When and for Whom are Roth Conversions Most Beneficial? A New Set of Guidelines...

SnowballCamper

Full time employment: Posting here.
Joined
Aug 17, 2019
Messages
691
RunningBum linked to this article in another thread, asking if it needs it's own thread. I think it does, so here it is.

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3860359

I found the following quote very informative (the base case is converting $100k at age 65 for a MFJ couple):

The worst case: early distribution of converted funds
This online paper considers a version of the base case where distributions from the Roth begin to be taken at age 80, with the amount set equal to the RMD...

Unfortunately, to begin distributions from the Roth at age 80 will seriously compromise the outcomes of an age 65 conversion. Pretax
breakeven is not reached by age 100...

Moral of the story: Roth conversions are designed to increase financial wealth. If used instead to increase consumption wealth, they may blow up.

The author emphasizes that many of the simplifying assumptions we make when considering conversion options don't universally hold. As a result, we continually have discussions about this issue :cool:
 
Last edited:
Upon more reading of that paper, I made this comment in the other thread yesterday:

---

I didn't make it through that paper. The tone seemed slanted to me, but maybe because I didn't agree with it. Mostly the problem I had is that it talked of a breakeven point. It did make me think more about whether my view that a breakeven is not valid, and I'm as convinced as ever.

First, it assumes that an amount in a tIRA has the same value as the same amount in a Roth. Thus, the breakeven happens when you've been forced to take enough RMDs and thus enough of that deferred tax liability to bring the conversion plan higher. But it's not barely past breakeven, because the tIRA still has a balance and deferred tax liability.

Second, it ignores that fact that the IRA lives on beyond your lifetime. It's not like the SS breakeven point where you really can collect more if you live long enough, or come out ahead if you die before the breakeven. If you die with a tIRA balance, your heirs still have to pay the deferred tax, and it's now in a condensed 10 year time frame. If your heirs are in a lower tax bracket you may need to prioritize which is more important. Some don't care about their heirs, but dying with a tIRA balance means you left some money behind that you could not spend until you withdrew and paid the deferred liability.

Since the breakeven seemed to be the crux of the paper's conversion decision, I just saw no reason to really dig into it. Maybe I missed something valid. But my course is already set and my remaining conversions will be at 0, 10 or 12%. So my motivation to try to find a nugget in the paper is low.

If you want to leave your unused tIRA balance to charity, or do QCDs, that's a different story. If you have a certain amount you want to leave, rework your conversion calcs to leave that balance. If you want to leave any amount you don't need, then don't do any more conversion than you expect to need, and leave the rest for the charities, because the charities will receive more from the tIRA than your Roth or taxable account because they don't have to pay the tax you deferred.

---

Another poster noted: The author of this paper has a PhD in social psychology and an undergrad degree in interdisciplinary studies from Evergreen State. He is a retired marketing professor whose research and teaching are not in the area of Roth conversions.

---

The author took part in an exchange in a Bogleheads forum thread, but I only skimmed it.

Edit: I'm not complaining about the thread being opened, especially since I suggested it. I'm just saying that something that looked potentially useful lost its interest to me. If others find it useful, have at it.
 
Last edited:
I understand I don't need to take out MDR for a Roth IRA.

But what happens to a Roth IRA if you never touch and it bequest it to your heirs. Do they have to take MDRs once you pass?
 
I understand I don't need to take out MDR for a Roth IRA.

But what happens to a Roth IRA if you never touch and it bequest it to your heirs. Do they have to take MDRs once you pass?

They have to follow the rules in the SECURE Act, which generally means that they have to empty it by 12/31 after the 10th anniversary of the date of death.
 
They have to follow the rules in the SECURE Act, which generally means that they have to empty it by 12/31 after the 10th anniversary of the date of death.

Right, but do they pay income tax or capital gains on it. Or is it all tax free like it is for the original investor?
 
I agree with OP and Running Bum that breakeven points are a wrong headed way to look at this. Years ago, when we could have made a concerted (albeit, taxwise expensive) effort to convert I looked at it with a breakeven view and decided that we were already in too high a bracket to bother with it. Might as well put the big taxes off a decade rather than pump them up for the whole period. What I didn't think through was the advantage to our heirs. They would have potentially benefitted from the conversion.
 
