only.

So in the above, is the "same" method being used for computing accrual rates in both plans?

--------------------

As an aside, I just remembered, when testing an aggregated DB/DC plan in one of the commercial softwares I worked with, one simply entered the DC account balances in certian fields without entering anywhere the benefiting service in the DC plan. So the DC's Equiv rates ended up being computed using the same testing service as the service used for DB plan - doesn't look right!

[AndyH]

I don't see anything that requires that the denominator of the fraction be the same for for DB and DC plans. I think the combined rate is the sum of separate calculations.

What would you do, for example, if one plan type were established years later than the other?

And I don't know of any justification for using service for the DB fraction if the formula and accrual method are based upon participation. They must relate to one another.

[Mike Preston]

I'm not sure I agree as to the denominator. Once you are using the accrued-to-date methodology, it is predicated on, well, what you have accrued to date. The divisor should be the period over which those benefits have accrued. Of course, that is just an English derivation of how I thought the rules were supposed to operate. Add to that the fact that I think adding up fractions runs the risk of using double integration and I have always added benefits before performing a single division. I have LOD's on the methodology so it seems to be an acceptable method. Of course, that doesn't mean much.

[AndyH]

Mike, just for my sanity, when you get a minute, would you take a look at 1.401(a)(4)-9(b)(2)(ii)(A), which in part reads

"...by treating all dc plans ...as a single plan; and all db plans ...as a separate single plan; and determining an allocation rate and equivalent normal and most valuable allocation rates under sections ...1.401(a)(4)-2©(2) and 1.401(a)(4)-8©(2), respectively. The employee's aggregate normal allocation rate is the sum of the employee's allocation rate and equivalent normal allocation rate determined in this manner ....."

I cannot see how this says anything but determine the rates under the separate db and cross tested dc rules and then add the percentages together.

But throughout the regs there are references to determining rates "in a reasonable manner" and I would think that your method meets the reasonable manner rule but does not exactly match the section I cited.

And flosfur, going back to your original cite of 1.401(a)(4)-11(d)(3)(i)(B) , I think this simply says that a plan can credit pre-participation service. I don't think this in any way says you can use pre-participation service as testing service. Elsewhere it says that you can but only if such service is considered under the plan's benefit formula.

[Mike Preston]

That is certainly what it says. Note that each section has a reference back to imputation of permitted disparity. Taken literally, it might appear that multiple imputations are permitted. I don't think that is possible.

However, I agree with you that it appears the fractions should be added, rather than the numerators added.

Kind of makes you wonder why they put it in that way.

But ours is not to reason, I guess.

[Blinky the 3-eyed Fish]

Andy, we have been down this road before regarding the testing service.

http://benefitslink.com/boards/index.php?s...=22649&hl=point

[AndyH]

Thanks, Blinky. I was wondering when you would chip in. Now I know why you didn't. Deja vu all over again.

And, Mike, thanks as always to you. My sanity has been recovered for the next couple of hours.

[flosfur]

...

What would you do, for example, if one plan type were established years later than the other?

I don't mean the "same years" of service but rather the "same type" of service, e.g. benefiting service or any other permitted type of service.

.....

And I don't know of any justification for using service for the DB fraction if the formula and accrual method are based upon participation.  They must relate to one another.

Well, the jury is still out on that - see the Testing Service thread I started. I can give the philosophical justification for it but that is not at issue here.

As I have said before, if the pre-plan service could not be taken into account for Testing Service, then I cannot see how those 250% for the Owner and $100 for the Other type of plan designs presented at the conferences and seminars could pass the 401(a)(4)?

[flosfur]

Andy, we have been down this road before regarding the testing service.

http://benefitslink.com/boards/index.php?s...=22649&hl=point

It is not the same road. On that thread we were travelling down one road, namely non-aggreagted DB plan only and the testing service that can be used.

Here we have two roads merging into one, DB & DC plans aggregated - so the issue is not the same.

As an analogy:

For the DB & DC separately, one can use the annual method or accrued to date method (as well as the projected mehotd for the DB plan).

But when aggregated, same method should be used for both plans.

[AndyH]

flosfur, I am not familiar with the 250% owner/$100 employee example that you reference. I'd be interested in exploring it if you wish to elaborate. And it would be useful to know who the originator was if you know, as that may lend some insight.