Luke Bailey

Right. You can argue for a negative conclusion based on a "two separate potential taxable events" theory as I outlined previously. But I don't exactly understand the rest of this. If I don't do a Roth conversion, but rather roll to a pre-tax account, then sure, any amount rolled after 2020 triggers refund. That's actually not in the statute, but rather the way IRS practically reconciles the fact that the statute tells you to recognize the income over three years, and requires income inclusion as early as the first year, but then allows you to treat the amount, including what you may have already included in income, as rolled over to the extent you roll over within three years. (Note that the actual statutory language is "contribute," while "repayment" is in the caption.) It seems to me that it is also not inconsistent at all with either the language or, really, spirit of the statute to say that in the case of a Roth conversion (again, "conversion" of course not in the statute) of a CRD, you are on a parallel track that also works. First, by rolling to the Roth you have a "contribution" to a Roth IRA, at some point within the three year period. All the statute (2202(a)(3)(B)) tells you is that you can treat a "contribution" of a CRD to an "eligible retirement plan" as a timely direct rollover if it is made within 3 years. So you did that. Your conversion is timely (i.e., by contrast, if you had taken a distribution, waited 61 days, and then rolled to a Roth, your conversion would be no good; you would owe the tax, but would not have a Roth). Done. Second step: Since I rolled to a Roth (i.e., converted my pre-tax money to after-tax), I have to pay tax. When do I pay it? Well, since I rolled my money to a Roth, it is just my "any amount required to be included in gross income for such taxable year" under 2202(a)(5), thus I include it in 2020-2022. The only way I can see the IRS arguing that this does not work is to say that when Congress references 408(d)(3) in 2202(a)(3)(A), it really just meant 408(d)(3), not 408(d)(3) as modified by 408A(d)(3)(A)'s "notwithstanding," which seems to me to make 408(d)(3) inclusive of 408A(d)(3), especially when you take into account that 408A(a) tells you to treat a Roth IRA as a pre-tax IRA except as otherwise provided in 408A. That is just a variant on the "two separate potential taxable events" theory, and maybe that's what they will do. Since this issue does seem to be present in the prior disaster provisions, I'm still surprised it is not addressed in guidance.

Thanks for the explanation, RBG. But do you agree that if there was going to be a rule that "repayment" did not include repayment to a Roth, IRS would have included it in 8915?

That seems right to me, Belgarath. I would note that unlike some other issues one could think of, where a rule would be hard to administer for a plan of a different employer, in this case it would not seem any more burdensome, because the plan with the loan knows what it needs to know, i.e., that there is a loan that qualifies for the relief permitted under the statute. I have struggled with understanding the impact of 2020-23 on loans. Doesn't this just amount to the offset occurring on July 1 instead of June 30? If so, is that a big deal? For posterity, the Code section is IRC sec. 402(c)(3)(C). Yeah, in the absence of regulations or other guidance, which I don't recall any existing, we're just guessing, but "failure to meet the repayment terms of the loan from such plan because of the severance from employment of the participant" probably was intended to mean, basically, a loan acceleration clause where employer only accepts repayment if done through payroll withholding, and the withholding agreement terminates on severance from employment. Here, the first employer let the loan continue to be paid after termination, and the reason it is accelerated is the former employee's failure to make payment, which, even if termination with the second employer counted, which it probably doesn't, might or might not be "solely by reason of" the separation from second employer. Anyway, there's a good chance the borrower will get to treat the loan offset distribution as a CRD, right? Gives him or her even more time, I would think.

mming, what the CARES Act seems to say, mostly, is that the tax that would otherwise be owed on a CRD is going to be spread over 2020-2022, unless the participant elects otherwise. What I was theorizing is that an alternate interpretation (which seems to me less likely, but what do I know) is that where you do a Roth rollover, you have two different taxable events, the initial distribution (which gets the three-year spread) and then the Roth conversion, which doesn't. I think (but again, even more so in this instance, what do I know) that that is maybe sort of what MoJo is thinking should be the rule, although if it's not I'm sure he will let us know. What I was pointing out is that IF one did want to apply such a "two taxable event" analysis, it could be defeated by waiting until the third year to roll everything over to the Roth. Nothing that I have said or tried to explain, for whatever it's worth, addresses state taxes, which of course vary from state to state.