I understand I don't need to take out MDR for a Roth IRA.

But what happens to a Roth IRA if you never touch and it bequest it to your heirs. Do they have to take MDRs once you pass?


MDRs is that a typo or an acronym I don't yet know?
 
I understand I don't need to take out MDR for a Roth IRA.

But what happens to a Roth IRA if you never touch and it bequest it to your heirs. Do they have to take MDRs once you pass?

I think you mean RMDs, your spouse can inherit and keep inherited the inherited IRA and use it as their own. For non-spousal heirs, while there are no year by year RMDs on any inherited IRA, the heirs have to withdraw it within 10 years (prior to the SECURE act of 2019, people could stretch it out per a life expectancy table).

An inherited Roth IRA is not taxable when they withdraw it (so it's best for them to keep it all 10 years). It has a 10 year clock as congress didn't want an inherited Roth to be some sort of multi-generational tax free trust fund.

An inherited traditional tax deferred IRA also has to be withdrawn in 10 years, but optimum withdrawal timing is more complicated as withdrawals count as taxable income.
 
His premise seems to be that the assumption of a "tax torpedo" putting you in a higher bracket post retirement may be false for most people.

That is probably true, but possibly less true among this group.

My analysis has been that Roth conversions at lower tax rates (say up to 22 percent)
may make sense, but beyond that it becomes more muddled, due to IRMAA, the width of the brackets, unknown levels of future stock market gains/losses, and heirs' tax rates.
 
MDRs is that a typo or an acronym I don't yet know?

I interpreted it as minimum distribution requirement, more commonly referred to (and referred to by the IRS) as required minimum distribution or RMD.
 
We jumped the other thread pretty hard and talked about this quite a bit. Over at bogleheads, there is a big thread on this too. In there, I asked Prof McQuarrie about the point RunningBum raises and the Prof meant the use of ROI is a convenient shorthand to compare the benefit to the converted amount. But RunningBum is right, it just confuses the discussion for people trying to learn it for the first time. The only way to lose money doing a Roth is to convert into a tax bracket you would otherwise never have faced in your lifetime.

So you have to start the analysis by looking at the No Roth conversion case in something like i-orp to see if you even have a problem. For instance, if you're never out of the 12% bracket with pensions, SS, & RMDs, then converting beyond the 12% bracket would lose money.

The paper studied the simplest case, where the taxes on the conversion were paid from tax deferred. Then he looked at that situation for people in various tax brackets. The government owns a portion of your tax deferred, though the size of that portion is somewhat unknown until end of life. I believe the Prof's analogy if you do not do a Roth is that the government's share "silts up" your low tax brackets over time, pushing you into higher brackets. That effect compounds over time, so like any compounding, the dollar amounts start small and grow exponentially.

My favorite reframing of the way ROI’s that the Prof presented is as fee reductions. An effect the paper discussed as a 1% ROI can be thought of as a 1% reduction in fees on the converted amount. If we saw an account that the same investments for 1% lower fees, we would certainly jump all over that.

The paper discussed but did not focus on the case of paying taxes on conversions from taxable. His math was that if you hold tax efficient funds, there are still some dividends and the typical tax drag from those dividends is about 0.3%/year on the taxes paid. If you do use taxable to pay the conversion taxes, the 0.3%/yr on the taxes paid is additive to the base Roth benefit.

Other positives of doing a Roth conversion that, being a finite paper, he did not really go into but could be really significant:

-If you are MFJ filing status now and one spouse dies, the other will be paying in single status, so higher rates, possibly for a long time.

-If you leave a lot for your heirs in tax deferred, their tax bracket on their income plus the withdrawal of inherited funds may be high or the inherited money may interfere with their ability to do Roth conversions on their own IRAs.

-You can do asset location optimization, putting stocks in Roth and bonds in tax deferred

-Future tax rate risk seems high with the TCJA tax cuts set to expire after 2025 and little sentiment in Washington to keep rates low for higher incomes. (Of course, if Washington gets hungry enough, they can come after Roth accounts too).

