Plan Design - Safe Harbor 3% 401(k) and Cash balance Plan

[Gary]

Considering a plan that provides 3% 401(k) SH, and 2% profit sharing for total of 5% (which meets TH) for NHCEs.

Cash balance plan is to provide 3% (with a minimum dollar amount of Employer dollars to pass non discrimination) for NHCEs.

Much more provided for several owners.

7.5% gateway is met, so plans are cross-tested for non-discrimination.

In the plan design consideration, in order to pass ND tests it required one NHCE getting a larger cash balance credit.

1. Is there a problem with one or say a few NHCEs getting more than the rest of the NHCEs? That is, does some classification of employment need to be created to avoid the effect of name enumeration?

2. I need to check this point out further myself, but to what extent can the above design be implemented for 2005 calendar (plan) year? Do we have to wait until 2006 for the SH plan structure? What can be done for 2005?

Thanks.

[Tom Poje]

"7.5% gateway is met, so plans are cross-tested for non-discrimination."

maybe, maybe not. or at least that would be my understanding.

you have to convert the DB rate to an equivalent dc allocation rate.

you do get the advantage of taking the average of the NHCEs (1.401(a)(4)-9((b)(2)(v)(D)(3)

[Blinky the 3-eyed Fish]

To illustrate Tom's point how a 3% cash balance contribution is most likely NOT 3% for gateway purposes consider this example of a 35 year-old retiring at 65:

$20,000 * .03 = $600 CB

Convert $600 to an annuity using AE (say 5% & 94 GAR)

$600 * (1.05)^(65-35) / 11.794088 = 219.87

Now convert the annuity to an equivalent allocation rate using the testing assumptions (say 8.5% and 83 IAF)

219.87 * 9.615584 / (1.085)^(65-35) = 182.92

$182.92 / $20,000 = 0.9% - that is the value of your 3% contribution for gateway purposes

As to your other questions, you can have allocation groups within a cash balance plan much like a cross-tested designed DC plan. You could also utilize a -11(g) amendment to increase somebody as needed to pass testing. There isn't an issue with giving one NHCE more than others other than if his/her co-workers find out and wonder what the deal is.

If none of this has been implemented yet, you just need a 3 month year for the SH 401(k) piece. The DB can be adopted 12/31.

[Gary]

Thanks for the responses.

So my understanding is that the plan can favor one or two NHCEs over others, but do we need to try and put them (or him) in a certain class, like certain administrative-type employees or something to officially make it appear to not be benefitting a person by name?

Regarding the 3-month aspect for 401(k) plans, it seems we can wait until 1/1/2006 or amend the profit sharing plan to 1) implement the 401(k) safe harbor feature and 2) amend the plan year to be from say 9/1/2005 to 12/31/2005 and then revert back to the calendar year plan year for 2006?

[Tom Poje]

Blinky provided a very good example - taking it one step further, in his example the ee ended up an accrual of 219.87

or 219.87 / 20,000 = 1.099 EBAR

the good rule of thumb for the 'meaningful' benefit is that it needs to be at least .5.

(You have to have that to pass minimum Participation as well)

..............

if all you have is a profit sharing plan, you can amend the plan to be a safe harbor anytime during the year as long as people can defer for at least 3 months. I don't know if I would say you are 'reverting back' to a calendar year in 2006, but rather the first year of the 401k feature was simply less than a year.

watch out, if the 3% safe harbor only covers 3 months of comp then you would fail the gateway minimum.

...........

there is no problem naming an individual by name for purposes of class. the govt has said you can do that. some argue that creates a CODA, but it appears you are still able to do this. You could have accomplished the same thing by setting up individual plans for people, so I think the there is a valid argument for saying you haven't created an artificial CODA. If, however that ee receives zippo, and the plan fails the ratio % test for coverage you can't pass using the nondiscrim classification test