I’ve written a paper titled, “When and for Whom Are Roth Conversions Most Beneficial?” It can be downloaded from ssrn.com here.
[update 10/5/2021: I have started to revise this paper. A new spreadsheet is available for download here ]
An example of the spreadsheets I used to explore conversion outcomes in the paper is here. Using it, you can plug in your own numbers and assumptions if you like. The executive summary follows.
If you are not quite ready for the paper, here are some media mentions that may help you decide whether to proceed: marketwatch 1, marketwatch 2, bogleheads forum, investment news. A few posts on this site may be of interest: here, here, and here.
Executive Summary
Much has changed since penalty-free Roth conversions were inaugurated in 2010. Tax rates have gone up and down. The re-characterization provision went away. Heirs can no longer stretch out inherited Roth accounts over a lifetime. Medicare surcharges were expanded and began to adjust for inflation. The age to begin Required Minimum Distributions was pushed out to age 72 and the IRS changed the RMD divisor tables to further slow the pace of distribution. Given these developments it seemed worthwhile to re-examine the rationale for Roth conversions. That effort exposed multiple flaws in conventional wisdom:
• Future tax rates need not be higher for a conversion to pay off;
• Nor is it all that helpful to pay the tax on conversion from outside funds;
• Nor are Roth conversions especially beneficial for top bracket taxpayers as compared to middle class taxpayers;
• Rather, the greatest benefit accrues to taxpayers who can make the conversion partly in the zero percent tax bracket, i.e., during a year with no other taxable income.
While the benefits from a Roth conversion are often small and slow to arrive, a Roth conversion will almost always pay off if given enough time, i.e., for life spans that extend past 90 and so long as annual distributions from converted amounts are not taken. Roth conversions work because of compounding, which requires the conversion to be left undisturbed for a long time. The paper elucidates the role played by the mathematics of compounding in underwriting the success of Roth conversions.
One item not addressed is the fact that a Traditional IRA inherited by a spouse that is taking RMDs may end up having higher taxes because they will go from filing Married and jointly to single filing status.
That is correct. I am considering making an already too long paper longer still by addressing that case. It is a modelling challenge: yes, the survivor becomes taxed at the single rate for the same amount of RMD income. But half of social security goes away, reducing income, half of the Medicare expense goes away, half of medical expenses go away, yada yada. More tax will be paid; but will after-tax, disposable income be reduced? Don’t know yet, that is what makes the spreadsheet work fun.
For this stylized couple, SS will be cut in half, but their RMDs overwhelm their guaranteed income. So, short of them dying together in a car crash, one of them is certainly going to be paying higher taxes late in life. Perhaps you can find data on average time a spouse spends being a widow/er and run that example.
I realize that in your neck of the woods, there are a lot of affluent couples, and maybe this paper is just for them. But for most retired couples, one spouse has better SS income than they other and RMDs usually don’t change significantly when one person dies. It is hard to imagine a scenario where the loss of one spouse cuts income in half. So, your base case really should factor in a few years of taxes at the single rate rather than MFJ.
The widow case has proved to be the most significant omission from the current draft. Careful, though: yes the widow will have almost as much income as before, but be in the single tax bracket, hence, more tax dollars are paid. But does her after-tax disposable income go down? That’s my next spreadsheet challenge.
Have you modeled married filing joint returns where both spouses collect a pension and that pension is guaranteed to the surviving spouse? Upon the death of one of us, the joint income will not differ much as the income for the survivor will be reduced by only the loss of the deceased’s benefit. However, the pension income will remain unchanged while the surviving spouse will now be filing as a single taxpayer with its lower standard deduction and higher tax rates. Our pension place us solidly in the 22% bracket and our two Social Security benefits will push us close to $170k before even taking any withdrawals from deferred accounts. And IRMAA taxation is just too close for comfort. I look forward to playing with your spreadsheet. Thank you for making it available.
Do watch out for IRMAA. Per the widowed survivor, see the last few comments/replies. Single IRMAA brackets are half of MFJ and might indeed motivate a conversion.
Our (my wife and I) plan is to have our Roth IRA’s for an inheritance for our 2 children. We have about 400K in traditional IRA’s and about 120K in our Roths. We are 68 in good health and live comfortably off our pensions and Social Security. We have other savings also for emergencies. Does it make sense for us to convert some of the IRA’s to a Roth, pay the taxes now (we can afford it) for the purpose of inheritance? Also, the money in the Roth will earn more over time so that additional amount won’t be taxed for our children to pay.
You can run your numbers in the spreadsheet to see. Generally, Roth conversions succeed when the motive is a bequest. Reason: now it has lifespan + ten years to catch up, so more compounding per the paper.
Forgive my naive question; I am just starting to learn about Roth conversions. IRS’s Uniform Lifetime Table (Publication 590-B -Appendix B) shows a Distribution Period of 25.6 for age 72. Why does the spreadsheet use 27.3 instead of 25.6 as the RMD divisor for age 72?