Offsets/Risks
-If there are lower returns than anticipated for a long time or a Japan style crash and no recovery, Roth conversions mean you have paid taxes on anticipated gains that never happened

- Capturing the secondary benefit of using taxable to pay taxes on the Roth can incur LTCGs that would otherwise get a step-up basis on death. So those LTCGs paid really are a true investment in the Roth conversion and need time before death of the last surviving spouse to pay out. If you do a $100K conversion in the 24% bracket, you owe $24 K. If you have 50% LTCG on the money to pay those taxes and a 15% LTCG rate, your taxes are $1,800, so a couple years of the benefit.

-Capturing the location optimization benefit can also run into the need to pay taxes on capital gains in taxable once the Roth gets large and full of stocks. The consumer level analysis tools really don’t let you look at location optimization.

-Various subsidies like ACA can be significant enough that they may affect the Roth math. Also, complexities like IRMAA tiers, amount of SS that is taxable, etc. need to be considered.

All this means you need a powerful program to look at the situation, home grown spreadsheets are likely to miss important things.
 
Thanks for this nice summary, and pointing out some of the many issues not addressed in the paper. I think this statement over simplifies things a bit...

The only way to lose money doing a Roth is to convert into a tax bracket you would otherwise never have faced in your lifetime.

Two ways it isn't true, is dying before the break even point, or using the converted funds prematurely, as in the quote from the article I started with.

To me, the conversion problem is interesting because the answer comes down to the individuals assumptions about the future
 
Thanks to Exchme for the nice summary and the pointer to the BH thread on this paper. I have read quite a bit (but not all) of the paper, and about half the BH thread. My takeaway is very similar to Exchme. The following points stand out to me:

As I offered in @Time2's thread that started tackling this paper, the main driver for my Roth conversion plan is the likely scenario of one of us (probably DW) filing single instead of MFJ:
My main driver for the decision is my wife's situation after I pass (assuming I go first). Pensions and SS will use up most of whatever the low brackets are. Without conversion, she will probably be in IRMAA Tier 4, and whatever the 24% bracket will be (presumably 28%). With conversions, I think she will be in IRMAA Tier 2, and maybe in the 22% (-> 25% ) bracket. Meanwhile, (before SS), I will be converting in the 12% and 22% brackets. I acknowledge this is not life-altering, but seems like a worthwhile thing to do for her. (Of course, maybe she will remarry and get back to MFJ! :) )
The author of the paper acknowledges that this issue is important, but he was not able to consider it, to keep the scope of the paper reasonable. But, to my mind, that means you cannot make any conclusions about the advisability of RoCo's based on this paper!

Exchme alludes to another thing that, frankly, kind of astounds me. The paper concludes that the gains from RoCo are modest. In his metrics (using ROI, which is problematic), in realistic, unfavorable cases, a RoCo strategy delivers a ROI of "only" an annualized 1% (and more favorable cases it is more like 2% or more). And from this, the author and many others conclude that this gain is too small to worry about. What!? It is an annualized 1%! If we think that is trivial, shouldn't we stop urging people to leave Edward Jones? Shouldn't we blithely use mutual funds with ERs of 1%?
 
dying before the break even point

I think this is one of the larger points where people talk past each other.

In my mind, all dollars in all tax-deferred accounts have that embedded tax liability which will be paid eventually either when the original owner takes a distribution, either elective or RMD, or by the beneficiaries, typically within 10 years after death of the original owner. (I suppose in the case of QCDs or excessive medical expenses or charitable bequests this tax liability can effectively become zero.)

Since I have three children who are beneficiaries of my IRA, I include in my analysis and thoughts the tax liability they'll pay. So even if I die before the breakeven point, if my kids' tax situation means that collectively we had a tax savings, that's good enough for me.

I think some of the posters here draw their analysis more tightly and only consider what happens during their lifetime. I think that's a reasonable point of view to take even if it differs from my own. It's especially reasonable if one has no heirs or has reasonable plans to make the tax liability zero (like leaving their traditional IRA to charity).

But I think that how big of a circle one draws when approaching this problem definitely can have effects on the resulting analysis. In particular, someone dying with a $4M IRA and a 25% tax bracket is either including or excluding $1M of tax liability, which is probably a significant enough sum to affect the conclusions drawn.
 
