How aggressive should Cheryl be with her Roth conversions?
Voluntary withdrawals and small Roth conversions only work for Cheryl 1.0 because her TDA was small enough, at $1.25 million at age 68. In this post I’ll introduce a more fortunate Cheryl 2.0 who has a larger TDA. (Or if you like, a Cheryl who was even more remiss in not doing Roth conversions early and often.)
In the projection below, I’ve increased Cheryl’s starting balance by about $200,000, to $1,447,000, and kept the living expenses and the pre-IRMAA conversions as before; net, Cheryl 2.0 now hits IRMAA exactly when RMDs begin. C’est la vie: Cheryl version 2.0 is somewhat better off, enough so to wipe out the effect of the voluntary withdrawals and no-IRMAA Roth conversions that were so successful for Cheryl 1.0 (no QCDs considered on the initial run).
Back to square one: If nothing is done, Cheryl 2.0 will pay IRMAA from the get-go.
Sanity check: Is this level of wealth plausible for a single person age 68? I’d say neither more nor less plausible than Cheryl 1.0. The larger TDA of Cheryl 2.0 could reflect a somewhat more aggressive strategy, say a 60/40 portfolio; or retirement after a relatively more favorable period for investing; or a failure to do Roth conversions early and often.
In any case, now we have to look into what her brother Charles and cousin Celia suggested. They told Cheryl 1.0 to convert up through IRMAA #4 at least (Charles), or possibly until she converts enough of her TDA, in 2023 and 2024 (Celia), to never face IRMAA throughout some reasonable time horizon. Will these maneuvers succeed?
To begin, note the slightly paradoxical element here: “I detest future IRMAA s-o-o much that I’m willing to pay it right now, several times over, voluntarily and unnecessarily.” M’k …
The paradox only resolves if, for instance, one can pay four IRMAA in 2023, and four more IRMAA next year, to save, say, a dozen IRMAA down the road; and even then, only if those twelve future IRMAA, some over a decade out, have a discounted present value greater than eight present-day IRMAA paid this year and next. Depending on the discount rate, a ratio of 8 : 12 might go either way; an 8 : 20 payoff ratio would be a more definitive win / more comfortable expectation. But that may be too steep a hill to climb. No way to tell until we run the numbers.
Next wrinkle: although a substantial portion of Cheryl’s RMDs do fall into the SS tax torpedo, subject to a marginal rate of 40.7%, to be at risk of IRMAA #1 when RMDs begin she had first to obtain enough income to
exit the torpedo. On torpedo exit she returns to the 22% rate seen in IRS tables. This necessary exit from the torpedo before IRMAA is an artifact of current IRMAA thresholds, the current 22% / 24% income tax threshold, the current maximum SS payment, and the current (and not inflation-adjusted) thresholds used to determine the taxable fraction of SS payments. A future legislative change to any of these values could potentially create a situation where the IRMAA #1 threshold gets passed while still inside the tax torpedo; but at present, that does not happen in any standard case.*
*BTW, the dollar gap between torpedo exit and IRMAA onset is larger in the couples case, because the fixed SS thresholds for couples, of $32,000 and $44,000, are not 2X those applied to singles.
**If Cheryl only had $40,000 in SS payments, not the $50,000+ shown, she would exit the torpedo at a still lower income, producing a still larger dollar gap between torpedo exit and IRMAA onset.
My point: to escape IRMAA #1 by income reduction through a Roth conversion is to save that IRMAA payment plus 22% of the dollar reduction as income tax savings—not 40.7%. If one gets to the point of saving 40.7%, as occurred for Cheryl 1.0, then one must have already dropped below the IRMAA threshold and cannot expect to save any IRMAA payment by further reductions in income, i.e., by Roth conversions that further reduce the TDA balance.
Next, the general rule for Roth conversions requires that future tax rates move higher for the conversion to produce a gain in after-tax wealth. A conversion by Cheryl today that triggers IRMAA #1 but stops short of IRMAA #2 will be taxed partly at 22% and partly at 24%, net about 23% (the IRMAA #1 dollar span straddles the 22%/24% income tax threshold). Pay 23% now to save 22% later …?
Fine, call it a breakeven conversion, one percent here, one percent there, whatever.* This simply indicates that the payoff from an IRMAA-triggering Roth conversion today comes down to the present value of any future IRMAA averted thereby; it won’t be boosted by income tax savings if the future rate remains 22%.
*Or you could argue that TCJA will expire as scheduled, with the 22% rate reverting to 25%, so that a conversion today at 23% or even 24% will benefit from a mild boost in income tax savings. I repeat: whatever.
