AndyH

Mike, under what scenario might you suggest that? Where you are granting past service? Not accruing on participation I assume?

Dittos. I was wondering the same thing.

There isn't much that I know of. The best source that I found was in ASPA's C-4 exam study material. The reading compendium is called "Current Topics for the Retirement Plan Consultant". The article that was helpful to me started on page 579 of the Fifth edition. It is called "Planning Opportunities to Maximize Benefits for Key Employees Under Qualified Defined Benefit Plans". It is written by Maria Sarli and Dennis Coleman of Kwasha Lipton. It is apparently a reprint of an article first in Tax Management Compensation Planning Journal August 5, 1994. Although it contains a lot of DB stuff, there is a lot of cross testing information including cross testing of DC plans. It goes through the component plan testing process pretty much step by step.

Plus, does the document anywhere say that the lump sum is based upon the immediate annuity? I think most plans calculate the lump sum on the deferred benefit. If somehow there is a reference to calculating a lump sum on an immediate annuity, I would think there would be a second minimum calculation based upon the 417(e) on a deferred basis. Third question, is this benefit at the 415 limit, causing 5% to override the applicable interest rate?

I've run across a lot of those documents that Mike describes in takeover plans. They are typically attempts at prototype cross tested plans, maybe envisioned before the IRS decided not to allow prototype cross tested plans. By now most of them have been re-written, but it is absolutely something to watch for. I would not be surprised if there were volume submitter plans out there that specifically state that the plan is to be tested on a benefits basis. This might preclude the opportunity to test part of the plan on a contributions basis, which is typically done with component plan testing.

No, I've heard and read a number of people raise it as a possible issue, but not the IRS (yet at least). Since they lost the new comparability war, I get the feeling they're hoping for Bush's proposal to wipe out everything but IRAs and (for now) DB plans. Many of these one person allocation groups are set up on behalf of physicians who effectively control the decision of what goes to whom, so the CODA line of reasoning works for me. In fact I don't see a difference for an owner-employee. I happen to work in a relatively high volume atmosphere with clients, contacts, and even employees geographically dispersed, so lower maintenance and low risk are more important to me than they might be elsewhere.

I think the consensus is that it can be done, but has risks and burdens that make it less than advisable. One issue is a possible impact on coverage. If an allocation group receives 0%, is that different than excluding that person by name? That would then raise the question as to whether the eligibility criteria is reasonable. If not, the average benefits percentage test may be unavailable for coverage testing. Some would also question whether the plan could be considered at least in part a CODA. Third issue is just practicality, communication, and efficiency. But, having said all of this, it can be done and I have seen a few of these animals. p.s. Each of these issues is minimized with a group of only HCEs, however.

possible: Yes legal: Maybe. maybe not. wise: Some would say no problem. I would advocate against any explicit use of age in an allocation formula that creates the possibility of someone older receiving less than someone younger based solely on age, even if a specific current census makes that unlikely.

I don't understand the last response, but here's mine: Your client might be best served by keeping the child in an allocation group with NHCEs and then breaking the plan into to component plans. Test one component on a benefits basis and the other, which includes the HCE child, on a contributions basis. To do this, each component must separately satisfy the ratio/percentage coverage test. The HCE-child will have the same allocation percent (e.g. 5%) as the others in his component, so there will only be 1 rate group in that component with lots of NHCEs. If you do a search for component plan testing in this section, there are a number of detailed explanations. Your second option is to put the HCE children in their own allocation group, defined as perhaps owners through attribution. And give them a low enough allocation to allow it to pass. But you should do this only if you are unable to do the first. Your eligibility exclusion idea could perhaps be done but it has some potential problems, so I would advise against it.

You can offer distributions to be "in-kind" in a DB termination just like any other plan that does not have self directed investments. The participants would all need to be offered the same opportunity and would obviously need to elect a lump sum and elect to have part of their distribution in the form of stock (or any other security).

Yes, you are right. A target works just the same way as a simple MP plan in this manner. I don't know a source document cite off hand, but any reference guide (ERISA outline book for example) will have examples of the allocation rules for MP plans and a target is an MP and no different in this respect.

Yes, Blinky, right on point (page 9.85 of the 2002 Book). Thank you both.

I looked back at the language I was reading and I find it vague. I think the prior actuary mis-interpreted the existing language (and the regulations applicable to safe harbor plans). It is a Benetech volume submitter, if anybody out there who cares to comment has one of those, but I have no FDL so I have no way of knowing if it is an approved version.

