AndyH

Merlin, that sounds like a very odd situation. You aren't counting income attributable to the balance existing at the measurement start date, right? See 1.401(a)(4)-2©(2)(iii) if you are.

Well, we can't see Blinky unless he divulges his true identity, or unless he actually looks like that!

Thanks, Mike. Good to hear from you.

Well, certainly the incorrect payments need to be corrected through reimbursement of one plan to another, and then I would think you need to put the plan which incorrectly made the payments in the position it would have been in had the error not ocurred (i.e. income on the money incorrectly paid). I'd probably be inclinded to tell the sponsor to reimburse the income in an amount equal to the greater of what the money would have generated in EITHER plan, although you could run afoul of deduction rules if that were deemed a contribution and not the correction of a fiduciary breach. But if the income would not amount to a lot of money that is what I'd do anyway. And how much income would have been generated in the last couple of years in this market? So maybe there is no lost income. Just some gut reactions; not legal advice. The correct answer is always to run this sort of thing by an ERISA attorney, of which there are many on these boards.

More details are needed. What is the time frame, how was this discovered, who discovered it, and why did it happen? Obviously it must be corrected, but the income issue may depend upon whether or not it is a prohibited transaction.

Gilmore, you answered your own question correctly. That is how to do it. The way that you defined the owner's group is very common. But your premise is false. The allocation groups do not have to be "reasonable". The rules that govern eligibility for the plan do, not eligibity for a particular allocation group. You could name people in the groups if you wanted to, but it would not be a good idea to name people eligible for the plan.

I think that we can choose enough people to give QNECS to make it pass, but I think that the group of people must meet the reasonable classification criteria, otherwise when we go back to test the enlarged group I won't get by the reasonable classification requirement to use the Average Benefits Test. In other words, I can't test a plan under average benefits that includes all employees of Company A plus the 100 youngest, lowest paid employees of Company B. I don't think that'll fly past the reasonably classification requirement of the Average benefits test. I could do this if I could pass the ratio/percentage test, however. And, yes, if we don't do the same thing for the match, that disaggregated "plan" will still fail. Nonelective contributions are not required to be fixed in the same rigid (and expensive) manner. That can be fixed by giving people enough to be considered "benefitting", plus of course the fix must have "substance" and meet whatever else 11-(g) requires. But, no, we don't need to merge. We just need to amend the plan year on a go forward basis so they can be aggregated (assuming that the tests then pass). Seems like a stupid rule to me.

May I interject a related question?: Is there or is there not a requirement that the normal cost be prorated for the portion of the year that the plan was active versus the portion of the year in which it was frozen (unless of course you choose to ignore the post-valuation date freeze)?

Tom, my numbers quoted aren't exact; the NHCE count is actually higher. Somebody other than me is still tweaking those but it does look like the concentration percentage is 99.1, so I think we're dropping to 99%, and it is a bit lower than you noted, but I agree with you that the target should be perhaps the unsafe harbor percentage of 20% anyway. But I do appreciate all the comments. I'm both looking for a magical solution and trying to make sure we're not overlooking something. This can't be the first such situation where the unreasonableness of the "same plan year" requirement is being exposed.

Won't help. 1750 people including 19 HCEs.

Unfortunately, no. The HCEs causing the failure are a stable group, with none of them hired recently. The failing plans provide for immediate eligibility, and, no, the otherwise excludable rules do not help.

Appreciate the suggestion but the QSLOB doesn't work. What is strange about this is there could be two identical 401(k) plans except that they have different plan years. If one has a substantially higher HCE percentage than the other then it fails, even if benefits are identical. So one group gets big QNECS? My situation is actually more complicated than I've let on because I have the same problem with some DB and PS plans in the mix, but 11-(g) permits a much easier correction than for a k or m plan. I just need to give people enough to be considered benefitting; whereas with a k or m plan a QNEC is specified equal to the average deferral or match percentage. Tough medicine. So at this point we're even entertaining legal arguments outside of the 11(g) fix.

ratio % 15% safe harbor 20.75% It would mean bringing in more than 100 people.

