I have a union-DB plan (not multiemployer) that is going to freeze it's plan and there will be replacement 401(k) plan that offers a 3% non-elective contribution as well as a 100% match of the first 3% of comp.

The employer has determined that there are certain older participants, however, that will not have enough time before retirement to benefit under the new 401(k) plan to make it worthwhile to them. The employer had a third party run numbers and use reasonable assumptions and came up with about 25 participants who will be offered the right to make a 1-time election to continue to accrued benefits under the union-DB plan. Only 1 of those 25 participants are HCEs.

Further, after the analysis, the employer and the two locals that represent these participants, agreed upon this arrangement.
I think there should be an MOA to reflect this.

Are there any anti-cutback or discrimination issues with this? I would appreciate any input on this, including the citation to any authority you have. Thanks so much.

Look at 401(a)(26)

Consider whether the selection of the individuals eligible to make the election is consistent with the ADEA

Yes.
Plan amendment that creates the "partial freeze" must be adopted before its effective date (but you knew that).
Also, see IRS Reg 1.401(a)(26).
There could be coverage and/or 401(a)(4) issues, including the question of whether the grandfathered EEs will also be covered by the new (k) plan.
I've probably forgotten something.

It would seem that Sections 401(a)(26) and 401(a)(4) would not apply because we're only talking about union employees who should be excluded from consideration. Is there an issue which I may have overlooked? Is there any more to do then freeze the plan and provide a 204(h) notice?

Andy -- There is not a blanket exemption from IRC Section 401(a)(26) testing for union employees (as there is for IRC Section 410(b) and 401(a)(4) testing), so plans with only union employees are subject to Section 401(a)(26) testing. (Treas. Reg. Section 1.401(a)(26)-2(d)(i); 1.401(a)(26)-6(b)(4) & (5); 1.401(a)(26)-1(b)(2).) But, non-union employees and employees of other unions would be excludable. So, as long as there is an HCE in the plan, I believe that Section 401(a)(26) still applies in this situation. (Treas. Reg. Section 1.401(a)(26)-1(b)(1)(i).) Note that one of the examples in the regs. contemplates a Section 401(a)(26) test for a plan with only union members as participants (Treas. Reg. Section 1.401(a)(26)-6(b)(6), Ex. 3). So, I believe that the DB plan will be subject to Section 401(a)(26) testing, and that might have an adverse impact since all union members will not be participating (accruing benefits) in that plan. (Of course, this analysis does not apply to the 401(k) plan.)

Thanks everyone for the good responses.

Turns out the plan actually encompasses 2 local unit members as well as NBU members. So it was not a union-only employee plan, as it turns out.

However, the only participants in the plan that will be allowed the one-time election to continue to accrue benefits are from the 2 locals, not any NBU members. One person in one of the 2 locals is an HCE. So does Code § 401(a)(26)(B) effect a participation requirement exclusion for the two collectively bargained units (since this is not a union-only plan)? Or does the permissive disaggregation rule in TR § 1.401(a)(26)-2(d)(2)(i) treat this DB plan as "three" separate plans which still require each of three separate plans (as it were) to go through the 401(a)(26) testing?

I don't think there is a 411(d)(6) issue because we're not talking about an amendment that reduces an accrued benefit or eliminates an optional form of benefit. I think the only issues are the discrimination issues but those benefitting (only bargaining unit members) seem to have requirement exclusions under 401(a)(4), 410(b)(3)(A) and 401(a)(26)(B).

Seems to me that you can permissively disaggregate under IRC Section 401(a)(26)(B) and under the -2(d) regs, but disaggregation is not required (as it is under 410(b) & 401(a)(4)). And, there are permissive excludable employees under the -6(b) regs (and, I think you have a union-only plan when it is reconfigured, since union members are the only ones continuing to accrue a benefit). So I think you can have either one plan (excluding all non-union members), or two plans (one plan excluding all non-union members and members of union #1, and the other plan excluding all non-union members and members of union #2), or one plan (excluding no one), or perhaps even other combinations of tests. I would think your best chance of reaching 40% would be to at least exclude non-union members. If you test each union separately, then one union's "plan" will automatically pass 401(a)(26) because there will not be an HCE participating. Since this is all permissive aggregation and permissive exclusion of certain employees, you have all sorts of combinations and permutations to try.

Sounds clear as mud to me--I hope it sounds as clear to you . . . .

Thanks Larry. Actually it's very clear. I'm trying to get the participant breakdown from the employer of the 2 union participants and the NBU participants. By doing the permissive aggregation testing, that might not still reach the 40% threshold.

But I wonder if there is authority for including the 401(k) (3% QNEC and 3% match) which is technically replacing the DB accruals for most of the other DB participants that are not part of the group that will be given a one-time election to stay in the DB plan. Again, the only DB participants that were given the one-time election to continue to accrue benefits in the plan were union employees (from both unions) that were determined by a third-party actuary who were too old to benefit from the DC plan and would do better at the DB plan upon normal retirement age.

Unfortuantely, Bearlee, contrary to IRC Section 410(b), permissive aggregation of other plans in order to pass IRC Section 401(a)(26) is not permitted.

Seems to me that if you are giving participants a choice then your disclosure requirements go through the ceiling. It might only apply to cash balance conversions, but there were some regs released 5 or 6 years ago that discussed the requirements if you are giving participants a choice between a db and dc. I believe you will need to demonstrate the impact of their decision under various scenarios in order for their elections to be valid.

It may be easier to just draw a line and say these stay in the db and these are out.