Leaderboard

If I had a nickel for every takeover where the coverage testing had only considered the employees eligible for the plan instead of all the employees of the Employer under 414(m) etc., I’d have a bag full of nickels.

Plan A coverage = (9/20 NHCEs) / (2/2 HCEs) = 45% Plan B has no HCEs covered, so plan B coverage is an automatic pass, and we assume it has the same safe harbor as plan A to allow aggregation with plan A if necessary to get plan A to pass coverage. Plan A’s 45% result is greater than the safe harbor percentage for coverage, which is 27.50% based on the concentration percentage, but it’s less than 70%. So without aggregation of plans, the average benefit test for coverage is necessary. Does plan A satisfy the nondiscriminatory classification test of 1.410(b)-4? Meaning, does it cover a reasonable business classification? If not, and you aggregate the two plans to get plan A to pass coverage, you must now also aggregate the two plans for purposes of nondiscrimination testing, and if allocations are tested on a benefits basis (cross-tested), then the gateway minimum allocation does apply. Shut you want plan A to pass coverage without aggregation with B so you can avoid the gateway in B. If you do have a reasonable business classification defining who is covered by plan A, then you run the average benefit test. You can run that on a benefits basis or on a contributions-basis. You need that to be 70%, and if it is, then Plan A passes coverage without aggregation with plan B, and no gateway applies to Plan B even if Plan A is cross-tested.

Sure! There are tons of GPTs in ChatGPT and other AI apps. For example, there's an excel GPT in ChatGPT that is focused on excel, so your answers are going to be tailored to excel. I'm creating a few custom ones for my own use or internal use. For example, an EPCRS GPT. I have limited it to using the current version of EPCRS and my instructions (which are expanding and help narrow answers). It's still in early stages, but so far its a good way to find answers or the correct section quickly. Another one im playing around with is a Controlled Group GPT. Instead of feeding it all the data you think it needs, it prompts you for input. In a nutshell, the CGGPT starts by asking you how many entities you are working with. Lets say you answer "3" The next question it asks is how many owners between all three entities. After you answer, it asks about each owner's equity in each entity. And so on. The basic idea is to have the GPT ask you for all the relevant data instead of you asking it to make a determination based on the data you give it. The more prompts you pre-program, the more questions are asked. You can get super detailed or very basic. It's a fun process to go through (for us pension nerds 🤓) and hopefully you get a useful tool in the end. I also have another one I have started uploading different plan documents to and building instructions in order to determine where certain provisions are and how they are similar or different by vendor. I know of at least one company put there working on a document AI to help you map from one vendor to another. It's exciting stuff as long as you have guidelines you use it properly.

The statute’s provision with the heading “no tax on tips” is not an exclusion from income; it’s a deduction. Internal Revenue Code of 1986 § 224, added by Act § 70201 [attached]. Likewise, “qualified overtime compensation” is not an exclusion from income; it’s a deduction. Internal Revenue Code of 1986 § 226, added by Act § 70202 [attached]. A retirement plan might be unlikely to exclude from whatever otherwise would be within the plan’s measure of compensation, for whichever purpose, an amount the plan’s administrator and a participating employer might not know without obtaining information from the participant. Also, the Internal Revenue Service has sometimes interpreted 26 C.F.R. § 1.401(k)-1(a)(3)(iii)(A) and earlier interpretations to preclude a § 401(k) elective deferral from an amount “available” to the participant without handling through the employer—for example, a tip a diner paid, in currency, directly to the waiter. This is not advice to anyone. Internal Revenue Code 224.pdf Internal Revenue Code 226.pdf

RMDs from ROTH are no longer required so they do not satisfy the RMD requirements.

If you put it in the sense that Plan A covers 9/20 NHCEs rather than 9/9 NHCEs, then I had the feeling it wasn't going to pass gateway or average benefits test. That was originally what I thought. Thanks

The Treasury department’s interpretive rule does not specify how to determine when a plan “exists”. I imagine many of us might often suggest interpretations that not only follow reading the statute’s and the interpretive rule’s texts but also follow an assumed purpose of not allowing the employer’s new plan’s participants to make elective deferrals until 12 months after the last of the final distributions from the terminated plan. 26 C.F.R. § 1.401(k)-1(d)(4)(i) https://www.ecfr.gov/current/title-26/part-1/section-1.401(k)-1#p-1.401(k)-1(d)(4)(i). But if a decision-maker or adviser is considering possible interpretations and evaluating risks about when the next retirement plan “exists”, consider this: The consequence from an “alternative defined-contribution plan” that “exists” too soon falls on the terminated plan. It’s the terminated plan that will have provided a distribution without waiting for severance-from-employment, age 59½, or some other distribution-allowing condition because the plan’s administrator believed that the distribution was a § 401(k)(10) termination distribution. So, it’s the terminated plan that would have had a provision that resulted in ostensible elective deferrals that might not have tax-qualified as a § 401(k) cash-or-deferred arrangement (and further might have tax-disqualified the terminated plan). This is not advice to anyone.