Hi: RMD schedule was changed in November 2020, effective 2022 and after; tax nerds may enjoy this explanation: https://www.federalregister.gov/documents/2019/11/08/2019-24065/updated-life-expectancy-and-distribution-period-tables-used-for-purposes-of-determining-minimum
Dr. McQuarrie,
Thank you for this wonderful paper that promises to be widely read and appreciated for decades to come.
I have one question. My wife is 26 and I am 29; we have a combined income of $155,000 and are both covered by workplace savings programs, both of which have a limited selection of mutual funds with load fees and eye-watering expense ratios.
Although your paper bolsters the case for taking the deduction up front, it is for the reason listed above that we have only been contributing just enough to take advantage of the match at our workplace savings programs, and plowing the rest into Roths. There is a seemingly unlimited selection of securities from which to choose and no transaction fees.
Since our AGI exceeds the $125K income limit where married filing jointly individuals covered by workplace savings plans may still take a deduction for Traditional IRA contributions, we opt for the Roth.
Am I missing something here? Are we allocating our contributions in the most tax savvy manner possible?
We think that any deduction to be had by directing excess contributions above the match to our 401(k)s is far less than the upside reaped from having a wide assortment of low cost funds and ETFs at our disposal through our Roths.
Our Roth IRAs have enjoyed an annualized rate of return 15 percentage-points higher than our workplace savings plans over the last five years.
Any light you could shed on this would be greatly appreciated.
Sincerely,
Christopher T. Black
Dear Christopher: if the workplace funds are bad enough in selection and cost, I understand your decision to go with the Roth instead. You might look into the following: 1) make a stink with management and get a better selection of funds. Now is the time, with businesses struggling to retain workers. My wife’s employer added a brokerage option awhile back–that solved the problem of limited selection. Remember, in your bracket you have to earn $9000 to contribute $7000 to the Roth. Whereas, a workplace traditional contribution of $9000 only costs you $7000 of lost spending power.
2) next, plans vary in how easy it is to transfer funds out of the plan into an IRA before termination. If you could get the funds out easily, you could fully fund the workplace plan each year, get the deduction, and then transfer funds out to a traditional IRA at the end of each year.
The paper “aims high.” What about “aimling low” to where you can reduce your pre-tax accounts and hence the RMD’s low enough to reduce tax on social security?
Additionally, many retirees will not be able to live comfortably without withdrawing from their retirement accounts until age 72, particularly if they have taken advantage of the liberal contribution limits and as a result have fewer assets in taxable accounts.
I’ve always said that the responsibility of a volunteer is to make every participant 100% happy 100% of the time – sarcastically, of course – in other words, no good deed goes unpunished. That said, there are a mind-numbing number of moving parts, including but not limited to Obamacare subsidies (for those under age 65 who need health coverage), Medicare surcharges, capital gains taxes on liquidating taxable accounts, ages that both spouses live to. There is actually a simple solution to all this, a couple should divorce and each marry someone who is unemployed and their grandchildren’s age! For those not open to such a radical maneuver, all I can say is that I consider my spreadsheet skills to be in the top 1%, and I still haven’t even tried to build a model accounting for all these factor. Kudos to the original paper.
Yes, there are indeed a lot of moving parts–that’s what made it interesting to me, as a retired individual who needs to stay sharp 🙂
A word about Obamacare subsidy limits for the pre-65 set: same cliff edge and same modeling approach: a conversion today that takes you above the Obamacare cut off, requires payment of income tax and “payment” of the loss of subsidy. Depending on the dollar amounts, the two together may constitute a rather high marginal rate, arguing against any conversion that would take you over the subsidy cut off. After Medicare, the IRMAA threshold is rather higher, so might wait to convert until then.
At the back end, the paper / spreadsheet gives credit to conversions today that reduce RMDs below an IRMAA threshold later (=better, sooner payoff). Although not considered in the paper, at lower income levels a conversion today that in the future reduces RMDs enough to remove some portion of Social Security from taxation would also be more beneficial than one that simply reduced ordinary taxable income.
Unless I missed something your paper did not address the impact of Roth conversions on estate taxes. Surely some in your readership will still need to pay estate taxes at the current high exemption and many more when and if the federal exemption is lowered. The federal estate tax ranges from 18-40% and many states will tack on an additional state estate tax.
Some states currently do not tax retirement assets but that could change in the future or you may move and retire in a state that does tax retirement benefits .
As you have pointed out predictions are difficult but considering these additional factors would help people assess their own situation .
I think you have done an important service describing the various scenarios and the impact on one’s taxes.
[saw your second comment but wanted to address the estate tax point]
I actually had a section in the paper on what happens to the Roth conversion outcome in the event there is estate tax liability, but cut it, for a couple of reasons.
1. I didn’t think there were enough cases where the TDA portion would be over $27.6 million in 2027. Far from zero cases of course, but not enough for an already long paper addressed to a general audience.
2. Because TDA balances begin to decline in the 80s as RMDs accelerate, even a couple with, oh, say $30 million in 2027 at age 72, would fall out of the estate tax range after a decade or so (with inflation, the estate tax boundary would have increased to $38 million by 2037). So the problem is self-liquidating until you get to the centi-millionaire level, given decent longevity.