Sure, MoJo. Just wanted to understand the argument. Maybe RBG could elaborate more, and perhaps I'm misunderstanding his earlier post, but it seems to me like he's right if he's saying it's odd this has not already been dealt with, one way or the other, in the instructions to the various Forms 8915. It does seem to be the same language and it does seem like it should have come up before as a problem needing guidance.

But presumably no abuse where someone takes the CRD, uses it for living expenses, then sees a turnaround in their financial life and can do a rollover in 2021, but chooses a Roth rollover instead of pre-tax. So the three-year spread is ok there, right? So what you want is a rule that would say you need to include in income 1/3 in 2020, plus whatever you rolled to Roth in 2020, etc.? Just trying to understand the argument.

I suppose it's possible that the IRS could view the taxable event consisting of the coronavirus-related distribution (i.e., the actual exiting of pre-tax funds from the plan into the control of the participant, which is the "any distribution" referred to in 2202(a)(4)(A)) as separate (for purposes of the three-year averaging rule) from the taxable event that occurs when the funds are converted to a Roth by being rolled to a Roth IRA (i.e., the inclusion that occurs under 408A(d)(3)(A)(i)), but that seems like a stretch. And couldn't it be defeated (albeit with a later Roth start date if the taxpayer did not already have a Roth) by the taxpayer's simply waiting until 2021 to roll to the Roth? And why is this any more of an abuse than all the cases where folks who qualify for CRD distributions they don't really need take them anyway? What if they do need them, but don't by 2021, and instead of doing a pre-tax roll, they do a Roth roll because it works better for them based on their 2021 circumstances?

But RatherBeGolfing, that does not affect my point, right? So say you took $40k from pre-tax account in 401(k), and $60k from after-tax account, including $10k of earnings. Sure, only $50k would (if nothing rolled over) be includable in income and get the 3-year spread. So if you don't roll over the $50k that was pre-tax, that is what the "any amount" would apply to in that case, i.e., the amount that was taxable because not rolled over. But suppose you rolled that $50k to a Roth. That $50k would still be the "any amount" on which tax was required, just for a different reason. I think that in order to get an interpretation of the phrase that excludes the Roth rollover from the 3-year spread you would need to read "any amount" as if it were "any amount [includable in gross income because not rolled over to an eligible retirement plan]," and that's just not what it says.

I think IRS will have a hard time saying it doesn't work. 2202(a)(4)(C) of CARES Act defines "eligible retirement plan" to which CRD may be rolled by reference to IRC sec. 402(c)(8)(B), of which 402(c)(8)(B)(i) is an IRA described in 408(a). 408A(a) says a Roth IRA = an IRA except as otherwise provided in 408A (there's a run-around between account and "plan" and 7701(a)(37) is involved, but that's what it amounts to; check it out). 2202(a)(5)(A) says "any amount" required to be included in income on account of a CRD gets the 3-year spread, no distinctions as to whether must be included because not rolled, or because rolled to Roth. Also, take a look at 2202(a)(5)(B), which refers to the old 408A(d)(3) Roth rule when they first allowed conversion without income limits in 2010, although I have to say that one has me a little baffled. Basically, under 408A(d)(3), if you did a Roth conversion in 2010 and were supposed to include the income ratably in 2011 and 2012, but took a distribution from the Roth in 2010 or 2011, you accelerated your tax (i.e., you lost all or part of the 2-year spread, because you did not leave it in the Roth). Why that would matter for a CRD (whether the roll is to a Roth or to another type of plan, if, as seems possible, Congress's reference in CARES intended the 408A(d)(3) rules to be extended to non-Roth IRAs and plans as well) is beyond me, i.e., if you took the CRD, thought you could afford to roll it back, then end up taking it out again because your needed expands with pandemic, I don't see why that should accelerate the tax, even if the roll was initially to a Roth. 408A(d)(3) also has a rule for accelerating if the recipient of a CRD passes away before the last year of the spread, which would make more sense. Perhaps that latter part of 408A(d)(3) is all Congress had in mind. Anyway, it seems hard to believe that Congress would have specifically referenced an old Roth anti-abuse rule in 2202(a)(5)(B), but not intended the 3-year spread generally to apply to CRDs rolled to Roths.