What are you testing, and why? Nondiscrimination or coverage? If coverage, don't you pass the ratio/percentage test? Does the discretionary contribution need to be general tested, or is it a safe harbor (i.e. same pct of pay)? I'd be happy to comment if you would clarify what you are trying to do. If you are general testing an employer contribution, then you have left out the Average Benefits Percentage Test, where the average % for NHCEs needs to equal or exceed 70% of the average % for HCEs. Maybe that answers your question? Passing the midpoint only lets you progress to the ABPT.

I heard Jim Holland say (about 2 years ago) "No comp, not an employee. Not in the test. Any test."

Member of controlled group discovers (in 2003) that it fails 410(b) for 2002. Controlled group has 4 other members with plans. Problem is that they have different plan year ends and therefore cannot be aggregated for 2002. Two have 12/31 year ends and 2 have 3/31 year ends (why??). Aggregating the plans that do have the same year ends does not work. Aggregating a 12/31 year end with a 3/31 would work, however. How to fix for 2002? 1.401(a)(4)-(11)(g) says correction can be done retroactively (by 10/15) by making QNECS to expanded group equal to average NHCE deferral (and same for ACP) but that would mean bringing in large number of employees of another company who already participate in a 401(k) plan. It would also mean substantial $$$. Is there an alternative? Anybody found a different solution? Can it be argued with merit that the QNEC can be offset by the employee's actual deferrals/match in their 401(k) plan? Big problem with costly solution due only to plan year ends being different (and problem discovered late). Ideas? (The free pass for coverage change is not available.)

Thank you for your interesting and informative comments. Reasonable people can disagree. You lost me, however, as a non-skeptic when you stated that you have sold tons of annuities in the past and that they can be a "fantastic" investment in or out of a retirement plan. I could not disagree more. And, I have nothing against brokers or people who sell insurance. I have close relatives that do. But I think that annuities are usually sold to take advantage of uninformed people. Just another opinion.

Not a bad idea, but it's hard to take away something now even if it is little. I'd guess there may have been a concern about 401(a)(26) at the time, but as I understand it that is no longer an issue.

Controlled group has three companies, A, B, and C, each of which have a 401(k) plan with a match and profit sharing contribution. Anything preventing aggregating the profit sharing components of A and B for 410(b) and 401(a)(4) purposes, while aggregating the 401(k) and 401(m) components of B and C for 410(b), ADP and ACP testing? I think this is fine. Anybody disagree?

DB plan provides annual accrual of 7% to owner and .5% to all other NHCEs. Some HCEs are excluded. DC plan provides 7% to all NHCEs and most but not all HCEs. The pv of the DB accrual for NHCEs averages somewhere between .5% and 1% on an contributions basis. That is subtracted from the 7.50% requirement. Part of aggregated plan (old HCEs included) tested on benefits basis and part (young HCEs included) on a contributions basis. Plan was designed by a very well known person at the forefront of this technique.

If that is the question, then I have tested for several years a DB/DC combo that needed a 5.5%-6% DC contribution to pass. Once the gateway rule came into effect, the gateway was around 7% (you can use the pv of the average NHCE db accrual toward the gateway), and of course it passes much easier. This obviously depends entirely on the design and demographics.

Back to Lori for a minute, Lori I'm curious about your reference to the cumulative permitted disparity limit. That's a concept more common in DB plans or general testing of DC plans, not something pertinent to regular integrated DC plans. It has to do with the use of more than 35 years of service. Do you want to discuss that reference?

To copy someone else's words (who we haven't heard from in a while), I don't disagree. And, don't forget important Doctor plan design-use a 3 year cliff vesting schedule and there'll be 100% staff turnover every 2.49 years. That'll invalidate the 3 year cycle for sure.

You don't have trouble getting timely census (with everyone on it) from Doctor's offices?

Isn't that the three year testing cycle?