3. and last, although I think I treated income with respect to a decedent correctly in the spreadsheet in that now-cut section, that level of tax knowledge is way beyond my pay grade. So I wasn’t comfortable putting it in the paper.
PS: Peter Thiel may single-handedly cause Roth funds to become taxable at death; just another Congressional risk to keep in mind.
Dear Dr. McQuarrie
Thanks for your reply. You seem to be only considering TDA (Tax Deferred Account) balances in your estate tax scenarios. I believe a more realistic situation to be a combination of TDA assets and taxable assets forming an estate. The combination of taxable and TDA assets makes a scenario where estate tax is paid to be more likely.
The TCJA increased the estate tax exclusion from $5,450,000 to $11,400,000. This exemption has a sunset provision, which means it will revert back to the 2017 exclusion amount in 2026. .
In addition the Biden administration has proposed a tax at death and an increase in capital gains rates. The Biden administration also has proposed rolling the exemption back to the 2009 exemption of 3.5 Million. None of this is news to you.
As you have said it is very difficult to predict future rates and I agree. I do think it is very likely that the current exemption will roll back to at least the 2017 amounts and possibly lower. If the step up in basis is lost and if the assets will be deemed to have been sold at the date of death, the estate could be taxed at the highest rates.
All the above would make it more advantageous to do Roth conversions at the current tax rates. A series of conversions that occur between retirement and age 72 potentially would shield at least some of the estate from high tax rates imposed at death. As you have pointed out Roth conversions make sense if you goal is to pass on assets to your heirs. Unfortunately taxpayers have to make their decisions based on current laws. Waiting to see what happens has its downsides as well.
By the way, I very much appreciated the list of “Stupid Roth Tricks”. I appreciate the risk in long term tax planning when the rules can change at any time.
I’m a bit late to the party, but oh well. Kudos to Ed on the paper and spreadsheet. I have been thinking about trying to create something similar, but was daunted by the many many moving parts. (Despite being recently retired from a data analytics job and considering myself a reasonable spreadsheet jockey.)
I’m considering doing significant Roth conversions (e.g. filling up the 22% or 24% tax bracket) primarily because of the Oregon estate tax of 10-16% on everything over $1M (including the combined value of your house, investment accounts, retirement accounts, etc.) I expect my IRA will be primarily left to my kids, and OR will immediately take 10+% and then both Fed and OR will tax the distributions as income. Seems to me like a big win to do aggressive Roth conversions and avoid the future double taxation.
I’m not eager to do the spreadsheet work to model this (and I don’t expect you to), but my intuition tells me that aggressive Roth conversion is a clear win. Any thoughts?
If heirs are involved (even without estate tax) Roth conversions tend to make more sense. Converting up to the first IRMAA boundary ($182,000 in 2022) is almost always an easy ‘yes.’ But as soon as you start climbing up the IRMAA ladder unnecessarily (the equivalent, at the top of each IRMAA, of an extra 4% or 5% on the rate), and climb through NIIT at $250K, then the payoff may be delayed or smaller.
Since you’ve got the skills, try expressing the OR estate tax in present value terms, relative to the cost of 2,3 or 4 IRMAA (which you incur by going to the top of the 24% bracket).
My apologies. you did address the state tax issue
Thank you for freely sharing the paper and spreadsheet. It is very helpful, especially the idea to reconsider paying taxes from the converted amount if the recovery horizon is long enough or main intent is bequest. It was generous to share your work.
Thanks. I have everything I need to live well–so why not share results, and hope for the largest possible audience? Still, nice to hear that public availability is appreciated.
I have an interesting twist to your paper. I retired early using the little known IRS Rule of 55. With 17 years before RMDs hit, My gut tells me the extra 10 years of conversions (filling the 24% tax bracket) and compounding tax-free growth should have an earlier ROI. Do you have something more concrete that can prove this out?
I tend to agree: the earlier the conversions, the greater the ROI at a given later age, 80 etc. Later in this thread there is an exploration of how much difference it makes to convert earlier: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=358688&sid=eb6c2e0c2859bb6d6de5b86f6f453c71
Be careful when you get to IRMAA age (63): the 24% bracket takes you through #1, #2, #3 and potentially #4; you may wish to drop back to the top of the 22% bracket just below IRMAA #1
Thank you for your ultra-quick reply as well as the boglehead link. My thinking is if the market does so-so and I live a long time, I break even (of my grandparents lived to over 95 years old with one hitting 100). If I stay in stuff that grows like or better than the S&P500 the growth might outpace my ability to Roth Convert. But IRMAA for my wife at the moment is mid 5 figures a month, so mid 4 figures per year. If we continue to have gangbuster years like the last 3, I want to be on the 0% tax side of this equation. I see this as paying 5 figures in taxes and 4 figures in Medicare in order to save 6-7 figures per year in future taxes.
…typo…IRMAA is mid 3 figures per month