Begin: Convert and pay IRMAA #1 in 2023 but stop short of IRMAA #2
This will entail a conversion of $26,000, the span that separates the IRMAA #1 threshold from IRMAA #2. It will increase her 2023 income from $97,000 (just below the IRMAA #1 threshold) to $123,000 (just below the IRMAA #2 threshold). At an estimated 4% of that span, she can expect to pay an IRMAA dollar penalty of $1,040. Split across the 22% and 24% brackets, she’ll pay about 23%, or $6000, in income tax. With tax paid from inside the conversion, she’ll have a new Roth balance of just under $19,000, after IRMAA + income tax charges totaling $7,040.
As shown below, this 2023 conversion lowers her 2027 RMD by $1,102, putting her below the IRMAA threshold in 2027. But it does not move the needle enough get her out of IRMAA the following year.
Oops: pay one IRMAA now to avoid (only) one IRMAA later. That doesn’t look too promising.
It gets worse. With 3% inflation, the 2027 IRMAA payment is expected to rise to $1171 (1.03 to the 4th power times $1,040). That doesn’t sound so bad, until one examines the counterfactual of no conversion at all. In that event, the $1,040 IRMAA payment
not made in 2023 would have grown at the 6% rate projected for the TDA, to $1313.
Cheryl gave up a future TDA balance of $1313 to avert a future IRMAA payment valued at $1171. That doesn’t sound … smart.
Oh well, at least she saved $242.44 in income tax on the $1,102 reduction in 2027 RMD. That goes part of the way toward defraying the $6000 she paid in income tax to do the conversion.
Wait, what? She’s still in the hole on the conversion, in income tax terms, by $5,758?!? How can that be, when the 2027 tax rate is only one percentage point below what she paid to convert in 2023?
And now we come to the dirty little secret of Roth conversion payoff calculations (at least as I’ve done them in the past; I wouldn’t want to accuse anyone else). The typical payoff calculation applies the future tax rate to the
entire balance of the counterfactual conversion amount, i.e., the future value of the $26,000 if the conversion had not occurred and the entire amount remained in the TDA to grow at the indicated return, 6% in the running example.
The amount stated above, of $242.44, is the income tax savings only on the RMD reduction that took place in 2027—the bird in the hand. Next year there will be another bird in the hand, and the year after, another. And then at some point Cheryl will die. If she has no heirs, then the accumulated birds in hand and the remaining birds flushed out of the bush by death (i.e., tax on the remaining balance) will show the payoff received, if any, from the Roth conversion undertaken many, many years ago.
It is never necessary to liquidate any portion of a TDA other than the required minimum distribution. Even at death, anticipated for twenty-five plus years from now for Cheryl, if there are heirs they will have ten more years to distribute, and need only liquidate 10% per year.
Realized tax savings from a conversion intended to reduce RMDs only accrue as those distributions are received. Eventually, the TDA will all be distributed, and eventually, the conversion will pay off in keeping with the difference between the tax rate at conversion and the tax rate(s) applied to the distributions over time. If niece Janice has to pay 48% a year on ten distributions commencing twenty-five years from now, she will be quite happy that Aunt Cheryl chose to do more Roth conversions rather than fewer. But Aunt Cheryl only earns back her Roth conversion taxes bit by bit, year by year, as RMDs take place. Her own RMD withdrawal won’t hit 10% until age 94, approximately her life expectancy, hence her realized tax savings will poke along for quite a while.
*
historical note: the early calculations of Roth conversion outcomes, beginning two decades ago, tended to assume a single conversion and a single lump sum distribution. Example: If $10,000 is converted today at 12%, $8,800 will be placed in the Roth. Some years later, suppose the investment has tripled. The Roth can be liquidated, free of tax, to produce $26,400, enough to purchase a modest full-size sedan. If the future tax rate has risen to 25% and no conversion had been made, the $10,000 would have become $30,000, but only $22,500 would have been available to spend after-tax (i.e., a subcompact basic automobile).
The payoff is the difference between the tax rate prevailing at the time of payoff versus the tax rate at conversion, multiplied by the future value of the counterfactual (here, [25% - 12%] * $30,000, or $3,900 more future dollars from converting).
RMDs, which will continue for as long as Cheryl lives and for ten years afterwards, drastically slow down the conversion payoff relative to this old-fashioned liquidation analysis. Only 3.77% must be distributed when RMDs begin, and the pay off that year is, in the running example, [$30,000 * 3.77% * (25% - 12%)], or $147. (Payoffs will continue to accrete each year, of course).
And if Aunt Cheryl, alas, needs assisted living or even nursing home care for her last few years of life, she may draw down most of her remaining TDA balance at a tax rate near zero, if the expenses are deducted as medical expenses. In which case, in the final sum up, the Roth conversion probably will have failed (from the standpoint of saving Cheryl taxes; Cheryl’s niece Janice may still be better off, if there’s anything left to inherit).
Did you know that about Roth conversion payoff calculations under RMDs?
Ahem, moving right along: brother Charles didn’t say to do one single puny conversion into the IRMAA zone—he said to do a bunch of them (and cousin Celia went further still).
Maybe Cheryl can make it up on volume. Next post investigates.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.