Now I'm confused. Would somebody kindly state a conclusion for the benefit of the class? Mike, I'm with you except for your #2: "2) If TH, then permissively aggregate another plan and test again. If not TH, all is well." Didn't you conclude that if you aggregate then you need to apply the gateways and test together under a(4)? The above seems to me to contradict what I thought was the conclusion. You're not actually done, i.e. all is NOT well. Right?

what kind of plan is it? If it's a DB plan, then you should consider moving this question to the DB section; otherwise you might want to try Retirement Plans in General unless of couse it is a 401(k) plan in which case "nevermind" comes to mind.

Thanks for the kind words, Lori. I learned a lot of what I know from reading things here and then cross checking that information myself, so it's good to be able to pass it on. If you were to look at my first posts, I've learned a great deal since then, much of it here.

You must calculate and protect the accrued benefit through the date that timely ERISA 204(h) Notice becomes effective. Start with IRC 411(d)(6) and ERISA 204(h), including the regs just issued.

If the plans separately satisfy coverage, the NCT test can be done by ignoring the other plan if you wish. If you choose to do so, however, and any rate group falls below 70%, you must proceed to the Average Benefits Percentage Test, and you must include the other plan in the ABPT if either plan is cross tested for 401(a)(4). For example, if you test a PS plan on a contributions basis (i.e. not cross testing), then for all testing purposes including the ABPT you can ignore the DB. But if you are cross testing the PS (i.e. on a benefits basis), then the DB accruals must be included in the ABPT regardless of whether you choose to aggregate that plan for the NCT test. Regarding software, you only need to calculate the MVAR if the DB is included in the NCT test, which would happen if you are aggregating plans for 401(a)(4) testing purposes. There is no MVAR in the ABPT. And the DB/DC gateway rule only applies if you do choose to aggregate the plans for testing.

Takeover plan (unit accrual) effective in 2000 defines compensation as three year average starting with date of participation. Two principals at the 401(a)(17) limit shown in 2/28/2002 (EOY) val with average comp of $170,000 which would be $166,666.67 if not for the pre-participation comp exclusion. Another person (with 10 YOS) paid out based upon post-participation comp only after two years in the plan. I think that this is not permitted for a safe harbor plan unless each participant has a compensation history at least equal to the number of years in the average. So, in other words, pre-participation comp can be excluded once a participant has three or more years of participation. Cite is 1.401(a)(4)-3(e)(2)(i) "be no shorter than the averaging period" This is less than 100% clear, however. Opinions? The plan has no FDL.

You are of course right. Scary stuff out there.

Blinky, I agree with you. My second paragraph makes no sense; consider it withdrawn. TGIF.

Merlin, I would think that the deferrals are irrelevant other than to generate the need for the top heavy contribution. They are mandatorily disaggregated for 401(a)(4) testing purposes. The gateway would be based upon pay as defined for nec purposes, although to be safe I would make sure it is at least 1% of pay for the entire year (i.e. 1/3 of 3% assuming 3% is the top heavy minimum).

Understood. We both know that you aren't exactly going out on a limb agreeing with Mike! And I did think that you and I understood these rules the same, so we'll blame the Schedule C'ers.

Blinky, we have set some up with one participant per group and also taken some over, but I discourage this design. I ran across a ALI-ABA/IRS Q&A a couple of months ago that basically says that an argument can be made that a one person group who receives 0 has the same effect as excluding the person in that group by name, which is not a reasonable classification for 410(b) purposes. This, as the argument goes, makes the average benefits percentage test unavailable for coverage testing purposes. This is not normally a problem because the 0 is usually an HCE but a circumstance can arise (assume a new doctor for example) where that person could be an NHCE in the first year with potential problems resulting. This plus the potential for a CODA argument makes me leary of this. And another problem is communication, both client/employees and TPA/client. We've seen classes by dates of birth, e.g. physicians born on ______. Strange but true. But, having said that, we will permit this in very limited circumstances. And we take over a lot of plans with one person groups. I attended an ASPA session a couple of years ago when a Milliman actuary (well known but his name escapes me at the moment) says he once designed a plan for 300 employees with each employee in his/her own group, and he said it is perfectly acceptable. But he said it was administratively impractical and would never do it again!

Keith, it looks like you have to go into your Control Panel email setting and re-check "Enable 'Email Notification' by default?" to YES to get notifications. The same thing was happening to me and your comment caused me to go look at that.