Right. A deemed distribution (employee defaults, but there is no independent distributable event, so he/she gets a "dry" 1099-R, so to speak, showing only the amount of "air" distributed) can be paid back, because the loan obligation still exists as a commercial transaction between participant and plan's trust, even though IRS says it's time to pay tax. Repaid amounts create basis, and in essence are like after-tax contributions, but not subject to limits or 415(c). "Loan offset" distributions are actual distributions, ending the repayment obligation to the plan, and cannot be paid back, although as Sully above stated there is an extended rollover deadline under, what, TCJA I think. There has been so much legislation recently it's hard to keep track.

Jakyasar, the amounts paid to the plan as interest should be deductible in the same way and on same deadlines, and subject to same restrictions, as contributions.

OK, Larry. Agreed.

Larry, yes, but.... The exception to privilege that you are referring to in ERISA cases is based on the idea that when an attorney represents a plan, he or she is really representing it participants, and therefore the participants can be viewed as clients for purposes of privilege. Here, if the attorney were engaged and paid by the company and/or its owners, and the engagement is only to determine tax and civil protection for the owners, and not to represent the plan, and the engagement letter makes that clear, the communications between attorney and client (the company and/or owners), e.g. regarding their tax and civil exposure and how to protect against those, would be privileged in any civil suit by the plan's participants against the company or owners. If the company decided to go the VCP route, then representation of the plan in the VCP (even though not paid for by the plan), whether by the same or a different attorney, would probably not be protected by privilege from the participants in a civil suit. If you're talking only tax, then practically speaking CPAs and EAs have significant privilege under IRC sec. 7525.

Sure, Larry. More facts would be good, and if the funds were borrowed, there is an offsetting liability. And of course, typically book value is just one component of value, along with market comparables and discounted cash flow.

Of course $400k in cash that belongs to the company is part of its value. If the cash is lent out for appropriate business reasons, the debit to cash will be offset by equal credit to loans or receivables as asset. They'd better be.

Purplemandinga, completely agree with CuseFan on substance, but note that CuseFan is not in fact saying the plan does not have to be amended, just that it does not need to be amended either immediately or permanently. It does need a temporary, or "pop-up" amendment, as it were, adopted by the end of 2022, but effective only for 2020.

This has been thought of before, and implemented in varying fashions. I think a big problem with the concept is that it does not fit within either the classic ERISA 404(c)(1)(A) self-directed or the "new" 404(c)(5) default investment safe harbors.

That's a question an actuary should answer, Mike.

Whop, it probably can be fixed in VCP, but the client must have an estimate of the cost of correction (largely in terms of the cost of additional contributions, as it does not appear to be complex from a legal/tax standpoint), as well as their exposure to IRS if they don't correct, so that they can weigh the alternatives.

Santo Gold, based on your description, I had a similar case a number of years ago. We got it through VCP.

As Larry said, based on the description, the plan is not qualified. The only way to make it qualified would be IRS Voluntary Correction Program. You have MANY issues surrounding as well.

Yes, that is what I mean. Just pointing out that the preclusion of "disinheritance," as it were, of children of prior marriage only protects up to the point that the benefit might be taken in form of QJSA. But sure, very few spouses ever force that, even when they would not consent to having someone else named for portion of pre-retirement death benefit.

Right. OK. Absolutely. So the plan administrator is back to the problem of deciding whether there is a spouse or not, and how much risk it runs if it takes participant's word that there isn't one. This is a question that involves discretion, and therefore a fiduciary decision that service providers can advise on, but ultimately decision is left with plan administrator.

Right. Basically, for many reasons, including history and chance, our system has painstakingly complex rules for qualified retirement plan nondiscrimination, but a crazy-quilt when it comes to health.

Larry, you mean so that you have a 50% QPSA and the participant can name beneficiaries for the other half without spousal consent if dies before annuity starting date? But if participant dies after ASD and spouse insisted on QJSA, kids of prior marriage still disinherited, right? But still works for many fact patterns.