Table 7 in the linked McQuarrie paper sums it up for me. Average age of break even is 90 years of age. Way too old. Average “wealth” impact is 1%-2%. Far too little juice. Far too late in life. I’m supposed to be dead before age 90.

Roth conversions are a no-go for me. Roth conversions would make me poorer, rather than richer. I do not consider myself unique and therefore Roth conversions are assuredly a no-go for many others.

On Day 1 after a Roth conversion the investor is impoverished, to wit: made poorer, in the amount of the income tax paid on the conversion. Therefore payback is exactly the right type of decision making approach for Roth conversions. The retiree is investing the tax paid for hope of future benefit. The future benefit comes very late in life and is hardly worth talking about - low single digits percentage gained.
 
Thanks for this nice summary, and pointing out some of the many issues not addressed in the paper. I think this statement over simplifies things a bit...



Two ways it isn't true, is dying before the break even point, or using the converted funds prematurely, as in the quote from the article I started with.

To me, the conversion problem is interesting because the answer comes down to the individuals assumptions about the future

Planning for the future does require assumptions of rates of return, life expectancy, future tax rates for you and your heirs. But to make sure we don't confuse casual readers, "dying before the breakeven point" is not a valid concern and "using converted funds prematurely" might not be.

Roth conversions are not a conventional investment where you are under water and later finally reach a breakeven point. The liability of taxes on the t-IRA balance exists, the Roth conversion is simply an attempt to even out taxable income to avoid high tax brackets at a later time. In a progressive tax system, that's the best plan.

In order for a properly sized Roth conversion to actually be a bad deal, you have to have something like:

-Bad market performance and no recovery, so the growth you expected never materializes
-A magical tax cut from our benevolent Congress so the tax rate you anticipated never arrives
-Heirs tax rates much lower than assumed
-Or maybe you counted on dying early so you thought your spouse would have a high tax rate as a single and you manage to live instead. Hopefully your spouse agrees getting you for a long time instead of extra cash was a good trade. :)

To the specific instance of you (and your spouse) dying early, there is less total time to get the money out of the t-IRA than there would be if you lived a long time, plus your heirs have to add this inherited IRA withdrawal on top of whatever else they are doing in their lives. Since your heirs may still be working or have their own Roth conversions to do, their effective tax rate on that unexpected inheritance is quite likely higher than if you lived a long time and took it out a bit each year in RMDs and then passed along a reduced amount to your heirs.

At the risk of repeating an illustration I used in another thread, if you convert an amount and then immediately die, you might think you are behind, but it's easy to see that's not correct as your heirs have yet to pay taxes on the t-IRA balance. If your heirs fully liquidated immediately, they would be even whether there was a Roth conversion just before death or not. Roth or no, somebody owed the taxes. In a real inheritance, they of course don't liquidate immediately, they get 10 years so the Roth has time to pull ahead unless your heirs are in lower tax brackets than you were even though they have your inheritance on top.

As for "using converted funds prematurely" making the Roth a bad deal, I would say the answer is "it depends". If it's a deductible loss like an uninsured medical emergency, the tax deduction can partially offset the taxable withdrawals from the t-IRA, so in that case, having done a Roth could be wrong. If the premature need arose because rates of return were bad, then that's an already acknowledged point.

But there are other situations where doing the Roth comes out ahead, even if you need the funds early. For instance, i-orp and its inhuman optimizer will sometimes do Roth conversions early and then use some of the converted money to cover the cash shortage that often exists before RMDs and SS kick in, as maximizing the ($ x time) of the Roth account helps the portfolio.

If returns were in line with expectations and someone just used the t-IRA for a non-deductible purpose, then to get around to needing the converted money, they had to have paid taxes on whole t-IRA balance already. That would require withdrawals at even faster rate than your plan (since it's rarely right to fully deplete the t-IRA) and therefore those unplanned withdrawals were likely at a higher tax bracket than your early Roth conversion, so the Roth would be ahead.

There's no hiding from the tax man, the tax liability exists in all cases and properly planned Roth conversions are the best chance of maximizing your after tax money.
 
Back
Top Bottom